DNA-RNA complexes that might represent transient attachment sites of

nuclear DNA to the matrix

CHRISTOS PATRIOTIS, MARIANA ANDREEVA, MARY PASCALEVA, VESELIN IVANOV and LALIO

DJONDJUROV'

Institute of Molecular Biology, Bulgarian Academy of Sciences 1113 Sofia, Bulgaria

Summary

In this study we describe DNA-RNA complexes inmatrix DNA of Friend cells. The presence of suchunusual structures is confirmed by the followingevidence. When a preparation of matrix DNA iselectrophoresed in agarose an RNA componentalways migrates together with DNA. There shouldbe a close interaction between DNA and RNA insuch a preparation because the presence of the RNAcomponent causes resistance of DNA to DNase Iand Exo m. An intimate, hybrid-type association ofpart of the RNA component with DNA is indicatedalso by the fact that about 20% of this RNA issensitive to RNase H. By specific inhibition of theRNA synthesis with a-amanitin and actinomycin Dit was shown that the bulk of associated RNA istranscribed by RNA polymerase III. Hybridizationexperiments showed similarity between the DNA

sequences isolated from the complexes and thosefrom the base of dehistonized DNA loops obtainedby high-salt extraction of nuclei. This observationsuggests that the complexes might represent attach-ment sites of nuclear DNA to the matrix: possibly,the attachment is mediated via the RNA component.Experiments with induction of erythroid differen-tiation indicated that a profound reorganization ofthe nucleus, accompanying terminal differen-tiation, leads to a striking reduction in the numberof complexes and thus in the number of attachmentsites. This suggests that the complexes should func-tion as transient attachment sites.

Key words: DNA attachment, DNA-RNA complexes,matrix DNA, nuclear matrix, snRNA.

Introduction

Three levels of DNA organization have been shown inthe eukaryotic nucleus: basic nucleosomal chromatinfibers, 30 nm thick chromatin fibers and DNA loopdomains. The existence of DNA loop domains was firstproposed by Cook and Brazell (1975). It was foundfurther that the loops are attached at the base to com-ponents of the nuclear matrix (Vogelstein et al. 1980). Anumber of investigations have been focussed on theorganization of attachment sites (Hancock and Huges,1982; Berezney and Basler, 1982; Berezney, 1984; Zehn-bauer and Vogelstein, 1985; Gasser, 1988). In respect ofthe nature of DNA at the point of attachment, uniquesequences (Nelkin et al 1980), repetitive sequences(Razinef al. 1978, 1979; Matsumoto, 1981) or sequencesthat do not differ from the bulk DNA were found(Pardoll and Vogelstein, 1980; Basler et al. 1981). Inrespect of the matrix components to which DNA isdirectly anchored only proteins have been identified(Bodnar et al. 1983; Razin et al. 1981; Patriotis andDjondjurov, 1989). Rest et al. (1986) reported a firm

Journal of Cell Science 95, 667-674 (1990)Printed in Great Britain © The Company of Biologists Limited 1990

association between matrix DNA and RNP but no detailson the interacting macromolecules were provided.

In our opinion, all experiments describing the attach-ment of nuclear DNA to the matrix might be consideredin the following way. We believe that in living cells twobasic types of attachment exist. The first one is constitut-ive: it is responsible for constraining relatively largesupercoiled loop-units, the structure and topology ofwhich are permanent and do not depend on the functionalstate of the cells. Candidates for such attachment are, forexample, the tightly bound DNA-protein complexesdescribed by Bodnar et al. (1983) and Patriotis andDjondjurov (1989) as well as the putative replicationcomplexes proposed by Pardoll et al. (1980) and Berez-ney and Buchholtz (1981). The second type is faculta-tive: it ensures the formation of small, operational loopsthat are part of the large constitutive domains. Theirappearance and functioning should be local and transient,and should depend on the events in which DNA isinvolved at the moment. Since nuclear DNA is subjectedto transcription, repair, cyclic spiralization and relax-ation, each of these events might cause such facultative,

667

transient attachment. We assume that these attachmentsshould be realized through the complexing of DNA withvarious matrix components. According to our under-standing, such a type of attachment has been describedby Laemmli's group (Gasser and Laemmli, 1986; Mirko-vitch et al. 1988; Izauralde et al. 1988) and by Loc Phi-Van and Stratling (1988).

In this study we describe an unusual organization ofsome of the matrix-associated DNA. It seems that itforms complexes with RNA that might function astransient attachment sites.

Materials and methods

Cell culture, inhibition of RNA synthesis and inductionof erythroid differentiationF4N Friend cells were used in all experiments. They weregrown in MEM supplemented with 10% calf serum andmaintained to exponential growth by daily subculturing in freshmedium. For a specific inhibition of RNA synthesis the cellswere treated with 2 or 100 figml" of cr-amanitin (BoehringenMannheim GmbH), 0.04 or lOf/gml" of actinomycin D(Sigma) (Marzluff and Huang, 1985). Erythroid differentiationin Friend cells was induced as described by Tsvetkov et al.(1989).

