Mature mRNA is selectively released from the nuclear matrix by an ATP/dATP-dependent mechanism sensitive to topoisomerase inhibitors.

Ovalbumin mRNA precursors were found to be almost quantitatively associated with the hen oviduct nuclear matrix. On the other hand, only one-third of the mature ovalbumin mRNA of whole nuclei was recovered in the nuclear matrix fraction. The binding of both the high molecular weight mRNA precursors and the mature-sized mRNA to the matrix displayed no difference in stability against salt, urea, or detergents. The mature mRNA, however, was found to be released selectively from the matrix by ATP. In contrast, the mRNA precursors remained completely bound to the nuclear substructure in the presence of ATP. Detachment of mRNA from the matrix also occurred in the presence of ADP, AMP plus pyrophosphate, or ATP analogs that contain nonhydrolyzable alpha, beta and beta, gamma bonds. Contrasting with the ATP-induced effect, addition of poly(A), ethidium bromide, or the copper chelator 1,10-phenanthroline to oviduct cell matrices caused an unspecific liberation of both mature and immature ovalbumin messengers. The release of the mature mRNA by ATP was found to be strongly inhibited by both nonintercalative and intercalative inhibitors of type II topoisomerase. These results suggest that the selection of the mature mRNAs for nucleocytoplasmic transport occurs at the release stage from the matrix (i.e. before translocation through the nuclear pore) and that reactions hitherto known to cause changes in the DNA secondary structure are associated with the detachment of mRNA from the nuclear substructure.


Mature mRNA Is Selectively Released from the Nuclear Matrix by an ATP/dATP-dependent Mechanism Sensitive to Topoisomerase Inhibitors*
Heinz C. Schroder, Dieter Trolltsch, Ursula Friese, Michael Bachmann, and Werner E. G . Muller$

From the Institut fur Physwlogische Chemie, Uniuersitat, Duesbergweg, 0-6500 Mainz, Federal Republic of Germany
Ovalbumin mRNA precursors were found to be almost quantitatively associated with the hen oviduct nuclear matrix. On the other hand, only one-third of the mature ovalbumin mRNA of whole nuclei was recovered in the nuclear matrix fraction. The binding of both the high molecular weight mRNA precursors and the mature-sized mRNA to the matrix displayed no difference in stability against salt, urea, or detergents. The mature mRNA, however, was found to be released selectively from the matrix by ATP. In contrast, the mRNA precursors remained completely bound to the nuclear substructure in the presence of ATP. Detachment of mRNA from the matrix also occurred in the presence of ADP, AMP plus pyrophosphate, or ATP analogs that contain nonhydrolyzable CY,@ and @,r bonds. Contrasting with the ATP-induced effect, addition of poly(A), ethidium bromide, or the copper chelator 1,lO-phenanthroline to oviduct cell matrices caused an unspecific liberation of both mature and immature ovalbumin messengers. The release of the mature mRNA by ATP was found to be strongly inhibited by both nonintercalative and intercalative inhibitors of type I1 topoisomerase. These results suggest (i) that the selection of the mature mRNAs for nucleocytoplasmic transport occurs at the release stage from the matrix (Le. before translocation through the nuclear pore) and (ii) that reactions hitherto known to cause changes in the DNA secondary structure are associated with the detachment of mRNA from the nuclear substructure.
Transport of mRNA from nucleus to cytoplasm is an ATPor GTP-dependent process which is thought to be mediated by a nuclear envelope-bound NTPase that is stimulated by poly(A) (for reviews, see Agutter, 1986;Schroder et al., 1987). This enzyme can be solubilized by Triton and purified to homogeneity (Schroder et al., 1986b). Experimental evidence, however, suggests that the NTPase and also other membranebound components of the nuclear envelope mRNA translocation apparatus cannot be solely responsible for nuclear RNA restriction since most of the nuclear RNA is still bound to membrane-depleted nuclei obtained after Triton treatment * This work was supported by Grants Schr 277/2-1 and Mu 348/7-5 from the Deutsche Forschungsgemeinschaft. