Detergent fractionation with subsequent subtractive suppression hybridization as a tool for identifying genes coding for plasma membrane proteins

$Page 1: Detergent fractionation with subsequent subtractive suppression hybridization as a tool for identifying genes coding for plasma membrane proteins$
Detergent fractionation with subsequent subtractivesuppression hybridization as a tool for identifyinggenes coding for plasma membrane proteins

Andreas Lange1,2, Claudia Kistler1,2, Tanja B. Jutzi1,2, Alexandr V. Bazhin1,2, Claus Detlev Klemke2,

Dirk Schadendorf1,2 and Stefan B. Eichmuller1,2

1German Cancer Research Center, Skin Cancer Unit, Heidelberg, Germany;2Department of Dermatology, University of Heidelberg, University Hospital Mannheim, Mannheim, Germany

Correspondence: Stefan Eichmuller, PhD, German Cancer Research Center, Skin Cancer Unit (G300), Im Neuenheimer Feld 580, D-69120

Heidelberg, Germany, Tel.: +49 0 6221 42 33 80, Fax: +49 0 6221 42 33 79, e-mail: [email protected]

Accepted for publication 28 October 2008

Abstract: The identification of tumor-specific proteins located at

the plasma membrane is hampered by numerous methodological

pitfalls many of which are associated with the post-translational

modification of such proteins. Here, we present a new

combination of detergent fractionation of cells and of subtractive

suppression hybridization (SSH) to gain overexpressed genes

coding for membrane-associated or secreted proteins.

Fractionation of subcellular components by digitonin allowed

sequestering mRNA of the rough Endoplasmatic reticulum and

thereby increasing the percentage of sequences coding for

membrane-bound proteins. Fractionated mRNAs from the

cutaneous T-cell lymphoma (CTCL) cell line HuT78 and from

normal peripheral blood monocytes were used for SSH leading to

the enrichment of sequences overexpressed in the tumor cells.

We identified some 21 overexpressed genes, among them are

GPR137B, FAM62A, NOMO1, HSP90, SLIT1, IBP2, CLIF, IRAK

and ARC. mRNA expression was tested for selected genes in

CTCL cell lines, skin specimens and peripheral blood samples

from CTCL patients and healthy donors. Several of the detected

sequences are clearly related to cancer, but have not yet been

associated with CTCL. qPCR confirmed an enrichment of these

mRNAs in the rough endoplasmic reticulum fraction. RT-PCR

confirmed the expression of these genes in skin specimens and

peripheral blood of CTCL patients. Western blotting verified

protein expression of HSP90 and IBP2 in HuT78. GPR137B could

be detected by immunohistology in HuT78 and in keratinocytes

of dysplastic epidermis, but also in sweat glands of healthy skin.

In summary, we developed a new technique, which allows

identifying overexpressed genes coding preferentially for

membrane-associated proteins.

Key words: cutaneous T-cell lymphoma – digitonin treatment –

membrane proteins – subtractive suppression hybridization –

tumor-associated antigens

Please cite this paper as: Detergent fractionation with subsequent subtractive suppression hybridization as a tool for identifying genes coding for plasma

membrane proteins. Experimental Dermatology 2009; 18: 527–535.

Introduction

During the last decade, antibody therapy has evolved to a

rational treatment option in tumor therapy and Paul

Ehrlich’s magic bullets have become a sharp sword for the

eradication of cancer cells (1). Much research has been

spent in the further development of therapeutic antibodies,

especially in terms of stability and arming, but surprisingly

the number of proteins targeted is still limited. This might

at least in part be because of some inherent properties of

proteins recognized by antibodies in vivo: They should be

overexpressed, membrane-bound and accessible from the

outside of the cell. If they are identified by antibodies, the

frequent post-translational modifications of membrane

proteins (e.g. glycosylation) need to be considered. Various

approaches dealing with target identification hamper from

not fulfilling one or the other prerequisite of this list.

In this study, we have focused on cutaneous T-cell lym-

phoma (CTCL) because of the lack of tumor cell-specific

antigens. CTCLs are characterized by skin infiltrates of

malignant, clonally expanded T cells. The major variants

are mycosis fungoides (MF) with patches, plaques and

tumors confined to the skin and Sezary syndrome defined

by erythroderma, generalized lymphadenopathy and circu-

lating tumor cells in the peripheral blood (2). It is a rare

disease, which runs a chronic course over years and even

up to decades. CTCL cannot be cured, but a number of

DOI:10.1111/j.1600-0625.2008.00821.x

www.blackwellpublishing.com/EXDOriginal Article

ª 2009 The Authors

Journal compilation ª 2009 Blackwell Munksgaard, Experimental Dermatology, 18, 527–535 527

targeted therapies including therapeutic antibodies (3,4)

are under investigation. Immunotherapy has apparently

become a promising alternative to palliative treatments;

CTCL was even termed ‘paradigm for biological therapies’

(5). The T-cell receptor of the malignant clone has been

proven to harvest HLA-dependent epitopes, which can be

targeted in an individualized therapy (6,7). Until now, only

very few tumor antigens specific for CTCL tumor cells are

known (8,9). None of these is both membrane-bound and

expressed specifically on the tumor cells (10). Only few

CTCL-specific antigens (11,12) and tumor-associated anti-

gens located at the plasma membrane (8,13,14) have been

identified so far. Therefore, there is a strong demand

for the identification of membrane-bound shared tumor

antigens in order to improve diagnosis and treatment of

CTCL.

Here, we present a new combination of methods allow-

ing the identification of genes, which are differentially

expressed between experimental specimen and code for

proteins predominantly localized at the plasma membrane

or being secreted.

Material and methods

Tumor cell lines, patients’ material and mRNA ofhealthy tissues and PBMCThe following CTCL-lines were used: HH (aggressive T-cell

lymphoma), MyLa (MF), SeAx and HuT78 (both Sezary

syndrome). Biopsies of four MF patients and PBMC of five

Sezary Syndrome patients were used as well as skin samples

and PBMC of healthy donors for expression analysis of

SLIT1, CLIF, plasminogen activator inhibitor-1 (PAI-1),

TIC2 and apoptosis repressor with caspase recruitment

domain (ARC). Peripheral blood mononuclear cells

(PBMC) of healthy donors were isolated by Ficoll density

gradient centrifugation from buffy coats stored at 4�C.

