Identüication O€ Nuclear Matrix Proteins Crossliaked to DNA by cis-

DDP 1A sirir in Human Bnast Caucer CeU Lines

by

Virginia Spencer

A Thesis Submitted to the Faculty of Graduate Studies

The University of Manitoba

In Partial Fulfilment of the Requirements for the

Degree of Master of Science.

National Library me1 OfCamda Bibliothèque nationale du Canada

Acquisitions and Acquisitions et Bibliographie SeMces services bibliographiques

The author has grrmted a non- exclusive licence allowing the National Lhrary of Canada to reproduce, loan, dismibute or selI copies of this thesis in microform, paper or electronic formats.

The author retains ownership of the copyright in this thesis. Neither the

L'auteur a accordé une licence non exclusive pemnettant à la Biblotheque nationale du Canada de reproduire, prêter, distriiiuer ou v&e des copies de cette thèse sous la forme de microfiche/nlm, de reproduction sur papier ou sur format électronique.

L'auteur consewe la propriété du droit d'auteur qui protège cette thèse.

thesis nor substantial extracts fiom it Ni la thèse ni des extraits substantiels may be printed or othemïse de c e k i ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.

TBE UNIVERSITY OF MANITOBA

FACULN OF GRADUATE STWDES

COPYRïGHï PERMISSION

A Thesis submitteà to the Faculîy of Graduate Studits of the University of Manitoba in partial fulTil1ment of the tcqpircments o f the degree of

Permission has betn granted to the LIBRARY OF THE UNIVERSï'N OF ktLXïïOBA to knd o r sen copies of thit thesis, to the NATIONAL LIBRARY OF CANADA to mictotilm this thesis and to lend or seU copies of the Rlm, and to ZMLVERSrrY MICROFILMS to pubüsh an abstract ot this thesis.

This reproduction or copy of this thesis bas beea made available by authority of the copyright owner sokly for the purpose of private study and rtsmrcb, and may only be reproductd and copiai as permitteâ by copyright hm or with express writtea authorization fmm the copyright owner.

Af*no.rkdrriacnLP

I wodd like to express my suicere gratitude to my supervisor Dr. J. R Davie for his

guidance and suppon during the course of my studies. 1 would also üke to thanL Dr.

Genevieve Delcuve for ber thoughtful advice in overcoming some of the challenges I have

encountered in my studPs. Furthemore, 1 am gntefbi to Dr. Leigh Murphy and Dr. Pe ter

Cattini for their williogness to sit on my advisocy cornmittee and ovenee my progress.

1 would WEe to extend my appreciation to Dr. Shanti Samuel for her excellent

instruction of two-dimensional electrophoresis and for supplying me with the breast cancer

ceb required for the a n a l . of DNA-binding NM protein prof3es over b m s t cancer

hormoaal progression. In annition, 1 am ihanldul to Ms. Amanda Coutts for supplying me

with the breast cancer cek neeûed to examine the effects of estrogen on cytokeratïn

expression over breast cancer hormonal progression.

1 would also Lilre to thanic the students of the biocbemistry depaftment, particuiar1y

Amanda Coutts, Debbie Chadee and Mariko Moniwa for their fiiendship. and support. But,

most importantly. I would like to thanL my mother and father. 'l'heu patience. understanding

and support sustained me through the course of my studies.

1 &O wish to acknowledge the National Science and Engineering Research Council

(NSERC) for providing me with a fellowship that aiiowed me to complete these studies.

........................................................................................................................... Results 36

Ceii Lines and Cell Culture ................................................................................. 36

CrosslinLing of DNA to proteins by cis-DDP in sitic in human breast

cancer ceüs lines .................................................. .-...O.. -38

Analysis of prote* crosslinLed to DNA by cis-DDP in situ in breast

cancer ceii iines ......... ... ........................................................................... 39

.................. Identification of common DNAnossIuiked pro teins ...39

Identification of di£fetences in DNA-msslinked protein

.................... patterns o f t k various breast cancer cell ünes studied 41

Cornparison of NM protein and DNAnosslinked protein two-

dimensional patterns .................................................... d

.......... Estrogen regdation of cytokeratin-DNA interactions in breast cancer celis 48

Discussio il ..................................................................................................................... 59

................................................................................................................... Refée~lces -79

Eirurn Figure 1.

Figure 6.

Figure 7.

Figure 8.

Figure 10.

Il&umm= u

...................................... Formula of cis-DDP ....-.....-. 17

Proteins crosslinked to DNA by cis-DDP in situ fkom the human

bras t epithelial ceit line MCF- 1OAl ....................................................... -50

Proteins crosslinked to DNA by cis-DDP in situ tïom honnone-

dependent breast cancer c d iines ............................................................ 5 1

Proteins crossiinked to DNA by cis-DDP in situ fiom homone-

independent breas t cancer ceii lines.. ........................................................ -52

Proteins crosslinked to DNA by cis-DDP in situ that are common

to breas t epithelial and breast cancer ceil lines.. ...............es..................... 3 3

T47D human breast cancer nuciear matrk proteins and proteins

crosslinlred to DNA by cis-DDP in situ. ..................................................... 54

Nuclear matrix prooles of hormone-dependent breast cancer

Nuclear matrix profiles of hormone-independent breast cancer ceil ünes

and a pseudo-normal breast epithelial cd line ....... .;.. .............................. -56

Cytokeratins crosslinked to DNA by cis-DDP in T-47D5 and TS-PRF

human breast cancer cek in estrogen-replete and estrogen-depiete . . condiuons ............................................................................................... 5 7

Two-dimensional gel patterns of pro teins crosslinked to DNA by

3 rnM cis-DDP in T47D5 and TS-PRF human breast cancer ce&

in estmgen-replete and esuogen-depiete conditions. ............. ......... .. S 8

nk mdt

Table 1. Major characteristics of human breast cancer ceii b. as weU as a

...................................... human breast epithelial ce8 IM.. ............... ... -37

Table 2. Major characteristics of early and late stage breast cancer ....................tics... 38

Table 3. Moiecufar weight range and isoelectric pont range of four DNA-

crosslinled protein clwters common to pseudo-nomal breast epithelial

..................................................................... and breas t cancer ceil lines, -40

Table 4. cis-DDP DNA-crosshked proteins of hormone-dependent and

............. ....*. hormone-independent haman breas t cancer c d lines.. .... -42

Table 5. Statu of cis-DDP DNA-crossünked proteins of hormone-dependent

and hormone-independent breast cancer ceii lines in the NM pronles

................................................................... of 7 of the 8 ceil ünes studied 46

bP

CAPS

C~S-DDP

CSK

DNA

DNase

Dm'

ECM

EDTA

EûTA

ER

FBS

IF

HAP

adenine

angstrom

absotbance at 260 nanometers

arginine bucyrate

apoüpoprotein

base pair

2-(cydohexylamino]- 1 -propanesulfonic acid

cis-diammiaedicblocopIatinum@)

c ytoskeleton

deoxynbonucleic acid

deoxyriinuciease

ditbiothreitol

extracellular matrix

ethylenediamine tetraacetic acid

ethylene glycol-bis@-amiaoethyi ether)-N. N, Ntetraacetic acid

estrogen receptor

fetai bovine serurn

intermediate filament

hydroxyapatite

heterogeneous nuclear ribonucîeic acid

hetecogeneous nuclear ribonucleoprotein

vii

kb

kDa

LIS

MAR

mA

mM

nM

NM

NM-IF

NS

PAGE

PI

PMSF

RNA

Nase

SAR

SDS

T

TBS

Cig

PM

v

kilo base

kilodalto n

iithium diiodosaücylate

matra attachment region

milliamps

milümolar

nanomolar

nuclear matrix

nuclear matMt-intemediate fjiament

nuclear scafTold

polyacrylamide gel eiectrophoresis

isoeiectric point

phenylmethylsulfonyl fluoride

ninucleic a d

n i n u c h s e

sc&old attachent region

sodium dodecyl sulfate

thymine

tris- buffered saline

microgram

miro meter

voltage

iuskaa

The nuclear ma& (NM) phys an important role in the structural and hnctionai

organization of the ceii nucleus. Human breast cancer is a dsease that is thought to progress

from a weU-dinerentiated, hormone-depudent state to a poorly-differentiated, hormone-

independent state. NM ptotein composition, and the interaction of NM proteins with

intermediate filament proteins varies over human breast c a m r progression. suggestuig that

NM proteins may be progwstic madcers of cancer progression. Cell lims representing each

stage of human breast cancer progression were treated with cis-diamminedichloroplatinum(II)

to crosslialr proteins to n u c k DNA in situ. and the DNA-bound proteins were isolateci.

Two-dimensionai profiles of DNA-crosslinked proteins from the ceii hes investigated

showed a rernarkable simkity to one another as well as to each cell line's respective N M

prom. However, differences in the types of DNA-crosslinked proteins were observeci

between ceIl lines representative of each stage of disease progression. thus adding to the

importance of the N'M. and, more specincally, NM-DNA interactions in bmst cancer

develogment. As weii, the association of some DNA-crosslinked proteins such as

cytokeraiins 8, 18 and 19 was dependent on estrogen in a hormonedependent humaa breast

cancer celi Le, but not in a hormone-independent human b m s t cancer ceii line. Thus, the

acquisition of a hormone-independent pheno type may interfere with the ability of estrogen

to regdate intermediate füament-nuclear DNA interactio w in human breast cancer celis.

These hdings suggest that some of the various cellular events leading to malignancy in

human breast cancer involve the NM and the cytoskeleton (CSK). Furthemore. the

characterization of NM pm teins and pro teins crosslinked to DNA by cû-DDP in situ, which

are prDaarily NM proteins, are two complementary approaches to identify NM proteins that

are informative in cancer diagnosis.

The nucleus is a cornplex, highly organized organeile capable of perfonoing a

multitude of cellular fiinctions. In p s t years, a great deal of infocmation has emerged denning

the individual structurai cornponents of the nucieus such as the nucleosomes, chromatin loops,

n u c k envelope and nucleai- hmnia Despite tbis wealth of laiowledge, cesearchers have yet

to assemble a pictute iiiustrating how the interactions between these components forrn a

hurtioning nucleus. What i s knowa. however, is that I vivo the genomic DNA is organized

into nucleosomes wbich assemble into 30 nm filaments (Felsenfeki and McGhee, 1986;

McGee et ai., 1983; Simpson, 1978). These filaments are then folded into negatively

supercoiled loop domains which appear to be anchond at their bases to proteins belonging

to the skeletai framework of the nucleus called the n u c h matrix (Cook and Brazelil, 1976;

Paulson and Laemmli, 1977; Vogelstem et aL, 1980; Gerdes et ai., 1994). The loops range

in size fiom 5 to 200 kb pairs in length (Jackson et ai., 1990). The constraint of DNA hto

these supercoiied units by the N M is believed to stnicturaily organize and regulate DNA

replication and transcription (Cook, 1989). thereby Muencing the transcriptional activity of

DNA domains. Rocesses such as tmmxiption (Mattem et aL, 1996; Mirkovitch et aL. 1984;

Pienta et al., 1991; Jackson and Cook, 1985; Ciejek et al, 1983), DNA replication

(McCready et ai., 1980; Beremey and Buchholtz, 198 1; Replogle and Pienta, 1996). and

RNA processing and transport (Scluoder et al., 1987; Zeitlin et al.. 1987; Mattern et al.,

1996) have been found associated with the NM. Moreover, Mirkovitch and colleagues

(1984) have propsed that the NM brings the promo ter and upsmam regulatory elements of

active genes physicdy close together, focming a subcompartment rich in DNA-binding

sequences able to associate with regulatory factors and nuciear enzymes such a s RNA

polyrnerase. Such evidence impiies that the NM bas a role in gene expression.

The Nuciezv Matrix

By definition, the NM is the structure that rernains after the treatment of interphase

nuclei with nucleases to digest the genomic DNA, and high salt extraction to remove

chromami and looseiy bound proteins (Mattern et ai.. 1996). This structure is composed of

three main regioas: a surrouDdmg d u a l nuclear envelope or nuckar hmma. residual

nucleoli and an extemive fibrogranalar mternal matrix (Berezney, 199 1; Fey et aL, 1986). As

much as 98.4% of the NM is composed of high rnoiecular weight, nonhistone proteins

(Berezney and Coffey, 1974). many of which remain unidenrined to date. Some of these

proteins are common among Wnent ceR types, whiie otbers are tissue- and ceCtype specific,

refkcting tk state of âifférentiation and H o m a t i o n (Fey and Penman, 1988; Dworetzky

et al, 1990; Cben et al, 1996; Samuel et al, 1997; Mattem et ai., 1996; Mattern et aL, 1997;

Stuurman et ai., 1990).

NM proteins identined to date include n u c k lamms A (Nakayasu and Berezney,

199 l), B (Stuurman et ai., 1990), and C (Nakayasu and Beremey, 199 1 ), nucleolar protein

B23 (Sm- et aL, 1990; Mattem et al.. 1996 and 1997; Nakayasu and Berezney, 199 l),

nuciear matrins (Nakayasu and Berezney, 199 L), the nuciear-mitotic apparatus protein (Zeng

et ai., 1994) and vario us heterogeneo us nuclear ribonucieo pro teins (hnRNPs) (Ho luoann et

al, 1997; Mattem et ai., 1996 and 1997). The actual composition of the NM appears to Vary

according to the method of isolation and the degree of scrutiny used when comparing two-

dunensionai NM pmtein pronles ofdinerent cells (Stuman et al., 1990). According to Fey

and Penman ( 1988). only those proteins detected exclusiveiy in the NM kaction are

designated as true N M proteins. In addition. any protein requiring intact DNA or RNA for

its association with the NM is no t considered part of this structure (Fey et ai.. 1986). in this

study, NM proteins were hW by treaiiog nuciei with DNase 1.0.25 M ammonium sulfate,

and RNase A followed by extraction of the remahhg intemediate filaments (Es) using an

IF disassembly and assembly process. hoteins such as actin. Wnentin, virnenrin-related

proteins and ribonucleoproteins (RNP) complexes were not considered NM since they were

fouad m otber œiiuiar fractions (Fey and Penrnan, 1988). The nmaining proteins unique to

only the NM were geaerally present m low abundance. This creates the possibility that these

proteins were actuaiiy present in other cellular fiactions but were masked by the more

abundant proteins (Fey and Penman, 1988). Also, the use of high salt to isolate the NM

rnight strip away mie NM proteins, cashg them to appear in other cellular fhctions (Capco

et ai., 1982).

Besides these problems, a major concem with this protocol is the selective removal

of IF proteins and heterogeneous nuclear ribonuclear RNA (hnRNA) and RNPs. Evidence

îÏom various studies suggests that bnRNP proteins may ako be present in the NM (Nakayasu

and Beremey .l991; Mattem et aL, 1996; Mattem et aL, 1997). The exact huiction of

hnRNPs is unclear but they are probably involved in many aspects of RNA metabolism

(Dreyfuss et ai., 1993). Even though Fey and Penman (1988) believed that hnRNPs are not

part of the NM, they were still unable to completely remove these proteins fiom the NM.

'Ibis suggesis that either the hnRNPs were so tightly associated CO the NM that they could no t

be efficiently removed or that the NM does inâeed contain a smd portion of hnRNPs. In

addition, stripping of hnRNPs nom the matrix by EtNase treatrnent distorted NM shape,

thereby suggesting a role of bnRNPs in organizing the naclear interior (Fey et ai., 1986).

The protein. hnRNPK, has been shown to have ail the properties of a transcription factor

(Michelotri et al. 1996). The fiiM:tional properties of bnRNPK in RNA processing and

iransrription, and the observarioos that RNA professing and transcription are associated with

the NM (Schroder et al, 1987; Mirkovitch et aL, 1984; Pienta et ai., 199 1; Jackson and

Cook, 1985; CjepL et al., 1983). M e r support the hypoihesis that hnRNPs are part of the

NM. Moreover, NM-IF isolation from nuclei stabiüzed with sodium tetrathionate in the

absence of RNase A provîded internai matrin structures that were predomuiantly cornposeci

of hnRNP proteins (Mattern et al., 1996 and 1997).

As weii, RNA may be an important constituent of the NM. As much as 70% of

hnRNA rernained associated with the NM after treatment of nuclei with DNase and salt

extraction (Fey et ai., 1986). Digestion of NMs with RNaseA markedly disto~ed the NM

shape and was ody 97% etticknt for removing NM-associated*bnRNA and compkxed

pro teins (Fey et aL, 1986). This indicates that nuclear RNA must play an important role in

maVix organization (Fey et aL, 1986).

Several O bsemtions suggest that IF pro teins may be part of the NM: (1) Fey et ai.

(1984) showed a stable association of IFS with the NM; (2) addition of a IF disassernbly-

assembly step to the NM-IF isolation procedure couid not completely remove aii IF proteins

(Fey and Penman. 1988); (3) the crosslirnking agent. cis-diamminedichloro platinum(Il) (cis-

DDP), which crosslinks DNA preferentially to associated NM proteins (Ferraro et al., 1992),

crosslinks r t in (MiUer and Costa, 1989) anci cytokeratins to DNA in situ (Ward et al., 1984:

Ohksi et al., L987); (4) intermediate filament-like structures have been observed in the NM

(He et al.. 1990; Nickerson et aL, 1992); (5) antibodies directed agauist lamin A have

bcali;ted to NM nbres Wozak et ai., 1995). Furti~rmore, because of the intirnate association

of the IFS with the NM, removal of IFS causes parts of the N M to be removed (Capco et ai.,

1982).

Even if the IFS are not part of the NM, evkience exists showing that these proteins

ineuence NM structure (Maniotis et ai., 1997). IFS containhg Wnentin, desmins, keratuis

or nemfihment proteins, abng with actin-contaihg microtilameats, and tubulin-coniaining

microtubules make up the CSK of eukaryotic ce& (Ingber, 1993). The actin-associated

proteins of the CSK bind to specitk ~ansmembrane receptors for extraceiiular matrix (ECM)

components, thereby physicaûy connecting this structure wïth the ECM (Turner and B d g e ,

1991). Based on evidence kom many studies, the CSK also appears to extend as far hto the

nucleus as the NM. Imrnunogold siaining of cytoskeleral and NM proteins in resinless

electron miccoscopy sections dispkys 10 nm intermediate nkmnts that stretch between the

nuclear lamina and the ceii peripbery (Nickerson et al, 1990). Transmission eIectron

micrographs (Fey et aL, 1984) abng with in vitro binding studies (Wang et ai., L 996b; Ward

et al, 1984) show an association of IFS with the NM and nuclear DNA, respectively. In

addition, cytokeratins are preferentiaiiy crossiiniced to DNA by cis-DDP (Ward et aL, 1984).

The purpose of the linkage between the nucleus. CSK. and the ECM might be to

stabiliz nuclear fonn and htegrate ceii and nuclear structure (Ingber, 1994). The CSK is

thought to be organized as a tensegrity network (Wang et al., 1993). According to Ingber

( MM), a tensegrity nehvork consists of a series of isolated. compression-resistant stmts that

can be physicaily altered by co~ect ions with a continuous series of tension eiemenu. To

maintain its stability, thig network must generate an internai tension. in ceiis, s t n i c t d

stabiiity depends on mechanical temion generated within the contractile micronlamenu of the

CSK. This tension is then transmitted across transmembrane receptors on to the ECM which,

m tum, nsists tbe added force and causes a force balance. Appiication of a mechanical stress

to a ceil o&ts the force balance and causes a number of coordinated events to occur in the

CSK and all extended areas without disniption or loss of tensional mtegrity (Ingber, 1994).

Therefore, pressure exericd on to the surrounding ECM cm be trammitteci to the CSK

A consequence of this force imbalance could be global rearrangernents in the CSK

mgber, 1994) which may alter the interactions of molecules physicaliy linked to this structure

, 1 9 8 ) In support of this stress-inducad CSK reamgement is the observation ihat

puliing integrins by rnicromanipuiating microbeads bound to the ceil surface causes

cytoskeletal nlaments to ceorient, nuclear shape to becorne distorted and nucleoli to

redistribute dong the axis of the appW tension (Maniotis et ai., 1997).

Aiterations in CSK organhtion have also been induced in ceils transformeci with the

ras oncogene (Pienta and Coffey, 1992). The presence of the ras oncogene may stimulate

the RAS signal transduction pathway which, in hun, wodd activate transcription factors such

as AP-1 and ETS (Karin and Hunter, 1995). This activation may lead to the aitered

expression of cytokeratins 8 and 19 (Pankov et ai., 1994).

Thus, changes m CSK organization resuitïng kom either physical or chernical changes

within a ceii may impact NM architecture by causing reamgements in NM proteins. These

rearrangements could reiieve the constrainu that prohibit DNA ftom unwinding (Ingber et

al., 1994). As a result, transcriptiody inactive domains may become decondensed into

active ones. If the tensepnty mode1 accurateiy explains a cers architecture then IFS may piay

an important role in gene expression by iduencing N M structure. anci, thecefore. chromatin

bop organization.

The loop mode1 for chromatin organization predicts a non-random folding of the

chromatin fiber (Gasser and Laemmli, 1986a) hto b o p of approximately 5 - 2 0 kb (Jackson

et aL, 1990) which are attacheci at th& bases to the nonhistone NM proteins (Cockerill and

Garrard, 1986a; Mirkovitch et al., 1984). Ache genes appear to be under torsionai stress

(Villeponteau et al., 1984). Tbus, the periodic attachent of these genes to the NM maintains

the torsional stress (LevyWdson and Fortier, 1989). When the histone octamer is removed,

DNA is released as negatively supercoiled Ioops (Vogelstein et ai., 1980). Each loop is

commoniy thought to represent an independent unit of transcription and replication that is

insulated h m the regdatory influences of neighbouring loop domains (Kalos and Fournier,

1995). The wuences thought to be responsible for the formation of these loops are termed

maaix- or scaffoki- attachment regions (MARS or SA&) (MHkovitch et aL, 1984; Cockerill

and Garrard. 1986a). Thus, the discovery of MARsISARs has been an important step

to wards hirther unders tanding the hnctionai and structural aspects of chromatin-loo p

org anization.

MARSlSARs often conskt of at least 200 bp of AiT-rich DNA (Cockerill and

Gamrd, 1986a; Kas et aL, 1989). have topoisornerase iI sites (Yu et aL. 1994; Adachi et al.,

1989), and contain one or more copies of the sequence AATATITï or similar sequences

(Cockeriü and Ganard, 1986a). However, not ai i MARS are AIT cich (Boulüras, 1995: Levy-

Wilson a d Fortier, 1989) or contain topoisornerase II sites (LRvy-Wilson and Fortier, 1989).

In addition, some MARs contain nucieation sites that becorne stably base-unpaireci under

negative superheiical tension (Bode et aL, 1992; Kohwi-Sbigematsu and Kohwi, 1990). Such

a feature may be important in MAR fiinction.

Structurai analysis shows tbar MAR sequences contain fatufes such as single-

suandedness (Luderus et aL, 1994)- double-strandedness flsutsui et aL, 1988; Probst and

Herzog, 1985; Luderus e al., 1994). a narrow minor groove resulting fiom oligo @A)-

oligo(dT) tracts (Kas et ai., 1989; Luderus et ai., 1994), and bending (Homberger, 1989).