Isolation of nuclear matricesNuclear matrices were isolated according to the conventionalprocedure (Djondjurov et al. 1986). Briefly, collected Friendcells were washed in 146 niM sucrose, 100 mM KC1, 10 mMTris-HCl, pH7.0, 1.5 mM MgCl2 and then resuspended for5min in the same buffer containing 0.2% Nonidet P40. Aftercentrifugation at 2000 revs min~ for 5 min the nuclei werewashed in 146 mM sucrose, 100 mM KC1, 10 mM Tris-HCl,pH 7.7, 5 mM MgCl2. 0.5 mM CaCl2, and digested for 45 min at4°C with 500ng ml DNase I (Sigma) in the same buffer at adensity of 5xlO8 nuclei ml"1. The digested nuclei were layeredon a 1.0 M sucrose cushion in 100 mM KC1, 10 mM Tris-HCl,pH7.7, 1.5 mM MgCl2, and centrifuged at 3000 revs min"1 for10 min. The pellet was extracted three times with 10 mMTris-HCl, pH7.2, 0.1 mM MgCl2, 0.1 mM CaCl2, and twicewith 2 M KC1, 10 mM Tris-HCl, pH7.7, 1.5 mM MgCl2.

Experiments with DNAFor purification of DNA the nuclear matrices were resuspendedin 0.5% SDS, 10 mM Tris-HCl, pH6.8, 1 mM EDTA, andincubated for 3 h at 37°C. Deproteinization and precipitation ofDNA was performed as described elsewhere (Patriotis andDjondjurov, 1989). Usually, the precipitated DNA was treatedwith RNase A (Sigma) and deproteinized again.

Purified matrix DNA was redigested at 37°C with 50^gml~'DNase I (Sigma) in 10mM Tris-HCl, pH7.0, 100mM KC1,5 mM MgCl2, 0.5 mM CaCl2, or with 10 units ^g"1 DNA Exo IIIin buffer recommended by the supplier (Pharmacia). ForRNase A (Sigma, SO^gml"1) treatment the samples weredissolved in buffer for DNase I. Digestion with RNase H(Pharmacia) was performed with 10unitsng~l DNA in bufferrecommended by the supplier. In some experiments matrixDNA was digested with 2 units fig~ DNA of micrococcalnuclease (Sigma).

//; vitro labeling of DNA was performed according toManiatis et al. (1982). [a--32P]deoxiribonucleoside triphos-phates were purchased from Amersham (3000Cimmol~ ).Hybridization was carried out as described by Allegra et al.

(1987). Radioiodination of the proteins tightly bound to thematrix DNA was carried out according to Patriotis and Djond-jurov (1989).

DNA was electrophoresed on 1 or 1.5% agarose gels inTris-acetate buffer as described by Maniatis et al. (1982) andvisualized by staining with ethidium bromide and ultravioletillumination.

In some experiments the matrix DNA containingDNA-RNA complexes was purified by ultracentrifugation inCsCl on a Beckman VTiSO rotor at 45 000 revs min"1 for 18 h.

Radio-iodinationIn order to study the presence of tightly bound proteins inmatrix DNA fragments with lengths of 250-150 base pairs (bp),after their electrophoretic separation in low-temperature gellingagarose (0.5%, Sigma), the corresponding zone was excisedunder the control of ultraviolet illumination and DNA wasrecovered, which was then dissolved in 50 fd distilled water andradioiodinated as described (Patriotis and Djondjurov, 1989).

Preparation and fractionation of residual structuresretaining large, dehistonized loopsNuclei (7X107) were extracted with 2ml of 2M NaCl for10 min. As our previous data show this leads to the appearanceof large, dehistonized loops attached to residual matrix struc-tures (Patriotis and Djondjurov, 1989). The released DNAloops were digested with HaelW and EcoRl as described byPharmacia and centrifuged at 15 000 revs min"1 for 15 min on anEppendorf microfuge. In order to reduce the size of thefragments associated with the matrix structures the pellet wasredigested mildly with 25^gml~' DNase I for 15 min at 4°Cand centrifuged. Both the supernatant, referred as fraction S,and the pellet, referred as fraction P, were deproteinized, thendigested for 15 min at 37°C with SOmgml"1 RNase A anddeproteinized again.

Using the above conditions nuclei isolated from inducedFriend cells were also extracted. A control by electron mi-croscopy on spread preparations confirmed the presence oflarge, intact DNA loops.

Electron microscopyFor electron microscopy the matrices were fixed with 2.5%glutaraldehyde in 0 . 1 M cacodylate buffer, pH7.4, postfixedwith 2% osmium tetroxide, dissolved in the above buffer andembedded in Durcopan ACM (Fluka). The ultrathin sectionswere stained with uranyl acetate and lead citrate (Reynolds) andexamined under a JEM 100B electron microscope.