PBMC of 11 healthy patients were pooled for subtractive

suppression hybridization (SSH). To evaluate the expres-

sion of genes in different tissues, we reverse transcribed

RNAs of skeletal muscle, heart, brain, lung, liver, kidney,

spleen, colon, placenta and testis obtained from a commer-

cial source (BD Biosciences Clontech, Palo Alto, CA, USA).

Detergent fractionation of cellsDetergent fractionation was carried out with HuT78 cells

and PBMCs from 11 donors according to Lerner et al. (15)

with some modifications. Briefly, individual cell pellets

were resuspended in cytosol buffer (pH 7.5) containing

150 mm CH3COOK, 20 mm HEPES, 68 mm NaCl, 75 mm

KCl, 2.5 mm (CH3COO)2Mg, 1 mm phenylmethylsulfonyl-

fluoride and 2 mm DTT, and incubated with 0.1 mg ⁄ ml

digitonin (Sigma Aldrich, St. Louis, MO, USA) for 5 min

on ice. After centrifugation, supernatants contained cytosol

including free polysomes and pellets comprising

membranes, nuclei and cell organelles including the rough

endoplasmic reticulum (ER).

Isolation of RNA and generation of cDNARNA isolation was achieved with TriFast (Peqlab, Erlangen,

Germany), containing phenol and guanidine isothiocyanate.

RNA pellets were treated with DNAse. Subsequently, phe-

nol–chloroform extraction and precipitation of RNA were

performed. Quality of total RNA was controlled by agarose

gels. Total RNA was reverse transcribed using the iScript

cDNA Synthesis Kit (Biorad, Hercules, CA, USA). For SSH

experiments, mRNA was isolated from total RNA using the

Oligotex Kit (Qiagen, Hilden, Germany) and reverse tran-

scribed by the AMV reverse transcriptase (BD Biosciences

Clontech).

Suppression subtractive hybridizationTo enrich genes that are overexpressed in CTCL, SSH was

performed using cDNA generated from fractionated

HuT78 and normal PBMCs by the BD PCR-Select and

Subtraction Kit as described by the manufacturer (BD

Biosciences Clontech). Briefly, double-stranded (ds) DNA

was generated by second strand synthesis and digested

with Rsa I. HuT78 dsDNA was divided into two parts and

separately linked with individual adaptors. Both portions

were first hybridized with digested cDNA of PBMCs

and then with each other. During the subsequent PCR

and nested PCR using adaptor-specific primers, only

sequences that are overexpressed in CTCL were amplified

exponentially.

SequencingPCR products of the SSH were cloned into the TOPO

Vector (TOPO TA Cloning Kit, Invitrogen, Carlsbad, CA,

USA) according to the protocol of the manufacturer and

transfected into TOP10 competent Escherichia coli (Invitro-

gen). Plasmids were isolated with the QIAprep Spin Mini-

prep Kit (Qiagen) and inserts were sequenced using the Big

Dye Terminator Cycle Sequencing Ready Reaction Kit and

an ABI PRISM TM 310 Genetic Analyzer (Abimed, Lange-

feld, Germany). Sequences were analysed with the HUSAR

package (Heidelberg Unix Sequence Analysis Resources)

and NCBI programs.

RT-PCRExpression profiling was performed by RT-PCR in at least

two independent experiments (annealing for 1 min at vari-

able temperatures; 35 cycles). Primers were selected using

appropriate prediction programs or taken from the litera-

ture in the cases of cycle-like factor (CLIF) (16), PAI and

uPA (17). Table S1 shows primer sequences, length of PCR

products and annealing temperatures.

Lange et al.

ª 2009 The Authors

528 Journal compilation ª 2009 Blackwell Munksgaard, Experimental Dermatology, 18, 527–535

Real-time RT-PCRTo compare the amounts of mRNAs of the identified genes

as well as housekeeping genes in the two analysed fractions,

we performed real-time RT-PCR using the ABsolute�QPCR SYBR� Green Mix (ABgene House, Epsom, UK)

and iTaq DNA polymerase (Bio-Rad) with the primers

shown in Table S1 on a MyiQ cycler (Bio-Rad). For each

gene, amplification curves were produced and threshold

values (Ct value) were obtained. Standard curves for

extrapolation were performed using specific PCR products

with determined copy numbers. Standardization of samples

was achieved by measurement of the endogenous reference

gene hydroxylmethylbilane synthetase (HMBS). By using

the DCT method, diagrams of all genes of interest were

established. Primer specificity was confirmed by melting

curve analysis and gel electrophoresis. The cycle parameters

for these transcripts and for the housekeeping genes HMBS

used for normalization were as follows: denaturing for 15 s

at 95�C; annealing and extension for 60 s at specific

temperature (see Table S1) for 40 cycles.

Western blottingTo control cell fractionation and protein expression,

Western blotting analysis was performed. Proteins were

separated by SDS-PAGE in a 12.5% polyacrylamide gel,

blotted onto nitrocellulose membranes and incubated with

the primary antibodies for 1.5 h at room temperature.

After incubation with appropriate secondary antibodies

(goat anti-rabbit or goat anti-mouse, peroxidase-coupled;

both Santa Cruz Biotechnology, Santa Cruz, CA, USA),

specific binding was visualized by usage of the Enhanced

Chemiluminescence System as described by the manufac-

turer (Amersham Biosciences, Piscataway, NJ, USA).

The following primary antibodies were used: Anti-Caln-

exin (Santa Cruz Biotechnology; 1 ⁄ 500) as a marker for the

membrane fraction; anti-b-actin (MP Biomedicals, Irvine,

CA, USA; 1 ⁄ 10 000) as marker for the cytosol; mouse IgG1

anti-HSP90 (Biovision, Mountain View, CA; 1 ⁄ 200), which

recognizes both alpha and beta form; rabbit IgG anti-insu-

lin-like growth factor binding protein-2 (IGFBP-2) (syno-

nym: IBP2; Cell Signaling Technology, Danvers, CA, USA;

1 ⁄ 1000]. The antibody against Slit-1 (Santa Cruz Biotech-

nology, Sc16616, goat IgG, 1 ⁄ 200) did not work with the

positive control (brain) and the antibody against Arc

(Neuromics, Edina, MN, USA; rabbit IgG; 1 ⁄ 500) stained

bands in the wrong size. Both antibodies were excluded

from the analysis.

