Immunodetecion of Poly(A) Binding Protein II in the Cell Nucleus

EXPERIMENTAL CELL RESEARCH 214, 75-82 (1994)

Immunodetection of Poly(A) Binding Protein 11 in the Cell Nucleus

SABINE KRAUSE: STAN FAKAN,t KARSTEN WEIS,:j: AND ELMAR W AHLE'"

*Biozentrum der Universität Basel, AbteiLung Zellbiologie, Klingelbergstrasse 70, CH-4056 Basel, SwitzerLand; tCentre de Microscopie Electronique, Uniuersite de Lausanne, 27 rue du Bugnon. eH-1005 Lausanne, Switzerland; and t;European Molecular BioloRY Laboratory, Postfach 102209, D-69012 Heidelberg, Germany

tail and strongly stimulates its extension. The protein is During the polyadenylation of pre-mRN A in vitro~ also involved in the mechanism that limits the length of poly(A) binding protein 11 (PAB 11) bind. to tbe growing poly(A) tails to 200-250 nucleotides [4, 6]. poly(A) taH, stimulating its extension. Tbe subcellular The proposed role of PAB 11 in mRNA polyadenyla localization of PAß 11 was investigated witb an anti tion prediets that P AB 11 is a nuclear protein. To test body affinity-purified from rabbit serum raised against this, a polyclonal antiserum was raised against purified the purified protein. Immunoßuorescence microscopy bovine PAB 11. Antibodies were affinity-purified from detected PAß 11 exclusively in tbe cell nueleus, botb in a the erude serum and used for in situ immunodetection widespread staining and in more intensely stained of PAB 11 in HeLa eells. The experiments showed the "speckles." PAß 11 was excluded from the nucleoli. ßy anticipated nuclear localization of the pro tein and an electron microscopy, PAß II was also found almost ex intranuclear distribution common to many factars in clusively in the nucleus, predominantly in clusters of volved in pre-mRNA metabolism. interchromatin granules, likely corresponding to the speckles observed by immunofluorescence microscopy, MATERIALS AND METHODS and in perichromatin fibrils, wbich represent nascent transeripts and probably tbe sites of pre-mRN A process Cell cuLture. HeLa cells were grown on glass coverslips in Joklik's ing. In addition, electron microscopy also detected P AB modified Eagle's medium supplemented with 10% fetal calf serum 11 in nucleoli. Tbe distribution corresponds largely to and antibiotics (Seromed, Biochrom KG, FRG). Cells were used at that of other factors involved in the processing of pre 60-80% conftuency. Cells were treated with drugs as described [7]. mRNA and is thus in agreement witb the proposed role For cordycepin treatment (3'-deoxyadenosine, Sigma) the cells were placed in fresh medium supplemented with cordycepin (10-S00 /.tg/ of the protein in polyadenylation. © 1994 Aeademie Press, Ine. ml). Incubation was continued for 2 h prior to fixation. HeLa cell nuclear extract was prepared according to [8). Antibodies. A New Zealand White rabbit was primed with 100 Jlg INTRODUCTION highly purified poly(A) binding protein from calfthymus [51 emuIsi fied with complete Freund's adjuvans (Difco) and Arlacel A (Sigma). Boosts were given 4, 8, and 11 weeks after the initial injection with the Two proteins have been deseribed that can bind spe same amount of protein mixed with incomplete Freund's adjuvans eifieally to the poly(A) tails of mRNA. The highly eon (Difco). Twelve days after each boost, the antibody titer was tested on served 68-kDa poly(A) binding protein I (PAB I) has Western blots. For affinity purification of PAB II-specific antibodies, been cloned from several species from yeast to man [re 10 /.tg of purified PAB II was loaded onto a SDS~polyacrylamide gel viewed in 1]. Bioehemieal and genetie analysis in yeast and transferred to nitrocellulose. The hlot was stained with Ponceau S, and the band of PAB 11 was cut out and used for affinity purifica indicates that this protein is essential for translation tion according to (91. Other antibodies used in this study were a mono and also involved in poly(A) tail shortening and, proba cional antibody (3A7) directed against the 64-kDa subunit oftbe hu bly, the regulation of mRNA stability [2, 3J. A second man cleavage stimulation factor [10], a monoclonal antibody against poly(A) binding protein, PAB 11, has a moleeular weight the splicing factor SC-35 [11], p80 coilin-specific buman autoimmune