Analytical proceduresProtein was determined according to Lowry et al. (1951) andDNA by the method of Burton (1956). Radioactivity wascounted in an Beckman scintillation spectrometer after dissolv-ing the samples in toluene/Triton X-100 scintillation mixture(Anderson and McClure, 1973).

Results

Matrix DNA resistant to DNase I and Exo IIIFig. 1 illustrates the size of DNA fragments recoveredfrom matrices. Interestingly, as revealed by staining withethidium bromide and ultraviolet illumination, if depro-teinized matrix DNA is digested with RNase A the size ofthe fragments is reduced to a mean size of 200 bp. Whensuch RNase A-treated and 32P-labeled matrix DNA is

668 C. Patriotis et al.

a b c

— 983

Fig. 1. Migration of matrix DNA in 1.5% agarose. Lane a,matrix DNA before treatment with RNase A. Lane b, matrixDNA after treatment with RNase A. Lane c, matrix DNAtreated with RNase A and digested with RNase H. Lane d,nuclease-resistant fragments obtained after extensive digestionof total DNA with DNase I. It can be seen that their size(shown in bp) is comparable to that of the original matrixDNA.

digested with DNase I and Exo III (Fig. 2) a significantportion of the label can be recovered as PCA-precipitablematerial: the same is true even if additional enzymeactivity is added at the time when a stable level ofdigestion is reached. This observation indicates thepresence of either DNA resistant to the above nucleasesor PCA-precipitable oligonucleotides. To distinguishbetween these two possibilities, samples of the same 32P-labeled DNA shown in Fig. 2, digested with nucleasesfor 40min, were run on agarose gels and the gels werethen dried and autoradiographed. For DNase I- and ExoIll-digested samples the results confirm the presence ofnuclease-resistant fragment migrating as undigested con-trol (Fig. 3A). Since we have loaded on each slot digestswith equal radioactivity (which means equal amounts ofthe initial matrix DNA) but the intensity of the signal ofDNase I-digested samples was always reduced it seemsthat the resistance to this nuclease is lower. In contrast tothe above two nucleases the micrococcal nuclease digeststhe matrix DNA completely (Figs 2, 3).

In order to see further whether this DNA represents aspecific fraction or is heterogeneous in nature we per-formed the following experiment. Isolated matrix DNAwas 32P-labeled and hybridized to DNA of two differentnuclear fractions obtained by extraction of DNase I-digested nuclei with hypotonic buffer and high salt,

20Time (min)

Fig. 2. Kinetics of digestion of matrix DNA by nucleases. Inthese experiments, purified matrix DNA treated with RNaseA was labeled with P by nick translation and thenredigested with nucleases. At intervals 100-/«l samples weretaken, which were precipitated subsequently with equalvolumes of 1.5wPCA and 1 % BSA (w/v). Arrows indicatethe time of addition of enzyme activity. A illustrates theresults for redigestion with DNase I; B, the results for ExoIII; and C, the results for micrococcal nuclease. (O O)The experiments without hydrolysis with KOH; ( • • )the experiments in which the labeled matrix DNA wastreated, before redigestion, with KOH to remove RNA; notethe restoration of the usual sensitivity to DNase I and ExoIII.

respectively (see the procedure for isolation of matrices).On the basis of the hybridization signals obtained(Fig. 4) one can conclude that the matrix DNA is rather afraction. This conclusion, considered together with theobservation that it is less accessible to nucleases, raisesthe question as to whether it is not generated duringDNase I digestion as fragments that are entrapped in thematrix. As is seen from the control experiments describedin Table 1 such a possibility sounds unreasonable.

The resistance to DNase I and Exo III is due to acomplexing of matrix DNA with RArAOne of the reasons for such unusual resistance might becomplexing of the fragments with proteins. To check thispossibility the matrix DNA was analysed for the presenceof tightly bound proteins, which survived the harshextraction with phenol (Neuer and Werner, 1985). Inorder to eliminate any possible interference with theDNA-protein complexes already described for Friendcells, migrating in agarose above the zone of 400 bp(Patriotis and Djondjurov, 1989), the 200 bp area of the

DNA-RNA complexes 669

d

A

d

B

».» I -200

Fig. 3. Electrophoretic evidence showing that the matrixDNA contains fragments resistant to DNase I and Exo IIIbut sensitive to micrococcal nuclease. In this experimentsamples of the matrix DNA from Fig. 2 with equalradioactivity, 32P-labeled and redigested with nucleases, wereelectrophoresed in 1 % agarose gels and the gel was thendried and autoradiographed; the data presented areautoradiograms. A illustrates the experiment withouthydrolysis with KOH, while B shows the experiment inwhich the labeled matrix DNA was treated with KOH beforeredigestion. In both panels: lanes a, represent undigested,control samples; lanes b, DNase I-digested samples; lanes c,Exo Ill-digested samples; and lanes d, micrococcal nuclease-digested samples. The value on the right is in bp.