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ T o whom correspondence should be addressed: Institut fur Physiologische Chemie, Abt. "Angewandte Molekularbiologie," Universitat, Duesbergweg, D-6500 Mainz, West Germany. ~ ~ (Agutter and Suckling, 1982a). It is well established that essentially all hnRNA is associated with a skeletal nuclear substructure termed nuclear matrix, nucleoskeleton, or nuclear cage (Faiferman and Pogo, 1975;Miller et al., 1978;Herman et al., 1978;Long et al., 1979;Van Eekelen and van Venrooij, 1981). This structure has been shown to be involved in important nuclear events, such as DNA replication (Pardoll et al., 1980), transcription of RNA (Jackson et al., 1981;Jackson and Cook, 1985), and RNA processing (Mariman et al., 1982a(Mariman et al., , 1982bAgutter, 1986). Actively transcribed genes, like the ovalbumin gene of chicken oviduct cells, have been shown to be associated preferentially with the nuckar matrix (Robinson et al., 1982). In addition, enzymes involved in DNA or mRNA metabolism have been found to be bound to the matrix as DNA polymerase 01 (Smith and Berezney, 1980) or poly(A) polymerase (Schroder et al., 1984). It is known that predominantly (or exclusively) mature mRNA is transported out of isolated nuclei (Jacobs and Birnie, 1982;Patterson et al., 1985). In the laying hen oviduct, the ovalbumin RNA represents the major mRNA species (Oka and Schimke, 1969). By analysis of Northern blots with an ovalbumin-specific probe, Ciejek et al. (1982) showed that the high molecular weight ovalbumin mRNA precursors are exclusively attached to the nuclear matrix, whereas only onehalf of the mature, nuclear mRNA is bound to this structure. In the present report we demonstrate that, like passage of mRNA through the nuclear pore complex, detachment of mRNA from the nuclear substructure, preceding mRNA translocation, also requires the presence of ATP. In addition, it is shown for the first time that, by ATP, only the maturesized mRNA but not the immature mRNA is released from the matrix. RNA release does not require hydrolysis of P,rphosphodiester bond and is sensitive to inhibitors of DNA topoisomerase reaction. units/mg), and bacterial restriction nucleases from Boehringer Mannheim (Mannheim, Federal Republic of Germany); a,P-methylene ATP and P,y-methylene ATP (purified by high performance liquid chromatography), cordycepin 5'-triphosphate, ouabain, quercetin, proflavine, oligomycin (65% A, 20% B, and 15% C), colchicine, podophyllotoxin, phalloidin, novobiocin, coumermycin A,, nalidixic acid, ethidium bromide, and PMSF from Sigma; DNase I (bovine pancreas, LSOO 02172) from Worthington; etoposide from Bristol Arzneimittel (Bergisch Gladbach, FRG); nitrocellulose sheets (BA 85) from Schleicher and Schuell (Dassel, FRG); agarose (standard low M,, 162-0100) from Bio-Rad; messenger-activated paper from Medac (Hamburg, FRG); and Kodak X-Omat XR-5 x-ray film from Eastman Kodak. m-AMSA (NSC 249992-0) was obtained from the Drug Synthesis and Chemistry Branch, Division of Cancer Treatment, National Cancer Institute (Bethesda, MD). Plasmid pBR322 DNA was isolated from the bacteria by lysis with SDS and by CsCl equilibrium gradient sedimentation in the presence of ethidium bromide according to Maniatis et al. (1982). Herring sperm DNA, isolated according to the method of Zahn et al. (1962), was a gift of H. Mack (Illertissen, FRG).
Cell Culture"L5178y mouse lymphoma cells were grown in Eagle's minimum essential medium supplemented with 10% horse serum in spinner culture (Muller et al., 1975). Cultures of 500 ml were initiated by inoculation of 2 X 10' cells/ml and incubated for 48 h; after this period, the cell concentration was 5.2 X 106/ml. Subsequently, the cells were collected by centrifugation (700 X g; 5 min; 37 "C) and resuspended in 320 ml of culture medium at a concentration of 3.9 X lo5 cells/ml. 2 pCi of [5,6-3H]uridine/ml then were added, the cultures were incubated in roller tubes at 37 "C for 0.5 h, and then the cells were harvested.
Preparation of Nuclei and Nuclear Matrices-Nuclei were isolated of Blobel and Potter (1966) except that 1 mM PMSF and 5 mM 2-from L-cells or the oviducts of mature egg-laying hens by the method mercaptoethanol were included in all the buffers used. Membranedepleted nuclei were prepared by treatment of isolated nuclei for 10 min at 0 "C with 2% Triton X-100. The nuclei then were washed with incubation medium without ATP. During this procedure, nuclear envelope-associated NTPase, providing the energy for nucleocytoplasmic RNA translocation, is removed (Schroder et al., 1986b). Nuclear matrices were prepared as described by Comerford et al. (1987). The nuclear matrices were stored at -20 "C in 50 mM Tris-HCl, pH 7.4, 5 mM MgC12, 1 mM EGTA, 250 mM sucrose, and 1 mM PMSF. Isolation of Total RNA and Poly(A)+ RNA-Total RNA was extracted from the nuclei, nuclear matrices, and supernatants by the method of Cathala et al. (1983). Poly(A)+ RNA was isolated from total RNA by binding to messenger affinity paper as described earlier (Messer et al., 1986). The concentration of RNA was determined spectrophotometrically (1 A,,, = 37 pg/ml).