of 49 kDa and is distinet from PAB I by physical and serum (a gift from Dr. A. Lamond), anti-Sm human autoimmune serum (Küng) [12], and a monoclonal antibody (IGElO) against hu funetional criteria [4, 5]. PAB 11 was identified by its man PAB I [13]. role in the 3'-end processing of pre-mRNA in vitro. In Western bloUing. Proteins were separated on 10% polyacryl this reaction, the RNA is first cleaved endonucleolyti amide-SDS gels [14] and biotted to nitrocellulose membranes by tbe eally at a partieular position downstream of the eoding semidry procedure [15]. Blots were blocked in 20 mM Tris~HCI, pH sequence. The upstream cleavage fragment then re 7.5,150 mM NaCl, 0.05% Tween 20, 5% nonfat dry milk. Incubation with antibodies and washing were done in the same buffer lacking dry ceives a poly(A) taU. PAB Il binds the growing poly(A) milk. Proteins were detected by peroxidase-conjugated swine antibod ies directed against rabbit immunoglobulins (DAKO, Denmark) and chemiluminescence (ECL kit, Amersham). 1 Ta whom correspondence and reprint requests should be ad Immuno{luorescence staining 0/ cells. Immunolabeling was per dressed. Fax: 41-61-2672078. formed as described [16]. The specimens were examined in a ftuores-

123456 RESULTS AND DISCUSSION

kDa Antiserum against PAB II was produced by repeated injections of the purified protein into a rabbit. On West 116 ern blots, a protein comigrating with purified P AB 11 97 - was the major protein recognized by the crude serum in 66 cell extracts (data not shown). PAB II purified to near ..... --- homogeneity [5J was run on a preparative SDS-poly 45 acrylamide gel and transferred to nitrocellulose. After staining, the band of P AB II was cut out and used for affinity purification of the antibody. Antibody obtained 29- in thiR way reacted with purified PAB n. A single promi nent band comigrating with purified PAB 11 was de FIG. 1. Western blot detection of PAß 11. A dilution series of tected in HeLa cell nucIear extract (Fig. 1). A weak sig HeLa cell nuclear extract, corresponding to 4, 2, 1,0.5, and 0.25.ul was nal at 40 kDa was also seen, which was estimated by separated on a 10% SDS-polyacrylamide gel (lanes 2-6). 240 ng of serial dilution of the extract to be about 10-fold weaker purified PAß 11 were run in parallel (lane 1). The gel was blotted to nitrocellulose, and PAß 11 was detected with affinity-purified anti than that of PAB 11 (Fig. 1). In a purification of PAB 11 body. we have observed a protein of this molecular weight that had the same activities as P AB II, poly(A) binding and stimulation of polyadenylation, but was separated from it during ion-exchange chromatography. This protein eenee microscope (lM35 or Axiophot, earl Zeiss, FRG) equipped with also reacted with affinity-purified antibody (E. Wahle epiftuorescence. Photographs were taken on Fujichrome 100 film. Al ternatively, the modular cOllfoeal microscope at the EMBL [17] was and S. Krause, unpublished data). Thus, the second used as described [16]. strongest signal detected by the affinity-purified anti Electron microscopy. Monolayers of exponentially growing HeLa body very likely represents a protein closely related to cells or fragments of mouse liver were fixed with 4% paraformalde P AB H, possibly adegradation product. Additional sig hyde in Sörensen phosphate buffer, pH 7.4, for 2 h on ice. The speci nals were even weaker and could only be seen with long mens were then dehydrated in ethanol and embedded in LR White resin following the manufacturer's recommendations. Some liver exposures of the Western blot. specimens were cryofixed, cryosubstituted with acetone, and embed A crude estimate of the abundance of P AB II was ded in the same resin [18]. Ultrathin sections were incubated with obtained by Western blot analysis of aliquots from a affinity-purified anti-PAB 11 antibody according to a standard proto total SDS Iysate of a known number of He La cells along co] [191, in most cases in such a way that both sides ofthe section were with known quantities of purified PAB II. Under the labeled. Goat anti-rabbit IgG coupled with 15-nm colloidal gold parti des (Aurion) was used as a marker. Control sections were incubated assumption that the HeLa cell protein reacts