Fig. 4. Distribution of the matrix DNA sequences in variousnuclear fractions. Isolated matrix DNA was labeled with 32Pand hybridized to the following fractions: A, matrix DNAused as internal control; C, DNA from the fraction extractedfrom DNase I-digested nuclei with hypotonic buffer (see theprocedure for preparation of nuclear matrices); D, DNAfrom the high-salt extracts of DNase I-digested nuclei; andB, nuclease-resistant DNA fragments recovered from totalDNA after extensive digestion with DNase I. The obvioussimilarity of the matrix sequences with the nuclease-resistantsequences obtained from the total DNA can be seen. 1 and 2,mark the slots loaded with 5 and 10 f.ig DNA, respectively.

gel in which a sample of matrix DNA was separatedunder the usual electrophoretic conditions was excised,and the DNA was recovered and radiolabeled with I25I.As is seen in Fig. 5, however, no radiolabeled proteins inthe zone of 200bp are detected.

On the other hand, the matrix DNA is copurified withan RNA component that is partially sensitive to RNase A(Fig. 1) and might be removed completely by hydrolysiswith KOH (Fig. 6). The results given in Fig. 6 empha-size, moreover, that both components, DNA and RNA,

Table 1. Secondary binding of added chromatinfragments and deproteinized matrix DNA to residual

nuclear structures

Stage ofaddition

Digestion withDNase 1

Extraction withhypotonic buffer§

Extraction withhigh-salt§

Ctsnun bound to the finalmatrix preparation (%)•

Chromatinfragments^

0.13

0.39

0.24

Deproteinizedmatrix DNAJ

0.02

0.05

0.04

• The data are mean values of two experiments.•(•Proliferating Friend cells were labeled with 0.25 ^Ciml"1

[3H]thymidine for 16h. After digestion of 1X107 nuclei with DNaseI, the labeled chromatin fragments from hypotonic and high-saltextractions were pooled and dialyzed against corresponding buffersolutions. The labeled chromatin fragments were added to 1X106

unlabeled nuclei at different stages of matrix isolation. The ctsmin"bound to the final matrix preparation were calculated as a percentageof the added radioactivity.

% Deproteinized matrix DNA from 1X107 nuclei, labeled asdescribed above (1) was added to 1X106 unlabeled nuclei, whichwere processed further for isolation of matrices. The radioactivityrecovered from the final matrix preparation was presented as apercentage of that added initially.

§ See 'Isolation of nuclear matrix' in Materials and methods.

comigrate in agarose. An intimate, hybrid-like associationof part of the RNA component with the matrix DNA issuggested also from the fact that about 20 % of this RNA(by radioactivity) might be removed by digestion withRNase H (Table 2); it seems that such treatment does notchange the electrophoretic characteristics of the matrixDNA (Fig. 1).

The question that arises from the above results iswhether this RNA, because of complexing with DNA, isnot responsible for the observed resistance to nucleases.In order to answer this question, samples of the same 32P-labeled matrix DNA shown in Fig. 2 were first hydro-lysed with KOH and then digested with DNase I and ExoIII: the results given in Fig. 2 and Fig. 3B illustrate ausual sensitivity. Thus the nuclease resistance of thematrix DNA is due to complexing with RNA. Such aconclusion is supported strongly by the fact that themicrococcal nuclease that catalyzes the cleavage of bothcomponents disintegrates the complexes completely(Fig. 2, 3).

Table 2. Removal of RNA in the final matrix DNApreparation by treatment with RNase H and KOH

Treatment RNA left (%)•

NoRNase HKOH

10O±3.0781±3.891.2±0.09

We consider the final matrix DNA preparation to be matrix DNAthat is recovered after deproteinization and digestion with RNase A.

•Mean values of two experiments. The percentage was determinedby radioactivity.

670 C. Patriotis et al.

c

A

— 200

Fig. 5. Evidence that the resistance of matrix DNA toDNase I and Exo III is not due to complexing with proteins.In this experiment matrix DNA was run in low-meltingtemperature agarose, the zone of 200bp was excised, theDNA recovered and radiolabeled with 1 2 5 1 . The sample wasthen re-electrophoresed in 1.5% agarose and the gel wasdried and autoradiographed. Lane b shows the recovered andradiolabeled matrix DNA visualized by staining withethidium bromide and ultraviolet illumination, c. Anautoradiogram of b. Note the absence of any radiolabeledproteins in the zone of 200 bp. The arrow indicates thenegligible radioactivity retained at the start. Lane a showsDNA obtained by digestion of nuclei with micrococcalnuclease and run in parallel as a marker.