MARS have been found in different locations of various genes. They have been found to

cohabit with promo ters (Stein et al, 199 l), and origins of replication ( D i . e l and Hamiin,

1988). and they have been found adjacent to enhancers (Cocked and Garrard, 1986a).

Alrhough they do not share an extensive homology (Levy-Wilson and Fortier, 1989). MAR

sequences fiom dinerent types of ceUs and nom dinerent species have al1 been shown to

compete with each other for NM attachent sites in cornpetitive MAR binding assays

(Cockerili and G d , 1986b). As a result, the mechanism responsible for MAR binding to

the NM appears to be evolutionarily consend. One exception to this conservation is the 5'

p r o M human apolipoprotein (apo) B MAR, a sequence that associates with the N M only

in cells expressing the apoB gene (Levy-Wilson and Fortier, 1989).

The exact mechanisrns responsible for MAR DNA associations with the NM are

uncertain. in interphase nucbi, the majority of DNA loop attachent sites have been

localued by cytochemical and biochemicd assays to the internai N M and the remainuig are

present in the periphed n u c k iamina region (Zini et ai.. 1989; Phi-Van and Stratluig, 1988;

ïimmkie et al, 1988). The large variety of pro teins composing the NM would suggest that

there are diverse modes of recognition for DNA structures or sequences (Kay and Bode,

1994) .

As was preMousiy mentiod, most MARS contain A-T rich regions (Cockerill and

Garrard, 1986a; Ras et ai., 1989). Tbese regions are believed to faditate DNA unwinding

W . by hehss, torsionai saain. or sin&-stranded DNA-binding proteins (Bode et aL,

19%). Loop formation resuiîing wben MARS bàd to the matrix may induce local unwinding

of A+T rich squences. These unwound sequences couid then bind to the matrix and

reinforce the matrix interaction (Mieike et aL, 1990). The unwinding of MARS may be

stabiiized in the fonn of a plectonemic DNA structure of B-type, double-stranded, knotted,

negative superhelica (Bode et aL, 1996) or crucifonn (Hsieh and Wang, 1975; Lilley, 1980)

character. Thus ù is poss1'ble that MARS may assume some IMd of landmark structure that

aliows them to associate with the N M (Probst and Herzog, 1985; Tsutsui et ai., 1988).

Altematively, MAR recognition sites may in fact be the single-stranded DNA

segments resulting nom MAR unwinding events (Bode et aL. 1996). The observation that

supemiied, single-stran&d, and double-stranded DNA al compte for the same NM DNA-

buiding sites (Mielke et al, 1990, Tsutsui et al., 1988) combined with the knowledge that

many DNA binding NM proteins preferentiaiiy bind single-stranded DNA sites (Luderus et

aL, 1994; Michelotti et ai.. 19%; Probst and Herzog, 1985; Kay and Bode. 1994) supports

the theory that a single-straruied character may be important for MAR hinction. Fucthermore,

this theory is svengthened by the hdings that increases in transcriptional activity attributed

to MARS are correiated with the ab- of MARS to widergo separation into single-strands

under negafïve superhelical tension (Bode et aL. 1992). However. single-strandedness does

mt qply to all MAR sequences since one-haK of N M DNA binding sites are inaccessible to

single-stranded DNA (Kay and Bode, 1994). and since DNA- binding NM proteins SAF-A

(Romig et ai., 1992). SATBl (Dickinson et ai.. 1992) and ARBP (von Kries et al, 1991) do

no t recognbe single-stranded DNA.

Regardiess of the recognition mode. MARS bind to the NM in a manner that is

sanuable. and of hi& aflïnity (Luderus et aL, 1994). Once bound the stable base unpairhg

of these sequences wdet torsionai strain is believed to reiieve a transcriptional unit of this

strain (Bode et ai., 1992). The thennodynamic energy fiom the relieved strain couid be

aansmined to other regions of the domain and generate negative supercob at these remote

sites* By relieving negative superhelicity, MARS may open chmmatin domains (Zhao et al,

1993) to the aanscriptiod cnacbinery, and, therefore. augment transcriptional rates of stably

transfected genes ( A h et ai., 1993; Bode et al , 1995). Despite this function, MARS are

unabie to alter transcriptional rates in transient transfecton assays (Kalos and Fournier. 1995;

Wang et ai.. 1996a) . Because of this, there is a ciear distinction between MAR and enhancer

sequences. In addition, other studies prode evidence that contradicts the ability of MARS

to augment transcription. For instance. McKnight and cokagues (1992) observed no

mcrease in expression when a gene for whey acidic protein (WAP) and the chicken lysozyme

MAR were cohtegrated into a tcansgenic rnouse. Similarly, the apoB MAR did not enhance

the expression of a transgene containing the WAP promoter, the iate gene SV40 uitron, the

bovine growth hormone cDNA and the SV40 iate gene tenninator (Attal et ai., 1996).

Tbe inability of these MARS to enbuice tfanscription à bo th studies may be a result

of transgene copy number. Dorer and Henikoff (1994) have suggested that the presence of

multiple transgene copies in mammalian ce& ieads to pairiog of the repeat copies. These

copies fold together and produce stnictures that are recognizeâ by proteins specific to

heterochro matin. The recognition of these foidecl gene copies by heterocbro matin-specific

p r o t d Lads to the formaiion of heterochrornatin, and, therefore, gene skncing. This event

may explain the inabüay of MARS to enhame aaasgeae expression @I the study by McKnight

and colleagues (1992), because the ceils a n a l . in this study contained multiple copies of

a MAR-transgene constnict Furthemore, Attal et al (1996) have consistently observed a

higher Ievel ofexpression of low copy number transgenes than high copy number transgenes.

As was previously stated, chromatin is beiieved to be organized into loop domains in

interphase cek by LMARNM associations (Gerdes et aL, 1994; Cockerill and Garrard. 1986a;

Mirkovitch et aL, 1984), and each loop domain npresents one transcription unit (Cook,

1989). This arrangement may provide a mechanisrn for isoiating transcription units from the

intluence of neighbouring enhancers and silencers. MARS may act ik the domain boundaries

that partition chromosomes into independently regulated units (Kalos and Fournier, 1995;

McKnight et aL, 1992). On the other hand, evidence nom other studies contradicts the

theory that MARS isoiate transcription domains h m neighboursig DNA sequences. Variable

reporter gene expression was observed in constnicts containing the 5' dûtal and proximal apo

B MARs, the apo B second intron eohancer, and the apo B promoter that were stably

transfected into rnice (Brooks et aL, 1994; Wang et aL, 1996a). Despite the inability O€

MARs to completely insuiate transgenes boom position effecu. higher levels of expression

were genedy seen for constnicu contaïning MARS as opposed to those lacking MARS

(Brooks et aL, 1994; Wang et ai., 1996a). Thus, instead of hcreasing the level of transgene

expression in an individual ceii, MARS may increase the percentage of cells expressing a

transgene, and this uicrease will contribute to the overail elevated expression levels.

One exphution for the abiüty of MARS to increase gene expression couM be that

MARs support the targeting of enhancers to the NM and provide their prorrimity to

transcribed units (Boulikas, 1995; Yu et al., 1994). Once the enhancer has been brought

closer to the NM, the enbancex may increase the probability that a domain will achieve a

stable and active transcriptional state (Walters et aL, 1995). In support of the theory that

MARs work together with enhancers to increase the probability of'transgene expression are

the fkdings that certain enhancers are tightly associated with MARs (Gasser and L a e h ,

1986b). M O ~ W ~ ~ the Pglobin MAR by itself has little effect on transcription of a reporter

gene in tramformants but when linked to an enbancer, this consmct confers high kvel

transgene expression (Yu et ai., 1994). Sirnilatly, the transcriptional activation of the

immunoglobulin p heavy chah locus gene during normal lymphoid development also requires

a synergistic coUaboration between the enbancer and Oanking MARS (Forrester et ai., 1994).

Thus, MARS may b ~ g proximal fùnctional sequences closer to the NM and closer to the

appropriate trans-acting factors in order to facilitate transcription (Dijkwel and H m ,

1996; Bode et aL, 1996).

W A R Isoiatioa Techniques

Numerous in vitro studies have been perfocmed to detemine the nature of MARS

and their NM attachrnent sites (Mïrkovitch et ai., 1984; Cockeriii and Ganard, 1986a; Gasser

and Laemmii, 1986a; Imunalde et ai. 1988). In one very signi6cant study for MAR research,

the N M was isolated by treatment of nuclei with high salt (2 M NaCl) and DNaseI and then

recowtituted with spe& DNA sequences in the presence of cornpetitor DNA By

definition, a specinc DNA kgment that was preferenwy retained by nuclear matrices after

cornpetition contained an in vitro MAR (Cockerili and G d , 1986a).

In another important MAWSAR study, the nuclear scafEold (NS), which is essenMy

the same as the NM except it is içolated by treaanent of nuclei with a mild LIS detergent and

restriction enyms, was extracted and its associateci DNA sequences isolated. By definition.

a DNA fiagment that remained preferentially associated with the NS after restriction

endonuclease ciigest contained an in vitro SAR (Mirkovitch et aL, 1984). The tenns SAR

and MAR are synonymous. MAR-binding proteins identifid to date are: mutant p53

(Deppert, 1996; Muller et aL, 1996). sp120 (Tsutsuiet al., 1993), NMPlNYl (Guo et ai.,

1995). SATB l (Cunningham et al, 1994; Dickinson et al, 1992). attachent region binding

pro tein (ARBP) (von Kries et al., 199 1 ), sca@old attachment factor A (S AF-A) (Romig et

al , 1992), sca€Fold attachment factor B (SAF-B) (Renz and Fackehayer, 1996), nuclcolin

(Tlickinson and Kohwi-Shigematsu, 1995). p 1 14 (Yanagisawa et al.. 1996), topoisornerase

II (Adachi et ai.. 1989). histone H 1 (Izaunatde et aL. 1989; Ivancknko and Avramova, 1992)

, lamuis A (Ludems et ai., 1994). and B (Luderus et d. 1992). md actin (Ivanchenko and

Avramo va, 1992). Mutant p53 and p 1 14 are of particular importance for cancer research

since they are bund only in a subset of breast cancer ce&.

Aithough the use of these in vitro studies is wdespread, there are various aspects

about the mthods of NM/NS isolation that create uncertainty in the identification of MARS

and MAR-binduig proteins. For instance, treatment of nucki with high sait may severely

disturb chromatin structure, possibly causing the DNA to skie or rearrange itself along the

rnatrix anaichment sius (Miricovitch et al, 1984). In addition to altering chromath structure,

high salt treament aiso appears to remove some NM proteins and cause significant shrinkage

of the matrBr structure (He et ai., f99O). The high sait might also induce precipitation of the

transcription complexes ont0 the matrix, thereby enriching the isolated NM fiaction with

artefactual transcrip tion-dependent active genes (Mitkovitc h et aL , 1984).

Due to the potentiai DNA loop narraagement problems resulting fiom high salt

treatment, an alternative NM isolation procedure was developed (MirkoMtch et aL, 1984).

In this pmtocol, the NS was extracteci with LIS in an attempt to avoid signincant alterations

in NM morphology, and, therefore, changes in loop structure. However, LIS fded to

adequately separate chromatin proteins fiom the ma& (Fey et ai., 1986). In addition, ihis

detergent depleted the NS pnparation of RNPs (Fey et ai., 1986). The act of stripping these

proteins hom the NM wouid open up new mavix binding sites along the DNA. îherefore,

DNA sequences that are norrnaliy prevented fkom binding to the NM in vitro would be able

to associate with this structure. Such an occurrence codd lead to inaccurate identification

of MAR DNA-NM protein complexes. As a result, this alternative approach may be

inadequate for the biochemical analysis of SIMARs.

Ali previously mentioned pmtocols expose ceUs to hypertonie conditions that may ui

some way aiter NM morphology. Jackson and cokagues (1988) developed an alternative

ap proach that ailowed them to study chromatin loop-domain organization within ceUs in a

riear-physioiogÉai bder with ceduceci rkk of artefact formation. In this method, ceils were

encapsuiated in agarose microbevls that form a protective coat permeable to protein

complexes as large as 1.5 x lb Da, and not larger skeletai ekmenis or chromosomal DNA

Thus, the chromatin within the beads was protected tkom sbeving forces. accessible to

e ~ ~ z y m e ~ , and inhibited fiom aggregating in physiological b&r. Aggregation of chromatin

may influence accessibility of enzymes, salts and detergents to the sinichues targed for

study. After encapsutation, the ceiis were lysed with Triton X-100. Then the chmmatin was

digested with restriction enzymes, and pked under voltage for up to 5 houn to remove any

DNA not associated with the scaffold (Jackson et al , 1985; Jackson et al, 1988; Hempel and

Stratling, 1996).

Even though this ektroelution protocol has the distinct advantage of allowing MAR

identincation in physiobgiEal conditions (Jackson and Cook. 1985), it has the disadvantages

of incornpiete chromatin extraction (He et aL. 1990), and solubilization of essential nuclear

cornponents with Triton X-100 (Jackson and Cook, 1985; Jackson et ai.. 1988). In addition,

the b a e r used may have been physiological but cornpounds that are normal nuclear

constituents which huiction in chromatin packaging were omitted due to their potential

deieterious effects on the ceii (Hempel and S W g , 1996). As we& whether the application

of an electrical curent to cek over a period of up to 5 hours (Jackson and Cook. 1985;

Jackson et ai.. 1988) caused loop reamngements by stripping MAR-DNA from MM pro teins

still cemains to be detennïned.

The identification of MARS and theu NM binding sites becomes more complicated

when comparing the resuits of the N M and NS isolation procedures with those of

electroelution. In a study by Eggert and Jack (1991), a chromvin Gragment containhg the

Drosophihzftr SAR was eluted nom the nucleus as readiy as a hgment lacking this SAR.

Moreover, stably Ûansfécting this SAR into va~sgenic Ows did not have any sigdicant

effects on the chromarin structure. Similac resolts were obtained by Hempel and Stratlùig

(1996) who f o d that both the 5' chicken lysozyme MAR and the Drosophih Kc histone

SAR were electroeluted to the same extent as buk chromatin. These nodings m e r

strengthen t&e loop tearraogement hypothesis proposed to tesuit fkom high salt extraction

protocol ~ o i t c h et aL, 1984), as well as the theory that LIS extraction causes artefactual

binding of proteins to the matrix (Fey et ai., 1986).

The nurnerous disadvantages resulting fiom the three previously mentioned

procedures for studying DNA-NM interactions create the need for an alternative approach.

A more diable method would be to isolate the MAR DNA bohd to iis respective N M

protein in situ. Thus, pmblems of artefact formation nsulting fiom loop cearrangements and

protein stripping would be avoided. nie Ui situ isoiation of these NM-MAR DNA complexes

can be accomplished according to a method developed by Ferraro et ai. (1992) where living

œik are treated with cis-DDP, an agent that has been shown to preferentially crosslink N M

proteins to MAR DNA (Ferraro et ai., 1995). This method has the distinct advantage of

aiiowing matrbc pro tein-DNA complexes to be easily isolated. Once isolated. the crossiink

stabilizing these cornpiexes can be easiiy reversed.

Roperties of &DDP

cisaiamminedichloroplatinum 0 (cis-DDP) is a heavy metal cornplex containhg a

cenaal atom of platinum surrounded by two chloride atoms and two ammonia molecules in

the cici psidon (Johnson et al, 1980) (Figure 1). This compound is very electrophilic, and.

therefore, capable of ceacting with electroii-rich nucle'i acids and proteins. In both cases, cis-

DDP bding is thought to be controlled by the rate of hydrolysis via nucleophilic substitution

by solvent water (LeRoy and Thompson. 1989; Dedon and Borch. 1987). The resultuig

hydroiyzed fonn may then bind to DNA, proteins or sulphur-cont&ing compounds and by

some unLnown mechanism produce long-lived covalent DNA-DNA and DNA-protein

crossllliks (Ward et ai., 1984; Mïiier and Costa, 1989).

Figure 1. Fonnula of cis-DDP

In the case of nucleic acid binding, cis-DDP has been shown by x-ray pho toelectron

spectroscopy to bind to DNA at the C-6 carbonyl oxygen and the N-7 electron-rich sites of

the guanine nucleotide base kough its two chiorine atoms (Millard et al.. 1975). As weU.

th* agent has been show in many studies to produce intrastrand DNA crosslinks. Such an

interaction is thought to occur between the C-6 carbonyl oxygen or the N-7 of the guanine

base anci the C-2 carbonyl onygen of thymine (Morris and Gale, 1 973 1. The close promyminty

of these elecuopbilic areas to one another in DNA, and the knowIdge that cis-DDP interacu

with electron-nch substrates provides evidence supporthg this action (Morris and Gale,

1973)-

In terms of the buding of ciî-DDP to proteins, this agent bas been show to react

with methionk, histidine and cysteine (Pattanaik et ai., 1992), and to have a low propensity

for gemratng protein-protein crosslialcs (Lippard and Hoescheie, 1979). Once hydrolyzed,

cis-DDP is thought to bnd to proteins in two steps. Fit it bmds to a protein possibly

through the sulfhydryl group at a reiatively fat pace (Yotsuyanagi et al . 199 1; Chen et aL,

1994). This biniing tben causes changes to occur in the conformation of the afliected protein

wbkh expose more cis-DDP bsdmg sites (Yotsuyanagi et aL, 199 1), and ultimately produce

a crosslink that is stable to SDS, urea, and RNase A treatment (Miller and Costa, 1989).

A large volume of evidence explaios the interaction of this platinum agent with DNA

and proteins mdividually, but lit& is kwwn about the mechanism of cis-DDP protein-DNA

crossiinicing. nie observation that suifhydryi teducing agents reverse this interaction suggesu

that suühydryl linkages may be important (Watd et aL, 1984; Costa, 199 1).

For reasonr unknown, cis-DDP appears to preferentially mssiink DNA to nonhistone

nuciear proteins (Banjar et al.. 1984; Ward et aL. 1984; Miiler and Costa, 1990; Lippard and

Hoeschele, 1979; -ski et ai., 1983). Using electron microscopy, Khan and Sadler ( 1978)

have shown that thin agent primarily targets the nuchlus and inner side of the nuclear double

membrane. h t w s d y crosslinked to DNA by cis-DDP include nuclear larnins (Ferruo

et ai. 1996). actin (Milier and Costa, 1989), cytokeratins (Ward et ai.. 1984; Wedrychowski

et ai-. 1986) and NM proteins (Fetrw et al., 1992). In fact. preMous studies show that the

most abundant crossLinked proteins are cornponents of the NM (Fenaro et ai., 1995; Costa

1991). Histones are crosslùilued only when the concentration of cis-DDP is high (Foka and

Paoletti, 1986; Lippard and Hoeschele, 1979) (Le greater tban 2 mM) and the incubation

period exceeds 7 hours (Ward et ai., 19W). h one snidy, core histones and histone H 1 were

observed in crosslinked samples isolateci fiom ceils treated with 1 rnM cis-DDP for 2 hours,

however, the amount of histones crossiinlred represented only a fraction of the crossiuiked

proteins even though histones are more abimdant than any orher nuclear protein (Ward et al.,

1984). Similarly, Ward and cokgues (1984) observed that histone HI had a decreased

crosslinking eaiciency compared to cytokeratins.

In an attempt to understand the reason for cis-DDP's preference for NM proteins,

Bubiey et aL (1996) showed that fiôroblasts Uicubated in 150 p M cis-DDP for 1 hour had a

6 foici hcrease in adduct formation within NM-associated DNA compared to total ceIIular

DNA. These hdings are supported by results of previous studies which show cis-DDP to

preferentiany bind intemucleosomd DNA (Morris and Gaie, 1973; Hayes and Scoveii, 199 1).

Despite this evidence, binding of ch-DDP to NM-associated DNA was unexpected since

DNA sequences associated with the NM are usuaiiy A-T rich (Gasser and Laemmli, 19871,

and cis-DDP preferentiaüy binàs to guanine residues (Pinto and Lippard, 1985). As a result,

crossiinking of MAR DNA to NM proteins by cis-DDP mi@ be a funcrion of the MAR DNA

confomation (Bubley et al, 1996).

In addition, treatment of fibrobiasts with arginine butyrate (AB) resulted in a

concentration-cîependent increase in cis-DDP-DNA adduct formation compared to crlls

treated only with cis-DDP (Bubley et ai.. 1996). Ahhough AB aects many ceiiuk

processes, this agent pcimady inhibits hinone deacetylase (Boffa et ai.. 1978). Thk principd

action res& in the hyperaceSaiion of the lysines in histones HZA, H2B, H3 and H4 (Boffa

et ai., 1978; Schiake et al. 1994) which possibly gives the DNA an open codonnation for

mteractions with tmmaiption factors (Schlalre et ai.. 1994). cis-DDP may also then bind to

this aitered DNA Thus the increase in cis-DDP binding to this artincially opened DNA

conComation prodes m e r evidence to suggest tbat chromaiin configuration has a

significant effect on cis-DDP adduct formation (Bubley et aL. 19%).

Another factor that may be impkated in cis-DDP adduct formation is the presence

of NM proteins bound to the DNA In a study perfonned by Ho£ûnann et aL (1991). the

number and dismiution of clmg-DNA lesions was dinerent when DNA was exposed to cis-

DDP in the presence or absence of an Esherichiu coli singIe-stranded DNA-binding (SSB)

protein. DNA treated with cis-DDP in the presence of the SSB protein showed twice as

many cis-DDP adducts compared to DNA treated in the absence of protein (Pinto and

Lippard, 1985). Moreover, the majority of DNA sequences that bound to cis-DDP were

already identined as p u t e sites for cis-DDP-single-srranded DNA interactions. As a resuit,

when associahg with DNA, the SSB protein may have altered the distribution of DNA in

such a way as to expose additionai cis-DDP DNA binding sites. Thus, the presence of the

SSB may have facilitated the interaction of cis-DDP to regions of DNA not nomaiiy

accessible to ihis h g in its absence.

Alternatively, the SSB may have had the ability to melt DNA hairpin structures,

thereby aiiowing c&DDP to bind to DNA sequences that were othenvise less accessible

( U o f f m a ~ et al., 199 1). In this study, no evidence was presented to suggest that cis-DDP

crosslùil<s DNA to the SSB protein. However, the possibility exists that the binâing of thû

agent to DNA and the subequent alterations in DNA distribution or structure caused by this

interaction, may precede further interactions between cis-DDP and the SSB protein. This

niccbanisrn of cis-DDP pro tein-DNA ctosslinkllig may apply to other single-stfanded DNA

bbding proteins such as hnRNPK, a NM pro tein involved in transcription (Miche10 ttî et ai.,

1996) and DNA replication (Alberts, 1990).