ImmunohistochemistryImmunohistochemical analysis was performed as described

(8). Briefly, paraffin sections were dewaxed, rehydrated and

heated in 10 mm citrate buffer (pH 6 for 10 min) for

antigen unmasking. After blocking of unspecific binding

using the avidin ⁄ biotin kit (Linaris, Wertheim, Germany)

followed by 5% normal goat serum, sections were

incubated with the primary antibody (anti-TM7SF1, Acris,

Hiddenhausen, Germany; diluted 1:200 in PBS-Tx) at room

temperature overnight. Detection was carried out using a

biotinylated secondary antibody and the ABC kit (Linaris)

with alkaline phosphatase as enzyme. Cytospins were fixed

in cold acetone ()20�C for 10 min) and processed as the

rehydrated paraffin sections.

Results

For the identification of membrane-associated, overexpres-

sed antigens, detergent fractionation (15) of cells was

combined with SSH (18). Incubation of cells with digito-

nin resulted in permeabilization and subsequent centrifu-

gation allowed to separate two fractions, which also

divided the mRNAs accordingly to their location; the cyto-

solic fraction contained mRNA located at free ribosomes,

while the pellets comprised membranes, nuclei and cell

organelles including the rough ER and its associated

mRNAs.

Subcellular fractionation was carried out separately with

PBMCs from 11 donors and the cell line HuT78. Using

antibodies against calnexin and b-actin, the successful sepa-

ration of cytosol versus membranes and organelles was

confirmed by Western blotting (example see Fig. S1). The

pellet fractions of HuT78 and the 11 PBMC pellets of

healthy donors were then used for total RNA and mRNA

isolation. Subsequently, an SSH was performed using the

cell line as tester and the control mRNA as driver. The

resulting, tumor-enriched cDNAs were cloned into a pCR4

vector and sequenced to identify the corresponding genes.

Twenty-one different genes were identifiedA total of 99 individual clones have been sequenced and

found to code for 21 different genes, which are summa-

rized in Table S2. These genes code for the membrane-

bound proteins GPR137B (19), FAM62A (20), NOMO1

(21), HSP90 (22) and TCR (7), the extracellular proteins

SLIT1(23), TIC2 (24) and IBP2 (25), and also the intracel-

lular proteins stearoyl-CoA desaturase (Scd) (26), AMSH-

LP (27), CLIF (16), interleukin-1-receptor-associated kinase

(IRAK) (28), ARC (29), CTAK1 isoform 2 (30) and

CDK5RAP3 (31), as well as the ribosomal proteins RL9

(32), RS3A (33), RS6 (34), RLA0 (35), RL36A (36), RL7A

(37) and one mitochondrial gene (38).

Several genes were represented by repetitive clones. The

sequences of 31 clones could not be assigned. Five of the

identified genes are coding for membrane-bound proteins,

three for extracellular and 14 for intracellular proteins.

Detergent fractionation with SSH for membrane protein identification

ª 2009 The Authors


Apart from IBP2 (39) and HSP90 (40), these genes have

not been associated with CTCL so far.

mRNAs coding for plasma-membrane or secretedproteins are enriched in membrane-boundpolysomesUsing quantitative real-time RT-PCR we evaluated, whether

mRNAs coding for differently located proteins are selec-

tively enriched in one of the two analysed compartments,

namely free versus membrane-associated polysomes. We

selected genes coding for proteins located in the plasma

membrane (GPR137B), secreted proteins (TIC2, IBP2, PAI)

and cytosolic proteins (the house keeping genes GAPDH

and HMBS). In fact, the distribution of the specific mRNA

was as predicted (Fig. 1): The average ratio of specific

mRNA at bound ribosomes versus free ribosomes was

1 ⁄ 4.4 for the housekeeping genes (GAPDH and HMBS)

and 13.3 ⁄ 1 for genes coding for membrane-bound or

secreted proteins. Ratios for genes coding for cytosolic pro-

teins were always below 0.6 and those for membrane or

secreted genes were always above 1. The highest values were

observed for PAI in MyLa (57 ⁄ 1) and GPR137B in SeAx

(34 ⁄ 1), while the lowest ratios were detected for HMBS

(1 ⁄ 55) and GAPDH (1 ⁄ 12) in HuT78.

mRNA expression pattern analysisTo confirm the SSH results and to get a basis for further

selection of candidate genes, expression analysis by RT-

PCR was performed for 13 of the identified genes and

additionally of PAI-1 and uPA (see below) on a small panel

of cDNAs: CTCL-lines HuT78, MyLa, HH and SeAx, as

well as PBMC and skin from healthy donors (Fig. 2). Genes

coding for ribosomal or mitochondrial proteins and the

T-cell receptor were excluded from further analysis. All

genes were detected in HuT78 and most of them were not

found in healthy PBMCs, confirming largely the SSH assay.

Exceptions were TIC2, as well as HSP90, IRAK and

CDK5RAP3, which were expressed both in healthy PBMCs

and skin (Fig. 2).

Next, we determined the expression of those genes,

which code for membrane-bound proteins, namely

GPR137B, FAM62A and NOMO1, in a panel of healthy

control tissues containing skeletal muscle, heart, brain,

lung, liver, kidney, spleen, colon, placenta and testis by RT-

PCR. All of them showed a widespread tissue distribution

(Fig. S2): GPR137B was positive in lung, kidney, spleen,

placenta and testis, and weakly positive in heart, brain

and colon indicating a broader tissue distribution than

described previously (19). Both FAM62A and NOMO1

were expressed in all healthy control tissues investigated

and thus excluded for further analysis.