— 200

Fig. 6. Evidence that the matrix DNA is copurified with anRNA component that migrates in agarose together withDNA. The results shown represent autoradiograms of driedagarose gels in which labeled samples have beenelectrophoresed. A. For this experiment the cells wereprelabeled with [14C]uridine. Lane a, matrix DNA withouttreatment with RNase A. Lane b, the same matrix DNA butafter treatment with RNase A. Lane c, the same as in b, butbefore electrophoresis the sample was treated with KOH.Note the elimination of the label. B. Lane a, shows matrixDNA treated with RNase A, then labeled with 32P by nicktranslation and electrophoresed in parallel with the samplesshown in A. Lane b, is the same as a but after hydrolysiswith KOH. C. Lane a, a nick-translated DNA purified fromnuclei digested with micrococcal nuclease and run in parallelas a marker. A 200bp repeat is indicated.

The nature of the RNA of the complexesThe nature of the bulk of the RNA, complexed with thematrix DNA, was clarified in experiments using specificinhibition of RNA synthesis. In these experimentssamples of Friend cells were preincubated for 30min ineach of the inhibitors shown in Table 3 and the incorpor-ation of [3H]uridine in RNA associated with the matrixDNA purified through CsCl was measured for another30min in the presence of inhibitors. The incorporationinto controls in which inhibitors were not added wastaken as 100%. From the results it is clear that noinhibition occurs upon treatment with 2figm\~ <x-amanitin or 0.04/igml~ actinomycin D, which meansthat the RNA of the complexes is not transcribed byRNA polymerase I or II. This RNA is inhibited almostcompletely by lOO^gmP1 a/-amanitin, indicating that itis transcribed by RNA polymerase III and thus shouldbelong to the family of snRNAs.

Nuclease-resistant fragments might be recovered fromtotal DNAOne can argue, however, that the sheer number of smallRNAs within the nucleus might contribute to the second-

Table 3. Inhibition of tyiithesis of the RNA componentin purified DNA-RArA complexes

Treatment Inhibition (%)*

Actinomycin D (0.04 ug mlcr-Amanitin (2//gml~ )(T-Amanitin (100/<gml~')

11.2±0.718.9±0.46

%.5±1.S2

In this experiment samples of Friend cells were preincubated for30 mm in each of the above inhibitors and the incorporation of[3H]undine into RNA of complexes, purified in CsCl gradients, wasmeasured for another 30min in the presence of inhibitor. 100%synthesis was taken to be the incorporation of labeled precursor intoRNA of two control samples in which inhibitor was not added.

•Averages of two experiments.

ary hybridization of these molecules to the matrix DNAduring the isolation procedure. In such a case it isimportant to clarify whether there is a pre-existingassociation between RNAs and matrix DNA. Since theusual deproteinization procedure does not cause a sec-ondary hybridization we have to look for a pre-existingassociation in total DNA, deproteinized from freshlyisolated nuclei. When such total, RNase-treated DNA

DNA-RNA complexes 671

S P Table 4. Changes in the percentage of the matrix DNAof proliferating and induced Friend cells under different

conditions of matrix preparation

Fig. 7. Dot-blot hybridization illustrating the similarity ofthe nuclease-resistant matrix DNA sequences to thesequences isolated from the base of dehistonized DNA loops.For this experiment the matrix DNA was redigested withDNase I and then the recovered fragments were nicktranslated and hybridized to P and S fractions. 1 and 2 markthe slots loaded with 5 and lOjug unlabeled DNA,respectively.

was digested for 30 min at 37 °C with 50 ̂ g ml~' DNase I,from the treated samples we recovered, by ethanolprecipitation, resistant fragments with lengths compar-able to that of the matrix DNA (Fig. 1). The amount ofthis resistant DNA represents 3.23 % of the total. If suchDNA is further subjected to dot-blot hybridization, itbehaves in a similar way to the originally isolated matrixDNA (Fig. 4).

The complexes might represent transient attachmentsites of nuclear DNA to the matrixSince the resistant DNA fragments are recovered frommatrices, it is reasonable to believe that they representattachment sites of nuclear DNA to the matrix. Thisassumption was confirmed in hybridization experimentsillustrated in Fig. 7, in which isolated matrix DNA wasredigested with DNase I and the resistant fragments werelabeled and then hybridized to P and S fractions,obtained as described in Materials and methods. Briefly,P fraction is enriched in sequences from the base of DNAloops released by treatment of isolated nuclei with highsalt, while S fraction is enriched in DNA from the freeaxis of the loops. From the hybridization signals obtainedwe can conclude that there is a great extent of similaritybetween the nuclease-resistant matrix sequences andthose isolated from the base of the loops.