Northern Blot Analysis-Probes were prepared from the plasmid pOV9.8, consisting of the complete chicken ovalbumin gene (9.5-kb DNA fragment extending from -1.5 kb 5' upstream of the leader to -1 kb 3' downstream of the end of exon 7) cloned into the BamHI site of the plasmid pBR322 (Breathnach et al., 1978). The 9.5-kb fragment, excised with BamHI from the pOV9.8, was labeled by nick translation with [w3'P]dTTP to a specific activity of 5-6 X lo7 cpm/ pg DNA according to the method of Rigby et al. (1977). RNA samples were denatured at 56 "C for 30 min with 6% formaldehyde in electrophoresis buffer (40 mM MOPS, 10 mM sodium acetate, 1 mM EDTA, pH 7.2) containing 50% dimethyl sulfoxide (Goldberg, 1980). Electrophoresis on 1.1% agarose slab gels containing 6% formaldehyde, blottransfer to nitrocellulose, and hybridization with the 32P-labeledprobe were performed according to Maniatis et al. (1982). Prehybridization was done at 42 "C for 12 h in 5 X SSPE (SSPE = 180 mM NaCl, 10 mM sodium phosphate, 1 mM EDTA, pH 7.4), 5 X Denhardt's solution, 50% formamide, 0.1% SDS, 100 pg/ml of heat-denatured and sheared herring sperm DNA, and 1 pg/ml of poly(A). Hybridization was done at 42 "C for 72 h in the same solution which contained, in addition, the nick-translated probe. The filters then were washed in 2 X SSC containing 0.1% SDS (25 "C, 30 min) followed by 0.2 X SSC containing 0.1% SDS (60 "C, 1 h), dried, and exposed to Kodak XAR-5 x-ray film with an intensifying screen for 2-6 days at -70 "C.
Assay for RNA Release-Nuclear matrices (22 pg of protein) or membrane-depleted nuclei (4.6 X lo6 nuclei) were incubated in a final volume of 100 p1 with 50 mM Tris-HC1, pH 7.5, 10 pM ATP, 1 mM MgC12, 100 mM KC1, 250 mM sucrose, and 1 mM PMSF at 22 "C for 30 min. The mixture then was centrifuged (20,000 X g, 10 min), and the acid-precipitable or poly(U) paper binding radioactivity released into the supernatant was determined as described earlier (Schroder et al., 1984(Schroder et al., , 1986aMesser et al., 1986). For determination of the influence of effectors on RNA release, the reaction mixture was preincubated with the effector in the absence of ATP for 10 min at 22 "C, and the ATP then was added. Sucrose Gradient Centrifugation-RNA samples were dissolved in a small volume of water, denatured in 25% dimethyl sulfoxide, 50% dimethylformamide, 10 mM EDTA, and 0.2% SDS for 2 min at 65 "C (Dubroff and Nemer, 1975) and then applied to 15-30% isokinetic sucrose gradients containing 30 mM Tris-HC1, pH 7.4, and 0.2% SDS. Gradients were centrifuged in a Beckman SW 40 Ti rotor at 285,000 X g for 14 h at 20 "C and then fractionated.
DNA Topoisomerase Assay-DNA topoisomerase types I and I1 were assayed by measuring the relaxation of supercoiled DNA according to Osheroff et al. (1983) with the following modifications. The standard assay mixture contained in a final volume of 20 pl, 0.6 pg of negatively supercoiled circular pBR322 DNA (about 90% of the DNAs were fully supercoiled and about 10% nicked circles) in 10 mM Tris-HC1, pH 8.0,l mM ATP, 5 mM MgC12,50 mM KCl, 50 mM NaCl, and 15 pg/ml of bovine serum albumin. For determination of topoisomerase I activity, ATP and MgC1, were omitted. Before adding to the reaction mixture, nuclear matrices were washed in 10 mM Tris-HCl, pH 7.2, 50 mM NaCl, 0.2 mM dithiothreitol, 0.1 mM EDTA, 0.5 mg/ml of bovine serum albumin, and 10% glycerol. Reactions were performed at 30 "C for 0-60 min and were terminated by adding 10 mM EDTA and 0.1% SDS (final concentrations). Proteins were then digested with proteinase K (50 pg/ml) at 37 "C for 20 min. After adding 10 pl of 100 mM Tris-borate, pH 8.0, 25 mM EDTA, 7.4% sucrose, 0.1% SDS, and 0.1% bromphenol blue to the samples (25 pl), samples were incubated at 70 "C for 5 min and subjected to electrophoresis on a 1.0% agarose gel in 100 mM Tris-borate, pH 8.0, 2 mM EDTA at 5 V/cm. Gels were stained for 10 min with 1 pg/ml of ethidium bromide and DNA bands visible by transillumination with UV light were photographed using a Polaroid type 665 positive/ negative film.