with the in the absence ofthe primary antibody. The sections ofparaformal antibody in a manner similar to the bovine protein, two dehyde-fixed specimens were stained with the EDTA method prefer experiments with independent lysates detected approxi ential for nudear ribonudeoproteins [20], whereas cryofixed and mately 2 X 10· and 3.4 X 10· moleeules of PAB H per cryosubstituted material exhibited such contrast already after stain cell. The amount of hnRNA per nucleus in mouse L cells ing with uranyl acetate and lead citrate. The preparations were ob served in a Philips CMIO electron microscope at 80 kV using a 30- bas been determined as 1.8 pg [21J. Ifthe average molec 40-~m objective aperture. ular weight of hnRNA is 5 X 10·, and 20% of the mole For quantitative estimates of the distribution of label, gold parti cules carry poly(A) tails [22], nearly 40,000 poly(A) tails des were counted over the interchromatin granule clusters, nucleoli, are estimated to be present in one nucleus. With a nucleoplasm (corresponding to the nuclear area with the exclusion of poly(A) tail length of 250 nucleotides and a packing clusters of interchromatin granules and nucleoli) and cytoplasm. For this purpose, 15 randomly chosen cells of each group (paraformalde density of one P AB II moleeule for every 25 nucleotides hyde-6xed HeLa cells and liver and cryofixed and cryosubstituted (5),400,000 moleeules ofthe protein would be required liver) were photographed and printed at the same magnification to coat this amount ofpoly(A). Thus, PAB II appears to (X28,800). Labeling densities (number of gold grains per square mi be present in sufficient quantities to coat all nuclear crometer) were determined for each cell compartment. The signifi poly(A), and possibly in excess. Its intranuclear concen cance of the labeling density differences between the different cell compartments within each group was evaluated using the Wilcoxon tration is in the micromolar range, three orders of mag signed rank test. nitude above the Kn for specific binding to an isolated

FIG.2. Immunoßuol"escence staining of HeLa cells. HeLa cells were ftuorescence-Iabeled with affinity-purified anti-PAB 11 (red, a) and anti-SC-35 (green, b) simultaneously. The secondary antibodies were labeled with Texas Red (PAB 11) and FITC (SC-35). The overlay (c) reveals colocalization of PAB II and SC-35 in speckles (yellow). Double labeling of HeLa cells with anti-PAB 11 (red, d) and a monoclonal antibody against CStF (green, e), and false color overlay (yellow, O. Double labeling of cells with anti-PAB 11 (red, g) and with anti-pBO coilin antibodies (green, h) reveals no colocalization in the overlay (i). Bar, 10 p.m. NUCLEAR LOCALlZATION OF PAß II 77 78 KRAUSE ET AL. oligo(A) binding site [5J. At this concentration, binding ported to be widely distributed throughout the nucleo to sequences other than poly(A) would be expected [5 J plasm [10J. We have confirmed this wide-spread distri (E. Wahle, unpublished data). bution. More intensely labeled regions were also visible. The PAB II signal in cytoplasmic extract was much Although so me of these appeared to correspond to the weaker than that in nuclear extract (data not shown) speckles stained by anti-PAB II, the colocalization was and may result at least partiaUy from leakage or break not perfeet (Figs. 2d-2f). Upon treatment of ceUs with age of nuclei during isolation. No signal was observed in a-amanitin, CStF was almost exclusively localized in nuclear or cytoplasmic extract in the 70 kDa region the same speckles that were recognized by anti-PAB II where PAB I would be expected. A monoclonal antibody (data not shown). Treatment with a-amanitin also in directed against P AB I also did not crossreact with P AB creases the association of snRNPs with interchromatin II (data not shown). granules [7J. The only partial overlap of P AB II and Immunofiuorescence staining of HeLa ceH mono CStF may be related to their functions: CStF (also layers with the affinity-purified PAB II antibody caUed CF1) is thought to interact with an RNA se showed an exclusively nuclear reaction (Figs. 2a, 2d, and quence downstream ofthe cleavage site [25J and is prob 2g). Little or no staining was detected in the nueleoli. In ably