The observation that some of the matrix fragments arecomplexed with RNA suggests that some of the DNAloops might be attached to the matrix structures viaRNA. Previous data have clarified that the intensivenuclear matrix is composed of peripheral lamina—porecomplexes, internal fibrogranular network and nucleolarmatrix (Bouteille et al. 1983; see also Fig. 8A). A numberof investigators have specified, moreover, that both thefibrogranular network and the nucleolar matrix containRNA as a component (Herman et al. 1978; Long et al.1979; Van Eekelen et al. 1981; Werner etal. 1988; Harrisand Smith, 1988). The following data support the notionthat a significant number of attachment sites are locatedat these structures. When matrices are isolated by simul-taneous digestion of nuclei with DNase I and RNase A,the percentage of the matrix-associated DNA is dramati-cally reduced (Table 4). Since this reduction is not

Preparation

1. Matrices from proliferating cellsisolated according to the conventionalprocedure

2. The same as in 1 but during DNase Idigestion RNase A was added

3. Matrices from induced cells isolatedaccording to the conventional procedure

4. Residual structures obtained fromproliferating cells by direct extractionof nuclei with 2 M NaCl and successivedigestion with REf and DNase I

5. Residual structures as in 4 butduring DNase I digestion RNase Awas added

6. Residual structures obtained

Matrix DNA (%)•

1.01±0.12

0.10±0.03

0.17±0.05

7.22±0.49

0.5710.07

0.7310.09from induced cells as in 4

•Averages of five experiments.•(•Restriction enzyme.

accompanied by a change in the size of DNA (notshown), it might be supposed that RNase A digests thesupport for a significant number of attachment sites. Thisassumption is in accord with parallel ultrastructuralstudies (Fig. 8) showing in simultaneously digested nu-clei a diminution of the internal fibrogranular and nu-cleolar components.

One of the intriguing questions in our study was todetermine how stable are the attachment sites realized bythe DNA-RNA complexes upon profound reorganiz-ation of the nucleus. This point was checked in inducedFriend cells, committed to erythroid differentiation.Under our conditions of induction, 72 h after treatmentwith dimethylsulfoxide the percentage of B + cells is morethan 72%. When matrix DNA was isolated from suchinduced cells its amount in comparison to that of control,noninduced cells was reduced almost five times(Table 4). When the above DNA was redigested withDNase I and Exo III practically no resistance was found,which suggests an almost complete absence of the RNAcomponent of the complexes. Moreover, we were not ableto recover complexes from differentiated cells after ultra-centrifugation in CsCl or DNase I-digested matrix DNA(not shown). Interestingly, a reduction of RNA associ-ated with matrices isolated from differentiated chickerythrocytes was reported also by Lafond and Woodcock(1983).

Discussion

In this study we describe DNA-RNA complexes inpreparations of purified matrix DNA. It might besupposed on the basis of the resistance to DNase I andExo III that a significant portion of the matrix DNA isorganized in such complexes (Figs 2, 3). At present it isnot possible to give a detailed description of their

672 C. Patriotis et al.

8A

j

>

_ ..̂

The presence of such unusual structures is confirmedby the following evidence. First, when a matrix DNAsample is electrophoresed in agarose an RNA componentalways migrates together with the DNA (Figsl , 6).Second, when matrix DNA is digested with DNase I orExo III most of the fragments exhibit resistance; theirsusceptibility is restored after treatment with KOH,removing the RNA component (Figs 2, 3). Third, asmight be predicted, incubation of the complexes withmicrococcal nuclease digests them completely (Figs 2, 3).

Since we have found the DNA-RNA complexes inDNA isolated from matrices we believe that they rep-resent attachment sites of nuclear DNA to the matrix.This supposition is supported by dot-blot hybridizationexperiments showing a great extent of similarity betweenthe sequences associated with the complexes and thoseisolated from the base of loops released by high-saltextraction of nuclei (Fig. 7). Further, we have presentedevidence suggesting that most of the sites of attachmentof nuclear DNA are located in the internal compartmentsof the matrix (Table 4, Fig. 8). Finally, by usingerythroid differentiation in Friend cells as a model systemwe have shown that the attachment sites realized by theDNA-RNA complexes are of the transient, facultativetype.

The results from our study confirm and further extendthe recent data from Penman's group (Fey et al. 1984;Nickerson et al. 1989). They suggest not only animportant role for RNA in matrix organization but also inattachment of the loop-units of nuclear DNA to theinternal structures of the matrix.

We thank Ms Totka Glavcheva and Vania Zoncheva fortechnical assistance, and Zdravko Apostolov for preparation ofthe manuscript.

Fig. 8. Electron microscopy of isolated matrix structures. Ashows a residual matrix isolated according to the conventionalprocedure. The main matrix components, peripherallamina-pore complexes, internal fibrogranular network andnucleolar matrix are clearly seen. B shows the characteristicreduction of both the internal fibrogranular network and thenucleolar matrix upon simultaneous digestion of the nucleiwith DNase I and RNase A. Bar, 1 /im.

organization. The DNA component represents shortfragments with a mean size of 200 bp (Figs 1, 3, 5, 6). Itseems that these fragments are at least partially double-stranded because Si nuclease does not digest them (notshown) and, in addition, they might be labeled efficientlyby nick translation (Figs 2, 3, 6). Since the complexessurvive the classical phenolization procedure, it might beconcluded that the linkage between the two componentsis strong. Some experiments with RNase H and RNase Aindicated that these structures are not organized asperfect DNA-RNA hybrids, nor as a simple DNA-RNAcomplexes (Table 2, Fig. 6). Specific inhibition of RNAsynthesis with cr-amanitin and actinomycin D has demon-strated that the bulk of the RNA component is tran-scribed by RNA polymerase III and thus should belongto the family of snRNAs (Table 3).