Evidemx that cis-DDP crossiinks linker DNA predominantly to non-histone proteins

of NM origin (Bubley et ai., 1996) and the observations that many croulinked proteins are

associated with matrR associatedlattachment regions (MARS), suggests that this agent cm

be used to isoiate MARS. By definition, a MAR is any region of DNA associated with the

NM ( C o c k d and Garrard, 1986a; Mitkovitch et al, 1984). MARS have been identined in

DNA sequences of cis-DDP crosslinked proieikDNA cornpiexes (Ferraro et ai., 1996). In

addition, M . DNA prepared by the U S detergent extraction pro tocol of Mirkovitch et al

(1984) recognises and binds to proteins crosslinked to DNA by cis-DDP (Ferraro et al,

1995).

Breast Cancer

Breast cancer is the leading cause of death for women between 40 and 55 y û u s of age

in North America. Despite its high prevalence, the pathogenesis of this disease remains

u n c h (Kuiler, 1995; Emster et al., L996). A current mode1 of the evolution of this disease

is that breast epithelial ceiis undergo a se& of poorly understood genetic changes that result

in the progression from hyperpiasia, dyspiasia, in situ carcinoma. invasive carcinoma and

Gnally metas tatic carcinoma (Ailred et al., 1993). The develo pment of this disease relies. in

part, on the pce~ence ofesvadiol (Elenderson et aL, 1988). as weii as on the presence of other

growth fictors and peptide hormones (Lippman and Dickson. 1989). In initial stages, human

breast cancer ceils are generally weU Merentiated, require estrogen for growth, and respond

to anti-estmgen therapy. However, as the cancer progresses. breast cancer ceOs become

poorly differentiated, and no longer requke estrogen for growth. At this stage, other

unhiown growth control processes are beüevad to take over the mitoge& function of

estrogen. In adnition, late stage breast cancer ce& are tesistant to anti-estrogen therapy

(Raymond and Leong, 1989a. 1989b; Clarke et al., 1990; Thompson et ai.. 1992).

A curent mode1 of breast cancer hormonal progression stipulates a correlation

between the loss of the ER and the development of a more aggressive phenotype (Chrke et

al., 1990)- The loss of estrogen receptor (ER) expression has been suggested to occur

through outgrowth of a receptor negative tumeur ceii population, or through the formation

of variant estrogen receptors (Murphy et aL, 1993). However, when passaged continuously

m the pnsence or absence of estrogen, ER+ breast cancer ceiis do not display a change in ER

phenotype. ER- breast cancer ail lines also have not been found to alter their ER phenotype

m culture (Robertson, 1996). Moreover, the ER+ MCF-7 hwnan breast cancer ceU iine has

been shown to develop an endocrine-resûtant phenotype without loss of ER (Murphy et ai.,

1990). Such evidence argues against a phenotypic drift of the ER fiom a positive to a

negative state during breast cancer evo lu tion to hormone-independence This sugges ts that

the progression of breast cancer ceUs to a homo ne-independent s tate may no t require the loss

of the ERT and h t other cellular events are most ükely involved in the development of this

late stage pheno type.

Even though the mechanisms leading to breast cancer hormonal progression are not

widerstood. a general bekf is that the genetic events responsible for the development of a

hormone-hdependent phenotype may involve activation of dominant oncogenes and

inactivation of dominant tumor-suppnssor genes. Such a theory i supported by evidence

ihat the ümeased expIession of tbe oncogene c-&B2/HER-2 is comlated with a resistance

to endocrine therapy (Wright et al. 1992). In addition, insertion of the v-Brus oncogene

into MCF-7 ce& resulted in ceils capable of producing tumors in the absence of estrogen

(Kasid et aL, 1985).

nie mechankm by which oncogenes may heip confer hormone-independence on

breast cancer celis is unknown. However, products nom src. kk, ras and mfoncogenes may

Sumulate the RAS signal transduction pathway (Bortner et ai., 1993). The stirnuiation of this

pathway may activate AP-1 ad ETS transcription fa~tors and alter the expression of certain

cytokeratins (ie. cK8 . cK18) (Panlrov et ai.. 1994). Since cytokeratins are part of the CS&

alterations in their expression could influence the structure of the CSK, and, therefore, the

structure of the entire celL Changing the c d structure could alter protein-protein and

protein-DNA interactions. Such aiterations could leaà to the constitutive expression of

proteins such as growth factors. which may. by some mechanisni. override the rnitogenic

control of esvogen on breast cancer cek.

Considerable evidence exists showing an altered nuclear shape to be a hallmark of

transfomation (Pienta et al, 1989; Replogle and Pienta, 1996; Getzenberg et aL, 199 1). For

instance. nuclei from breast tumor cek appear enlarged with a marked variation in shape

while normal breast epitheiiai ce& display nuclei t h are srnalier, and more round or oval

(Pienta and Cofky, 199 1). In addition. evidence shows that tumor cells nom node-positive

breas t cancer patients have a sharp Smeue in nuclear area compared to those nom node-

negative patients. T M suggests that there û a comiation between nuclear area and

metastatic potmtial (Pienta and Coffey, 1991; KomitowsLi and Janson, 1990; Komitowski

et ai., 1993)-

The observation that nuclear shape is altered by transCormation combineci with the

knowbdge tbat n w k shape is determined at least in part by the NM (Replogle and Pienta,

1996) suggests that alterations in NM pro- may be indicative of transfodon and more

specificaiiy, the state of transformation (Getzenberg et ai.. 199 1 ; Pienta et a l , 1989). This

h r y is supported by observations that no& and caacerous ceUs show differences in NM

protein profiles (Pienta and Coffey. 1992; Dworetzky et aL, 1990; Getzenberg et aL. 199 1;

Khanuja et aL, 1993; Partîn et aL, 1993; Getzenberg et ai.. 1996). Furthermore. breast cancer

ceils representing difkrent stages of disease progression also display slight dinerences in NM

pronles (Samuel et ai., 1997).

Along with changes in the pmtein composition of the NMT Taylor and coIleagues

(199 1) have observeci a signifiant increase in the average size of DNA Ioops associateci with

the NM in hurnan fibroblasts either stimuiated into celi division, or transformecl with the SV40

T-antigen. As we& increases in the sites of NM-associateci Ioops were evident in human

nbrosarcoma, lung carcinoma, and cervical carcinoma ceii lines compared to nomal human

fibroblasts. From this evidence. a suggestion was made that transformation may induce -

changes in DNA structure by altering the extent to which DNA within a particular replicon

is supercoiled- Since studies have shown differences in NM pronles between normal and

tumor cells (Partin et aL. 1993; and Getzenberg et ai., 199 1; Getzenberg et aL, f996),

*rations m NM composition could be mponsible for changes in DNA topology. B a d on

this evidence, the NM may be an important prognostic marker in the pathogenesis of breast

cancer.

A protein or group of proteins is considend a prognostic d e r when its presence

is comlated with other known prognostic fxtors or when its presence is correlated with

suMva and disease-fitee SUCYiVaL The sigaincance of a prognostk marker is determjned by

ïts comIation aRth other known prognostic factors. Rognostic or diagnostic NM pro teins

may be easiiy i d e t l m in the blood and urine of cancer patients, thus eliminating the need

for invasive tissue coktion. UsiDg mine samples fkom patients. this approach has already

been successfully used in a clinral setting tu detect uro theiiai cancer (Miyanaga et aL. 1997).

In kt, the diagnostic use of a N M profen unique to this cancer provideci more sensitivity for

disease iâentification than that of other diagnostic methods such as cytoscopy or voided-urine

cytology.

Alternatively, topobgical DNA changes afker transformation rnay instead result Erom

alte+ions in CSK organuatEon ùiduced by the transformulg agent. In support of this theory

is the obsewation that gene expression changes without apparent changes in NM composition

when the CSK is dismpted with cytochaasn D (Macoska et aL, 1994; Zarnbetti et ai-, 199 1).

Furthemore. changes in CSK organization have also been induced upon transfomation of

Kirsten kidney celis with the ras oncogene (Pienta and Coffey, 1992). Genes parriculvly

influenced during breas t cancer progression ap pear to be those that encode in termediate

k e n t s .

in k t tumor ah, the p ~ c i p a l intemediate nlament proteins are cK8, cK18, and

cK19, whereas nonoal breast epitheaal ce& predomindy express cytokeratins cK4, cK.5.

cK6. cK14 and cK17 (Tcask et ai., 1990). Based on tbs difZerential expression, and

observations tbat cK8 and cK18 expression m human melanoma cell lines ieads to an increase

m mvasive and metastatic pmperiies of these ceüs (Hendrix et a l , 1992). cK8 and cK18 may

be consldeted iate mdcers of caccitloma progression. Besides cytokeratms, the intermediate

filament protein vimentin may also have prognostic importance since it is expressed by some

hormone-independent breast cancer ceii hes and not in hormone-dependent ceii lines

(Sommers et al.. 1989)-

The N M plays an important roh in the structurai and huictional organization of the

œil nuckus. A large amount of evidence exists to suggest that the NM is associated with IFS

@y et al., 1984; Fey and Peman, 1988; Ferraro et al, 1992; Miller and Costa, 1989; Ward

et al, 1984; Oiinkski et aL, 1987; E k et aL, 1990; Nickerson et al., 1992; Hozak et al7 1995).

Since IFS are components of the CSK, and the CSK is associated with the ECM (Ingber,

1994), any extraceMar force or chemical signal that &ers the CSK structure wül, in tum.

alter the structure of the NM. Since active gens are associated with the NM, aiterations in

the arrangement of NM pro teins may alter the association of NM proteins with DNA and

relieve some of ihe constraints that prohibit DNA h m unwinding (Ingber et aL, 1994). Such

an event could alter gene expression-

Hypothds and Appmach

In a pcevious study9 changes were obsenred behveen the two dimensional NM protein

p r o f i ofceii IMs tbought to represent cüttèrent stages in breast cancer progression (Samuel

et a l , 1997). Tbe NM proteins obtained for this coqmison were isolated using the 2M NaCl

high salt extraction procedure previousiy mentionad. This rnethod. however, is prone to

artekt formation. Since tk agent, cis-DDP, bas beni shown to crossMc predominanrly NM

proteins to DNA, the hypothesis of this experiment is that cis-DDP crossluilàng can be used

as a complementary appmach to the high sait extraction procedure for identifjing putative

NM proteins that may be informative markers in cancer diagnosis.

To idenrify changes in DNA-NM protein interactions during breast cancer

progression, DNA-binding pmteins fkom several hormone-dependent and hormone-

independent ceii lPies wiîl be isolated using the crosslinking agent cis-DDP, aud analyzed by

two-dimehoal electrophoresk. To conkn that the maprity of ihese DNA-bïnding proteins

are NM protebu, the two-dimensionai profiles of DNA-binding proteins and NM proteins

fiom the same ceii liae wiU be compared. Proceeding this, the hormone-dependent breast

cancer ceii lineV T47D5, w d i be grown under noccnai, estrogendeplete, and estrogen- ceplete

conditions, and the hormone-independent cell line, TS-PRF, wiil be grown under estrogen-

depkte conditions. The DNA-binding NM proteins wiü be isolated fiom the three T-47D5

ueatments and the T5-PRF ceil line using cis-DDP and analyzed by two-dimensional

electrophoresk. Merences in N M protein levels over the three T-47D5 tieatments and the

T5-PRF ce1 Iine will be assessed to detennine ifestrogen influences the association of NM

proteins with DNA

CeU Lines and CeU Culture

To identify changes in DNA-binding protein pronIes throughout breast cancer

progression, we studied homne-dependent (MCF-7, T-47D. ZR-75), ER+* hormone-

independent fl5-PRF) and ERO. hormone-idependent (MDA-MB-23 1, MDA-MB-468, BT-

20) human breast cancer ceii liaes. MCF-LOA1, a spontaneously immortalized ceii line fkom

a reduction iriammoplasty ôreast tissue sampk riait et al., 1990) was used as a conuoL With

the exception of T5-PRF, MCF-1OAi and MDA-MB-468. aU ceil lines were maintained at

37°C (humidined aunosphere, 5% COJ95% Air) on 150 x 20 mm tissue culture dishes

(Nunc) in culture medium containing DuRecco's Modined Eagle's Medium (DMEM)

suppkmented wah 1% (vlv) GgIutamine. 1% (vlv) glucose, 1% penicillinlstreptomycin and

5% (vlv) fetal bovine senun (FBS) (Gibco, Grand hiand, New York). MCF- 1OAl was

maintained in DMEM supplemented with 5% home sem (Gibco), 1 % (v/v) glutamine, 1 %

penicWs trep tomycin, 30% glucose. 1 nM hydrocortisone (Sigma, S t. Louis, Missouri)

cholera toxin (100 pg/î) (Sigma). insuiin (10 mg/i) (Sigma) . epidennal growth façtor (20

p&n) (vpstate Biotecb, Lake PIacidT New York) in a C 0 2 - k environment. The MDA-MB-

468 oell line was maintainai in L- 15 (Leibovitz) (Gibco) culture medium supplemented with

10% FBS and 1 % peniciliïn / streptomycin in a CO#ee environment. At a confluence of

approxbmely 9046. cells were removed fiom the plates with a rubber policeman and fcozen

as pekts containing 1 x 107ceL at -70°C.

To determine the effit of esmgen on cytokeratin-DNA interactions T-47D5 and T5-

PRF ceii lines were useci. Both ceii hes have been described previously (Watts et ai.. 1992:

Coutts et al, 1996). nie cell line T47DS is ER+ and hormone-depndent. This ceU üne

was maintauleci in DMEM supplemented with 1% (vlv) L-glutamine, 1% glucose. 1%

penicinincitreptornycin and 5% (vlv) FBS. TS-PRF was devebped by passaging T-47DS cek

for at k t 60 tmies m DMEM-Pknol Red Free (PRF) (Sigma) supplemnted with 5% (v/v)

NviEe cbarcoal stripped FBS, glucose, L-glutamine, and peniciWstreptomycin as previously

mentioned (5% BS). Celis were routhely passagai at 70.8096 confluence. T-47D5 ceb

acutely depleted of estmgen were grown in 5% BS for one passage. T-47D5 ce& acutely

depleted and then repletai of estrogen were grown for one passage in 5% BS foliowed by one

passage m 5% BS containhg 10 nM eseadiol m an ethanol vehicie. AU ceUs were passaged

at 70-80s confluence asing Earie's EDTA solution. 1 x IO7 ce& were harvested at

approximately 80% confluence using a rubber policeman, and cell pellets were stored at -

70°C.

CbDDP C d n k i n g

Cis-DDP crosslinking was p e r f o d accordhg to the procedure of Ferraro et aL

(1991). Except for the initial 4 centriftgation steps, ail steps in this protocol were perfotmed

on ice unless othenvise mentioned. AU solutions used in this p-toc01 contauleci 1 mM

pheny~thyisdfonyl fluoride (PMSJ?). 1 x 10'ceiJs were washed a total of three tirnes in 30

ml of Hanks b&er and centrifiiged at 50 x g for 8 min at mom temperature in between each

wash.

C e b were then resuspnded in a lresh solution of either 1 mM or 3 rnM cis-DDP

solubilized in Hanits buffer contauillig 137 mM NaAc uistead of NaCL NaCl was exciuded

kom this solution in order to prevent the chioride ions nom competing with the cellular

proteins for cis-DDP and, thetefore, impair the crosslinlong reaction (Lippard. 1982). The

suspension (1Q ceWm was incubateci at 37°C for 2 hours with constant shaking. The

iength of incubation and the use of a 1 mM concentration of cis-DDP was based on

crosslinking conditions used in previous studies (Ferraro et aL, 1991; Ferraro et aï., 1992;

Rrraro et ai., 1995; E;erraro et al , 1996). In these studies, a cis-DDP concentration of 1 m M

and an incubation t h e of 2 hours appeared to be sufficient for optimal DNA-protein

aossiinking without sipifimt msslinknig of histones. However, since these studies did not

use breast cancer tek, a crosslinLing expriment using a 3 rnM ch-DDP concentration was

ais0 performed to determine if a 1 mM cis-DDP concentration was high enough to produce

the same two dimensional pattern of DNA-crosslinked NM proteins.

Following treatment, the ce& were ceneifuged ai 50 x g for 15 min at room

temperature, and the cell pellet was resuspended in 10 mi of lysis buffer (5 M urea, 2 M

guanidine-hydtocbloride, 2 M NaCi, 200 mM potassium phosphate buflier, pH 7.5). nie ceii

lysate was then combwd with hydroxyapatite (HAP: Bio-Rad. California) for 1 hour at 4°C

with continuous mixing. HAP was used to isolate ali cellular DNA in the cell lysate including

DNA crosslinked to proteins by cis-DDP. nierefore, the HAP was used to indirectly separate

the pro teins crosslinked to DNA fkom aU O ther non-cmsslinked proteins in the ceii lysate.

Che gram of HAP was used for every 4 mg of DNA in the ceil lysate since this ratio has been

shown in a previous study conducted within our lab to bind aU cellular DNA with

approximately 10046 eflkiency (Li et ai., 1993). To determine the arnount of hydroxyapatite

to combine with the ceU Lysate, 10 pl of ceii lysate was aliquoted into 990 pl of iysis buffer

and the A, reading was recordeci. This reading was appiied to the biiowing equation to

determine the totai mg of DNA present in the ceii lysate:

Il, x 50 pglml x LOO x 10 ml! 1000 pg per mg of DNA = total mg of DNA

One %, unit Epresents 50 pg of DNA per ml of ceîi lysate, thus the absorbante reading was

6rst muitiplied by 50 and then by the dilution factor (1 0) to detemine the pg of DNA in 1

ml of ceil iysate. nie temithg vahie was multiplied by the toial volume of cell lysate (10 ml)

to deremiine the total pg of DNA in the celi lysate and then divided by 1000 to convert this

value to mg. The total mg of DNA was M e d by 4 to determine the amount of HAP to

use. After incubating the EUW with the ceii lysate for 1 hou, the resin was isohted by

centrifiiging the celi lysate-HAP slurry at 3000 x g for 5 min at 4'C. The resin was washed

three times with 20 ml of lysis baer . The HAP was then resuspended in 10 ml of tysis

buffer containhg 1 M thiourea instead of 5 M tuea in order to reverse the crosshk between

the DNA and ihe NM proteins. The slurry was incubated at 4'C on an or bitron for 2 hours.

Foiio wing incubation, the thiourea solution containhg the DNA-binding pro teins was

separated fkom the HAP by cenaihigatfon at 10000 x g for 20 min at 4°C. This solution was

then dialysed for 24 hours at room temperature against 5 changes of double distüled water

(400 ml per sample per change). ).r dialysis, the solution was lyophiiized and the resulting

proteins resuspended in LOO pl of 8 M urea The protein concentration for each sample was

&termuied using the Bio-Rad Protein Assay (Bio-Rad) with bovine semm albumin (BSA) as

a standard. Samples were stored at -20°C and analyzed by either one-dimensional or two-

dunemional gel electroph~resis~ l"he separate crossluiking experiments were perfonned for

each ceil line studied.

Nuclear marrices wae isoiated accotdng to a procedure previously reported (Samuel

et al., 1997). Ceil peku containmg 1 x 10' ceils were resuspended in 10 ml of ice-cold

TNM butIér (100 mM NaQ 300 rnM sucrose, 10 mM Tris, pH 8.0.2 mM MgCb, 1% (v/v)

tbbd@coi) with 1 mM PMSF. The ceii suspension was homogenized 5 times with a Tenon

pestle on ice. Foiiowing homogenkation, cells were incubated on ice for 5 min. A 6nai

concentration of 0.596 (vlv) Triton X-100 was added to release iïpids and solubie proteins

whik nuclei remain intact. The suspension of nuclei was passed tbree times through a 18

gauge needle ami cokted by cenaifllgation at 1ûûû x g for 10 min at 4'C. The nuclei were

again resuspendeâ in ice-cold TNM bu&r with 1 mM PMSF, homogenized and pekted as

before. The nuclei wen resuspended to a concentration of 20 AJml in coid DIG (50 mM

NaCi, UX) mM sucrose, 10 mM Tris-HCi, pH 7.4,3 mM MgC4, 1 % (vl v) thiodiglyco~O.S%

(vh) Triton X-100) and digesteci with DNase 1 at a coacentration of 168 unitdrni for 20 min

at room temperature. Ammonium sultate (anal concentration of 0.25 M) was added with

stimng to faciltate chromath removal, and the NM (NM-IF) pekt was obtained by

centrifugation at 960 x g for 10 tnin at 4OC. The NM-IF peiiet contains NM associated with

IFS. This pellet was resuspended in ice-cold DIG with 1 mM PMSF, extracted by adding 4

M NaCl to a nnal concentration of 2 M, and incubated on ice for 30 min. The sarnple was

centfigeci at 9600 x g for 10 min at 4'C. The NaCl extracted NM pellet (NM2-IF) was

again resuspended Pi ice-cold DIG. extracted with 2 M NaCl and centriluged as before. The

NM2-IF was resuspended in Disassembly Bder (8 M ma, 20 mM 2[N-mo cp holino J ethane

suifonic acid, pH 6.6. 1 mM EGTA, 1 mM PMSF, 0.1 mM MgCb, 1% (v/v) P-

mercaptoethanol) and dialyseci overnight at room temperature against 2 liters of Assembly

Bunét (0.15 M K a 25 rnM noidazole, pH 7.1.5 mM MgCa 2 mM DIT. O. 125 M EGTA

0.2 mM PMSF). Dialysis aiiowed the una to be removed and the IFS to teassemble. The IFS

were removed by Sracenuifugation at 1509000 x g for 90 min. The resulting supernatant

containing NM proteins was removed carefiilly and lyophiüud. Lyopbilued samples were

resuspended in 8 M urea, aliquoted aud frozn at -20°C. Before two-dimensionai gel

eiectrophoresis, protein concentrations were determined using the Bio-Rad Assay with BSA

as a standard.

Quantification of Cytokeratin Levek in Crudiniceci Sampks

To find the relative ievek of the cytokeraths in the cis-DDP crosslinked pro teh

preparation fkom various ce& the proteins were resolved on 12% SDS (sodium dodecyl

sdh@ gels accordng to the method of Laemmli (1970). Gels were then stained with 0.0496

S e m Blue G (Sema Biochemicals, Westbwy, NY), and quantification of cytokeratin levek

was pecfonacd on the stained gel using scanning densitometry and Quantity One" version 2.7

software (PDI. Kingston Station, New York). Initially. a linear range of protein siaining

versus the protein bad was es tablished by loading increasing amounu of NM-IF proteuis on

a SDS geL Before determinhg the relative amount of cytokeratin in each preparation. we

checked that the amount of cytokeraturs ùi a prepmtion was within the linex range. The

amount of cytokeratin pet pg of pro tein was then caiculated.