Based on the described RT-PCR results and expression

data known from the literature (cf. Table S1), further six

genes were selected for analysing their mRNA expression

pattern in tumor and healthy skin specimen (Table 1):

ARC, IBP2, TIC2, PAI-1, SLIT1 and CLIF. cDNA was

obtained from biopsies of four MF patients and from

PBMC of five Sezary Syndrome patients, as well as skin

specimens of six and PBMC samples of seven healthy

donors. SLIT1 and IBP2 were detected only in tumor cells,

but not in healthy tissue or PBMCs. CLIF and PAI-1 were

also detected in normal PBMCs and ⁄ or skin, but less fre-

20

25

14

16

18

20 Cytosolic ER-bound

30

35

40

25

30

35

10

15

6

8

10

12

x 10

3 co

pies

/ as

say

x 10

4 co

pies

/ as

say

assa

yx

10 5

copi

es /

10

15

20

25

10

15

20

copi

es/a

ssay

GA PD H

0

5

HuT78 My La

SeAx 0

2

4

HuT78 My La

SeAx HuT7 8 My La

SeAx

GPR137B TIC 2

0

5

HuT78 My La

SeAx HuT78 My La

SeAx

PA I I BP 2 HMBS

0

5

HuT78 SeAx

Figure 1. Quantification of detergent fractionation (real-time RT-PCR). The expression of mRNA coding for a membrane-bound protein (GPB137B),

secreted proteins (TIC2, PAI-1, IBP2) and cytosolic proteins (GAPDH and HMBS) was determined by real-time RT-PCR in the membrane fraction or

cytosolic fraction of different CTCL cell lines.

Lange et al.

ª 2009 The Authors


quently. TIC2 and ARC mRNA were found in all MF and

Sezary Syndrome samples, while normal PBMCs displayed

only weak TIC2 and normal skin only weak ARC bands.

Because of their functional connection to CLIF, we addi-

tionally analysed the expression of PAI-1 and uPA. PAI-1

was expressed in HuT78 and MyLa, and uPA was expressed

in MyLa. In contrast, both were negative in PBMC and

normal skin (Fig. 2). CLIF was detected in all the four cell

lines, but not in the control PBMCs or skin.

Protein expression analysisTo determine the respective protein expression, we selected

proteins, against which commercial antibodies were

available, and used Western blotting analysis (Fig. 3).

HSP90 and IBP2 were both detected in the cell line

HuT78, which was used for the screening process. The

analysis of subcellular fragments of HuT78 revealed a pre-

dominant localization of HSP90 in the cytosol, but small

amounts could additionally be detected in the membrane

fraction (Fig. 3a). Furthermore, the cell line HH and one

of six CTCL specimens were positive for HSP90 (Fig. 3b),

while RT-PCR unravelled more frequent expression of

HSP90 mRNA. IBP2 was perceived only in HuT78, not in

the other cell lines and tumor specimen (Fig. 3c), which

was in accordance with the mRNA expression. The tested

PBMCs from normal donors were negative for HSP90 pro-

tein, but one was positive for IBP2. The antibodies against

SLIT1 and Arc did not work in our hands, as also the

positive controls (brain and HeLa) were negative.

The distribution of GBP137B was analysed by immuno-

histochemistry using sections of normal and diseased skin

(Fig. 4). In healthy skin, we found a dominant staining in

glands and their ducts, while the epidermis was largely neg-

ative. In contrast, in MF sections, we found a large number

of GBP137B-positive keratinocytes within the dysplastic

epidermis, while the tumor cells often were negative.

Discussion

Identifying membrane-associated proteinsVarious different methods are available to test the subcellu-

lar localization of known proteins, provided specific anti-

bodies are available. In contrast, screening methods for

membrane-associated proteins are much more restricted.

Strategies including 2D gel electrophoresis have difficulties

originating from typical properties of membranes proteins

like large size and bad solubility (41). We have previously

used polyclonal, polyspecific rabbit antibodies generated

against the membrane fraction of a tumor cell line in a

phage display assay (14). Although we could identify mem-

brane proteins, this method does not allow detecting

strongly post-transcriptionally modified proteins because of

the usage of a recombinant expression system.

In the present study, we have utilized the well-known

separation of mRNAs between free and ER-bound poly-

somes accordingly to the property of the coded proteins;

mRNAs coding for membrane proteins are preferentially

translated at the ER-bound polysomes. Although this

separation is by far not perfect and mRNAs coding for

cytosolic proteins can be detected at the ER (15), at least

an enrichment of mRNAs coding for membrane proteins

can be acquired. Furthermore, mRNAs harvested from the

rough ER are in the very act of translation in contrast to

mRNA freely floating in the cytoplasm. Digitonin treatment

(15) was used for the separation of both mRNA popu-

lations; digitonin solubilizes cholesterol and thereby

permeabilizes the cell membrane. Subsequently, the cytosol

including the free mRNA and free polysomes is released.

CTAK

FAM62A

GPR137B

NOMO1

1 2 3 4 5 6

ARC

PAI1

uPA

CLIF

SLIT1

IBP2

AMSH-LP

TIC2

IRAK

HSP90

GAPDH

CK5P3

Figure 2. mRNA expression analysis by RT-PCR. The identified genes

were expressed in CTCL lines and only some of these were also

detected in PBMC and ⁄ or healthy skin. Lanes: (1) HH, (2) HuT78,

(3) MyLa, (4) SeAx, (5) PBMC and (6) skin.

Table 1. Expression frequencies (%) of selected genes

MF SS Normal skin Normal PBMCs(n = 4) (n = 5) (n = 6) (n = 7)

SLIT1 75 60 0 0IBP2 0 40 0 0CLIF 100 100 17 0PAI-1 75 100 33 29TIC2 100 100 0 100ARC 100 100 100 0

Frozen specimen from Mycosis fungoides lesions (MF) and healthy

skin, or peripheral blood monocytes (PBMCs) from Sezary syndrome

patients (SS) and healthy donors were used for RNA isolation and

subsequent RT-PCR. The number of samples (n) is given in brackets.

All cDNAs were tested positive with primers against GAPDH (positive

control).


ª 2009 The Authors


The remaining cell residues represent the membrane frac-

tion, including cell membrane, cell organelles and among

them the rough ER with its polysomes. We used this frac-

tion as the source of mRNA for SSH. The efficiency of this

process was confirmed by Western blotting. SSH was per-

formed with ER fractions of the Sezary syndrome cell line

HuT78 versus peripheral blood mononuclear cells of

healthy donors. Most importantly, quantitative real-time

PCR confirmed the uneven distribution of mRNAs coding

for membrane or intracellular proteins between ER-bound

and free polysomes, respectively; mRNA coding for mem-

brane proteins is enriched at the ER-bound polysomes.

Our assay revealed 21 overexpressed genes in CTCL

(Table S2). Here, we discuss their possible role for CTCL.