Nuclear matrix-associated NTPase activity was measured essentially as previously described for nuclear envelope-bound NTPase (Schrder et al., 1986b). For determination of substrate specificity, tritium-labeled nucleotides (40 dpm/pmol) were used instead of [y-32P]ATP. Reaction products were analyzed on polyethyleneimineimpregnated cellulose (F1440/LS254 from Schleicher and Schuell) as described (Schroder et al., 1986b). The same procedure was also used for detecting myokinase (ATP-AMP phosph0transferase)-like activity. In this case, ATP was replaced by [3H]ADP (120 dpm/pmol) in the reaction mixture for assaying NTPase.
Analytical Methods-Protein was measured as described by Lowry et al. (1951) using bovine serum albumin as a standard.

Association of hnRNA and mRNA with Nuclear Matrices-
For preparation of nuclear matrices, a procedure was chosen that avoids aggregation of material by detergent extraction, artifactual precipitation of protein by calcium ions, and proteolysis by serine proteinases during isolation. Using the procedure of Comerford et al. (1987), which was performed in the continuous presence of EGTA and PMSF, a matrix preparation was obtained which contained about 55% of the rapidly labeled RNA from isolated L5178y mouse lymphoma cell nuclei (about 4.4 X lo4 dpm/106 nuclei; for conditions of prelabeling, see "Experimental Procedures"). In contrast to Ciejek et al. (1982), we found it unnecessary to perform the isolation of the nuclear matrices at -20 "C. When rapidly labeled RNA was isolated from L-cell matrices and analyzed on sucrose gradients, a sedimentation profile was found very similar to that obtained with RNA extracted from whole nuclei (Fig. 1). More striking was the observation that when oviduct cell matrices were used and the matrix-bound RNA was analyzed by Northern gel blots, no degradation of the matrix-bound ovalbumin RNA was detected (Fig. 2).
To study the distribution of a specific mRNA (ovalbumin) and its precursors among the matrix and nonmatrix nuclear fractions, we prepared nuclear matrices from laying hen oviduct and analyzed Northern blots of the extracted and elec- nitrocellulose. Blots were analyzed by hybridization with "P-labeled, cloned pOV9.8 (containing the complete chicken ovalbumin gene). The autoradiographs of the RNA blots after hybridization with the "P-labeled pOV9.8 probe are shown. Lane a, 2.4 pg of total RNA extracted from whole nuclei; lane b, 1.5 pg of total RNA from untreated nuclear matrices; lane c, 1.1 pg of total RNA from the nuclear matrix pellets obtained after incubation with 10 p~ ATP; lane d, 1.2 pg of total RNA from the supernatant obtained after incubation of nuclear matrices with 10 p~ ATP. The size of the RNAs were estimated by the use of EcoRI and Hind111 restriction fragments of phage X DNA as size standards. The kilobase values on the right refer to the lengths of the ovalbumin sequence containing RNAs. Besides the 2.0-kb RNA band, representing the mature ovalbumin mRNA, at least three ovalbumin sequence-containing bands of higher molecular weights (3.3, 5.0, and 7.9 kb) are visible in the autoradiographs which are apparently the three most abundant ovalbumin mRNA processing intermediates and were previously identified by Chambon et al. (1979) and Tsai et al. (1980). trophoretically separated RNA by hybridization to an ovalbumin-specific probe (nick-translated 9.5-kb ovalbumin gene DNA cloned into pBR322). Using this probe, the mature 2.0kb ovalbumin mRNA as well as high molecular weight ovalbumin mRNA precursors, including the 7.9-kb primary transcript, were detected in the electrophoresed and blot-transferred total RNA from nuclear matrices (Fig. 2). Densitometric evaluation of' the autoradiographs revealed that only about 30% of the mature mRNA is retained on the nuclear matrix after extraction of nuclei (compare lane a for whole nuclear RNA and lane b for matrix-bound RNA). The supernatant fractions obtained during matrix isolation were found to be apparently free of ovalbumin mRNA precursors as analyzed by Northern gel blots (data not shown). These findings confirm the results of Ciejek et al. (1982) showing that the nuclear ovalbumin mRNA precursors are nearly quantitatively associated with the nuclear matrix fraction.
As shown in Table I, the binding of both the hnRNA and mRNA to the nuclear matrix is very stable. The association of rapidly labeled RNA with the L-cell matrix was not affected by treatment with 4 M urea, 3 M NaCl, 50 mM EDTA, or 2% Triton X-100.