released after the pre-mRNA has been cleaved. In tbe nucleoplasm, there was a wide-spread staining in vitro, the factor is not required for the second step of addition to an intense spot-like pattern reminiscent of 3'-end processing. PAB 11, in contrast, is involved in the the "speckled" pattern described for antibodies directed latest stage of 3'-processing, the elongation of the against factors involved in the splicing of pre-mRNA poly(A) tail, and probably remains associated with the [11, 16J. The speckled nuclear pattern was evident in poly(A) tail until the RNA is exported from the nucleus. growing as weIl as in confluent ceHs. In mitotic ceHs, Several antibodies and antisense probes directed PAB II was present throughout the cytoplasm (data not against components of the splicing machinery recognize shown). When the labeling reaction was carried out in not only interchromatin granules but also "coiled bod the absence of the primary antibody, no fluorescence ies" [reviewed in 26J. Figures 2g-2i demonstrate that was detectable. With preimmune serum, only a weak the antibody against PAB II does not decorate the coiled general staining was seen without any specific nuclear bodies, which are identified by their reaction with a hu reaction. When the affinity-purified antibody was used man autoimmune serum against coilin, an 80-kDa pro in the presence of an exces!; of purified P AB II, aU fluo tein present in these structures [27, 28J. rescence was uniformly reduced to near background level (data not shown). Treatment of HeLa ceUs with drugs interfering with The P AB II staining pattern was compared to the transcription, actinomycin D, a-amanitin, or cordyce staining patterns of other nuclear proteins. Monoclonal pin (3'-deoxyadenosine), did not result in cytoplasmic antibodies against the SC-35 protein, a non-snRNP staining with anti -P AB Ir. Exclusively nuclear staining splicing factor, show a speckled pattern [11J. These was also observed in other ceU lines (HepG2, 3T3). In speckles obviously correspond to clusters of interchro BHK ceUs, PAB II was also predominantly localized in matin granules visible in the electron microscope [23J. the nucleus, but there was a higher cytoplasmic signal. PAB II and SC-35 were colocalized in the interchroma Immunoftuorescence experiments have independently tin granules as demonstrated by a superposition of been carried out with a human bladder carcinoma ceH pseudo-colored confocal images of HeLa ceU nuclei im line, T24, in the laboratory of Dr. Luitzen de Jong. The munolabeled with anti-PAB II and anti-SC-35 simulta experiments confirmed the high specificity of affinity neously. Outside the granules, staining was predomi purified antibody, the localization pattern described nantly by PAB II (Figs. 2a-2c). The same speckles that above, and the colocalization of PAB II and SC-35 in the were stained with anti-PAB II also reacted with a hu clusters of interchromatin ,granules (L. de Jong, per man anti-Sm autoimmune serum, which recognizes the sonal communication). snRNPs involved in splicing [24J. In this case, the stain The intranuclear distribution of P AB II described ing also overlapped throughout the nucleoplasm (data above is in very good agreement with light microscopic not shown). observations ofthe distribution ofpoly(A): After hybrid CStF, a factor involved in the 3'-cleavage of pre ization with biotinylated poly(dT), labeling is observed mRNA that precedes polyadenylation, has been re- both in a widespread nucleoplasmic distribution and in

FIG. 3. Localization of PAB 11 by electron microscopy. (a) Immunogold-Iabeled section of a paraformaldehyde-fixed HeLa cel!. The antibody is mainly associated with perichromatin fibrils (some indicated with small anows) and with clusters on interchromatin granules (large arrows). Few gold grains are seen on the nucleolus (n). X2B,BOO. (b) Immunogold·labeled section ofparaformaldehyde·fixed mouse liver. Labeling occurs mostly in association with perichromatin fibrils (small anows) and a cluster of interchromatin granules (large arrows). Perichromatin granules (open arrows) are rarely labeled. Some gold grains can be observed on the nucleolus (n). X38,BOO. NUCLEAR LOCALIZATJON OF PAB II 79 80 KRAUSE ET AL.