Genes encoding for membrane-bound proteinsFive genes were identified, which code for proteins located

partially (Hsp90) or exclusively at the plasma mem-

brane: GPR137B, FAM62A, NOMO1, HSP90 and the T-cell

receptor.

G protein-coupled receptor 137B (GPR137B) was first

described as a membrane protein, which is upregulated

during kidney development and an important role in cell-

type-specific differentiation-dependent signalling processes

was suggested (19). GPR137B was also found in Wilms

tumors (19) and differentiating osteoclasts (42). NODAL

modulator 1 (NOMO1) is a transmembrane protein that

antagonizes NODAL signalling (21). NODAL, a member of

the TGF-b family, was shown to play a pivotal role in deter-

mining left–right asymmetries in embryogenesis (43).

Antagonizing NODAL in embryonic stem cells influences

their differentiation (44). HSP90, known as an intracellular

protein, has recently been shown to be additionally localized

on the cell membrane and in the extracellular space (45).

HSP90 is a chaperone for MMP2, IRAK-1, NFjB, mutated

p53 and others (40) and thus contributes to tumor progres-

sion and invasiveness. Notably, IRAK-1 was one of the genes

we have identified. In MF, only rare expression of HSP90

has been reported (40), while no data are available for

Sezary syndrome. Because of its role in apoptosis regulation

(46) and promising studies with HSP90-inhibitors (47),

HSP90 might be a promising target for cancer therapy.

Genes encoding for secretory proteinsThree genes were identified, which code for extracellular

proteins: Slit-1, TIC2 (precursor of Testican-2) and IBP2.

SLIT1 belongs to the Slit protein family, which functions as

‘molecular guidance cues’ for both neurons and immune

cells and thus contributes to axon growth and leucocyte

chemotaxis (48). SLIT1 as a repellent for immune cells

could represent a new aspect of tumor-escape in CTCL.

Moreover, Slits are meant to play a role in angiogenesis

(49) and to be connected with integrin signalling (50).

SLIT1 is predominantly expressed in neuronal tissue. It

may play a role in tumorigenesis of glioma (51). SLIT1 was

also shown to be expressed in prostate tumors (52), mela-

noma, neuroblastoma and breast cancer (53). In addition,

this interesting molecule is connected to the WNT ⁄beta-catenin signalling pathway and c-myc, and thus has to

be considered as an anti-apoptotic, proliferative factor,

contributing to tumor progression (54,55). IBP2 contrib-

utes to proliferation (56). It is also able to induce cell

mobility by interacting with integrin alpha5 (57) and

increases the expression of MMP-2 (58). Overexpression of

IBP2 was proposed to enhance the malignancy of many

tumors (25). The molecule was further proposed as a bio-

marker for the response to therapy with inhibitors of

HSP90 (59).

150100

75

Cytosole Membrane Nuclei

1 2 3 4 5 6

kD

(a) HSP90

(b) HSP90

HuT78

100

75

37

(c) IBP2

25

kD

Po

s. c

trl.

Po

s. c

trl.

SeA

x

Hu

T78

MyL

a

HH

skin

MF

IIa

SS

MF

VIa .

MF

IIb

PB

MC

PB

MC

PB

MC

PB

MC

No

rm. s

kin

MF

Ib

Figure 3. Protein expression of HSP90 and IBP2 as determined by Western blotting. (a) A subcellular analysis of fractions derived from HuT78

revealed a predominant localization within the cytosole (arrowhead), but additionally a weak band in the membrane fraction (arrow). (b) Besides in

HuT78, HSP90 was detected weakly in HH and one CTCL specimen. (c) IBP2 protein was found in HuT78 and one tumor specimen.

Lange et al.

ª 2009 The Authors


Genes encoding for intracellular proteinsSeven genes coding for intracellular proteins were detected

in our assay: Scd, CLIF, ARC, AMSH-LP, IRAK, CTAK1 iso-

form 2 and CDK5RAP3. CLIF upregulates the expression of

the PAI-1 and additionally enhances cell proliferation (60).

It is predominantly expressed in neuronal and endothelial

cells (16). ARC is predominantly expressed in muscle (29)

and has been shown to inhibit caspase 8 triggered apoptosis

(61). Recently, decreased activation of caspase 8 was shown

in CTCL (62). ARC inhibits both extrinsic and intrinsic

apoptosis pathway and confers chemo- and radiation-resis-

tance (63). Upregulation of ARC has been demonstrated in

neoplastic cells (63,64). IRAK is involved in the TLR ⁄ IL-1R

and activates MAPK and NFjB (28). Overexpression of

NFjB was shown in the CTCL-lines HuT78, SeAx and MyLa

and in PBMC of CTCL patients (65). IRAK was shown to be

a target for decreasing the activation of NFjB (66).

Genes encoding for ribosomal and mitochondrialproteinsA number of genes have been detected, which code for

ribosomal (RL9, RS3A, RS6, RLA0, RL36A and RL7A) and

mitochondrial proteins. Besides their role in protein bio-

synthesis, ribosomal proteins are often associated with car-

cinogenesis. For instance, suppression of RS3A expression

leads to apoptosis (67) and RS3A was reported to be over-

expressed in various tumors and cancer cell lines (68) and

to enhance the malignant phenotype (69).

Identified genes and known relationship to CTCLIn a very recent publication, van Doorn et al. (70) reported

copy number alteration of several genomic regions for MF

and Sezary Syndrome. Two of the genes we identified are

located in regions with gains. The arc gene is located at

8q24.3 and was found to be amplified in 75% specimen of

Sezary Syndrome and 38% of MF patients. Cdk5p3 locates

to 17q21.32, for which a gain in 41% of MF patient has

been detected (70). We observed an expression of CLIF

with concomitant PAI-1. Interestingly, van Doorn et al.

described a direct gain of 7q21-7q22 (which contains pai-

1) in 55% of MF, but not Sezary Syndrome specimen (70).

Notably, the expression of several genes differed between

individual cell lines and specimen, as has previously been

reported for tumor antigens (8–10), and also for chromo-

somal gains and losses (70,71). Still, differences may

partially be attributed to recently acknowledged differences

between MF and Sezary Syndrome (70,72,73).