Release of Mature mRNA from Nuclear Matrices-Efflux of mRNA from isolated nuclei has been demonstrated to occur only after adding ATP to the external transport medium (for description, see Agutter et al., 1976;Schroder et al., 1984). Because nuclear matrices (Ciejek et al., 1982) as well as nuclei (Stuart et al., 1977) still contain most nuclear RNA including mature messengers, we investigated whether the release of RNA from these structures also requires the presence of ATP. To avoid depolymerization of actin-containing filaments which have been shown to be a component of the L-cell nuclear matrix (Nakayasu and Ueda, 1984) we performed our studies under incubation conditions favoring actin polymerization (see "Experimental Procedures"). It was indeed found that already small amounts of ATP were able to cause detachment of rapidly labeled RNA from mouse L-cell matrices ( Table 11). Half-maximal release of RNA was determined to occur at a concentration of MgATP of 0.14 PM (Fig. 3). The time course of RNA release from L-cell matrices is shown in Fig. 4. Under the assay conditions used, about 80% of the released RNA was determined to be able to bind to poly(U) paper (messenger-activated paper) and thus seems to consist of poly(A)+ RNA. Most important was the finding that, after incubation of oviduct cell nuclear matrices with ATP, only the mature-sized ovalbumin mRNA was released from the matrix (Fig. 2, lane d ) , whereas the immature mRNA precursors remained bound to this structure (lane c). Consistent  with this result, the bulk of the rapidly labeled RNA that was released from L-cell matrices in the presence of ATP sedimented in a range from 8 to 20 S (Fig. l ) , indicating that it is also composed primarily of mature mRNA species. As can be calculated from Fig. 4, about 20% of the labeled, matrixassociated RNA was maximally released in the presence of 10 p~ ATP; the remainder which is not detached from the matrix under these conditions may represent high molecular weight mRNA precursors (labeled material above 20 S; see  The reaction exhibited a relatively broad pH optimum (pH 6-7.5; Tris buffer). As shown in Table 11, dATP can partially substitute for ATP. GTP was less active, and the pyrimidine nucleotides UTP and CTP displayed no significant effect. In further experiments, RNA release from nuclear matrices of L-cells was determined in the presence of compounds known as inhibitors of NTPase, protein kinases, or phosphoprotein phosphatases. It was found that quercetin (20 pg/ml), proflavine (20 pg/ml), phalloidin (20 pg/ml), oligomycin (50 pg/ ml), podophyllotoxin (100 pg/ml), and ouabain (500 pg/ml) markedly inhibit the release of rapidly labeled RNA from Lcell matrices to 22.2, 16.2, 51.4, 38.1, 36.9, and 15.1%, respectively (control = 100%). Some inhibition was also obtained in the presence of colchicine (70.1% at 800 pg/ml). The latter compound has previously been shown to decrease RNA efflux from isolated, whole nuclei by a mechanism that seems not to involve the nuclear envelope NTPase (Agutter and Suckling, 1982b). Addition of 1 mM NaF resulted in a 38.4% inhibition of the reaction.
Because translocation of mRNA across the nuclear envelope requires hydrolysis of ATP by a nuclear envelope NTPase (Agutter et al., 1976;Schroder et al., 1986bSchroder et al., , 1987, the question arises of whether RNA release from the matrix occurs by a similar mechanism. The nuclear matrix preparation obtained by the procedure of Comerford et al. (1987) indeed contained a significant NTPase activity. The nuclear matrix NTPase, however, seems not to be identical with the nuclear envelope enzyme, particularly because of its inability to be stimulated by poly(A) (data not shown 47.4%; Cu2+, 47.2%; and Ba'+, 15.7%). Maximum activity was obtained with a ratio of Mg2' to ATP of 1:l and pH between 7.5 and 9.5. The enzyme activity was sensitive to inhibition by quercetin (inhibition to 89.2% of control at a concentration of 20 pg/ml), proflavine (68.5%; 20 pg/ml), phalloidin (61.9%; 20 pg/ml), oligomycin (30.7%; 50 pg/ml), and ouabain (29.0%; 500 pg/ml). Colchicine (800 pg/ml) and EGTA (2 mM) had no effect on enzyme activity, contrary to their effects on ATPpromoted RNA release from the matrix (see above). Preliminary results suggest that the nuclear matrix-associated NTPase activity is attributed to at least two different enzymes, as we concluded from the differential effects of inhibitors on the hydrolysis of different nucleotides. The properties of the ATP-splitting enzyme activity partially resemble those of the ATPase/dATPase identified by Berrios et al. (1983) in the Drosophila nuclear matrix pore complex lamina fraction.
In order to prove whether RNA detachment from the matrix is caused by matrix-associated ATPase, ATP was replaced by nonhydrolyzable ATP analogs in the incubation mixture. It was found that RNA release is also achieved by a,P-methylene ATP or P,y-methylene ATP (Table  11). The apparent K,,, values for both analogs were significantly higher than that for ATP; the apparent K,,, for a,P-methylene ATP was 10.1 p~ and that for P,y-methylene ATP 20.4 p~ (determined under standard conditions in the presence of an equimolar amount of Mg2+). As shown in Table 11, ADP and pyrophosphate also promoted RNA release to a significant extent. Because the nuclear matrix preparation used in our studies seems to contain no myokinase-like active (no detectable [3H] ATP after different incubation periods with [3H]ADP under the assay conditions used, in accordance with the results of Clawson and Smuckler, 1982), a conversion of ADP to AMP and ATP cannot be responsible for the release-promoting activity of ADP. AMP had only a little effect, and adenosine and adenine were ineffective. However, in the presence of an equimolar amount of pyrophosphate, AMP exhibited a higher efficacy than ADP. In these properties, RNA release from the matrix is clearly distinguished from RNA efflux from isolated nuclei. Translocation of mRNA through the nuclear pore does not occur in the presence of nonhydrolyzable ATP analogs or ADP (Yu et al., 1972;Agutter et al., 1976; and our own results; for contrary results, see "Discussion"). It therefore seems unlikely that nuclear matrix-associated ATPase (or protein kinase) activities are essential for mRNA detachment from the matrix structure. Moreover, the matrix-associated ATPase activity was insensitive to topoisomerase inhibitors (except m-AMSA; data not shown), contrary to ATP-induced mRNA detachment (see below).