FIG. 4. Section of a paraformaldehyde-fixed mouse hepatocyte incubated under contral conditions, in t he absence of the a nti-PAB 11 antibody. Symbols as for Fig. 3. The arrowhead points to a gold grain. X28,800. the same speckles that are strongly labeled with anti na nt nuclear and the weak cytoplasmic labeling. The Sm antibodies [29, 30]. analysis shows a highly significant difference in labeling The intracellular localization of PAß IJ was also in densities between interchromatin granules and other vestigated at the ultrastrueturallevel with .ffinity-puri eompartments (P '" 0.001). It thus confi rms interchro fied primary and gold-Iabeled seeondary antibody. In .11 matin granules as the structural component of the nu specimens, the majority of label was nuclear and asso cleus containing the most significant concentration of ei.ted with periehromatin fibrils .nd clusters of in PAß Ir. The labe1ing density in the nucleoplasmie area terehromatin granules (Figs. 3a and 3b). Periehromatin outside the interchromatin granule clusters appears rel granules showed only oeeasionallabeling (Fig. 3b). La atively low in comparison. However, perichromatin fi beling of control seetions ineubated without primary brils are the major struetural component strongly la antibody was negligible (Fig. 4). A quantitative estimate beled with anti-PAß II antibody in such nucleoplasmie ofthe labeling densities in HeLa and liver eells (Table 1) regions. Sinee the surfaee occupied by the fibrils is confirmed a significant difference between the predomi- rather limited and virtually impossible to evaluate, the overall labeling density in the nucleoplasm eertainly underestimates the association of PAß II with periehro matin fibrils. In eontrast to the immunofluoreseence TABLE 1 experiments, nucleolar JabeJing was clearly detectabJe Values for Gold Grain Densities (Grain Number per Square in the electron microscope. Micrometer ± SEM) in Different CeU Compartments after As in t he immunoßuorescence experiments, the dis Labeling with Anti-PAB " Antibody tribution of PAß II revealed in the eleetron mieroseope Nucleoplasm IG Nucleolus Cytoplasm was in good agreement with previous studies on the in t ranuclear distribution of poly(A), in that t he major la HeL. 7.42 ± 0.25 27.56 ± 1.91 8.75 ± 0.68 1.87 ± 0.23 beled structures were clusters of interchromatin gran Liver pf 6.44 ± 0.43 21.93 ± 2.04 7.27 ± 1.1 7 1.84 ± 0.20 ules and periehromatin fibers, while the periehromatin Liver cryo 8.46 ± 0.27 40.00 ± 3.19 13.92 ± 1.48 3.14 ± 0.25 granules were less strongly labeled [30]. The same re N ote. le, interchromatin granule clusters; pf, paraformaldehyde port [30] also found coiled bodies to be devoid of fixation; cryo, cryofixation and cryosubstitution. poly(A), in agreement with the immunofluoreseence re- NUCLEAR LOCALIZATJON OF PAß [] 81 sults presented above. The major discrepancy concerns Foundation (ta E.W.). RW. is the recipient of a Heisenberg fellow the presenee of PAB II in the nucleolus, in wh ich ship of the Deutsche Forschungsgemeinschaft. poly(A) has not been found [30]. Other nuclear RNA bin ding proteins involved in mRN A metabolism have REFERENCES been detected in the nucleolus as weil, even though no 1. Sachs, A. B. (1990) Gurr. Biol. 2. 1092-1098. nucleolar hmction of these proteins is known [31, 32]. 