Conclusion

Using a new combination of detergent fractionation and

SSH, we were able to identify overexpressed genes coding for

membrane and secreted proteins without the need of protein

methods during the screening process. The presented

method allows ignoring the impact of post-translational

modifications as well as notorious difficulties with large, par-

tially hydrophobic proteins during the screening procedure

and is easy in comparison to other screening systems. Fur-

thermore, the unravelled genes are in the very act of transla-

tion and are enriched in tumor cells. By this means, a

number of highly interesting genes previously not recognized

to be associated with CTCL have been identified. Because of

known functions or association with other malignancies,

future experiments should address IBP2, PAI-1, HSP90,

SLIT1, CLIF and ARC. Further analysis on the role of these

Ctrl Ctrl

HuT78 MF

MF MF(e) (f)

(d)(c)

(b)(a)

Figure 4. Cellular localization of GPB137B as revealed by

immunohistochemistry. Within the epidermis of normal skin, few

keratinocytes show a weak staining (a), while sweat glands and their

ducts are strongly labelled (b). (c) A cytospins preparation of HuT78

confirms the positivity of this cell. (d) In contrast, skin specimen of

Mycosis fungoides samples show strong epithelial labelling. A close-up

(f) delineates this labelling being mainly confined to epidermal

keratinocytes, but individual T cells within the dermis (closed

arrowhead) and epidermis (closed arrow) are found to be stained, while

most tumor cells (open arrow) are negative. (e) A Pautrier microabscess

with central, weakly stained tumor cells (closed arrowheads) and a layer

of strongly positive epidermal cells (arrows).


ª 2009 The Authors


proteins in CTCL might allow a rational selection of promis-

ing therapeutic targets for interfering with their function.

Acknowledgements

We are grateful to Elke Dickes and Anita Heinzelmann for excellent techni-

cal assistance. This work was supported in part by a grant from the Boeh-

ringer Ingelheim Foundation for Basic Research in Medicine to AL.

References

1 Zafir-Lavie I, Michaeli Y, Reiter Y. Novel antibodies as anticancer agents. Onco-gene 2007: 26: 3714–3733.

2 Willemze R, Jaffe E S, Burg G et al. WHO-EORTC classification for cutaneouslymphomas. Blood 2005: 105: 3768–3785.

3 Dummer R. Future perspectives in the treatment of cutaneous T-cell lymphoma(CTCL). Semin Oncol 2006: 33: S33–S36.

4 Klemke C D, Goerdt S, Schrama D, Becker J C. New insights into the molecu-lar biology and targeted therapy of cutaneous T-cell lymphomas. J Dtsch Der-matol Ges 2006: 4: 395–406.

5 Pichardo D A, Querfeld C, Guitart J, Kuzel T M, Rosen S T. Cutaneous T-celllymphoma: a paradigm for biological therapies. Leuk Lymphoma 2004: 45:1755–1765.

6 Winter D, Fiebiger E, Meraner P et al. Definition of TCR Epitopes for CTL-Med-iated Attack of Cutaneous T Cell Lymphoma. J Immunol 2003: 171: 2714–2724.

7 Zeng J, Muller-Berghaus J, Nguyen X D et al. Identification of HLA class Idependent immunogenic peptides from clonotypic TCR expressed in cutaneousT cell lymphoma. Int J Cancer 2006: 119: 2476–2480.

8 Hartmann T B, Mattern E, Wiedemann N et al. Differentially expressed genesin cutaneous T-cell lymphoma. Exp Dermatol 2008: 17: 324–334.

9 Hartmann T B, Thiel D, Dummer R, Schadendorf D, Eichmuller S. SEREX identi-fication of new tumour-associated antigens in cutaneous T-cell lymphoma.Br J Dermatol 2004: 150: 252–258.

10 Eichmuller S, Usener D, Thiel D, Schadendorf D. Tumor-specific antigens incutaneous T-cell lymphoma: expression and sero-reactivity. Int J Cancer 2003:104: 482–487.

11 Eichmuller S, Usener D, Dummer R, Stein A, Thiel D, Schadendorf D. Serologi-cal detection of cutaneous T-cell lymphoma-associated antigens. Proc NatlAcad Sci U S A 2001: 98: 629–634.

12 Usener D, Schadendorf D, Koch J, Dubel S, Eichmuller S. cTAGE: a cutaneousT cell lymphoma associated antigen family with tumor-specific splicing. J InvestDermatol 2003: 121: 198–206.

13 Poszepczynska-Guigne E, Schiavon V, D’Incan M et al. CD158k ⁄ KIR3DL2 is anew phenotypic marker of Sezary cells: relevance for the diagnosis and follow-up of Sezary syndrome. J Invest Dermatol 2004: 122: 820–823.

14 Lee M, Kistler C, Hartmann T B et al. Immunoscreening of a cutaneous T-celllymphoma library for plasma membrane proteins. Cancer Immunol Immunoth-er 2007: 56: 783–795.

15 Lerner R S, Seiser R M, Zheng T et al. Partitioning and translation of mRNAsencoding soluble proteins on membrane-bound ribosomes. RNA 2003: 9:1123–1137.

16 Maemura K, de la Monte S M, Chin M T et al. CLIF, a novel cycle-like factor,regulates the circadian oscillation of plasminogen activator inhibitor-1 geneexpression. J Biol Chem 2000: 275: 36847–36851.

17 Castello R, Estelles A, Vazquez C et al. Quantitative real-time reverse transcrip-tion-PCR assay for urokinase plasminogen activator, plasminogen activatorinhibitor type 1, and tissue metalloproteinase inhibitor type 1 gene expressionsin primary breast cancer. Clin Chem 2002: 48: 1288–1295.

18 Diatchenko L, Lau Y F, Campbell A P et al. Suppression subtractive hybridiza-tion: a method for generating differentially regulated or tissue-specific cDNAprobes and libraries. Proc Natl Acad Sci U S A 1996: 93: 6025–6030.

19 Spangenberg C, Winterpacht A, Zabel B U, Lobbert R W. Cloning and charac-terization of a novel gene (TM7SF1) encoding a putative seven-pass transmem-brane protein that is upregulated during kidney development. Genomics 1998:48: 178–185.

20 Morris N J, Ross S A, Neveu J M, Lane W S, Lienhard G E. Cloning and preli-minary characterization of a 121 kDa protein with multiple predicted C2domains. Biochim Biophys Acta 1999: 1431: 525–530.