References

ALLECRA, P., STERNER, R., CLAYTON, D. F. AND ALFREY, V. G.

(1987). Affinity chromatographic purification of nucleosomescontaining transcriptionally active DNA sequences, jf. molec. Biol.196, 379-388.

ANDERSON, L. A. AND MCCLURE, W. O. (1973). An improvedscintillation cocktail of high solubilizing power. Analvt. Biochem.53, 173-179.

BASLER, J., HASTIE, N. D., PIETRAS, D., MATSUI, S., SANDBERG, A.

AND BEREZNEY, R. (1981). Hybridization of nuclear matrixattached deoxyribonucleic acid fragments. Biochemistry 20,6921-6929.

BEREZNEY, R. (1984). In Chromosomal Xonhistone Proteins, vol. 4(ed. L. E. Hnilica), pp. 119-180. CRS Press, Boca Raton.

BEREZNEY, R. AND BASLER, J. (1982). Nuclear matrix organizationand DNA replication. In The Nuclear Envelope and the NuclearMatrix, pp. 183-197. Alan R. Liss, New York.

BEREZNEY, R. AND BUCHHOLTZ, C. A. (1981). Dynamic associationof replicating DNA fragments with the nuclear matrix ofregernerating liver. Expl Cell Res. 132, 1-3.

BODNAR, J. \V., JONES, C. J., COOMBS, D. H., PEARSON, D. AND

WARD, D. C. (1983). Proteins tightly bound to HeLa cell DNA atmatrix attachment sites. Molec. cell. Biol. 3, 1567-1579.

BOUTEILLE, M., BouviER, D. AND SEVE, A. P. (1983). Heterogeneityand territorial organization of the nuclear matrix and relatedstructures. Int. Rev. Cytol. 83, 135-183.

BURTON, K. (1956). A study of the conditions and mechanism of thediphenylamine reaction for the colorimetric estimation ofdeoxiribonucleic acid. Biochem. J. 62, 315-323.

DNA-RNA complexes 673

COOK, P. R. AND BRAZELL, I. A. (1975). Supercoils in human DNA.J. Cell Sci. 19, 261-279.

DJONDJUROV, L., IVANOVA, E., MARKOV, D., BARDAROV, ST. AND

SACHSENMAIER, W. (1986). Is the nuclear matrix the site of DNAreplication in eukaryotic cells? Expl Cell Res. 164, 79-96.

FEY, E. C , WAN, K. M. AND PENMAN, S. (1984). Epithelialcytoskeletal framework and nuclear matrix-intermediate filamentscaffold: three-dimensional organization and protein composition.J. Cell Biol. 98, 1973-1984.

GASSER, S. M. (1988). Nuclear scaffold and higher-order folding ofeukaryotic DNA. In Architecture of Eukaryotic Genes (ed. G.Kahl), pp. 421-471. VCH Verlagsgeselschaft mbH, D-6940,Weinheim.

GASSER, S. M. AND LAEMMLI, U. (1986). The organization ofchromatin loops: characterization of a scaffold attachment site.EMBOJ. 5, 511-518.

HANCOCK, R. AND HUGES, M. (1982). Organization of DNA in theinterphase nucleus. Biol. Cell 44, 201-212.

HARRIS, S. G. AND SMITH, M. C. (1988). SnRNP core proteinenrichment in the nuclear matrix. Biochem. biophvs. Res. Commun.152, 1383-1387.

HERMAN, R., WEYMOUTH, L. AND PENMAN, S. (1978).

Heterogeneous nuclear RNA-protein fibers in chromatin depletednuclei. J. Cell Biol. 78, 663-679.

IZAURALDE, E., MlRKOVITCH, J. AND LAEMMLI, U. K. (1988).Interaction of DNA with nuclear scaffolds in vitro. J. molec. Biol.2000, 111-125.

KAUFMANN, S. H., COFFEY, D. S. AND SHAPER, J. H. (1981).

Considerations in the isolation of rat liver nuclear matrix, nuclearenvelope and pore complex-lamina. Expl Cell Res. 132, 105-123.

LAFOND, R. E. AND WOODCOCK, C. L. F. (1983). Status of thenuclear matrix in mature and embryonic chick erythrocyte nuclei.Expl Cell Res. 147, 31-39.

Loc PHI-VAN AND STRATUNG, W. H. (1988). The matrix attachmentregions of the chicken lysozyme gene co-map with the boundariesof the chromatin domain. EMBOJ. 7, 655-664.

LONG, H. B., HUANG, CH. AND POGO, O. A. (1979). Isolation and

characterization of the nuclear matrix in Friend erythroleukemiacells: chromatin and hnRNA interactions with the nuclear matrix.Cell 18, 1079-1090.