As also shown in Table 11, the nucleotide requirements for RNA release from the nuclear matrix essentially parallel those for the release of RNA from Triton-treated nuclei. RNA release from membrane-depleted nuclei also seems not to require ATP cleavage, in contrast to RNA efflux from intact, untreated nuclei (cf. Agutter et al., 1976;Bachmann et al., 1984;Schroder et al., 1984Schroder et al., , 1986a. Triton treatment has been shown to remove the nuclear envelope-bound NTPase (Smith and Wells, 1984;Schroder et al., 1986b).
Interestingly, phalloidin is able to inhibit the release of RNA by ATP (see above). Moreover, it was found that cytochalasin is able to cause RNA liberation from the nuclear matrix also in the absence of ATP.' These results may indicate participation of actin-containing filaments in RNA attachment to the nuclear matrix. Actin has indeed been found in the nuclear matrix from oocytes (Clark and Merriam, 1977; H. C. Schroder, D. Trolltsch, R. Wenger, M. Bachman, and W. -E. G. Muller, submitted for publication. Gounon and Karsenti, 1981) and also in that from mouse Lcells (Nakayasu and Ueda, 1984). Nakayasu and Ueda (1985) reported that rapidly labeled RNPs are released from the Lcell nuclear matrix under conditions which cause depolymerization of actin filaments. Using the buffer system applied by these authors, we also obtained release of labeled RNA. However, the nucleotide specificity and kinetics of this process are different from those observed under the moderate salt conditions used by us, which favor actin polymerization. The release obtained under low salt conditions was found to be highly specific for ATP (no release with UTP, CTP, GTP, dATP, or dTTP) and to occur already at 0 "C, whereas that under moderate salt conditions also takes place in the presence of GTP, dATP, or dTTP and requires elevated temperatures. In our hands, bound RNA was also liberated to a marked extent when the low salt Tris-dithiothreitol buffer was used without ATP, Ca", and EDTA, whereas almost no release of RNA was found with the ATP-depleted incubation medium used by us. Although these results suggest that actin is involved in matrix association of RNA, from the properties of RNA release observed under our conditions we assume that further ATP-dependent reactions participate in RNA binding and/or release from the nuclear matrix.
Influence of Poly(A), Topoisomerase Inhibitors, and Metal Chelators on RNA Release-The 3' poly(A) segment and the double-stranded regions of the mRNA and its precursors have been assigned a potential role in mRNA binding to the nuclear matrix (Herman et al., 1978). Recently, we were able to solubilize and purify two polypeptides of 55 and 64 kDa from rat liver nuclear envelopes that bind to poly(A) in an ATPlabile linkage.' Interestingly, RNA release from the matrix was also achieved in the presence of poly(A), whereas other polynucleotides were ineffective (Table 111). However, the poly(A)-caused effect was less specific than that caused by ATP since both mature and immature ovalbumin RNAs were found to be detached from the matrix (Fig. 5, lane d ) . Doublestranded poly(A) .poly(U) displayed no effect (Table 111).