21 Haffner C, Frauli M, Topp S et al. Nicalin and its binding partner Nomo arenovel Nodal signaling antagonists. EMBO J 2004: 23: 3041–3050.

22 Eustace B K, Sakurai T, Stewart J K et al. Functional proteomic screens revealan essential extracellular role for hsp90 alpha in cancer cell invasiveness. NatCell Biol 2004: 6: 507–514.

23 Wong K, Park H T, Wu J Y, Rao Y. Slit proteins: molecular guidance cues forcells ranging from neurons to leukocytes. Curr Opin Genet Dev 2002: 12:583–591.

24 Nakada M, Miyamori H, Yamashita J, Sato H. Testican 2 abrogates inhibitionof membrane-type matrix metalloproteinases by other testican family proteins.Cancer Res 2003: 63: 3364–3369.

25 Hoeflich A, Reisinger R, Lahm H et al. Insulin-like growth factor-binding pro-tein 2 in tumorigenesis: protector or promoter? Cancer Res 2001: 61: 8601–8610.

26 Falvella F S, Pascale R M, Gariboldi M et al. Stearoyl-CoA desaturase 1 (Scd1)gene overexpression is associated with genetic predisposition to hepatocarci-nogenesis in mice and rats. Carcinogenesis 2002: 23: 1933–1936.

27 Kikuchi K, Ishii N, Asao H, Sugamura K. Identification of AMSH-LP containinga Jab1 ⁄ MPN domain metalloenzyme motif. Biochem Biophys Res Commun2003: 306: 637–643.

28 Janssens S, Beyaert R. Functional diversity and regulation of different interleu-kin-1 receptor-associated kinase (IRAK) family members. Mol Cell 2003: 11:293–302.

29 Stoss O, Schwaiger F W, Cooper T A, Stamm S. Alternative splicing determinesthe intracellular localization of the novel nuclear protein Nop30 and its interac-tion with the splicing factor SRp30c. J Biol Chem 1999: 274: 10951–10962.

30 Muller J, Ory S, Copeland T, Piwnica-Worms H, Morrison D K. C-TAK1 regu-lates Ras signaling by phosphorylating the MAPK scaffold, KSR1. Mol Cell2001: 8: 983–993.

31 Chen J, Liu B, Liu Y et al. A novel gene IC53 stimulates ECV304 cell prolifera-tion and is upregulated in failing heart. Biochem Biophys Res Commun 2002:294: 161–166.

32 Nika J, Erickson F L, Hannig E M. Ribosomal protein L9 is the product ofGRC5, a homolog of the putative tumor suppressor QM in S. cerevisiae. Yeast1997: 13: 1155–1166.

33 Song D, Sakamoto S, Taniguchi T. Inhibition of poly(ADP-ribose) polymeraseactivity by Bcl-2 in association with the ribosomal protein S3a. Biochemistry2002: 41: 929–934.

34 Vega F, Medeiros L J, Leventaki V et al. Activation of mammalian target ofrapamycin signaling pathway contributes to tumor cell survival in anaplasticlymphoma kinase-positive anaplastic large cell lymphoma. Cancer Res 2006:66: 6589–6597.

35 Barnard G F, Staniunas R J, Bao S et al. Increased expression of human ribo-somal phosphoprotein P0 messenger RNA in hepatocellular carcinoma andcolon carcinoma. Cancer Res 1992: 52: 3067–3072.

36 Kim J H, You K R, Kim I H, Cho B H, Kim C Y, Kim D G. Over-expression ofthe ribosomal protein L36a gene is associated with cellular proliferation inhepatocellular carcinoma. Hepatology 2004: 39: 129–138.

37 Zhu Y, Lin H, Li Z, Wang M, Luo J. Modulation of expression of ribosomal pro-tein L7a (rpL7a) by ethanol in human breast cancer cells. Breast Cancer ResTreat 2001: 69: 29–38.

38 Mambo E, Chatterjee A, Xing M et al. Tumor-specific changes in mtDNA con-tent in human cancer. Int J Cancer 2005: 116: 920–924.

39 Melendez B, Martinez-Delgado B, Cuadros M, Fernandez V, Diaz-Uriarte R, Beni-tez J. Identification of amplified and highly expressed genes in amplicons of theT-cell line huT78 detected by cDNA microarray CGH. Mol Cancer 2005: 4: 5.

40 Valbuena J R, Rassidakis G Z, Lin P et al. Expression of heat-shock protein-90in non-Hodgkin’s lymphomas. Mod Pathol 2005: 18: 1343–1349.

41 Watarai H, Inagaki Y, Kubota N et al. Proteomic approach to the identificationof cell membrane proteins. Electrophoresis 2000: 21: 460–464.

42 Nomiyama H, Egami K, Wada N et al. Identification of genes differentiallyexpressed in osteoclast-like cells. J Interferon Cytokine Res 2005: 25: 227–231.

43 Schier A F. Nodal signaling in vertebrate development. Annu Rev Cell Dev Biol2003: 19: 589–621.

44 Minchiotti G. Nodal-dependant Cripto signaling in ES cells: from stem cells totumor biology. Oncogene 2005: 24: 5668–5675.

45 Picard D. Hsp90 invades the outside. Nat Cell Biol 2004: 6: 479–480.46 Didelot C, Schmitt E, Brunet M, Maingret L, Parcellier A, Garrido C. Heat

shock proteins: endogenous modulators of apoptotic cell death. Handb ExpPharmacol 2006: 172: 171–198.

47 Sharp S, Workman P. Inhibitors of the HSP90 molecular chaperone: currentstatus. Adv Cancer Res 2006: 95: 323–348.

48 Wu J Y, Feng L, Park H T et al. The neuronal repellent Slit inhibits leukocytechemotaxis induced by chemotactic factors. Nature 2001: 410: 948–952.

49 Fujiwara M, Ghazizadeh M, Kawanami O. Potential role of the Slit ⁄ Robo sig-nal pathway in angiogenesis. Vasc Med 2006: 11: 115–121.

50 Stevens A, Jacobs J R. Integrins regulate responsiveness to slit repellent signals.J Neurosci 2002: 22: 4448–4455.

51 Dickinson R E, Dallol A, Bieche I et al. Epigenetic inactivation of SLIT3 andSLIT1 genes in human cancers. Br J Cancer 2004: 91: 2071–2078.