Two-DimeusionaD Cd Electrophods

Two dirœirnonai gel electropûoresis was pefionned on the isoiated proteins according

to the method of O'Famll(197!5). Separation of proteins by their isoelectric point was

perfonned by loading 80 pg of protein sample on to isoelectric focusing tube gels (1.5 mm

x 18 cm) contajning 2% pre-blended arnpholines of isoeiectric point (PI) values of 3.5 - 9.5

and 5 - 8 (Phmach BioTech, Uppsaia, Sweden). The gels were electrophoresed at 400 V,

400 mA for 16 hours and then 800 V, 400 rnA for 2 hours. Mer electrophoresis. the gels

were placed in a sample ducing bu&r contaking 3 % (wlv) SDS. 1.5 46 (wlv) Dm, 0.07

% (w/v) Tris-HCI, pH 6.7, 0.01 % (w/v) and Bromophenol Blue for 20 min at room

temperature. Following the incubation. the tube gels were layered on to SDS - 8% PAGE

resolving gels prepared and ran according to the method of Doucet and Trifaco (1988).

Eiectrophoresis was perfomed at a constant amperage of 50 mA per gel for 2.65 hours at

room temperature. The molecular mass and pI of the sample proteins was deterxnined using

two-dimensional SDS-PAGE standards (Bio-Rad) and carbamyhted carbonic anhydrase

(Phanaacia Biotech). Gels were stained with dver using the Pharmacia Silver Stain kit and

then dcied benmen sheets of gel drying film (Promega Corp.) at room temperature. Stained

gels wece scanned d g a PD1 3250E densitometer (PDI, Huntington Station, N.Y.) and the

data was analyzed with Image Master software (Pharmach Biotech).

Western Blot Anaiysis

Western blots were prepared by NnriOg DNAcrosslinked samples (30 pg) on a 12%

SDS gel according to tbe method of Laemmli (1970) at room temperature for 45 min at 200

V. Pro teins were then transferred 6rom the SDS gel on to a nitroceiiulose membrane (Bio-

Rad) in the presence of CAPS m e t baiffer (25 mM of 3-[cyclohexykminoJ-1-

propanesulfonic acid, pH 10,2W methano1 (vlv)). 'Ihe protein transfer took place ovemïght

at 30 V and 4 O C . FoUowiag transfer, the membrane was oicubated for 1 hour at room

tempe- in a bbcong solution containhg 5% slom milk and 0.2% Tween-20 in LX Tris-

Buffereâ Saiine (TBS: 100 rnM Tris-Ci, pH 7.5,0.9% NaCl). The membrane was then

washed two times m 1X TBS containhg 0.2% Tween-20, and immunochemically stained with

the human anti-vimentin antiidy (Monosan, Uden, The Netheriands), and the goat ami-

mouse antibody linked to horseradiîh perodase (Sigma) using the ECL (enhanced

chemiluminescence) detec tion sys tem (Amenham LXe Science Inc., Arling ton Heights,

nlinois).

Cell Lines and Cell Culture

The human breast cancer ceil lines used for investigating changes in the composition

of DNA-bLiding proteins over breast cancer progression were T47D, MCF-7,ZR-75, BT-

20, MDA-MB-231, MDA-MB-468, and T5-PRF. In addition, the c d ihe MCF-IOAI, a

non-tumorigenic human breast epithebi ceil line arising nom spontaneous irnmortalization

of bmst epitheiîai ceb obtained from a reduction mammoplasty, was chosen as the closest

representative to normal breast epithelial celis. The characteristics of these ceii lines are

shown in Table 1. According to the major characteristics oôsewed in early and iate stages

ohalignant progression in human breast cancer (Table 2). the ceii hes T-47D, MCF-7 and

ia-75 riiiimC the pbenotype of early stage breast cancer while the celi hes TS-PRF. BT-20.

MDA-MB-231 and MDA-MB-468 mimic late stage breast cancer. AU these cell hes were

cultureci by Dr. Shanti Samuel

The human breast cancer ce1 lines used for ïnvestigatkig the effects of estrogen on

cytokemth levels within human breast cancer cek were the T-47D5 and TS-PRF ceil lines.

The T-47D5 ceii iine was origindy thought to be a subclone of T-47D ce& but was later

identifleci by DNA fkgerprinting as an MCF-7-nlated cell iine (Watts et al, 1992). These

NJO cell hes were gmwn and h o r m o d y treated by Ms. Amanda Coutts.

Table 1. Mapr charaçteristics of human breast cancer ceil lines. as weiI as a human breast epitheiiai ceii iine.

CeN Tumor Tissue ER Esbogen Esaogen AnüesaPgen lmasive Metastatic Une Type" SourceD StatusV Deperidence .- .

Responsivmss serwMty

T470 IOC PE + + Y +" +* + + T4705 IDC PE +v +" +v +" ND ND MCF-7 IDC PE + +" +O +O u ++ DR-7s IDC Ascites + +c +O +' + + T5-PRF IûC PE + œV O +v ND V

C ND

MDA-ME231 AC PE - 9 P +++++ ++CC+

MOA-MB-468 AC PE O E 1 4 + +++ BT-20 IOCC OF ,< -c +" ND" ND

a a MCF-1OAl NP RW P O ND ND ND

Unlas othcxwisc spedicd, the histopathological diagnosis of the ceIl Iines was cletamined by Thompson et al. (1992): IDC, infiîtrating ductal amhoma; AC, adenocarCmoma; NT, non- huncinmc. Unlss otbawise specifIeà, tissue source was obtamd b m niompsoo et aï. (1992): PET plairal &ion; OT, original bnast m, RD, rCdlEdi011 mammoplasty. Uniess otbenuïse speciaexi, ER status was âcmmbd by Sommas et al., 1989. Acrivity ofeach cd line in Bo* chamôer chemomvasioa assay. Activity is graded as % MDA- MB-23 1: +, 0-2046; ++, 20-4046; w, 4060'ib; +-H+, 60180%; ++++, >8096. ND, not detennined- PerfOllbed by Thompson et al. (1992). Acciviy m Bo* chambcr chernotaxis toward fibroblast-conditioned medium. Activity is graded

% MDA-MB-23 1: +, 0020%; ++, 204%; +te, 40-6û%; 60-8096; ++++, >80%. ND, wt dctermined Paformed by Thompson et ai. (1992). Although this ctll lioes was not inchded m Bo* chacdxx cheminvasion assays, BT-20 has been shown in a previous study to have a higher grade of inaltrative growth into precuittired heart hgments than the MCF-7 human breast cancer ceU linc (Mandeville et al., 1987). f)etermined by Reddel et ai. (1984). Detennined by Coutts et al. (1996). Determined by Seii et al. (1983). Detennined by Suthaland ct al. (1983). Determiad by Engel and Young (1978). Determureai by Lippman et al. (1976). Detefmifled by Armstrong et al. (1992)- Detennind by Kenney et al. (1993). Detennitled by Soule et al. (1990) Dctefmined by Engel et al. (1978) Dctermined by Toi et al. (1992) Determined by Reddel et al. (1985); however, the œll üne BT-20 was much less sensitive to antiestrogen than the T 4 D , MCF-7 and ZR-75 cells Liaes.

Table 2. Major characteris tics of early and late stage breast cancer (Clarke et al., 1990).

Honnooe-Depeadeot Honaohe-Iadcpendeat Estmgen rtsponsive Estmgen aonresponsive Poorly hvasive Highiy invasive Poorly mctastatic tiigûlyamstatic '

Antiestrogca-sensitive Antiestcogea-insensitive

Crosslinh'ng of DNA to pmteins by d5-DDP Ui Jitu in brePPt canœr di €ines

DNAhimihg protein pro& fimm homonedependent (T-47D. MCF-7.ZR-75) and

ho rmone-independent 05-PRF, MDA-MB-23 1, MDA-MB-468, BT-20) human bceast

cancer c d lines were analpd to idenMy aiterations in the patterns of DNA-bSding proteins

over breast cancer progression @om a homonedependent to a hormone-independent stage.

A non-iumorigenic human breast epitheliai ceU üae arising fiom spontaneous irnmortalization

of breast epithelial ceh obtained 6rom a reduction mammoplasty was chosen as the closest

representative to normai breast epitfieiia In oder to obtain DNA-binding protein profiles,

each ce11 Inie was treated with 1 m M cis-DDP in an environment with a physiological pH and

temperature (37°C). and a very low Cl- ion (4 mM) concentration, and proteins crossiinked

to DNA in situ were isolated. niis cis-DDP treatment crosslinked proteins bound to DNA

U1 situ. As a resuit, a b w Ci- ion concentration of the crosslinlùng buffer was required since

the hydrolysis of cis-DDP into its active DNA-binding (Dedon and Borc h. 1987) and pro tein-

binding (LeRoy and Thompson. 1989) form is signincantly inhibited in high chloride

environments (Chu 1994). Roceeding isolation, the proteins crosslinked to DNA in each ceii

üne were resolveù by two dimension gel electrophoresis, and the resulting gels were stained

with dver. The silvw stained two-dimensional gels were then scanned using a densitometer

and analyzed using an Image Master System (Pbmch BioTech). Crosshkïng experiments

were perfonned three tims for each ce1 laie. Only proteins that consistently appeared in alI

rhree aossbked pmnles of a c d IM were considerad representative DNA-binding proteins.

Anaiysis ofpmtciir9 cmdhked to DNA by cbDDP ù, in breast cancer ceil liaes

(3nce the DNA-binding protem pro& of each ceii line was identined. a cornparison

was made among the DNA-biading protem profiles of the human breast epithelial ceil h e

(Figure 2). and the 3 hormone-dependent (Figure 3) and 4 hormone-independent (Figure 4)

human breast cancer ceii lines previously mentioned. From this cornparison, 4 distinct

clusters (A, B, C. D of Figure 5) were identifid as common to the profles of proteins-

crossluiked to DNA by cis-DDP in ai l eight ceii lines. Thus, ail pro- of proteins

crosslinked to DNA by cis-DDP were compand using clusters A, B. C, and D as alignment

markers. 'Lhe molecular weight and isoelecaic point coordinates of each cluster are kted in

Table 3. These 4 clusters account for the majority of proteins that were abundantly

crosslinlred to DNA. Thus, human breast epithelial, and h u m breast cancer ceUs display a

large similarity in their DNA- binding protein pronles.

Table 3- Molecular weight range and isoekctrk point range of four DNA- crosslinked protein clus ters common to pseudo-no rmai breas t e pit helial and breast cancer ce1 lines.

Protein Cluster MW range @Da) of p I range O t pro teins pro teins within cluster within cluster

A 57 5.4 - 5.5 B 59 5.0 - 5.25 C 66 - 71 5.25 - 5.45

A cornparison of the isoelectnc point and molecular weight coordinates of the

prominent cornmonly crosslùiled proteins with the coordinates of proteins identikd in

previous studies shows diat the transcription factor hnRNPK was cn>ssLinked to DNA to the

same extent in aü the ceU lines studied. Lamins A and C were also evident in ail crosslinked

preparations when higher loads (80 pg) of protein were used (Figures 2.3,4).

In addaion to lamms A and C, and hnRNPK, cytokeratins 8 (54 kDa, pI 5.4). 18 (45

kD4 pI 5.3). and 19 (41 ma, p14.9) also appeared as prominent proteins in the pronles of

proteins crosslinked to DNA in situ by cis-DDP for the ceii lines studied (Figures 2, 3,4).

l[bese 3 cytokeratins were identined in a previous study as abundant proteins in the NM-IF

thction fiom T-47D5 hormone dependent bnast cancer ceh by microsequencing

polypeptides excised b m a two dimension gel (Coutts et al, 1996); Therefore, this analysk

provided the preciîe locations of these thne cytokeratins in the two dimension crosslinked

protein profiles. Furthemore, cK8, cK18 and cK19 in the NM-IF preparation comigrated

with the three polypeptides p54, p45, ami p41 of the cis-DDP DNA-crosslinked protein

preparation resolved on two dimension gels (data not shown) .

Despite ihe large de- of siiaùarity in the DNA-bOdOig protein pronles amongst the

ce0 lines* ciifferences were observed betwem DNA-crossWed preparations of the pseudo-

n o d breast epitheiial, and the hormonedependent and hormone-independent breast cancer

ceu h e s (Figures 29 3.4). These di&rences are! summaraed in Table 4. To c o b that

these différe- were rot due to bading error, the protein representing hnRNPK in the two-

dimensional pmnles of an the ceU lines studied was used as an intemai loading controL This

protein appeared to be crosslinked to DNA to a simüar extent in a l i the ceii lines useci.

In the pseudo-normal breast epithelial ceii iine. MCF-1OA1. two DNA-binding

pro teins @Cl: 36.5 kDa, pI 4.3; BC2: 52 kDa, pI 5.9) were present at higher kvels than

those obsenied m &-DDP preparations oftbe other ceii iines (Figure 2). In addition* DNA-

crosslinked protein pronles of homonedependent T-47D and MCF-7 breast cancer cell Lines

showed an increase in abundance of proteins witbin cluster D compared to the crosslinked

profiles of other cell iines (F~gure 3). This increase was particuiarIy noticeable for proteins

within the molecular weight range of 48 to 53 kDa, and the isoektrk point range of 5.2 -

5-4. A cornparison of the isoelecaic point and molecuiar weight coordinates of these proteins

with those of IF proteins isolatecl fiom the human breast cancer ceîi lines MDA-MB-23 L

inckated that a significant portion of the cluster D proteins may be IF proteùis. Within this

cluster are three proteins (Ba: 47 ma, pI 5.25, BC4: 46 kDa, pI 4.2. BC5: 43 kDa. pI 5.1)

found predominantly in the T-47D and MCF-7 ceii lines that are either absent or barely

detectable in the DNA-crosslinked preparations of the other ceii iines.

+/- Indicates the presence (+) or absence (-) of a DNA-binding protein Gom a two-

dïmensio nal pro file of DNA-crossiinked pro teins.

+(L) S i C a a lower abundance of ths protein when c o m p d to levels seen in other ceii

lines. Abundance was determined using hnRNPK as a loding çontrol.

Sindu to the homiune-dependent bnast canccr and pseudo-normal breast epithc Lia1

ceIl lincs. the hormone-indrprndcnt hrrÿst ciuicer cc11 Ihes also displayed DNA-hinding

pnitcim ih;it wcre either pmicularly ahundrui~ or cxçlusivc IO n ht~rmonc-indcpcndcni hrc:isi

protein (BC6: 41 Id>& pI 4.65) was most abundant in the cis-DDP crossiïnked protein

preparatom of the 4 hormone-independent celi lines. whiïe one other DNA-buiding protein

(BC7: 46 ma, pl 4.9) was signiticantly more abundant in DNA-crosslinked protein

preparations of the BT-20 and MDA-MB-468 ceii lines cornpared to the other celi hes

studied. LikewW. three D N A - b u g proteins (BC8: 110 ma, pl 4.7; BC9: 44.5 ma, PI

4.1; BCLO: 34 kDa, PI 4.55) were found most abuadantty in DNA-crosslinked protein

preparations of both ttic MDA-MB-23 1 and MDA-M.-468 celi lines cornpared to the other

œil iines smdied In addition, tk BT-20 celï line displayai a cluster of DNA-binding pro teins

@Cl 1: 79 kDa, pl 5.2-5.4) that was not detected in DNA-crossIinked protein preparations

of any other ceIi line studied- Furthemore. one protein bearing srmilar coocdinates to

proteins within MDA-MB-23 1 IF preparations (BC12: 57 ma, pl 5.0) was most abundant

m the MDA-MB-23 1 cell line.

Western blot a n a l . . of a two dimensional profile of MDA-MB-231 proteins

crosslinked to DNA by cis-DDP using a vimentin antibody produced a signal at the same

molecular weight and isoeiectric point coordinates as the Be12 protein. However, the

strength of this signai was not quivalent to the intensity of the BC12 pro tein when stained

with silver, suggesting that either vimentin is CO-rnigrating with a protein of a similar

rnolecukr weight and isoekctric point or that the association of vimentin with DNA changes

the shape of Wnentin in a way ut hides the epitope recognited by the antibody. Besides

BC 12, two O ther pro teins were identined most abundantly (BC 13: 46 kDa. PI 4.6; BC 14:

44.5 kDa, pI 4.65) in the DNA-crosslinked protein preparations of the MDA-MB-23 1 ceD

line when cornpared to the crosslinked preparations of the other ceil lines studied. Like

BC 1 2, these proteins &O have rnolecular weight and isoelectric point coordinates that were

sunilar to proteins within MDA-MB-23 1 IF preparatiow.

AAer twodimensional pro* of proteins crosslinked to DNA by cis-DDP in situ in

the cell lines were cornpanxi to one another, the two-dimensional profile of cis-DDP DNA-

c r o s s W proteins fiom the T47D ceii line was compand to a two-dimensional profile of

N M proteins from this same cell b e @gure 6). The preparation and analysis of ail NM

pro& presented in this paper was perfonned by Dr. Shanti Samuel in a previous study

(Samuel et aL, 1997). whüe any cornparision of the NM pronles with DNA-crossiinked

pro% was p d o d by rnysef. The cornparison of the T47D N M and DNA-crossiinked

protein profiles showed a iarge simikrity in the composition of both pnparations. The NM

preparation displayed the four ciusters of abundant proteins observed in the DNA-crossluiked

protein preparations fiom this study (Figure 5). Since the 4 alignment clusters were not as

evkknt at bwer loads (40 pg) in the crossliniced preparation. a pro tein designated as NMP 1

(46 kDa; pl. 5.0) also servecl as a point of reference for cornparhg the two-dimensional

profiles of NM and DNA-crosslinked pro tein preparations.

In addition to the four clusters, the proteins hnRNPK, lamins A and C, and

cytokeratins 8, 18 and 19. which were prominent proteins in DNA-crosslinked preparations.

also appeared to be prominent NM proteins in the T-47D ceii he. Süniiar observations were

made when cornparhg the NM pronles of the other ceii lines used in this study with their

respective DNA-crosslinked pattern (Figures 2. 3, 4. 7. 8). Furthemore. this cornparison

showed that the majority of proteins (BC3, BC4. BC5. BC6, BClO. BCl 1. BC12, BC13.

BC 14) with a clifliecenthi abundance between hormone-dependent and hormone-independent

breast cancer celi lines were a h identined in tbe NM pronles of the respective ceii h e fiom

which they were isoiated. Table 5 summataes the presence or absence of DNA-binding

proteins with a differeneial abundance between the 3 hormone-dependent and 3 of the 4

hormone-indepeadent beast cancer ceii lines studied. The iarge sunilarity between NM and

ch-DDP DNA-crosslinked protein profiles suggests that most of the abundant proteins

crossiinkeà to DNA with cis-DDP are NM proteins.

Despite the large simihrity between the NM and DNA-crosslinked protein profi& of

the T-47D ceii line, severai proteins that were a b d a n t NM proteins in these cells did not

appear m the pro& of proteins crosslinked to DNA by cis-DDP in T-47D cells (NMF2:

49kDa. pI 5.0; NMP3:55 ma, p15.45; NMP4 52 kDa, pI 5.5) @gure 6). In addition, when

compacing the twodimensional profiles of DNAnosslinked proteins in the other ceii lines

with NM pro& of their respective cell line, many DNA-crossIinked proteins were either

bmly detectable or absent fiom NM preparations (Figures 2, 3.4.7,8). For instance, the

protein BC5 was easily identifhbk in the MCF-7 DNA-crosslinked protein preparation, but

could not be detected in the MCF-7 N M pro&. As weii, other proteins @Cl, BC2, BC7,

BC8. BC9) croulinked to DNA by cis-DDP could not be detected in the NM preparations

of the ceil lines nom which they were isolated.

Shtus ofcic-DDP DNA-çrosslinkrd proteins of hormonc-dcpcndent and

hormone-independent breûst cancer ceU lines Ïn the NM pro t k s of 7 of

the 8 ceii iines studieâ.

+I- Indicates tbc presence (+) or absence (-) of a DNA-binding pmtein from a two-

dimensional pro& of DNA-crossiinked pmteins.

+(L) Signines a bwer abundance of thk protein when c o m p d to levels seen in other ceii

ünes. Abundance was detennined using hnRNPK as a loading conuoL

in a previous study, Dr. Shanti Samuel compaced the two-dimensional profdes of NM

proteins of the folbwing c d lines: MCF- LOAI, T-47D, MCF-7,ZR-75, TS-PRF, MDA-MB-

23 1, and BT-20 (Samuel et aL. 1997). As a nsdt of thk analysis. 5 NMPs designated as

NMBC 1.2.3.4 anâ 5 were found most abundantly in the 3 hormone-dependent (T47D,

MCF-7.ZR-73 and the TS-PRF, hormone-independent human breast cancer ceii luies (Figure

7) as weli as in hormone-depndent breast tumon, whüe a NMP designated NMBC 6 was

most abundant m hormone-independent human breast cancer c d IùKs (MDA-MB-231, BT-

20) (F~gure S) and hormone-independent breast himors (Samuel et al., 1997). Thus, while

comparing Dr. Samuei's NM pronles of these bceast cancer c d Iines with their respective

DNAaosslinked protein profiles, the presence or absence of these 6 proteins within DNA-

crossiinked protein patterns was noted. Two dimension profiles of proteins crosslinked to

DNA by cis-DDP dispkyed NMBCl (57 ma, pI 5.5) and NMBC2 (65 kDa, pI 5-15} as

prominent DNA-binding proteins in the ail the ceii lim studied including MCF-IOAI,

wfaereas NMBC3 (40 kDa, pI 5.4). NMBC4 (41 kDa, PI 5.3). NMBCJ (39 kDa, pI 5.9, and

NMBC6 (52 kDa, pI 5.7) couid not be detected.

Ln addition to Dr. Samuel's identification of 6 N M proteins with potential prognostic

sig- in breast cancer, Khanuja and colleagues (1993) identifid three NM proteins O<,

Y, Z) that were specifk to malignant breast tissue. Of these three proteins, however, Dr.

Shanti Samuel and coiieagues could only identify protein Z in the NM preparations of

homonedependent and hormone-independent breast cancer ceU lines (except for TS-Pm,

and hormone-de pendent and hormone-inde pendent breast turnors. S Vnilarly, pro teins X and

Y CO uld not be detec ted in DNA crossiinked pro tein preparations O € the celi ünes s tudied,

while protein Z was present at ieveis j'ust above detection in di breast cancer ceii ihes except

for ZR-75. TS-PRF, and BT-20. However, in addition to the breast cancer ceiî lines, protein

Z was also identi6ed in the DNA-crosslinked protein preparations of the pseudo-normal

breast epithelial ceii line. MCF- LOAl .

Estmgen reguiation of cytokeratin-DNA intedons in breast cancer cek

A ptevious study showed that the leveIs of cytokeratins in the NM-IF fraction of T-

47D5 ceils were altered when ce& were grown in the absence of estrogen (Coutts et al.,

1996). Since cytokeratùis are some of the prominent pro teins crosslinked to DNA by cis-

DDP, this agent was useci to determine whether the interaction of cytokeratins with DNA in

situ was regulated by estrogens. T-47D5 ceiis grown in the absence of estrogen for one

passage weze treated with cis-DDP and the crossWed proteins fkom these ceb dong with

crosslinked proteins fkom T-47D5 ceUs cdtured in the presence of estrogens were

electrophoreticaiiy resolved on one dimension SDS geis (Figure 9) and two dimension gels

(Figure 10). Cornparison of these electrophoretic patterns showed that the abundance of the

cytokeratins 8, 18, and 19 bound to DNA in ceUs grown in the absence of estrogens was

reduced to levels 0.- 0.08, 0.64 t 0.02, and 0.59 0.1 1 (n=S, where three experiments

were pedomed and two of these three experiments were npeated). respectively, of the T-

47D5 ceUs grown with estrogen in complete media. When these estcogen-starved T-47D5

ce% were passaged into 5% BS supplemented with 10 nM estrogen for 1 passage (72 hours),

the amount of cytokeratins 8, L8, and 19 bound to DNA in these cells rebounded to kvek

higher than those found in T-47DS ceUs cultured in the presence ofestrogens (1.95 t 0.22.