Some drugs intercalating with double-stranded DNA (e. g. ethidium bromide, 25 p~) also promoted liberation of matrixbound RNA in the absence of ATP (49% of the amount that is released in the presence of 10 p~ ATP within a 30-min incubation period under standard assay conditions; Fig. 5, lanes e and f), whereas others (e. g. actinomycin D, 10 gg/ml) were ineffective. Among the intercalating drugs interfering with mRNA binding to the matrix, those known as inhibitors of type I1 DNA topoisomerase may be important (Table IV and Fig. 6). The fact that low concentrations of nonintercalating topoisomerase I1 inhibitors (etoposide and nalidixic acid) also affect RNA release from the matrix points partic-

Inhibition of ATP-induced release of rapidly labeled RNA from nuclear matrices and from membrane-depleted nuclei by topoisomerase inhibitors
Triton-treated L-cell nuclei or nuclear matrices were preincubated a t 22 "C for 10 min in the absence or presence of inhibitors of the topoisomerase reaction using the standard incubation medium without ATP. After addition of 10 p~ ATP, incubation was continued for 30 min, and the radioactive RNA left in the supernatant was determined. Mean values of four determinations are given. The S.D. was less than 9%. ularly to an involvement of a topoisomerase 11-like activity in RNA detachment. As shown in Fig. 7, nuclear matrices from hen oviduct indeed contain a substantial amount of type I1 and a less significant amount of type I DNA topoisomerase activity as measured by the relaxation of supercoiled pBR322 plasmid DNA in the presence or absence of ATP. In the presence of ATP, the tested drugs inhibited the ATP-induced release of both rapidly labeled RNA (Table IV) and mature ovalbumin mRNA (Fig. 6) from the matrix (similar results were obtained with membrane-denuded nuclei; see also Table  IV). The lower inhibition measured for the release of rapidly labeled RNA, compared to ovalbumin RNA, may be due to the fact that the prelabeled material in L-cell matrices does not exclusively consist of mRNA but also contains minor amounts of ribosomal RNA, tRNA, snRNA, and their precursors, which have been shown also to be partially matrix-bound Nuclear matrices were preincubated in the ATP-depleted release mixture and then incubated with ATP as described under "Experimental Procedures." Total RNA was extracted from both the matrix pellets (lanes a, c, e, g, i, and k ) and the incubation supernatants (lunes b, d, f , h, j , and 1) and analyzed as described in Fig. 2. Lanes a  and b, 8.3 and 7.9 pg of RNA from nuclear matrices preincubated with 40 pg/ml of m-AMSA; lanes c and d, 9.7 and 9.2 pg of RNA from nuclear matrices preincubated with 40 pg/ml of etoposide; lanes e and f , 10.1 and 9.3 pg of RNA from nuclear matrices preincubated with 200 pg/ml of nalidixic acid; lanes g and h, 8.2 and 7.9 pg of RNA from nuclear matrices preincubated with 200 pg/ml of novobiocin; lanes i and j , 10.3 and 9.2 pg of RNA from nuclear matrices preincubated with 100 pg/ml of coumermycin A,; lanes k and 1,12.2 and 11.3 pg of RNA from nuclear matrices preincubated without inhibitor (control).  (Ciejek et al., 1982). Liberation of rapidly labeled RNA in the presence of P,y-methylene ATP displayed the same responsiveness against inhibitors of type I1 topoisomerase as that in the presence of ATP (data not shown).

RNA released
A significant portion of rapidly labeled RNA from mouse L-cell matrices was found also to be solubilized in the presence of the copper chelator 1,lO-phenanthroline. In the presence of 1 mM phenanthroline, 33.1% of the amount of radioactivity released in the presence of 10 PM ATP was liberated (standard incubation conditions). The effect of phenanthroline on RNA matrix attachment was found to be partially reversible after addition of 100 PM CuC12 (by 64.2%) or 100 p~ CaC12 (by 20.1%). Addition of EDTA did not result in a significant release of bound RNA (Table I). The phenanthroline-caused release of ovalbumin RNA from oviduct cell matrices exhibited no specificity for mature mRNA molecules (Fig. 6, lanes  g and h), like that caused by poly(A) (lanes c and d ) or ethidium bromide (lanes e and f ) .

DISCUSSION
Nucleocytoplasmic transport of mRNA can be subdivided into the following steps (Agutter, 1986;Schroder et al. 1987): (i) release of mRNA from the internal nuclear matrix structure, (ii) translocation of mRNA through the nuclear pore complex, and (iii) binding of the transported mRNA to cytoplasmic cytoskeletal elements. Current evidence suggests that, during all of these steps, mRNA never appears in a freely diffusible form. Transport of mRNA seems most likely to involve sequential attachment and detachment processes at specific binding sites of the nuclear matrix, the nuclear envelope (pore-complex lamina), and the cytoskeleton. At the stage of translocation, this binding site may be the poly(A)recognizing transport carrier within the nuclear envelope structure (Bachmann et al., 1984;Schroder et al., 198613). At the cytoplasmic site, microtubules, actin filaments (Schroder et al., 1982), and/or intermediate filaments (Bachmann et al., 1986) might participate in poly(A)+ mRNA metabolism and transport. In order to elucidate the part that the nuclear matrix plays during overall nucleocytoplasmic mRNA transport, the requirements for release of total mRNA as well as of a specific, high abundance mRNA (ovalbumin) from nuclear matrices were determined. All of the ovalbumin mRNA precursors were found to be associated nearly quantitatively with the nuclear matrix fraction from hen oviduct. This result confirms earlier findings of Ciejek et al. (1982). On the other hand, the portion of the nuclear, mature mRNA, that is bound to the matrix was found to be only 30%. Nevertheless, the matrix-bound, mature mRNA displayed no difference in its strength of binding as compared to its precursors when the matrices were subjected to treatments with high salt, detergents, urea, or EDTA.