2. Sachs, A. B., and Davis, R. W. (1989) Cel/58. 857-867. Nonspecific binding of an abundant protein to ribo 3. Sachs, A. B. (1993) Cell74. 413-42l. somal RNA (see above) might be an explanation. 4. Wahle, E. (1991) CellS6, 759- 768. Association of PAB II with interchromatin granules 5. Wahle, E., Lustig, A .. Jenö. P., and Maurer, P . (1993) J . BioL and perichromatin fibrils observed by electron micros Chem. 268, 2937-2945. copy is also in good agreement with the immunoftuores 6. Bienroth. S .• Keller, W., and Wahle, E. (1993) EMBO J . 12, eence data. The nucleolar and the much weaker cyto 585- 594. plasmic labeling were not detectable by immunolluores 7. Carmo-Fonseca, M .. Pepperkok, R.. CarvaIho, M. T ., and La cence. This difference may largely rellect different mond, A. I. (1992) J . Gell Biol. 117,1-14. signal thresholds detectable by the two methods. 8. Dignam. J. D., Lebowitz. R. M., and Roeder. R. G. (1983) Nucleic Perichromatin fibrils contain hnRNP cOre proteins Acids Res. 11, 1475-1488. and snRNPs [33], Ui and U2 snRNAs [34], as weil as 9. Burke, B., Griffiths, G., Reggio, H., Louvard, D., and Warren. G. the splicing factor SC-35 [35]. Sinee the perichromatin (1 982) EMBO J. 1, 1621-1628. fibrils have previously been shown to represent nascent 10. T a kagaki , Y., ManIey. J. L., MacDonald, C. C., Wilusz, J., and Shenk. T. (1990) Genes Dev. 4, 2112-2120. pre-mRNA [36, 37], it is highly probable that splieing 11. Fu, X. -D., and Maniatis, 'f. (1990) Nature 343, 437-44l. takes place on these structures. As 3'-end processing is 12. Matter. L., Schopfer, K., Wilhelm, J. A., Nyfenegger, T., Pari thought to be initiated on nascent transeripts [38], the sot, R. F., and De Robertis, E. M. (1982) Arth. Rheumatol. 25, association of P AB 11 with perichromatin fibrils is con 1278- 1283. sistent with its proposed role in mRNA polyadenyla 13. Görlach, M., Burd. C. G., and Dreyfuss, G. (1994) Exp. Gell Res. tion. The interchromatin granule clusters contain only 211,400-407. low levels of rapidly labeled RNA [for review, see 39] 14. Laemmli, U. K. (1 970) Nature 227,680-685. and hnRNP proteins [33]. However, they do contain 15. Kyse-Andersen, J . (1 984) J . Biochern. Biophys. M elk 87, 203- poly(A) (see above) and large amounts of different 209. splicing factars [23, 33-35, 40]. Although it is generally 16. Carmo-Fonseca. M., T ollervey, D., Pepperkok, R., Barsbino, S., assumed that the interchromatin granule clusters do Merdes, A., Brunner, C., Zamore, P. D., Green, M. R.. Hurt. E., and Lamond, A. I. (1991) EMBO J. 10, 195- 206. not represent splicing organelles, their true function in 17. Stelzer, E. H. K., Wacker. I.. and De Mey, J. R. (1991) Semin. nuclear RNA metabolism remains unknown [see 30 for Gell Biol. 2, 145-152. a recent discussion]. 18. von Sehack, M. L., Fakan, S., Villiger, W., and Müller, M. (1993) The nuclear localization of PAß II demonstrated in Eur. J. Histochem. 37, 5- 18. this paper, its colocalization with factors known to be 19. Biggiogera, M .• Fakan, S., Kaufmann, S. H., Blnck, A., Shaper, involved in pre-mRNA processing, and the association J . H., and Busch, H . (l989)J. Histochem. Gytochem . 37. 