1.59 t 0.22, and L .6 1 t O. 17, respectively) (n=5. where three experirnents were performed

and two of these thne experiments were repeated).

The NM-IF h t i o n of long term chronic esvogen depkted TS-PW (ER positive,

hormone-independent) breast cancer ceUs coniasled elevated levels of cytokeratins 8. 18 and

19 compared to the parent T47D5 ceii line cuitured in the presence of estrogens (Coutts et

ai., 1996). The cytokeratin 8, 18 and 19 levels in the pnparation of proteins crossslined to

DNA by cis-DDP in TS-PRF ceils were greater than those in preparations from T-47DS cells

cultured with estrogens (2.87 t 0.9,2.0 t 0.41. and 1 -93 t 0.16, n=S. respectively).

nie proteins shown in Figures 9 and 10 were nom ceiis crosslinked with 1 mM and

3 mM cis-DDP, respectively. Scanning densitometry of SDS-PAGE gels containhg DNA-

aosslinlred protein sarnples obtained by treatrnent of t&e T-47DS and TS-PRF celi iines with

3 mM cis-DDP showed levels of cytokeratins 8, 18 and 19 in the three T-47D5 estrogen

treatments as weii as in the TS-PRF ceil iine that were similar to cytokerath leveis obtained

by crouiinking celk with 1 mM cis-DDP. Thus. the relative abundance of these three

cytokeratins in the crosslinked protein preparations from T-47D5 cells cuitured with or

without estrogens and nom TS-PRF cells remaineci t i ~ same as the relative abundance of

cytokeratins observeci in cells treated with 1 mM cis-DDP.

In addition to cytokeratins 8.18 and 19, changes in protein levels were also observed

for proteins in cluster D that wem imrneûiately sunoundmg the thee cytokeratins (Figure 10).

This ùdicated that proteins other than cytokeratins may ais0 be estrogen-regulated.

However, the identity of these proteins remains unlnown.

Figure 2. Proteins crossiïnked to DNA by cis-DDP in situ lrom the human breast epithelial ceii line MCF- IOAL. Eighty pg of pro teins crosslinked to DNA with 1 mM cis-DDP were electrophoreticaiiy resolved on two dimension gels. The gels were stained with silver. The position of the carbarnyiated f o m of catbonic anhyàrase is indicated by ca. The position of the mokular rnass standards (kDa) is show to the leA of the two dimension gel patterns. LA and LC show the position of lamin A and C, respectively. cK8, cK18 and cK 19 idente cytokeratins 8, 18 an 19. respectively. hK designates the position of transcription factor hnRNPK NMBC 1 and M C 2 represent proteins founcl at higher leveis in NM preparations of hormone-dapendent breast cancer ceii Lines in a previous study (Samuel et al.. 1997). Protein Z ideniines the location of a protein previously found to be specitic to malignant tissue (KhYiujaet ai.. 1993). BC1 and BC2 represent proteins found only in MCF-IOAI cis- DDP prepantions of proteins crosslinked to DNA in situ. BC 13 represents n protein most abundant in cis-DDP DNA-crosslinked preparations of hormone-independent human breut cancer cell Lines.

Figure 3. Proteins crosskiked to DNA by cis-DDP in situ fmm homonedependent bieast cancer ces fines. Eîhty p g of pmteins cmsslinked to ONA with 1 rnM cis-DOP were eiecbophoreücaiiy resoived on Dno dimension gels. The gels were stanied with silver. The posiüon of the cahamylated fons of carbonic anhydrase is indiited by ca. The position of the moleailar mas standafds wa) k shown ID îhe lelt of the Iwo dimension pl patterns. LA and LC show the position of lamin Aand C. respectively. cm. cK18 and cK19 idenûty cytokemüns 8.18 and 19. respecibely. hK designates the position of ûansciipbn facbrhnRNPK NMBC1 and NMBC2 repmsent proteins found at higher levels in NM preparations of hormone- dependent bieast cancer cdl hes in a previous study (Samuel et al.. 1997). Protein Z identifes the location of a NM protein previously found to be specifk D malignant tissue (Khanuja et al.. 1993). BC3. BC4. and BCS represent pmteins found piedominanlfy in hononedependent cis-DDP preparabns of pmteins crosslnked to DNA in sihr cornpared b cmsslinked piepara~ons of me other cell hes shidied. BC7. BC8. BC9. BCIO and BC13 represent pmteins found rnost abundandy in ONAcmsslinked preparations of hononeindependent human brsa~t cancer cell lines.

-sau!i ga3 ia3uol ~ s ~ a i q ueuinq ~uapuadepeuouuoq u! buepunqe asou punoj u g a d paqyssai%)fNa e sluasaidai s3g -au!! /la3 m u m aseaq lgz-fl~ -vafi eq u! buepunqe asou punoj supaaid ~u!~u!~-vNQ auesaidei p13g pue 6 1 3 '2139 -au1 lia3 i m u n lsearq OZlgeq 40 suogeieclwd u!e(oid Q ~uepunqe]sou punq qqad peqys013~a o siuasaidai L 139 seuy 1103 i e x m lseaiq LEHw-vafi pue egpgfi va^ a u u! buepunqe asoui punoj me a q qqad payu! lssat3~~~ luesaider 0138 pue '638 'ma -sey II= ~ m u w ~seaiq 8 9 ~ 8 ~ - V ~ W pue ozla u! punoj 4wpunqe asou s! letg u!a)old p e y y s s o m ~ a e queçarde~ 'pqprqs sw!~ [lm Jexm aseaiq juepuedapq-auouiroq Jnoj aq 40 awg u! spnq iaq6y le punoj u p p ~ d p e y u p ~ a r s ~ ~ e queseidai 938 -pqprqs seu~ lp iay io e q p wqlemdaid paqu!lssa3 01 peieduio3 nps w 01 pqu~ssai3 su!qwd 40 suqwedad d~a-s!:, ]uepuedepeuouuoy q A(luqucqmd punoj su!aamd lueseidai mg pue mg -(~661 "le l e @nueu) ensq JW@~UI 01 jypeds eq 01 punoj 4sno!naid W Q J ~ RN 40 u o g e ~ arg swuep! 2 uiegoy '(~661 "p ae l e n w ) 4 4 s snqad o sw F J - ~ W auepuadap -8uouw 40 suo!wed~d HN u! s ~ e ~ e l JW@Y ae puno4 su!e)ord awwda ~ H N pue IWWN '> ld~~w~alw P W!sod elll selw6!W y4 +4943eds9J '6 1 Pue 81 '8 S U W J W ~ 6 1W Pue 8 113 'W 'Ale43edS9J '3 W V u!W 40 W!sod W M W 31 Pm VI -sui@Pd le6 U W 0 W W erR l o W W Q W W q (m WWW mUlq4ou a4 10 uogsod su -m Aq peleqpu! s! ese~pAque n u o c p 40 suuoj pajejAweqie=, q 40 uq!sod au ' ~ 8 ~ s qjy pau.qs a m qe6 e u -SI& uopueqp OAJU uo peyosai maioqdcu lwe aiam d a m g RIA 1 y & ~ ma 01 p q u p o i 3 supaaid jo 6 rl446i3 -wu4 gm i m u w aseaiq luepuedepupeuouuoq uaj np u! dao-sg Aq OJ pqu!lssw sloagaid -9 an6y

. .y: . Y.'. . .. ', . - . . . . . :. ,. .......... :.' . . . . .y.:. ....... .... .........: A . :.. . . . .

Figure 5. Proteins crosslinked to DNA by cis-DDP in situ that are cornmon to brrast epitheid and breast cancer cell Lines. Eighty pg of proteins crosslinked to DNA with I m M cis-DDP fÏom a pseudo-normal h a s t epithelnl ceii h e were electrophoretically resolved on two dimension gels. The gels were stained with silver. The position of the cuhamylated b m of carbonic anhydriase is indicated by c l The position of the molecular mass standards (kDü) is shown t« the lrft of the two dimension pl patterns. A. B. C and D designaie prii rein clusiers c»mmonly sern in the pro fües of pro teins bound to DNA by çis-DDP in situ Cor thc cighr d l lincs.

Figure 7. Nuclear matrix profiles of hormonedependent breast cancer cell lines. Prepared by Dr. Shanti Samuel. Forly vg of protein was eiectrophoretically resolved on Mo dimension gels. The gels were stained with silver. The position of Me catbamylated lorms of caibonic anhydrase is indicated by ca. The position of the rnolecular mass standards &Da) is shown on the left side of each Mo dimension gel pattern. LA and LC show the poslion ol lamin A and C, respectiety, The whhe arrows show the location of fNe nuclear mat& proteins (NMBC 1-5) identifled by Samuel et al. (1997) as more abundant to the hormone- dependent breast cancer cell lines. NMBC-Z identifies the location of a NM protein prevlously lound to be specific to maügnant tissue (Khanuja et al., 1993). The localion of hnRNPK is indicaled as hK. BC3 BC4, and BC5 designate proteins mosl abundantly found in the cis-DDP preparations of proteins crosslinked Io DNA in hormonedependent human breast cancer cell lines. BC11, BC12, and BC14 represent proteins found most abundantly in cis-DDP DNA-crosslinked preparations of hormone-independent human breast cancer ceil ilnes.

Figure 9. Cytokeratins mssIinked to DNA by cis-DDP in T-47D5 and T5-PRF human breast cancer ceils in estrogen-repiete and estrogen-deplete conditions. Cek were crosslinked with 1 mM ch-DDP, and 10 pg of protein crossiinked to DNA was electrophoreticai'iy resolved on a SDS gel (iane 1-4). The gel was stained with Sema Blue. Lane 1, T-47D5 ceUs cuitured m the presence of estrogen- Lane 2, T-47D5 cek grown without estrogen for one passage. Lane 3, T-47D5 cek grown in the absence of estrogen for one passage and then cultured in the presence of estrogen for one passage. Lane 4. TS-PRF ceUs cultured Ui the absence of estrogen. Lane 5, 10 pg of protein isolated frorn NM-IF fraction of T5-PRF cclls. Thc position of the molecular weight standards (in ihousmds) is shown on the IcR sidc of the gcl. cK8, cK18. and cK19 idcntify thc cytokeratins 8, 18 and 19.

+ 4 - NMPS 4- NMPS *--- NMPS Y'

7

* '

Figure 10. Two-dimensional gel patterns of proteins crosslinked to DNA by 3 mM ch-DDP in T-47D5 and TS-PRF humiin breust cancer ceüs in estrogen-nplete and estrogen-depkte conditions. Twenty pg of protein crosslinkeâ to DNA was electrophoreticnlly resolved on two dimension gels. The gels were stained with süver. Panel A, T-47D5 ceUs cultured in the presence of esirogen. Panel B, T-47DS celis grown without estrogen for one passage. Panel C, T-47D5 ceUs were grown in the absence of estrogen for one passage and then cultured in the presence of estrogen for one passage. Panel D, TS-PRF ceüs cultured in the absence of estrogen. cK8, cK18 and cKL9 identw the cytokeratins 8, 18, and 19. Two NM proteins used as interna1 standards are shuwn us NMPS and NMP6.

At present the majority of evidence used to elucidate s~ucturaL and functional aspects

of DNA loop organization has been detived nom in vitro techniques. These techniques,

however, are prone to = k t formation since they involve NM extraction protocols that rnay

strip proteins away nom DNA, cause sliding of DNA over NM attachment points, or cause

the exchange of proteins b o d to DNA Such eveats c o d alter or eliminate original DNA-

protein interactions. therefore, causuig the misidentifkation of NM proteins and DNA

sequeires mvolwd m bop formation An alternative in situ approach whereby NM protein-

DNA interactions can be identintd m mm cellr or nucki before physicaily extracting the NM

or DNA fiom these entities would be ideal for stttdying DNA loop organization because it

would eIiminate the problerns of p0tentja.l artefact formation inherent to current in vitro

techniques. Such an approach can be accomplished through the use of cis-DDP, a heavy

metal cornplex shown to crosslhk DNA predominantly to NM proteins in intact ce& and

nuclei (Ferraro et ai., 1995; Costa, 1991). Indeed. the preference of this agent for NM

proteins associated with DNA was m e r validateci when a comparison of a NM protein

pro& with a cis-DDP DNA-crosslinked protein profile performed in this investigation

showed tbat the majority of proteins crossIinked to DNA in si& in T-47D cek were present

m the NM of the same ceii iine. Moreover. rhis agent appeared to have some speciacity for

particuiar NM proteins sHre a small subset of proteins (NMPZ, NMP4, NMPS). presumably

proteins not bound to DNA. were not detected in the cis-DDP DNA-crossIinked protein

preparation.

Despite the preference of cis-DDP for NM proteins. this agent stiU crosslinked

proteins not detected in NM preparations to DNA (eg. BC L. BC2. BC5, BC7. BC8. BC9,

BC 10). However, some of these proteins may have been NM proteins weakly associated with

the NM that were accidentaiiy nmoved by the high salt tteatment used in NM isolation. h

addition, the protocol used for MM isolation involved the sektive ternoval of IF proteins by

a disassembly and reassembly W o r precipitation step. Since a portion of IF pro teins are

tïghtry associated with tbe NM (Fey and Penman, 1988), it is possible that IF removal during

N M isolation may also lead to the removal of some NM proteins. In addition, some of the

DNA-crossJinked proteins not a p w g m NM prepaTatiOns may be IF proteins. As a result,

there is uncertainty in identifjing a DNA-crosslinked protein as a N M protein based on

molecular mass and isoeiectric point data obthed fiom two-dimensional patterns of NM

preparations. An additional approach that could co&m the identity of cis-DDP DNA-

c r o s s W proteins as NM proteins would be to immunostain the DNA-crossiinked protein

m question with a fluorescent antiidy in situ and to determine if this protein is exclusively

locaiïzed to the NM.

The preferential crosslinLing of cis-DDP to NM proteins would. therefore, suggest

that the N M contains IF proteins. It is possible that some DNA-crosslinked proteins not

found in NM preparations may not be N M proteins, and are. uistead, nuclear proteins sirnply

bound to DNA Therefore, caution must be exercised when using the cis-DDP crosslinking

technique to identify NM proteins associated to DNA.

Analysis of the two dimensional patterns of pro teins crossiinked to DNA in situ by

cis-DDP showed that most proteins crosslinked to DNA by cis-DDP were common to the ceil

Iules used in this study. In addition, the extent ofcrosslùiking with 1 m M cis-DDP appeared

to be optimal since the ievels of cytokeratins within each T47D5 treatrnent and within the

T5-PRF ceii iine did not change when the cis-DDP concentration was increased to 3 mM.

Roteins ïdentioed as c o m n to the ceil lines studied wete the transcription faftor

hnRNPK. and la- A and C. The protein hnRNPK is an interna1 mavix protein (Mattem

et ai., 19% and 1997) which firnctions as a tranxription factor that binds to specifïc single-

stranded regions of DNA (Michebtti et al. 19%). Since bnRNPK is fkom the NM, any DNA

sequence associated with this protein wouid be considemi a MAR Furthemore, nucIear

liunins A d C have beai identined m NM preparations (Nakayasu and Berezney, 199 1). and

lamin A has been shown to specifkaliy bind single-stranded MAR DNA sequences in vino

(Luderus et al., 1994). The crosslinking of nuclear lamins (espedy lamui C) has also been

observed in NoviLoff hepatoma ceUs incubated in 2 mM cis-DDP for 8 hours. The presence

of these 3 proteins in all cis-DDP preparations is consistent Mth the hdings of Tutano and

colleagues which show rbat MAR sequences are weil represented in DNA fragments

crosslinked to NM proteins by cis-DDP in si& (Ferno et aL. 1995 and 1996).

In addition to haRNPK and lamiru A and C. cK8. cK18 and cK19 also appeared in

the common subset of NM proteins crosslinked to DNA in situ by ck-DDP in aii ceil lines.

'Ihe identification of t h e three cytokeratins m human breast cancer ceUs agrees with hdings

that immortalized cells have increased levels of cK18 and turnor ce& have increased levels

of cK8, cK18 and cK19 comparai to normal epitheiiai cells (Trask et al.. 1990). In addition,

the crossünLUig of tbese 3 cytokeratins to DNA agrees with results of Hnilica and colieagues

who found the principal proteins cmssiinked to DNA by cis-DDP in si& in Novikoff

hepatorna nuclei to be cytokeratinr p39 (cK19). p49 (cK18) and p56 (cK8) (Olinski et al.,

1987; Ward et al, 1984).

Another IF pcotein Oenti6ed in cis-DDP DNA-croslinlted protein preparations in the

MDA-MB-23 1 ER- human bteast cancex ceU line was vimentin. The identification of this

protein as a DNA-binding protein agrees with the fïnduigs of a previous study that showed

Wnentin to sektiveiy bind DNA sequences in via0 (Wang et aL, 1996b). In this study, the

DNA sequences binding to vimentin had sequemes characteristic of MARS, sequences

recognized by transcription factors, or sequences whose structural properties are important

in recombination and gene expression. However, the crosslinking of pro teins to DNA in

human breast cancer œlls appears to be the tint case where the association of Wnentin with

DNA was shown in situ. The identification of this IF protein as a DNA-binding protein

provides m e r evidence of an association of cytoskeletal IFS with the nucleus.

Furthemore. the crossiinking of cytokeratios and vimentin to DNA by cis-DDP, a shoa

distance cross-linler ( a p p r o ~ l y 4 A) tbat predomhantiy crosslinks DNA to NM proteins,

strongly niggests that some or aü types of IF proteins are NM proteins. or that cytoskeietal

IFS are able to penetrate the nucleus in close pcoximity to DNA (Olinski et aL, 1987; Ward

et aL, 1984).

The presence of cytoskektal IFS in the N M is a concept that is continudy argued by

various researchen. Fey and Penman (1988) argue that proteins present in the NM as weii

as other ceiiular &actions are not N M proteins. Thus, their dennition of the NM excludes

cytoskeletal IF proteins such as vhentin, accin. and RNP complexes. B y creating thk

deonition of the NM, Fey and Penman fail to consider the possibility that particular proteins

such as cytoskeletal IF pro teins can be divided into two subsets: one subse t that exisu in the

cytosolic fkaction of a ce4 and one subset that exkits in the nucleus. Such a theory is feasible

since the NM extraction protocol used by Fey and Penman (1988) was unable to completely

remove cytoskektai IF proteins h m N M preparations. Thus, IFS are either tightly associated

to the NM or they are actuaiiy components of the N'M.

IF proteins help compose part of the CSK of eukaryotic ce& (Ingber, 1993). The

CSK is believed to be organized as a tensegrity network (Wang et aL. 1993) that provides a

celi with structural stability by maintainhg a force baiance between mechanical tension

generated nom withui the CSK and fiom the ECM (Ingbet, 1994). Application of a

mechanical stress to a cell ir beïieved to o k t the force balance, causing global

Rarrangements m the CSK (Ingbet, 1994), ad, tberefore, changes in IF protein organization.

Such an effect could alter gene expression by rearranging IF proteins bound to DNA, and,

therefore, altering chromatin loop organization. This theory m e r expiains the suggestion

of Getzenberg and Coffey (1990) that Merences in the distriaution of genes tbroughout the

nucleus result fiom changes in the nuclear structure responsible for controlling nuclear

organization. The involvement of the CSK in chromatin loop organization wouM be

consistent with the observation that alterations in a c h filaments by cytochaksin D inhibit

synthesis of most pro teins in mouse mammary epithelial c e h in a concentration-dependent

manncr (Seely and Aggekr, 1991). However, such an inbibitory effect rnay instead be a result

of the ability of cytochalasin to induce alterations in the stabiüty of the association of mRNA

with the cytoskeleton (Macoska et ai.. 1994).

Changes ui CSK (and IF) organization may &O occur through alterations in a cers

chernical composition Pmducts h m oncogenes such as src. lck. ras and raf. may stimulate

the RAS signal transduction pathway (Bortner et al.. 1993). Once stimulated. the Ras signai

transduction pathway may activate AP-1 and ETS transcription factors ihrough the

phosphorylatmg activity of Linases (Kani and Hunter, 1995). This activation rnay lead to the

petsistent expression ofpairicular cytokeratins such as cK8 and cK18 (Padcov et aL, 1994),

proteins which fonn part of the CSK 'Iae aberrant expression of cytokeratin genes in various

carcinoma ceii lines is believed to be related to the presence of abnonnal cytokeratin

filamentous structures often found associated with malignant transformation (Mol et al,

1982).

In addition to oncogene products, hormones such as estrogen also appear to alter

CSK ocganjzation. Tcearment of M W breast cancer cells with physiological concentrations

of estradio1 modined the cen surfke by covering it with numerous microvüli Moreover, this

hormone transfonned 40050% of these MCF-7 ceiis ïnto secretory cells with secretory

granules (Vic et al., 1982). However, both these effects were not observed in honnone-

mdependent ceii line (BT-20). In another study, Marchisio and coUeagues showed ihar

treatment of esmgen-starved MCF-7 breast cancer ceUs with estrogen causes the formation

of a network of keratin fibers dong with rearrangements in the actin m i c r o ~ e n t s and

keratin IFS. Such effects occurred independently of the abüity of estrogen to induce celi

proliferation (Sapino et ai., 1986).

Based on observations that rat vaginal epithelium treated with estrogen dispiayed an

mcrease in the synthesis of cytokeratins (Kmnenberg and Clark, 1985), it is quite possible that

the estrogen-induced CSK aiterations obserwd in the two preMously mentioned studies on

the MCF-7 cefi line may have proceeded through the abüity of this hormone to alter

cytokeratin levels. The effect of hormones on cytokeratin Ievels has also been observed in

a previous study wbat androgen. but not estrogen. repressed the leveis of cK8 and cK18 in

the rat ventrai prostate while anti-androgenic compounds bad opposite effects (Hsieh et al,

1992). Moreover. Coutts et ai. (1996) showed that tbe -1s of cK8, cK18 and cK19 present

within NM-associated IFS were dramatkaily Rduced in an ER+, hormone-dependent ceU h e

(T-47D5) grown in acutely estrogen-depleted conditions. Treatment of these estrogen-

starved ceh with estmgen restored the cytokeratins levels. However, ceils chronicdy

&pleted of estrogen over e x p d ai i 3 qtokeratins comparai to the T-47D5 parent grown

in the presence of estrogen. Thus, a i l 3 cytokeratins were estrogen regulated in honnone-

dependent T-47D5 human breast cancer cells, and this estrogen reguhiion was lost once these

ce& obtained a hormone-independent phenotype.