It has already been established that predominantly mature mRNA species are transported out of the nucleus (Jacobs and Birnie, 1982;Patterson et al., 1985). It was, however, unclear at which step during nucleocytoplasmic transport the selection of mature mRNAs for transport occurs. In the present report we demonstrate that, by ATP, the mature ovalbumin mRNA (but not its precursors) is specifically released from the oviduct cell nuclear matrix. RNA release also occurred in the presence of ATP analogs that contain nonhydrolyzable a,@ or P,r bonds. Therefore, ATP hydrolysis by nuclear matrix-bound NTPase (or phosphorylation of proteins by a nuclear matrix-associated protein kinase) seems not to be required for dissociation of RNA from the matrix.
Moreover, the properties of the nuclear matrix-associated NTPase activity (which is obviously different from the nuclear envelope NTPase implicated in mRNA translocation across the nuclear envelope) do not sufficiently correspond to those of the RNA release from the matrix, especially with respect to nucleotide specificity and response to some inhibitors (e. g. topoisomerase inhibitors). The nuclear matrix NTPase activity described by us also differs in some properties from the enzyme described by Clawson and Smuckler (1982). This may be due to the presence of a DNA topoisomerease I1 activity in our preparation, in accordance with the results of Berrios et al. (1985). A possible participation of nuclear matrix-associated poly(A) polymerase (Schroder et al., 1984) in the release of RNA is excluded by the fact that the release also occurred in the presence of a,P-methylene ATP which does not serve as a substrate for poly(A) polymerase (Moore and Sharp, 1985). The detachment of mRNA from the matrix also cannot be caused by a simple chelating effect of ATP on divalent cations, since EDTA or some other nucleotides (UTP and CTP) did not release a significant amount of RNA. Therefore, we assume that the release of RNA is caused by a conformational change of a nuclear matrix (or mRNP) component induced by ATP (or its derivatives) without cleavage of any high energy bond. However, the possibility cannot be excluded that ATP hydrolysis is required to restore the ability of this component to interact with another mRNA molecule again. ADP was also able to detach the mRNA from the matrix but with a lower efficacy than ATP. The effect of ADP cannot be explained by a myokinasecatalyzed conversion of ADP in AMP and ATP, since no such enzyme activity could be detected in the matrix preparation used (unlike the situation in the envelope; cf. Clawson et al., 1978;Agutter, 1980). AMP had only a slight effect on RNAmatrix association, but addition of an equimolar amount of pyrophosphate made the monophosphate form of adenosine almost as active as the higher phosphorylated forms. Adenosine itself and the free base (adenine) were ineffective. GTP could partially substitute for ATP, but not UTP and CTP. From these results, we conclude that the nucleotide binding site involved in RNA release recognizes both the base moiety (which must be a purine) and the triphosphate moiety of the stimulating nucleotide and that the simultaneous binding of both parts of the molecule is required to cause maximum effect. Stoichiometric calculations revealed that the number of ATP molecules necessary to release half-maximally the matrix-bound mature mRNA (8.4 x 10" ATP molecules/ assay) is in the same order of magnitude as the number of nucleotides building up these mRNA molecules nucleotides/assay, estimated from the data given by Tobin (1979) and our own results). Interestingly, a similar relationship has also been postulated by Clawson et al. (1978Clawson et al. ( , 1980 for the efflux of mRNA from isolated nuclei. Some inhibitors of efflux of mRNA from isolated nuclei also inhibited RNA release from the matrix. Their effects on transport, especially those of quercetin and NaF, have previously been attributed to the inhibition of the nuclear envelope NTPase caused by them (Agutter et al., 1976;Schumm and Webb, 1978;Clawson et al. 1980). The results presented in this paper, however, suggest that the mode of action of these compounds on transport is more complicated than has been assumed and could involve other components of the mRNA translocation apparatus as well. In addition, some of these effects on whole mRNA transport could also be caused by an inhibition of nuclear envelope-associated protein kinase, phosphoprotein phosphatase, and myokinase activities by some of these agents.
Although the inhibitor studies suggest some similarities between ATP-induced release of mRNA from the matrix and ATP-dependent efflux of mRNA from whole, isolated nuclei, there are important differences. The apparent K,,, for the release of RNA from the matrix (0.14 FM with respect to MgATP) is significantly lower than the apparent K,,, for RNA efflux from isolated nuclei or for nuclear envelope NTPase; the latter values were found to be 0.28 and 0.42 mM, respectively (Agutter et al., 1976;Schroder et al., 198613). Matrixbound mRNA is significantly released only in the presence of purine nucleotides, whereas the NTPase (and the efflux mediated by it) exhibits a rather broad substrate specificity (with some preference for ATP and GTP). Release of matrix-bound mRNA does not seem to require hydrolysis of high energy phosphodiester bonds, in contrast to mRNA efflux (cf. Agutter