137 1- with perichromatin fibrils are in agreement with a role 1374. of the protein in polyadenylation. In contrast, immuno 20. Bernhard, W. (1969) J . Ullrastruct. Res. 27, 250- 265. fluorescenee studies with PAB I in yeast and mamma 21. Brandhorst, B. P., and MeConkey, E. H. (1974) J . M ol. Biol. 85, lian cells show that the major localization ofthis protein 451-463. 22. Greenberg, J. R. , and Perry, R. P. (1972) J. M ol. Biol. 72,91-98. is cytoplasmic [1 3, 41) . This is consisten t with the ge 23. Martin, T. E., Barghusen, S. C., Leser, G. P ., and Spear, P. G. netic data showing an involvement in two cytoplasmic (1987) J . Cell Biol. 105, 2069-2082. reactions, translation and poly(A) shortening [2] .. The 24. Lührmann, R. (1988) in Structure aod Function of Major and localization of the t wo poly(A) binding proteins in two Minor Small Nucleer Ribonucleoprotein Particles (Birnstiel, different cellular compartments contirms their roles in M. L. , Ed.), pp. 71- 99, Springer, Berlin. different aspects of poly(A) tail metabolism. 25. Gilmartin, G. M., and Nevins, J. R. (1989) Genes Deu. 3, 2180- 2189. We thank the following eolleagues für antibodies: Matthias 26. Lamond. A. 1., and Ca rm o ~ Fonseca, M. (1993) Trends Gell Biol. GÖrlaeh. Gideo" Dreyfuss. Clinton McDonald. T om Maniatis, snd 3,198- 204. Eddy Oe Robertis. We are grateful to Veronique Mamin and Jitka 27. Andrade. L. R C., Chan. E. K. L. , Raska, 1. , Peebles. C. L., Roos, Fakan for excellent technical assistance. to Marie-Louise von Schack G ., and Tao, E. M. (1991) J. Exp. Med. 173, 1407- 1419. for providing cryofixed and cryosubstituted mouse liver specimens, to 28. Raska, 1. , Andrade, L. E., Ochs, R. L., Chan, E. K. , Chang, C. M., Matthias Gärlach. Gideon Dreyfuss, and Luitzen de J ong for commu nicating unpuhlished data, to Walter Keller, Angus Lamond, and Roos, G. , and Tan, E. M. (1991) Exp. Cell Res. 195,27- 37. Alsn Sachs for reading thc manuseript, snd to Walter Keller and 29. Carter, K. C., Taneja. K. L.. and Lawrenee, J. B. (1991) J. Cell Angus Lamond for support. This work was supported hy grants from Biol. 115, 1191-1202. the Swiss National Science Foundation (ta S.F. and E.W.) and by the 30. Visa. N .. Puvion-Dutilleul, F .. Harper, F., BachelJe rie, J.-P., and Kantons of Basel-Stadt and Basel-Land end t he Reche Research Puvion, E. (1993) Exp. Ce ll Res. 208, 19-34. 82 KRAUSE ET AL.

31. Hauber, J., Perkinsj A., Heimer, E. p" and Cullen, B. R. (1987) 36. Fakan, S" Puvion, E., and SpQhr, G. (1976) Exp. Cell Res. 99, Prac. Natl. Acad. Sei. USA 84, 6364-6368. 155-164. 32. Felber, B. K., Hadzopoulou-Cladaras, M., Cladaras, C., Cope 37. Fakan, S., and Hughes, M. E. (1989) Chromosoma 98. 242-249. land, T., and Pavlakis, G. N. (1989) Proc. Natl. Acad. Sei. USA 38. Nevins, J. R, and Darnell, J. E. (1978) Cell15. 1477-1493. S6,1495-1499. 39. Fakan, S. (1986) in Methods and Achievements in Experimental 33. Fakan, S., Leser, G., and Martin, T. E. (1984) J. Cell Biol. 98, Pathology (Jasmin, G., and Simard, R., Eds.), Vol. 12, pp. 105- 358-363. 140, Karger, Basel. 34. Visa, N., Puvion-Dutilleul, F., Bachellerie, J. P., and Puvion, E. 40. Puvion, E., Viron, A., Assens, C., Leduc, E. H., and Jeanteur, P. (1993) Eur. J. Cell Biol. 60, 308-32l. (1984) J. Ultrastruct. Res. 87.180-189. 35. Spector, D. L., Fu, X. D., and Maniatis, T. (1991) EMBO.J. 10, 41. Adam, S. A., Nakagawa, T., Swanson, M. S., Woodruff, T. K, 3467-3481. and Dreyfuss, G. (1986) Mol. Cello Biol. 6, 2932-2943.

Received May 25, 1994