Taking the study by Coum et al (1996) one step forward, proteins nom T-47D5 ceils

grown under the same conditions were crosslinked to DNA in sizu by cis-DDP in order to

determine if changes in the abiüty of estrogen to effmt the levels of cK8. cK18 and cK19

bound to DNA occur during progression fkom hormone-dependence to homone-

independe~lce. The observed decrease in leveis of DNA-crossiinked cK8, cKl8 and cK19 in

T-47D5 ceUs cultured in estrogen deplete conditions for 1 passage combùied with the

ciramatic imease upon passaging these cek hto estrogen-replete conditions shows that the

interaction of cytokeratins and DNA is regulated by estrogen in hormone-dependent breas t

cancer celis. This observation implies that a r e o r g h t i o n of nuclear DNA has occurred in

these cek when estrogens are removed.

fn contrasi, the ER positive human breast cancer ceii he, TS-PRF. that has acquired

the ability to grow nonnaiiy in a medium greatly depleted of estmgens had higher amounts

of cytokeratins associated with the NM and DNA than did the parent ceii line cultured with

atrogen. Furthemore, Coutts et aL (1996) showed that treatment of the T5-PRF ceii line

with estrogen does not kad to funher up-regdation of these cytokeratins (Coutts et aL,

1996).

The ekvarPon of DNA-bnding cytokeratin ieveis in the TS-PRF ceii h e suggesu that

acquisition of a hormone-independent phenotype may, in part, result fiom an increase in the

production of DNA-binding cytokeratim andor nom changes in the CSK structure that

ianease cytokeratin-DNA interactions. These events. however. may not be involved in ER-,

hormone-independent breast cancer ceUs since cell lines with these phenotypes (eg. MDA-

MB-23 1 and HBL100) do no t dispiay an upreguiation of cytokeratin levels in the presence

or absence of estrogen (Coutts et ai., 1996). Instead, the ceii lines, MDA-MB-23 1 and

HBL100, dispiay vimentin as a promuient IF. This suggesu that the mechanisnu responsibie

for hormone-independence in ER+ ceii lines most Bely dafer fiom those in ER- ceii lines.

However, despite the di&rent mechanisms involved for the di&rent ceii lines, the acquisition

of a hormone-independent p heno type most Urely involves alterations in the ceii svuc ture

through changes in the cytoskeletal IF component of the CSK and the NM.

'Ibese nndings are consistent with those of Vic and colleagues (1982) who observed

no effects of physiologkal concentrations of estrogen on the CSK of hormone-independent

cek. More specficaiiy, the T5-PRF ceii line has apparently lost the capacity of estrogen to

regdate cytokeratin intedon with nuclear DNA: the orgmintion of nucleûr DNA mediated

by cpokeratins in later stage breast cancer is mallitaineci regardless of whether estrogen is

present or absent. Sàiiihr changes m levels of other proteins crossiïnked to DNA within the

38.5-54 kDa range and within a pI range of 4.6-5.4 were also observed in both T-47D5

tmtments and in the TS-PRF ceii üne, however the idetltity of these proteins remains to be

de termined.

How the TS-PRF ceIl Ihe manages to express eievated levels of these estrogen-

reguiated proteks in estrogen-depiete coaditions stül needs to be determineci. In one of the

current models of breast cancer develo prnent, the effect of estrogen on disease progression

is bekved to be mediated through tbe estrogen receptor (ER) (Landers and Spelsberg, 1992).

However, the elevated expression of cK8, cK18 and cK19 in estrogen-deplete conditions

suggests that, during disease progression, human breast cancer ceils develop the ability to

activate the ER through pathways other than those involving interactions between estrogen

and the ER In support of this theory k the observation that estrogen-independence in breast

cancer ce* is associateci with changes in tbe expression of estrogen-repuiated genes ( B m e r

et a l . 1993). Altematively, estrogen-independence rnay be a nsult of the ability of the ER

to be activateci in a ligand-independent manner (Amnica and Katzenelenbogen, 1993).

However, the possibiiiry exisu that, instead of having iigand-independent mechanisms of ER

activation, the breast cancer ceil may become supersensitive to estrogen, and, thecefore, only

require minute amounts of this hormone for growth.

The observation that estrogen regulates the expression of DNA-binding cpo keratins

only in hormone-dependent cell lines is important in understanding breast cancer development

since the progression o l breast epithelial c e l to mrilignancy is accompanied with increased

expression of cK8. cK18 and cK19 (Trask et ai.. 1990). The presence of cytokentins only

in epithelial c e h (Osbom and Weber, 1982) dong with their difEerentiaî expression over

breast cancer progression (Trask et ai., 1990) d e s cytokeratuis one of the classical

diagnostic d e r s for breast cancer.

Whüe cytokeratms may be markers of malignancy, the IF protein vimentin may be a

marker for bceast cancer progression. Vimentin was detected only in DNA-crosslinked

protein preparations of the ER-, hormone-independent breast cancer ceiî Iùie MDA-MB-23 1.

'Ibese resuits agree with the hdïngs ofother smdies (Thompson et ai., 1992; Sommers et aL,

1989). However, the results of motber s ~ d y by Regenass and cokagues (1987) showed a

beterogeneous expression of vimentin in ER+ and ER- breast cancer ce11 lines, as weli as in

normal prirnary human mammary cultures. Because of t h , it was specuiated that the

betmgenehy of Wiientin expression over the various breast cancer ceïï bes was a cause of

the tissue environment and culture conditions (Curschellas et ai., 1987). In ano ther study,

WIlentin was only expnssed in s o u metastatic and primary epitheliai tumors originating fiom

human serous cavity fiuids when these ceUs were shed into body cavities (Ramaekers et a l .

1983). Thus, metastasis was associated with Waentin expression in epitheiïai cek. As a

result, virnentin expression was cowidered as a process occurrhg in situ when ceik Iose

contact with neighbouring cells or with solid tissue during metastasis (Ramaekers et al..

1983).

Despite these two studies, the expression of Wnentin has been associated with various

Features of aggressive tumors such as the absence of hormone receptors (Thompson et ai.,

1992), increased tumor ceii proüferation (Thompson et aL, 1992: Seshadri et al.. 1996).

Ïncreasd tumorigenicity (Hendrix et ai.. 1997) . p53 expression (Seshiri et ai-, 1996), a d

hormone-Uidependence (Sommers et al, 1989; Curscheilas et ai.. 1987). Moreover, in vitro

studies show that over expression of vimenth in the ER+, hormone-dependent MCF-7 cell

line, a vhentin-negative ceïï Iàe representative of early breasr cancer development, causes

enhanced ceii motility. and irmeased invasiveness without changing the celi's metastatic

poteniia (Hendrix et aL, 1997). During this conversion to a more invasive phenotype, the

aansfected MCF-7 cells dispiayed shifts in the abdance ofdBierent i n t e g ~ subunits, as

weil as an imeased migiation to laminin (Hendrix et al., 1997), a potent ECM signaihg site

with many different biological activities (Kobota et al, 1992). Since inte- interact with

IFS (Quaranta and Jones, 1991). and since IFS are believed to inauence n u c h structure by

manmisha of ECM signals (Jngber, 1994). Wnentin may influence the ceWs interpretation

of ECM signals by altering the shape, spreading and migration of b.ast cancer ceUs in such

a way that provides vimentin-positive cells with a seiective advantage over ceiis not

expressing this F pmtein (Hendrix et ai., 1997). However. the inability of virnentin alone to

increase the metastatic potentiai of MCF-7 ceik to levels observed in the virnentin-positive

ceii line MDA-MB-231 suggests that the presence of virnentin and cytokeratin IF proteins

alone is rot sufticient to induce the metastatic phenotype (Hendrix et ai., 1997). Therefore,

O ther ceiiulat evenu together with vimentin and cpo keratin expression mus t be res ponsible

for the aggressive behaviour of the MDA-MB-23 1 ce1 line.

Thus, the initiation of breast cancer from a normal epithelïal stage foiiowed by iu

progression kom a hormone-dependent to a hormone-independent stage involves a series of

ceilufar events that may include changes in the expression of IF proteiw such as cK8. cK18.

and cK 19, and vimentin. This is fbrther supportai by the observation that, in addition to

expressing vimentin, the hormone-independent breast cancer ceii he, MDA-MB-23 1,

expresçed two DNA-bndMg proteins in cis-DDP crossiinked preparations (BC 13 and BC 14)

that were e&r absnt or bareiy detectabie in all oiber ce11 lines which had molecular masses

and isoelectric points s h i h to proteinr f o d in MD A-MB-23 1 IF preparations. At present,

these two proteins have not been identined. Whether changes in IF composition initiate

andlor are a dhxt consequence of events involved in hormonal progression of breast cancer

remains to be determineci.

Abng wïth changes m IF protein expression, the expression of N M proteins in breast

cancer ceh may also serve as a marker of dïsease progression. Studies on prostate (Partin

et al , 1993; Getzmberg et ai., 199 1). bkdda (Getzenberg et aL, 1996), colon (Keesee et aL,

1994). and breast (Khanuja et ai. 1993) cancer bave ail shown ditferences in NM composition

between ceils or tissue in a normal and a diseased state. For instance, a recent study identined

3 proteins (X, Y, and Z) as being specific to rnalignant breast tissue (Khanuja et aL, 1993).

Fuebermore, Dr. Shanti Samuel and colleagues (1997) identiûed 5 NM proteins (NMBC1,

NMBCZ, NMSC3, NMBC4, NMBCS) tbat were detectable in homonedependent breast

cancer ceh and tumon, and 1 protein (NMBC6) that was only detectable in hormone-

independent breast cancer ce& and tumors.

Since cis-DDP preferentially crosslinks DNA to N M proteins (Fermo et ai.. 1995;

Costa, 1991). this agent was used to identify DNA-binding protein markers of breast cancer

progression. In preparations of proteins crosslinked to DNA by cis-DDP in situ in this s tudy,

protein 2, as weii as NMBC 1 and MC2 were identioed in di ceil Lines except TS-PRF and

BT-20. Thk wouid suggest that, in cis-DDP DNA-binding protein prepmtions. these three

NM proteins have no diagnostic or prognostic signincance for breast cancer development.

However, the presence of NMBCl and NMBC2 in the pseudo-normal MCF- LOAL DNA-

crosslinLeci protein preparation could be a consequence of changes in gene expression

incurred by spontaneous immortalization. Thus, the analpis of two-dimensional profiies of

DNA-c~osslinLPn proteins in nonrial breast epnbelial ceil luies needs to be perfonned in oràer

to confinn the prognostic significance of these two proteins in cis-DDP DNAnossluiked

protein preparations. The four other potential pmgnostic markers for breast cancer that were

identified by Dr. Samuel et ai. (1997) were not observeci in preparations of proteins

crossiinked to DNA by cis-DDP. tberefore, i d k a h g thnt these proteins do not bind to DNA

at signiscant levels for detection.

A cornparison of cis-DDP DNA-binding protein preparations iden- several

proteins that may be of either diagnostic or prognostic importance in breast cancer

development Hormonedependent human breast cancer celi hes dispiayed tbree prominent

pro teins tint were either absent or barely detectabb in DNA-crosslinlred preparations of

hormone-independent b m t cancer and pseudo-nod bnast epitheIial ceU ünes (BC3, BC4,

BC5). In addition. hormone-independent human breast cancer ceU lines displayed several

DNA-binding proteins that were most abundant in ceüs with a hormone-independent breast

cancer cell phenotype. Of these homone-independent DNA-buiduig proteins. ody one (BC6:

4 1 kDa. p I 4.65) was identifieci most abundantly in three of the four hormone-independent

breast cancer ceii lines, while four were most prominent in only two of the three honnone-

independent ceii lines compared to the other celi Luies studied (BC7, BC8, BC9, BC 10).

Other cis-DDP-isoiated DNA-binding proteins were found most abundantly in the hormone-

independent breast cancer ceii lines BT-20 and MDA-MB-23 1 (BC 1 1, BC 12. BC 13, BC 14),

as weli as the pseudo-nod breast epithelial ceii hne, MCF-1OA1 (BC1. BC2). However,

the detedon of these pro teins in only one of the seven celi ünes s tudied mates dinicdty in

disceming whether these proteins were present due to ceU-typk specinc dinerences or

hormonal growth requirements.

As was just previously stated, a higher level of abundance of the DNAnosslinked

proteins, BCdBC14, was observed in tbree of the four hormone-independent breast cancer

œii lines cornpanxi to the three hormone-dependent and the pseudo-normal breast epiuiem

œU Iaies. However, most of these proteins were not detectable in the hormone-independent

ceil line, TS-PRF. Such an obsemtion suggests that either the DNA-crosslinked proteins,

BC6-BC14 are only expressed m breast cancer cek with a hormone-independent phenotype,

or, as was previously stated earlier in ihis papr, that dinerent mechanisrns are taking place

in ER+ and ER- ceLi lines to give both these ce1 types a hormone-independent phenotype.

Alteniativeiy, the absence of ihese proteins fiom the TS-PRF ceil line could be a result of the

œil-type specinc Merences between the TS-PR. ceii line and ibe three ER- ceii lines studied.

Despite the higher level expression of the DNA-crosslinked pro teins. BC6-BC 14. in

hormone-independent breast cancer celi iines when compared to hormone-dependent ceii

Iuies, the prognostic importance of these proteins stiii remains questionable. These proteins

may &O be found in other cancer cell lines, and may not be indicative of breast cancer. To

address this concem, hirther analysis needs to be perfomied on the two-dimensio na1 cis- DDP-

crossluiked pro files of other cancer ceii lines. This pro blem may &O present itself for the

three proteins (BC3. BC4, B U ) found most abundandy in the hormone-dependent celi lines

s tudied.

In additiDn to tbs pmblem. the use of ceii lines r a k r than tumor samples to identiQ

DNA-bincüng proteiris of potentiai prognostic importance k also a probiem. Cek grown in

culture are not exposed to the same &tors as ceh grown in v i w . Therefore, the dinerences

observed among rhe various cell lines s a d p d may not be repmentative of the differences that

would be observed ijz vivo. As a result, fiirther d y s i s on cir-DDP DNA-crosslinked

proteins nom turnor samples of various phenotypes ne& to be performed foiiowed by

clinical trials on b t cancer patients.

When comparing the two-dimensional pro- of DNA-crossiinked protein

preparatiom with NM preparations, several DNA-bniding proteins with a potential prognostic

value in cis-DDP preparations of breast cancer celi lines thought to represent different stages

in the hormonal progression mode1 showed no potential prognostic importance in the N M

profiles of the same celi IPies @C3, BC4, BC5, BC6, BC11). Therefore, while a NM protein

may be present throughout breast cancer devebpment, it may only bind to DNA and. by sorne

unknown mechanism. influence gene expression at specifïc disease stages. However, the

presence of some of these putative prognostic DNA-bindhg proteins was not always

restricted to a piuticular breast cancer cell or breast epithelial ceii phenotype, indicating that

the change in gene expression for sorne DNA-binding proteins throughout breast cancer

development rnay be iess exmme than for others.

One important celiulv component that may be implicated in these changes in DNA-

bindng pro tein composition thcoughout the diffierent stages of breas t cancer deveio pment is

the CSK The CSK is beüeved to help stabiiize nuclear structure (Ingber, 1994). Thus,

changes in the nuclear structure may affect CSK organhtion and induce changes in the

expression of some DNA-bjnding proteins over breast cancer progression. n i e maience of

changes in nuclear shape on the DNA-binding NM pmtein composition is most Wrely

mediated through tbe IF component of the CSK (Ingber, 1994). Since IFS bave been show

to influence NM structure (Maniotis et al, 1997). structurai changes in the CSK induced by

aiterations in nuclear structure may inaience the shape of the NM. and, therefore, the

chromath loop organization. Altematively, structural changes in the CSK induced by the

ECM may, in hm, cause changes in the NM structure which couid lead to changes in the

nuclear structure. The end result of either scenario would be a rearrangement in chromatin

bop organizatior Altecations in chromatin loop organization could then lead to changes in

the association of certain NM proteins with DN& as well as changes in gene expression. The

observation that the distribution of proteins with molecular weight and isoelectric point

coordinates smiilar to those of iF proteins was dinerent between hormone-dependent and

hormone-independent human breast cancer cell lines provides evidence to suggest the

mvolvement of the CSK and the N M in the dinerential expression of DNA-binding proteins

during breast cancer progression.

Although cis-DDP predominately crosslinked NM proteins to DNA, several of these

potentiaily prognostic DNA-binding proteins could not be detected in NM preparations of the

same ceii type from which they wece obtained. Despite the possibility that cis-DDP may

crossluik some proteins not found in the NM, the use of this agent a p p a s to be ided for

identifying DNA-binding NM prote&. Previous studies have prïmady used in vipo

techniques for isohting NM proteins (Samuel et ai.. 1997; Féy and Penman, 1988; Mgkovitch

et al.. 1984). Ho wever. these s t u k have been perfomed under onon-physiologicai

conditions and have involved NM isolation protocols that may strip NM proteins nom the

NM (Samuel et ai., 1997; Fey and Penman, 1988) or hadequatelyseparate chromatin fkom

the NS (Mickovitch et al.. 1984). T& use ofcis-DDP to isolate NM protein-DNA complexes

ahws the isoktion of N M proteins in sim d e r physiological conditions. Thus, the

problem associateci with U, vino techniques are avoided. As a result, the characterization

of proteins msslinlced to DNA by cis-DDP in situ, which are primarily NM proteins, cari be

used as a complementary approach to the current NM isolation protocol for identifying NM

proteins of putative diagnostic or prognostic sipiticance in cancer development.

In summary. the hypothesis of this study was that cis-DDP crosslinling can be used

as a compiernentary approach to the high sait NM extraction pro toc01 currently used for

identifjing putative N M proteins that may be mfomative biomarkers for cancer diagnosis. To

prove this hypothesis, DNA-bincüng proteins îiom several hormone-dependent and hormone-

independent ceU lines were isolateci using the crosslinkuig agent cis-DDP, and analyzed by

two-dimensional elecuophoresis. Foliowing this. the DNA-crosslinked proteins were

putatively identifieci as NM proteins by comparing the two-dimensional profiles of DNA-

binding proteins and NM proteins h m the same ceil line. The ability of estrogen to innuence

the association of DNA-binding NM proteins with DNA was then studied using the hormone-

dependent breast cancer ceii line, T-47DS. grown under nomai. estrogen-deplete, and

est rogen-re plete conditions. and the hormone-independent ceii line, TS-PRF. gro wn under

esuogen-depiete conditions DNA-binding NM pro teins were crossiinked to DNA using cis-

DDP nom the three T47DS ueatments and the TS-PRF ceii iine and then analyzed by two-

dimensional eiectrophoiwis. DiEerences in NM prorein levels over the t h e T-47D5

treatments and the TS-PRF celi line were assesseci.

'Llie results of t&is sady combhed with those of previous studies (Ferraro et ai.. 1995;

Costa, 1991) show that cieDDP preferenrianycrosdhks N M proteins to DNA in intact cek.

However, the identity of these cis-DDP DNA-crosslinked proteins as N M proteins needs to

be confirmed by immunoflourescence to determine tbe location of these proteins in the celL

In addition to displayhg a krge similarity to NM preparations. the rnajority of cis-

DDP DNA-crosslinked proteins extracteci !Yom various human breast cancer cell lines were

.amilai-. However, ciiffierences in the types of DNA-crossiinked proteins were stili observed

between ceii lines thought to npresent diitkrent stages m the breast cancer progression modeL

These differences dernonstrate that NM proteins, and. more specincdy. NM-DNA

interactions may be of importance in cancer prognosis.

Despite these ûndings. additionai cis-DDP D N A - c r o s s ~ g studies m u t be

performed on breast epithelia and breast tumor samples to determine if N M proteins and

NM-DNA interactions have a propos tic value in vivo. DNA-crosslinking s tudies using cis-

DDP must also be perfonned on other cancer ceii lines to detemiine the speciocity of these

NM proteins and NM-DNA interactions in breast cancer ce&. Once determined. proteins of

a potential prognostic d u e can then be brther studied in cluucal trials in order to accurately

assess their meaning in breast cancer development.

In addition to observing changes in patterns of DNA-binding proteins between

hormonedependent and hormone-independent breast cancer ceil lines, the bekviour of some

DNA-aosslUilred NM proteins such as cK8, cK18 and cK19 appeared to Vary with hormone

growth requïrernenrrs. T k association of cK8. cK18 and cK19 with DNA was dependent on

estrogen in a hormone-dependent hwaan breast cancer ceU line, but not in a homone-

independent hurnan breast cancer cen lhe. The ability of estrogen to Muence cytokeratin-

DNA interactions in a honnone-dependent and not in a hormone-independent breast cancer

ceil line suggests that, during tbe homonai prognssion of breast cancer to a more advanced.

hormone-independent stage, altemative mechanisnu develop that d o w the cell to undergo

changes in structure and DNA organization without hormonal regdation.

These nndings combined with the large abundance of vimentin in the human breast

cancer ceii iine MDA-MB-23 1 are consistent with the idea chat some of the various ceff ular

events leado>g to mdignancy in breast cancer may involve changes in ceii structure that may

lead to changes in protein expression through aiterations in DNA organization.

The ability of cis-DDP to crosslidc cytokeraiins to DNA in situ suggested that

cytokeratins are NM proteins associated with nuclear DNA Similar observations have also

been shown by H d c a and coiieagues (Wedrychowski et al.. 1986; Olinski et al , 1987).

This hding was important since the current pro tocol for NM isolation attempts to remove

IF proteins; thus, cytokeratins present in NM preparations will be those that escaped

reassembly and/or precipitation steps perfonned to remove IF proteins. These observations

argue that informative NM proteins rnay be lost during the preparation of NM proteins. As

a result, the charafterization of NM pmteins and proteins crosslinked to DNA by ci.-DDP

In situ. which are primarily NM proteins, are two complernentary approaches that can be used

to idenufy putative NM proteins tint may be intorrnative markers in cancer diagnosis.

Adachi, Y.. &S. E.. & ~aemmü. u.K ( 1989). Preferentid, cooperative buidhg of DNA topoisornerase il to scaffold-associateci regions. EMBO J. 8,3997-4006.

Alberts, B.M. ( 1990). Recipes for repiication Naricre, Lond 346,s 14-5 15.

Men, G.C.. Hall, GE. Ir., Childs, L.C.. Weissinger, Ak, Spiker, S., & Thompson. W.F. (1993). Scaffold aitachment regions increase reporter gene expression in stably transformeci piant ce&. Plant CeLt 5,603-6 13.

Allred, D.C. <YConne& P., Br Fuqua, S.AW. (1993). Biomarkers in early breast neopiasia J. CeLL. Biochem. 176, 125-13 1.

Armstrong, D.K, bacs, J.T., ûttaviano, Y.L., & Davidson, N.E. (1992). Programmed cell deaih in an estrogen-independent hurnan breast cancer ceil line, MDA-MB-468. Cuncer Res. 52,3418-3424.