LOWRY, O. H., ROSENBROUGH, N. J., FARR, A. L. AND RANDALL,R. L. (1951). Protein measurement with the Folin phenol reagent.J. biol. Chem. 193, 256-275.

MANIATIS, T., FRTTSCH, I. AND SAMBROOK, J. (1982). In Molecular

Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory,NY: Cold Spring Harbor Laboratory Press.

MARZLUFF, W. F. AND HUANG, CH. R. (1985). Transcription ofRNA in isolated nuclei. In Transcription and Translation, aPractical Approach (ed. B. D. Hames and S. J. Higgins), pp.89-129. IRL Press, Oxford, Washington, DC.

MATSUMOTO, L. H. (1981). Enrichment of satellite DNA on thenuclear matrix of bovine cells. Nature, Land. 294, 481-482.

MiRKOvrrcH, J., GASSER, S. M. AND LAEMMLI, U. K. (1988).Scaffold attachment of DNA loops on multiphase chromosomes. ,7.molec. Biol. 200, 101-109.

NELKIN, D., PARDOLL, D. M. AND VOGELSTEIN, B. (1980).

Localization of SV40 genes within supercoiled loop domains. Nucl.Acid Res. 8, 5623-5631.

MIRKOVTTCH, J., GASSER, S. M. AND LAEMMLI, U. K. (1988).

Scaffold attachment of DNA loops in metaphase chromosomes. J.molec. Biol. 200, 101-109.

NEUER, B. AND WERNER, D. (1985). Screening of isolated DNA forsequences released from anchorage sites in nuclear matrix. J.molec. Biol. 181, 15-25.

NICKERSON, A., KROCHMALNIC, G., WAN, K. M. AND PENMAN, S.

(1989). Chromatin architecture and nuclear RNA. Proc. natn.Acad. Sci. USA. 86, 177-181.

PARDOLL, D. M. AND VOGELSTEIN, B. (1980). Sequence analysis ofnuclear matrix associated DNA from rat liver. Expl Cell Res. 128,466-470.

PARDOLL, D. M., VOGELSTEIN, B. AND COFFEY, D. S. (1980). A

fixed site of DNA replication in eukaryotic cells. 19, 527-531.PATRIOTIS, CH. AND DJONDJUROV, L. (1989). Tightly bound

DNA-protein complexes representing stable attachment sites oflarge DNA loops to components of the matrix. Eur.J. Biochem.184, 157-164.

RAZIN, S. V., CHERNOKHVOSTOV, V. V., ROODYN, A. V., ZBARSKY,

I. B. AND GEORGIEV, P. (1981). Proteins tightly bound to DNA inthe regions of DNA attachment to the skeletal structures ofinterphase nuclei and metaphase chromosomes. Cell 27, 65-73.

RAZIN, S. V., MANTIEVA, V. L. AND GEORGIEV, G. P. (1978). DNA

adjacent to attachment points of deoxynbonucleoprotein fibril tochromosomal axial structure is enriched in reiterated basesequences. Nucl. Acids Res. 5, 4737-4751.

RAZIN, S. V., MANTIEVA, V. L. AND GEORGIEV, G. P. (1979). The

similarity of DNA sequences remaining bound to scaffold uponnuclease treatment of interphase nuclei and metaphasechromosomes. Nucl. Acids Res. 7, 1713-1719.

REST, R., MULLER, M. AND WERNER, D. (1986). Disintegration of

nucleoskeletal elements by metrizamide 2M salt isopykniccentrifugation. Expl Cell Res. 167, 144-156.

SMALL, D. D., NELKIN, B. AND VOGELSTEIN, B. (1982). Nonrandom

distribution of repeated DNA sequences with respect tosupercoiled loops and the nuclear matrix. Proc. natn. Acad. Sci.U.SA. 79, 5911-5915.

TSVETKOV, ST, IVANOVA, E. AND DJONDJUROV, L. (1989). Metabolic

behaviours of the core histones in proliferating Friend cells. ExplCell Res. 180, 94-105.

VAN EEKELEN, C. A. G. AND VENROOU, W. J. (1981). HnRNA andits attachment to a nuclear protein matrix. .7. Cell Biol. 88,554-563.

VOGELSTEIN, B., PARDOLL, D. M. AND COFFEY, D. S. (1980).

Supercoiled loops and eukaryotic DNA replication. Cell 22, 79-85.WERNER, D., REST, R. AND NEUER-NITSCHE, B. (1988). In

Architecture of Eukaryotic Genes (ed. G. Kahl), pp. 449-459,VCH Verlagsgeselschaft mbH, D-6940, Weinheim.

ZEHNBAUER, B. A. AND VOGELSTEIN, B. (1985). Supercoiled loopsand the organization of replication and transcription in eukaryotes.BioEssays 2, 52-54.

(Received 23 August 1989 — Accepted, in revised form,15 January 1990)

674 C. Patriotis et al.