52 Latil A, Chene L, Cochant-Priollet B et al. Quantification of expression of net-rins, slits and their receptors in human prostate tumors. Int J Cancer 2003:103: 306–315.

53 Rossi M R, Huntoon K, Cowell J K. Differential expression of the LGI and SLITfamilies of genes in human cancer cells. Gene 2005: 356: 85–90.

54 Katoh Y, Katoh M. Comparative genomics on SLIT1, SLIT2, and SLIT3 ortho-logs. Oncol Rep 2005: 14: 1351–1355.

55 Yang F, Zeng Q, Yu G, Li S, Wang C Y. Wnt ⁄ beta-catenin signaling inhibitsdeath receptor-mediated apoptosis and promotes invasive growth of HNSCC.Cell Signal 2006: 18: 679–687.

56 Hettmer S, Dannecker L, Foell J, Elmlinger M W, Dannecker G E. Effects ofinsulin-like growth factors and insulin-like growth factor binding protein-2 onthe in vitro proliferation of peripheral blood mononuclear cells. Hum Immunol2005: 66: 95–103.

Lange et al.

ª 2009 The Authors


57 Wang G K, Hu L, Fuller G N, Zhang W. An interaction between insulin-likegrowth factor-binding protein 2 (IGFBP2) and integrin alpha5 is essential forIGFBP2-induced cell mobility. J Biol Chem 2006: 281: 14085–14091.

58 Miyake H, Hara I, Yamanaka K, Muramaki M, Gleave M, Eto H. Introductionof insulin-like growth factor binding protein-2 gene into human bladder can-cer cells enhances their metastatic potential. Oncol Rep 2005: 13: 341–345.

59 Zhang H, Chung D, Yang Y C et al. Identification of new biomarkers for clini-cal trials of Hsp90 inhibitors. Mol Cancer Ther 2006: 5: 1256–1264.

60 Yeh C T, Lu S C, Tseng I C et al. Antisense overexpression of BMAL2 enhancescell proliferation. Oncogene 2003: 22: 5306–5314.

61 Koseki T, Inohara N, Chen S, Nunez G. ARC, an inhibitor of apoptosisexpressed in skeletal muscle and heart that interacts selectively with caspases.Proc Natl Acad Sci U S A 1998: 95: 5156–5160.

62 Braun F K, Fecker L F, Schwarz C et al. Blockade of death receptor-mediatedpathways early in the signaling cascade coincides with distinct apoptosis resis-tance in cutaneous T-cell lymphoma cells. J Invest Dermatol 2007: 127: 2425–2473.

63 Mercier I, Vuolo M, Madan R et al. ARC, an apoptosis suppressor limited toterminally differentiated cells, is induced in human breast cancer and conferschemo- and radiation-resistance. Cell Death Differ 2005: 12: 682–686.

64 Wang M, Qanungo S, Crow M T, Watanabe M, Nieminen A L. Apoptosisrepressor with caspase recruitment domain (ARC) is expressed in cancer cellsand localizes to nuclei. FEBS Lett 2005: 579: 2411–2415.

65 Sors A, Jean-Louis F, Pellet C et al. Down-regulating constitutive activationof the NF-kappaB canonical pathway overcomes the resistance of cutaneousT-cell lymphoma to apoptosis. Blood 2006: 107: 2354–2363.

66 Sims J E. IL-1 and IL-18 receptors, and their extended family. Curr Opin Immu-nol 2002: 14: 117–122.

67 Naora H, Takai I, Adachi M, Naora H. Altered cellular responses by varyingexpression of a ribosomal protein gene: sequential coordination of enhance-ment and suppression of ribosomal protein S3a gene expression induces apop-tosis. J Cell Biol 1998: 141: 741–753.

68 Goodin J L, Rutherford C L. Characterization of human ribosomal S3a geneexpression during adenosine 3¢:5¢ cyclic monophosphate induced neuroendo-crine differentiation of LNCaP cells. Regulation of S3a gene expression inLNCaP. Mol Biol Rep 2002: 29: 301–316.

69 Hu Z B, Minden M D, McCulloch E A, Stahl J. Regulation of drug sensitivity byribosomal protein S3a. Blood 2000: 95: 1047–1055.

70 van Doorn R, van Kester M S, Dijkman R et al. Oncogenomic analysis of myco-sis fungoides reveals major differences with Sezary syndrome. Blood 2008:Oct 1. [Epub ahead of print].

71 Vermeer M H, van Doorn R, Dijkman R et al. Novel and highly recurrent chro-mosomal alterations in Sezary syndrome. Cancer Res 2008: 68: 2689–2698.

72 Klemke C D, Fritzsching B, Franz B et al. Paucity of FOXP3(+) cells in skin andperipheral blood distinguishes Sezary syndrome from other cutaneous T-celllymphomas. Leukemia 2006: 20: 1123–1129.

73 Booken N, Gratchev A, Utikal J et al. Sezary syndrome is a unique cutaneousT-cell lymphoma as identified by an expanded gene signature including diag-nostic marker molecules CDO1 and DNM3. Leukemia 2008: 22: 393–399.

Supporting information

Additional Supporting Information may be found in the online version of

this article:

Figure S1. Validation of subcellular separation. Western blot validating

the subcellular separation by detergent fractionation. Calnexin was prefer-

entially detected in the membrane fraction (M), while b-actin was enriched

in the cytosolic fraction.

Figure S2. mRNA expression of GPR137B, FAM62A and NOMO1 in

control tissues. GPR137B is positive in lung, kidney, spleen, placenta and

testis and weekly positive in heart, brain and colon, while FAM62A and

NOMO1 are detectable in every tissue. Lanes: (1) skeletal muscle, (2) heart,

(3) brain, (4) lung, (5) liver, (6) kidney, (7) spleen, (8) colon, (9) placenta,

(10) testis.

Table S1. Primers.

Table S2. Identified genes, their function and their expression patterns.

Please note: Wiley-Blackwell are not responsible for the content or func-

tionality of any supporting materials supplied by the authors. Any queries

(other than missing material) should be directed to the corresponding

author for the article.


ª 2009 The Authors


Documents

Detergent fractionation with subsequent subtractive suppression hybridization as a tool for identifying genes coding for plasma membrane proteins