Aronica, S M , & Katzeneknbogen, B.S. (1993). Stimulation of estrogen receptor mediated transcription and alteration in the phosphorylation state of the rat uterine estrogen teceptor by estmgen, cyck adenosine monophosphate, and insuün-like growth factor- 1. MOL EndocrirtoL. 7,743-752.

Attal, I., Cajero-Juarez, M., Petitclerc, D-, Theron, MC, Stinnakre, M.G., Beanotti, M., Kann, G., & Houdebine, L M (1996). The effect of ma& attacheci regions (MAR) and speciaaed chromatm structure (SC) on th: expression of gene constructs in cultured ceils and in transgenic mice. Mol. Biol. Rep. 22.37-46.

Banjar, Z.M., Hnilica, L.S., Briggs, R.C., Stein. J., & Stein, G. - (1984). cis- and truns- diamminedic hloroplatinum(n)-mediated cross-linking of c hromosomal no n-histo ne pro teins to DNA in HeLa ce&. Biochemistry 23, 192 1- 1926.

Bereu~y, R & Coffey, D.S. (1974). Identification of a nuclear matrix protein. Biochem Biophys. Res. Commun. 60, 14 10- 1417.

Berezney, R (1991). The nuclear matrix: A heuristic mode1 for investigating genomic organization and function in the cell nucleus. Cell. Biochem 47, 109-123.

Berezney. R, & Buchholtz, LA (198 1). Dynamif association of replicating DN A h g ments with the nuclear mauix of regenerating iiver. &p. Cefi Res. 132, 1- 13.

Bode, I-, Kohwi, Y-, Dickinson, L., Ioh. T., Klehr, D., Mieke, C.. & Kohwi-Shigematsu. T. (1992). Biological signincance of unwinding capabiiity of nuclear mauix-associating DNAs. Science 255, 195- 197.

Bode, J., Schlake, T., Rios-Ramirez, M., Mielke, C., Stengert, M., Kay, V., & Kkhr-Winh, D. (1995). S c a o Wmatrix-attached regions (SIMARs) : structurai properties creating tmmxiptionally active loci In Jeon, KW.. Befezney, R. (ed): "Structurai and Functional Organizatioa of the Nuclear Matrix - International Review of Cytoiogy", Vol L62& Chap.8. Orlando: Academic Pnss, pp. 389-453.

Bode, JOT Stengert-Iber, M., Kay, V., Schlake, T., & Dietl-Pfeilstetter, A. (1996). S c s o WmaemC-anaicbed regions: TopoIogral switcbes with multiple regdatory functions. Crit. Rev. Eruk Gene Enp. 6, 1 15- 138.

Boff' L.C., V i d a G., Mann, RS., & Allfirey, V.G. (1978). Suppression of histone deacetyiation in vivo and in vitro by sodium butyrate. J. BioL Chem 253,33643366-

Bortner, D.M.. Langer, S.J., & Ostrowski, M.C. (1993). Non-nuclear oncogenes and the regdation of gene expression h uansfonned ce&. Cri& Rm. Oncog. 4, 137- 160.

Boulilcas, T. (1995). How enhancers work: Iuxtapositioning of DNA control elements by synergistic interaction of MARs (review-hypothesis). Int. J. Oncol. 6,13 13- 13 18.

Brooks, A.R-, Nagy, B.P., Taylor, S., Simonet, W.S., Taylor, LM., & Levy-Wilson, B. (1994). Sequences containing the second-intron enhancer are essential for transcription of the human apolipopmteh B gene m the Livers of transgenic mice. Mol. Cell. Biol. 14, 2243-2256.

Bubley, G.J., XU, J., Kupiec, N., Sanders, D., Foss, F., O'Brien, M., Emi, Y., Teicher, B A , Br Patiemo, S.R (19%). Effect of DNA conformation on cisplatin adduct formation. Biochem Phannacol. 51,7 17-72 1.

Capco, D.G., Wan, KM. & Penrnan, S. (1982). The nuclear mate: Tbree-dimensional architecture and protein composition. Ce11 29, 847-858. .

Chen, D., Ohta, N., Ukai, M., Masuda, M., & Yotsuyanagi, T. (1994). Binding and aggregation of human y-globulin by ci id ' uvaminedichloroplatinwn (II) through disulphide bond. Biol. Pham Bull, 17, 156 1 - 1566.

Chen, KY., Sun, LM., HendzeL, M.J., Rattner, J.B.. & Davie, J.R. (1996). Changes in the nuclear mauix of chicken erythrocytes that accompany maturation. Biochem J. 320, 257-265.

Chu, G. ( 1994). Cellukr responses to cisplatin: The role of DNA-bindhg pro teins and DNA repair. I. Biof. Chem 269,787-790.

Ciejek, E.M., Tsai, MJ., & O'Malley, B.W. (1983). Actively transcribed genes are associateci with the nuclear rnatrïx. Nature, Lund. 306,607-60.

Qarke, R, Dickson, RB., & Bmnner, N. (1990). The process of malignant progression in human breast cancer. Am. Oncoi. l,4O 1-497-

Cockerill, P.N., & Garrard, W.T. (1986a). Chromosomal loop anchorage of the kappa irnmunoglobulùi gene occurs next to the enhancer in a region containhg topoisocnerase II sites. Cell 44,273-282.

Cockerin, PN., & Garrard, W.T. (1986b). Chrornosomal loop anchorage sites appear to be evolutionarily consemed. FEBS Len. 204,5-7.

Cook. P.A. & BrazeU, LA (1976). CharaEterization of nuclear structures containhg superhekal DNA. J. Ceil Sci. 22.303-324.

Cook, P.R (1989). The nuceloskeleton aad the topology of transcription. Eur. J. Biochem 185,487-50 1.

Costa, M.. (1991). DNA-protein complexes induced by chromate and other carchogens. Environ Hldr Perspen 92,4542.

Coutts, A-S., Davie, I.R., Dotziaw, H., & Murphy, L.C. (1996). Estrogen regdation of nuclear mauix-intemediate filament proteins in human breast cancer ceh. J. Cell. Biochem 63, 174- 184.

Cunningham, J.M., Purucker, M.E., Jane, S.M., Safer, B., Vanin, E.F., Ney, P. A, Lowrey, CH., & Nienhuis, kW. (1994). The regdatory e h e n t 3' to the " y-globin gene binds to the nuclear matrix and interacts with special A-T-rich bhding protein 1 (SATB 1). an SAR/MAR-associating region DNA binding protein. Bbod 84, 1298- 1308.

Curscheh, E., Matter, A., & Regenass, CT. (1987). Immunolocalization of cytoskeletal elements in human mammary epitheliai celis. Ew: /. Cancer C h Oncol. 23, 15 17- 1527.

Dedon, P.C., & Borch, R.F. (1987). Characterization of the reactions of piatinum antitumor agents with bioiogic and nonbiologic sulhir-containing rycleophiies. Biochem Phamcol. 36, 1955- 1964.

Deppert, W. (1996). Bùding of MAR-DNA elemenu by mutant p53: possible implications for its oncogenic hinctions. I. Cell. Biochem 62, 172- 180.

Dickinson. LA, & Kohwi-Shigematsu, T. ( 1995). Nucleoh is a matrix-attachment region DNA-binding protein thît speciucaiiy recognizes a region with high base-unpaûing po tential Mol. CelL Biol. 15,456-465.

Dickinson, L.A. Joh, T., Kohwi, Y., & Kohwi-Shigematsu, T. (1992). A tissue-specSc MASBAR DNA-binding protein with unusuai binding site recognition. Cell 70.63 1- 645.

Dijkwel, P A , & Hamiin, I.L. (1996). Sequence and context eEecu on origin function in mammalian ceils. J. Cell. Blochem 62,210-222.

Dijkwei, P.A. & Hamlin, J.L. (1988). Maaix attachment regions are positioned near replication initiation sites, genes and an interampücon junction in the amplified dihydrofohte ductase dotnain of Chinese hamster ovary cek Mol. Cell. Bio 2. 8,5398- 5409.

Dorer, Dk, & Henikoff, S. (1994). Expansions of traiisgene repeats cause heterochromatin formation and gene siîencing in Drosophiia. Cell 77,993- 1002.

Doucet, J.P., & Trifaro, LM. (1988). A discontinuous and highiy porous sodiwn dodecyl sdfàte-potyacrSamide slab gel system of high resolution. Anui. Biochem 168,265-27 1.

Dreyfuss, G-, Matunis, M.J., Pinol-Rom, S., & Burd, CG. (1993). hnRNP proteins and the biogenesis of mRNA Annu. Rev. Biochem 62,289-32 1.

Dworetzky, S.I., Fey, EG., PeNllan, S., Lian, J.B. Stein, J.L. & Stein, O.S. (1990). Progressive changes m the protein composition of the nuclear rnavix during rat osteobiast differentiation. Proc Natl. Acad Sci. USA 87,46054609.

Eggert, K. &Jack, R (1991). An ectopic copy of the Drosopbila ftt associateci SAR neither reo rganizes local chromatin structure nor hinders elution of a chromatin h g m e n t fkom isolated nuclei EMBO J. 10, 1237-1243.

Engei, L. W., & Young, N.A (1978). Human breast carcinoma cek in continuous culture: A review. Cancer Res. 38,4327-4339.

Engel, L.W., Young, N-A., Traika, T.S., Lipprnan, M.E., O'Brien, S. J., & Joyce. M. J. (1 978). Establishment and characterization of three new continuous ceU iines derived ftom human breast carcinomas. Cancer Res. 38,3352-3364.

Emster, V.L., Barclay, J., Kerlikowske, K. Grady, D., & Henderson, C. ( 1996). Incidence of and treatment for ductal carcinoma in situ of the breast. JAMA 275,9 13-9 18.

Féisenfeld, G., & McGhee, J.D. (1986). Structure of the 30 nm chromatin fiber. Ce11 44. 375-377-

Ferraro, A., Grandi, P., Eufemi, M., Altieti, F., Cervoni, L., & Turano, C. (1991). The presence of N-gfycosylated proteins in ce& nuclei Biochem %ophys. Rex Commun. 178, 1365- 1370.

Ferraro, A., Grandi, P., Merni, M., Al- F., & Turano, C. (1992). Crosslinlong of n u c k proteins to DNA by cis-diamminediçhloro piatinum in intact cells: uivolvement of nuclear muUr proteinS. FEBS Lefi 307,383-385.

Ferraro, A-, Eufemi, M., Cervoni, L., &ri, F., & Turano. (1995). DNA-nuclear matrix interactions analyzed by cross-linking ceactions in intact nuclei nom avian iiver- Acta Biochi'm Pol. 42, 145-152.

Rrraro, A, Ceoron& L., Eufémi, M.. Altier& F., & Tutano, C. (1996). Cornparison of DNA- protein interactions in intact nuclei from avian liver and erythrocytes: A cross-linking study. /. CeU Biochem 62,495-505.

k y , E.G., Wan, KM., & Penman, S. (19û4). Epithelial qtoske1etai fhmework and nuclear- ma&-intermediate filament scaffold: Three-dimensional organization and protein composition. J. Ce11 Bi01 9û. 197% 1984.

ky , E.G., Krochmalnic, G., & Penmyi, S. (1986). The nonchromatin substructures of the nucleus: The nïnucleoproteia @NP)-containing and RNP-depleted matrices analyzed by sequential fkacbonation and resiniess section electron microscopy. Ce11 Biol. 102, 16% 1665.

Fey, E.G. & Penman. S. (1988). Nuciear rnatrix proteins reflect cell type of ongin in cultured human cells. Proc. Natl. A c d Sci. USA 85, 12 1- 125.

-ski, I., Kahn, KW., & Bonnet, W.M. (1983). Differential crosslinking of histones and non-histones in nuclei by cis-Pt@). F E N Le#. 152, 105- 108.

FO ka, M., & Paoletti, J. (1986). Interaction of cis-diamminedichlorophtinumCI1) to chromatin: Specifkity of the dnig distribution. Biochem Phamcol . 35,3283-329 1.

Forrester, W.C., van Genderen, C., Jenuwein, T., & Grosschedl, R. (1994). Dependence of enhancer-mediated transcription of the immunoglobulin p gene on nuclear m a h attachment regions. Science 265, 122 1 - 1225.

Gasser, S.M., & Laernmli, U. K. (1986a). The organisation of chromatin loops: chcuactetization of a sc&old attachent site. EMBO J. 5.5 1 1-5 18.

Gasser. SOM-, & Laemmh. U.K. (1986b). Cohabitation of scaffold-binding regions with upstreamlenhancer elernents of three developmentally reguhted genes of D. meianogaster. CeIf 46. 52 1-53O.

Gasser, S.M., & Laemmli, U.K. (1987). A glimpse at chromosomal order. Trends Genet. 3, 16-22,

Gerdes, M.G., Carter, KT Moen, P.T., & Lawrence, J.B. (1994). Dynamic changes in the higher-level chromatn orgaimaeon of specinc seque- revealed by in situ hybridization in nuclear balos. Celi Biol. 126,289-304.

Geaenberg, RH., & Co- D.S. (1990). Tissue speciocity of the hormonal response in sex accessory tissues is associated with nuciear matrix protein patterns. Mol. Endocnnol. 4,1336-1342.

Getzenberg, RH., Pienta, KJ., Hua~lg. E.Y.W., & Coffey, D.S. (1991). Identincation of nuclear mat& in the cancer and normal rat prostate- Cancer Res. 51,65 14-6520.

Getzenberg. RH., Konety, %.RaT Oeler, TA, Quigley, MM.. Hakam, A, Becich. M.J., & Bahason, RR (1996). B W r cancer-associated n u c k matrbt proteins. Cmcer Res. 56,1690-1694.

GUO, B., Odgren, PORT van Wijnen, AL, Last, T.J., Nickerson. J., Penman, S., Lian, I.B.. Stein, J.L.. & Stein, G.S. (1995). The nuciear protein NMP-1 is the transcription factor YY1. Proc. Nari. Acad Sci. USA . 92, 10526-10530.

Hayes, I., & Scoveil, W.M. (199 1). cis-DiamminedPchloroplaWium@) modioed chromath and nucleosomai core particle. Biochim Biophys. Acta. 1089,377-385.

He, D., Nickerson, LA. & Penman, S. (1990). Cote [ilaments of the nuclear rnauix. L Cell Biol. 110,569-580.

Hempel, IC, & Stratling, W.H. (1996). The chicken lysozyme gene 5' MAR and the Drosophila histone SAR are electroelutabie 6-om encapsulated and digested nuclei. J. Cell Sci. 109, L459- 1469.

Kenderson, B.E., Ross, R, Br Bernstein, L. (1988). Estrogens as a cause of human cancer: the Richard and Hinda Rosenthal Foundation Awad Lecture. Cancer Res. 48,246-253.

Hendrix, M.J.C., Sefior, E.A., Chu. Y.W., Sefior, R.E.B., Nagle, R.B., McDaniel KM., Leong, S.P.L. & Yohem, KH. (1992). Coexpfession of virnentin and keratins by human melanoma tmor cells: correlation with invasive and metastasis potentiaL J. Natl. Cmcer inst. 84, 165-L74.

Hendrix. M.J.C., Seftor, E.A., Seftor, R.E.B., & Trevor, KT. (19971- Experimentd co- expression of vimentin and Leratin intemediate Waments in human breast cancer ceils results in phenotypic interconversion and increased invasive behaviour. Am J. Pathol. 150,483495.

NU, T.L. (198 1). Microtilament or microtubule assembly or disassernbly against a force. Proc. NatL Acad Sei. USA 78,5613-56 17.

Hoffmann, LS., Hubscher, U., Michot, B., & Vilhni, G. (1991). Interaction of cis- diarnminedichioropia~um@) with single-stranded DNA in the presence or absence of ficherichia CO& singie-stmded buding protein. Biochent Pharmacol. 42, 1393- 1398.

Holniann, K., Korosec, T., Gerner, C., Grimm, R, & Sauermann, G. (1997). Identification of human common nuclear-mat& proteins as heterogeneous nuclear ribonucleoproteins H and ET by sequencing and m a s spectrophotometry. Eur. J. Biochem 244.479-486.

Homberger, H.P. (1989). Bent DNA is a structural feature of scaffold-attached regions in Drosophiia melanogaster interphase nuclei Chromosoma 98,99- 104.

Hozak, P., Sassevik, MLA, Raymond, Y., & Cook, P.R (1995). Lamin proteins fom an intemal nucleoskeleton as weiî as a peripherai lamina in human ce&. J. Ce11 Sci. 108, 635-644.

Hsieh, T-, & Wang, K. (1975). Tbermodynamic properties of superhelical DNAs. Biochemistry 14,527-535.

Wh, J.T., Zhau. RE., Wang, X.H., Liew, C.C., & Chung, L.WK (1992)- Regdation of basal and luminal ceil-spedk cytokeratin expression in rat accessory sex organs. J. Biol. Chem 267,2303-23 10.

ïngber, D.E. (1993). Cellular tensegrity: dethhg new cules of bio&ical design that govern the cytoskeleton. J. Ceil Sci- 104.6 13-627.

Ingber, D.E., Dike, L., Hairîen, L., Karp, S., -y, 8, Ma~otis, A, McNamee, H., Mooney, D., Plopper, G., Sims, J., & Wang, N. (1994). Cellular tensegnty: exploring how mechanical changes in the cytoskeleton cegulate cell growth, migration, and tissue pattern during morphogenesis. Int. Rev. Cytol. 150, 173-224.

Ivanchenko, M.. & Avramova, 2. (1992). Interaction of MAR-sequences with nuclear mauk pm teins. .L CelL Biochem 50, 190-200.

izaurralde, E., Mirkovitch, J., & Laemmli, U.K. (1988). Interaction of DNA with nuclear scaffolds in vitro. J. Mol. BioL 200, 1 1 1 - 125.

bumide, E., Kas, E., & Laemmlï, U.K. (1989). Highly preferential nucleation of histone H 1 assernbly on scaffold-associateci regions. Mol. Bi01 210.573-585.

Jkckson, D A , & Cook, P.R. (1985). Transcription occurs at a nucieoskeleton. EMBO 4,9 19-925-

Jackson. D-A, Yuan. J., & Cook, P.R (1988). A gentle method for pceparing cyto- and nucelo-skeietons and associateci chromatin, J. Cell Sci. 90,365-378.

Jackson, DA, Dickinson, P. & Cook, P.R. (1990). The sïze of chromath ioops in HeLa ~ells. E h Jt 9,567-57 1.

Johnson, N.P.. Hoeschele, J.D., & Rahn, RO. (1980). Kinetic analysis of the in vitro binding of radioactive cis- and trm-dichlorodiammineplatinum~ to DNA Chem Biol. Interact. 30, 15 1- 169.

Kabs, M.. & Fournier, R.E.K. (1995). Position-independent transgene expression mediated by boundary elexnents h m the apolipoprotein B chromatin do& Mol. Cell. Biol. 15, 198-207,

Karin, M.. & Hunter, T. (1995). Transcriptional control by pro tein phosphorylation: signal transmission î?om the ceU surface to the nucleus. Curr. Biol. 5,747-757.

Kas, E., Izaurralde, E., & Laemmli, U.K. (1989). Specific inhibition of DNA biding to nuclear d o l d s and histone H 1 by distamycin The role of oligo(dA)-oligo(dT) tracts. J. Mol. Biol. 210,587-599.

Kasid, A., Lippman, M.E., Papageorge. AE., Lowy, D.R., & Gelmann. E.P. (1985). Ttansfêction of v-rasH DNA into MCF-7 human h a s t cancer celis bypasses dependence on estrogen for tumorigenicity. Science 228,725-728.

Kay. V., & Bode, J. (1994). Binding speciticity of a nuclear scaEold: supercoiied, single- stranded, and sceold-attached-region DNA Biochemistry 33,367-374.

Keesee, S.K. Meneghini, MD., Szaro, R.P., & Wu, Y-J. (1994). Nuclear matrix proteins in humm colon cancer. Proc N d Acod Sci USA 91, 19 1% 19 16.

Kenney. N.J., Saelci, T., Gottardis, M., Kun, N., Garcia-Morales, P., Matrin. M.B., Nomanno, N., Ciardieiio, F,, Day, A., Cutier, M.L., & Salomon, D.S. (1993). Expression of transfonning growth factor alpha antisense mRNA inhibits the uuogen- induced proliferation of TGF-alpha and estmgen-induced prolifera tion of estrogen responsive human breast cancer ceh. J. Cell. Physiol. 156,497-5 14.

Khan, M.U.A. & Sdler, P.J. (1978). Distribution of a piatinum anti-tumor dru% in HeLa cells by andytical electron microscopy. Chem Biol. Interact- 21,227-232.

Khanuga, P.S.. Lehr, LE.. Soule, H.D., Gelmi, SIC, Noto, AC., Choudhury, S. Chen, R., & Pinta, K.L (1993). N u c h maaix proteins in normai and b m t cancer ceiis. Cancer Aes. 53,33943398-

Kobota, S., Tashiro, K., L Yamada, Y. (1992). Signahg site of lamulin with mitogenic activity. I' BioL Chem 257,4285-4288.

Komitowski, D., & Jansoa, C. (1990). Quantitative features of chromatin structure in the prognosis of breast cancer. Cmcer 65,2725-2730.

Kohwi-Shigematsu, T., & Kohwi, Y. (1990). Torsiod stress stabilizes extended base unpairhg in suppressor sites BanLing immunoglobulin heavy chah enhancer. B i ~ c h e ~ s n y 29,955 l-g%û.

Komitowski, D.D., Hart, M M , & Janson, C.P. (1993). Cbromatin organization and breast cancer: Two-dimensiod land thrieedimensional image analpis. Concer 72, 1239- 1246.

Kronenberg. M.S., & CIark, J.H. (1985). Changes in keratin expression during the estrogen- mediated differentiation of rat vaginal epithelim EndocrU>ology 117, 14804489.

Kuiier, L.H. (1995). nie etiology of breast cancer - nom epidemiology to prevention. Public Heulth Rev. 23, 157-2 13..

Laemmli, KK. (1970). Cleavage of smcturai proteins during the assembly of the head of bacteno phage T4. Nature, Lund. 227,680-685.

Landers, I.P., & Splsberg. TC. (1992). New concepts in steroid hormone action: Transcription factors, proto-oncogenes, and the cascade mode1 for steroid regdation of gene expression. Crit. Rat. Euk Gene Exp. 2, 19-63.

LeRoy, A.F., & Thompson, W.C. (1989). Binding kinetics of tetrachloro-1.2- diammùiocyclohexaneplatinum(n3 and cisdiamminedichloroplatinum~ at 37 O C with human plasma proteins and with bovine serum albumin. Does aquation precede protein binding? J. Nutl. Cmcer lnst. 81.427-436.

Lv).-Wion, B., & Fonier, C (1989). The tirnits of DNase 1-sensitive domain of the human apol@pmteh B gene coincide with the locations of chromosomal anchorage loops and define the 5' and 3' boundaries of the gene. J. Biol. Chem 264,2 1 196-2 l2O4.

Li, W., Nagara. S., Dekuve, G.P., HendzeL Ml, & Davie, J.R. ( 1993). Effects of histone acetytation, ubquitination and variants on nucleosome stability. Biochem J. 296,737- 744.

Lilley, D.M.J. (1980). nie invened repeat as a nrognizabk structural fature in supercoiled DNA molecriles. Proc. Natl. A c d Sci. USA 77,64684472-

Lippard, S.I., & Hoeschele, J.D. (1979). Binding of cise and truns-- dichbr-latuiutnm to the nucieosome core. Pmc. Natl. A c d Sci USA 76, 609 1-6095.

Lippard, SJ. (1982). New chemistry of an old molecule: cis-[Pt(NH,),ClJ. Science 218. 1075- 1082.

Lipprnan, M., Bo@ G., Monaco, M., Pidms, L., & Engels. L. (1976). Mode1 systems for the study of estrogen action in twue cuihm. I. Steroid B i o c k 7, 1045-105 1.

Lipprnan, M.E., & Dickson, RB. (1989). Mechanisnu of growth control in normal and malignant breast epithellum. Rec. Prog. H o m Res. 45. 383-340.

Luderus, M.E.E., de G d , A, Mattia, E., den Blaauwen, I.L., Grande, M A , de Jong, L., & van Driei, R (1992). Bmdng of maDix attachment regions to lamin B,. CeU 70,949- 959.

Luderus, M.E.E., den Blaauwen, IL, de Srnit, O.J.B., Compton, D.A. & van Driei, R. (1994). Binding of matrix-attachaient regions to kmin poiymers involves single-stranded regions and the minor groove. Mol. Cd. Biol. 14,6297-6305.

Macoska, J., Hoover, C.N., & Pienta, KI. (1994). Alterations in ceiluiar gene expression without changes in nuclear ma& protein content I. Ce12 Biochem 56.502-509.

Mandeviüe, R, Berrada, F., Ajdukovic, L. Sidrac-Ghali, S., & Zeiechowska, M. (1987). An in vitro mode1 for cancer invasion: 1. Grading and quantitation of aggressiveness. Anticancer Res. 7, 1133-1 142.

Maniotis, AI., Bojanowskï, K., & Inbger, D.E. (1997). Mechanicd continuity and reversible chromosome disassembly within intact genornes removed fiom iiving ceh. J. Cell. Biochem 65, 114430.

Mattem, K.A. de Jong, L., & van Driei, R. (1996). Composition and structure of the interna1 nuclear rnatrix. In Bird, RC., Stein, G., Lian, J., & Stein, J. (eds):"Nuclear Structure Gene Expression." Academic Press. San Diego, USA.

Mattem, K A, van Goethem, REM., de long, L., & van Driel, R. (L997). Major internai nuclear matrix proteins are common to ditferent human celi types. J. Ceil. Biochem 65. 42-52.

McCready, S.J., Godwin, I., Mason, D.W., BcazeU, I.A. & Cook, P.R. (1980). DNA is repiicated at the nuclear cage. Ceil Sci. 46,365-386.

McGhee, J.D., Nickoî., LM., Feknfeld, G, Br b u , D.C. (1983). Higher ocder structure of chromatin: orientation of nucleosomes withul the 30 ~n chromatin solenoid is independent of spies and spacer length. Ceil 33,83 1-841.

McKnight, RAT Shamay. AT Sachan, L., WaL. W., & Hennighausen, L. (1992). Mauix- attachent regions can impart position-independent reguiation of a tissue-specinc gene in transgenic mice. Proc. Nad. Acad Sci. USA 89.6943-6947.

Michelotti, E.F., Midielotti, G.A., Aronsohn, AI.. & Levens, D. (1996). Heterogeneous nuclear ribonucleoprotein K is a transcription factor. Md Celf. B i d . 16,2350-2360.

MieIke, C., Kohwi, Y., Kohwi-Shigemtsu, T., & Bode, J. (1990). Hierarchicai binding of DNA fhgments derived b m scafEoId-attacheci regions: correlation of pro perties in vitro and function in vivo. B i o c h e m i s ~ 29.7475-7485.

Millard, MM., Macquet, J.P., & Theophanides, T. (1975). X-ray pho toelectron specmscopy of DNA-Pt cornpiexes. Evideoce of 06(Gua). N,(Gua) chelation of DNA with cis-dichlorodiammineplatinum (II ) . Biuchim Biophys. Acta 4û2, 166- 170.

Milier, CA, &Costa, M. (1989). Analysis of pmteias cross-iinked to DNA after treatment of cek with fomaidehyde. chromate, and cis-diammlliedichloropIatinum(rn. Mol- Toxicol. 2, 1 1-26.

Miller, C.A., & Costa, M. (1990). Immunodeteetion of DNA-protein crosslinks by slo t blo tting. Mutation Res. 234,97- 106.

Mirkovitch, I., Mirauit, M.E., & Laemmli, U.K (1984). Organization of the higher-order chromatin loop: Specific DNA attachment sites on nuclear scûnold. Ce12 39,223-232.

Miyanaga. N., Akaza, Ho, Ishikawa, S., Ohtani, M., Noguchi, R., Kawai, K. Koiso. K., Kobayashi, M., Koyama A, & Takabashi, T. (1997). Clinical evaluation of nuciear ma& protein 22 (NMP22) in urine as a novel marker for uro thelial cancer. Eur. Urol. 31, 163-168,

MOM. Re, Franke, W.W.. Geiger, B., Krepier, R., P.L. & Schiller. ( 1982). ï h e catalog of human cyto keratins: Patterns of expression in nomal epithelh, tumors and cultured celis. Ce11 31, 1 1-24.

Morric. C.R.. & Gale, G.R. (1973). Interactions of an antitumor platinum compound with deol<yribonucieic acid, bistones, L-amino acids. poly-t-amino acids, nucleosides and nucleo tides. Chem BioL interact. 7,305-3 15.

Muller, B.F., Paulsen, D., & Deppen, W. (1996). Specinc bïnding of MARfSAR DNA- elements by mutant p53. Oncogene 12. 1941-1952.

Muqhy, C.S., Pink, JJ., & Jordan. V.C. (1990). Characteristics of a receptor-negative hormone non-responsive cbne dermd fiom T47D human breast cancer ce1 line kept under oestrogen fke condüions. Cmcer Res. 50,7285-7292.

Mmhy, L. C., Dotzlaw, U, Hamaton, J., & Schwan, J. (1993). Investigation of the origin of variant, uuncated estrogen receptor-ijke mRNAs identifieci in some human breast cancer biopsy sampIes. Breast Cmcer Res. Treat. 26, 149- 16 1.

Nakayasu, HOT & Beremey, R. (199 1). Nuclear Matrins: Identification of the major nuclear ma& pro teins. Proc. Nad. Acod Sei. USA 88, 103 12- 103 16.

Nkkerson, J A , Kr~~lonahiic, G.. Dacheng, EL, & Pe- S. (1990). Immunolocalization in ihree dimensions: immunogold staining of cytoskeletal and nuclear matrix proteins in cesinless electron microscopy sections. Proc. Nd. A C S c i USA û7,2259-2263.

Nickenon, J . k T Krocknialnic, G., Wan, KM., Tumer, CD., & Penman, S. (1992). A nonna@ masked nuclear maaR antigen that appears at mitosis on cytoskeieton fïiarnents adjoining chromosomes, centrioles, and midbodies. J. Cell Bioi. 116,977-987.

O'Farrell, P.H. (1975). High cesolution two-dimension electrophoresis of protek. J. Biol. Chem 250,4007-40S 1.

Ohski, R., Wedrychowski, A, Schmidt, W.N., Briggs, R.C., & Hniiica, L.S. (1987). In uivo DNA-protein cross-linking by cis- and nans-diamminedic hlo ro platinum(IJ). Cancer Res. 47,20 1 -205,

Osborn, M., & Weber, K. (1982). Intenilediate nlaments: ceil-type-specific markers in dinerentiation and pathology. CeLi 31,303-306.

Pankov, R., Umezawa, AT Maki, R., Der, C.J., Hauser, C A , & Oshima, R.G. (1994). Keratin 18 activation by Ha-ras is mediated through ETS and Sun binding sites. Proc. Nd. Acud Sei. US.A 91,873-877.

Partin, AW.. Getzenberg, RH.. CarMiCheai, M.I., Vindivich, D.; Yoo. I., Epstein. 1.1.. & Coffey, D.S. (1993). Nuclear maVix pmtein patterns in human brnign prostatic hyperphsia and prostate cancer. Cancer Res. 53,744-746.

Pattanaik, AT Bachowski, G.. Laib. J., Letnluil, D., Shaw, FOT Petering. D.H., Hitchcock, A., & Saryan, L (1992). Ropertks of the naaion of ckdichiorodiammuieplatinum(~ with rnetdothionein. I. BiuL C h 267, 16121-16128.

Paulson. J-R-, Br Laemmü, U.K. (1977). The structure of histone-depleted metaphase chromosomes. Cell 12,8 17-828.

Phi-Van. L-, & Stratiing, WH. (1988). The ma& attachment regions of the chicken Lysozyme gene CO-map wïth the boundaries of the chromatin do& EMBO 1. 7,655- 664.

Pienta, KJ., Partin, AW-, & Cofky, DS. (1989). Cancer as a disease of DNA organization and dynamic cell structure- Cancer Res. 49,2525-2532.

Pienta , KI. & Coffky, D. (199 1). Comlation of nuclear morphometry with progression of breast cancer. Cancer 68,20 12-2016.

Pienta, K J., Getzenberg, R-H-T & Coffey, D.S. (199 1). Celi structure and DNA organization. CRC Cr& RN. Euk Gene Exp. 1,355-385.

Pienta, KI, & Cofiy, D-S. (1992). Nuclear-cytoskeletal interactions: evidence for physical connections between the nucleus aid cen periphery and their alteration by transformation. J. Cell. Biochem, 49, 357-365.

Pinto, A.L., & hppad, S.J. (1985). Sequence-dependent termination of in vitro DNA synthesis by cis- and @am-diamminedichloropIatinum 0. Proc. Nad. Acad Sei. USA 82,4616-4619.

Robst, H., & Herzog, R (1985). DNA regions associated with the nuclear mavix of Ehrlich ascites celis expose single-stranded sites after deproteinization. Eur. J. Biochem 146, 167-171.

Quaranta, V., & Jones, JJCCRR (199 1). The internai affairs of an integrin. Trends Ce11 Bioi. 1,2-4.

Ramaekers, F.C.S.. Haag, D.. Kant. A, Moesker. O., Jap, P.H.K. & Vooijs, G.P. (1983). Coexpression of keratin- and Woentin-type intermediate nlaments in human metastatic carcinoma ce&. Proc. Nad. Acad. Sci. USA 80,26 18-2622.

Raymond, W-A, & Leong, AS. (1989a). VmKntin - a new prognostic parameter in breûst carcinoma? J. Paihoi. 158, 107- 1 14.

Rayrnod, W.A., & Leong, AS. (L989b). Co-expression of cytokeratui and Woentin mtenaediate nhment proteins in benign and neoplastic breast epithelium- I. Parhol. 157. 299-306.

Reddei, ER. Murphy, LC.. & Sutherland, RL. (1984). Factors a&rting the sensitivity of T-47D human breast cancer cells to tarnorrifen. Cancer Res. 44,2398-2405.

Reddel R& Murphy, L.C., Hall. RE., & Sutherland, RL. (1985). DSrential sensitivity of human breast cancer ceil IuKs to the growth-inhibitory effects of tamoxifen. Cmcer Res- 45, 1525-1531.

Rem, A, & Fackeirnayer, F.Q. (1996). m a t i o n and molecular cloning of the scsold attachment factor B (SM-B), a novel human nuciear protein rhat specificaiiy bhds to S/MAR-DNA Nue. Acih Res. 24,843-849.

Replogle, T.S.. & Pienta, KJ. (1996). Role of the nuclear matrix in breast cancer. In R- Dickson and M. Lippman (eds), Mammary Tumour Ceii Cycle, Daerentiation and Metastasis, Kluwer Academic PubIishers, pp. 127-140-

Robertson, J.F.R. (1996). Oestrogen receptor: a stable phenotype in breast cancer. Br. J. Cmcer 73-50 12-

Rornig,, H., Fackehayer, F.O., Renz, A, Ramsperger, U., & Richter, A (1992). Characterization of SAF-A, a novel nuclear DNA binding protein fiorn HeLa ceils with high afnnity for nuclear matrix/scaffold attachent DNA elements. EMBO J. 11, 343 1- 3440.

Samuel, S.&, Minish, T.M., & Davie, J.R. (1997). Altered nuclear ma& pro tein pro& m oncogene-transfomexi mouse fibrobiasu e x h i h g high metastatic potentid Cmcer Res. 57, 147-151.

Samuel, S. K., Minish, T.M., & Davie, J.R. (1997). Nuclear matrix proteins in weli and poorly dinerentiated human breast cancer ce1 lines. J. Celi. Biochem 66.9- 15.

Sapino, A, Pietribiasi, F., Bussolati, G., & Mmhisio, P.C. (1986). Estrogen- and iamoxifen-induced cearrangement of cytoskektai and adhesion structures in breast cancer MCF-7 ceiis. Cmcer Res. 46,2526-253 1.

Schkke, T., Klehr -Wi D., Yoshida, M., Beppu, T., & Bode, 1- (1994). Gene expression within a chromah domain: The role of core histone hyperacetyiatïo n. BiochemStry 33, 4197-4206,

Schroder, KC.. Troîitsch, D., Friese. U.. Bâcbmann, M., & Muller, W.E.G.. (1987). Mature mRNA is selectively releaîed fiom the nuclear mavDr by an ATPldATP-dependent mechaniSm sensitive to topoisornerase inhibitors. I. Biol. Chem 262.89 17-8925.

k l y , KA.. & Aggelei, J. (199 1). Modulation of minE protein synthesis through alteration of the cytoskeleton in mouse mammary epithelial ceils cultureci on a reconstituted basement membrane. J. Celf. PhysioL 146. 1 17- 130.

Seibert, K. Shak, SM, Triche, TJ-, Whang-Peng, J.J., O'Brien, SJ.9 Toney. J.H., Huff, KKT Br Lippman, M.E. (1983). Cancer Res. 43.2223-2234.

Seshadri. R., Raymond. W.A, Leong, AS-Y-? Horsfidi, D.J., & McCaui, K (1996). Viintin expnssion is not associated with p o t prognosis in bnast cancer. Int. Cancer 67,353-356-

Simpson, RT. (1978). Structure of the cchromatosome, a chroma^ particle containing 160 base pairs and ai i the histones. BiochemStïy 17,5524-553 1.

Sommers, CL., Waker-Jones. D.. Hecldord, S.E., Woriand, P., Valverius, E., Clark, R, McCorxnick, F., Stampfer, M, Abdamch, S., & Gelmann. E.P. (1989). Vhentin rather than keratin expression in some hormone-independent breast cancer ceil lines and in oncogene-transfonned mammary'epitheiial cek. Cmcer Res. 49,425804263.

Soule, H.D., Mabney, TM, W o u S.RT Peterson, W.D. Ir., Brenz, R-, McGrath, CM., Russo, J-, Pauley, RJ.. Jones. RF., & Brooks, S.C. (1990). Isolation and characterization of a spontaneousiy immortaüzed hwnan breast epithelial ceK hes, MCF- 10'. Cancer Res. 50.6û75-6086.

Stein, G.S., Lian, J.B., Dworetzky, S1. Owen, TA, Boneli, R, Bidweii, J-P-? & van Wijnen, A J. (199 1). ReguhtEon of transcription-factor activity during gro wth and dinerentiation: mvolwment of the nuclear matrix in concentration and localization of porno ter binding proteins. J- Cell. Biochem 47, 300-305.

Stuurman, N., Meijne, A M L , van der Pol AI., De long, L., van Driel R., & van Renswoude, J. (1990). The nuclear ma& fiom ceh of difkrent origin: Evidence For a common set of matru< pro teins. J. Biol. Chem 265,5460-5465.

Sutherland. R.L.. Haii, RE., & Taylor, LW. (1983). Celi proWention kinetics of MCF-7 human mammary carcinoma ce& in culture and effects of tamoxifen on exponentially g o wing and plateau-phase ceh. Cmcer Res. 43,3998-4006.

Tait, L., Soule. KD., & Russo, I. (1990). [ntrastructural and immunocytochemical charactekation of an immorialized human beast epithelial ceii iine, MCF 10. Cancer Res. 50,6087-6094.

Taylor, Y.C., Zhang, X., Parsian, AJ-, & Duncan, P.G. (199 1). Image analysis-based measurement of DNA supercoihg changes in transfomeci and nonvansfomed human ceil lines. Qp. Cell Res. 197,222-228.

Thompson, E.W.. Pa&, S., Brunner, N., Sommers, CL., Zugrnaier, O., Clarke, R, Shima, T.B.9 Tom I.. Donahue, S., L i p p u ME., Mania, G.R-, & Dickson, RB. (1992). Association of increased basement membrane invasiveness with absence of esuogen receptor and expression of vimentin in human bceast.canzr ceii lines. I. Cell. Physiol. 150,534-544.

Toi, M., Bickneil, R, & H d , AL, (1992). Iiihiiition of colon and breast carcinoma ceii growth by interieukin-4. Cancer Res. 52,275-279.

Trask, D.KT Band, V., Zajchowski, DA, Yawsen, P., Suh, T-, & Sager, R. (1990). Keratjns as markers that distinguish normal and tunior-derived mammary epithelial cek. Proc. Natl. A c d Sci. USA %7,23 19-2323.

Tsutsui, KT Tsutsui, K., & Muller, MT. (1988). The nuciear scaEold exhibits DNA-binding sites selective for supercoiieà DNA. J. Biol. Chem 263,7235-7241.

Tsutsui, IC, Tsutsui, K., Okada, S., Watarai, S., Seki, S., Yasuda, T-, & Shobmori, T. (1993). Identification and characterization of a nuclear scaffold protein that binds the mauix attachment region DNA. L Biol. Chem Mg, 12886-12894.

Turner, C.E., & Bumdge, K. (1991). Transmembrane molecular assembiies in ceii- extraceiiular maVix interactions. Curr. Opin. Cell. Biol. 3.849-853.

Vic, P., Vignon, F., Derocq, D., 8r Rochefort, H. (1982). Effect of estradiol on the ultrastructure of the MCF, human bceast cancer ceUs in culture. Cmcer Res. 42.667- 673.

Villeponteau, B., Lundeii, M., & Martinson, H. (1984). Torsionai stress promotes the DNase 1 sensitivity of active genes. Cell 39.469-478.

Vogelstein, B.. PYdoli, D.M., & Coffey, D.S. (1980)- Supercoiled Ioops and eukaryotic DNA replication. Cell 22.79-85.

von Kries, I.P.. Buhrmester, H.. & Stratling, W.H. (1991). A rnairix/sc~old attachment region binding pro tein: identification. putitication, and mode of binding. Ceil 64, 123- 135.

Walters, MC, Fiering, S., Eidemiller, J., Magis, W., Groudine, MOT & Martin, D.L.K (1995). Enhancers increase the probability but not the kvel of gene expression. Proc- N d AC& SC^. USA a2,7 125-7 129.

Wang, D.M., Taylor, S., & Levy-W'ion, B. (1996a). Evaluation of the hction ofthe human apolipoprotein B gene aucLar mairix association regions in transgenic mice. I. Lipid Res. 37.21 17-2 124.

Wang, N., Butler, JS., & Ingber, DE. (1993). Mecbanovansduction across the ceU surface and though the cytoskeleton. Science 260, 1 124-1 127.

Wang, X, Toistonog, Go, Shoemari, RL, & Tmb, P. (1996b). Selective bindùig of specific mouse gcnomic DNA fragments by mouse Wnentin nlaments in viizo. DNA Ce11 Bio. 15,209-225.

War4 W.S, Schmidt, W.N., S~hmidt~ C.A., & Hnilica, L.S. (1984). Association of cytokeratin p39 with DNA in mtact Novikoff kpatoma ce&. froc. Nat1 A c d Sci. USA 81,4199423-

Watts, CKW., Handel, ML., King, WB., & Sutberland, RL. (1992). Oestrogen receptor gene structure and hction in breast cancer. J. Sterdd Biochem Molec. Biol. 41,529- 536.

Wedrychowski. A, Schmidt, W.N., Ward, W.S., & W c a , L.S. (1986). Crosslinking of cytokeratins to DNA in vivo by chromium sait and cis-diarnrninedichlorophthm(a). Biochemistry 25. 1-9.

Wright, C.. Nicholson, S.. Angus, B.. Sainsbury, I.R., Farndon, JeT Cairns. J., Harris, AL., & Horne. C H (1992). Relationship between c-erbB-2 protein product expression and response to endocrine therapy in advanced breast cancer. Br. % Cancer 65, 1 1 8- 12 1.

Yanagisawa, J., Ando, I., Nakay;uaa, J., Kohwi, Y., & Kohwi-Shigematsu, T. (1996). A matrix attachent region (MAR)-binding activity due to a pl 14 kilodalton protein is found only in human breast carcinomas and not in normal and benign breast disease tissues. Cancer Res. 56,457-462.

Yang, C. H., Lambie. EL, & Snyder, M. (1992). NuMA: an unusually long coiled coii rehted protein in the mamnialian nucleus. I. Cell Biol. 116, 1303- 13 17.

Yotsuyanagi, T., Ohta, N., Futo, T., Ito, S., Chen, D., & Ikeda, K. (1991). Multiple and irreversibk buduig of ciSdiamminedichioroplatinum~ to hwnan s e m albumin and its efféct on warfarùi binding. Chem P h a n BuL 39.3ûû3-3ûû6.

Yu, J., Bock, J.H., Slightom, J.L., & Villeponteau, B. (1994). A 5' P-globin ma&- attachment region and the polyoma enhancer together confer position-independent transcription. Gene 139, 139-145.

Zarobetti, G., Ramsey-Ewing, A, Boail, R, Stein, G., Br Stein, 1. (199 1). Disniption of the cytoskeleton with cytochahin D induces c-fos gene expression Erp. Ce11 Res. 192. 93-101,

Zeitlin, S.. Parent, A, Silyerstein, S., & EEstraW, A. (1987). Pre-mRNA spking and the nuclear mat& Mol. Ce& Bid. 7, 1 1 1-120.

Zeng, C., He, Dm, Br Brinkiey, B. (1994). bcalization of NuMA protein isofonns in the nuciear matrix of mammalian ceb. Cell. Motif. Cytoskeleton 29, 167- 176.

Zbao, KT Ras, E*. Go~mkz, E.. & Laemcnli, U.K (1993). SAR-dependent mobilization of histone Hl by HMG-UY in vitro: HMG-IN is enriched in Hl-depieted chromatin. EMBO J. 12,3237-3247.

Zini, N., Mazzotti, G., Santi, P., Rkzo'' RT Galami, A, Rana, R., & Maratdi, N.M. (1989). Cytochemral1Ocalization of DNA Ioop attachent sites to the nuciear lamina and to ihe inner nuciear matrix. Hutochemi~by 91,199-204.