INTRODUCTION

Zinc finger proteins of the TFIIIA/Kruppel type belong to the largest known family of transcription factors (1, 2). These proteins are characterized by Cys₂-His₂ zinc finger motifs often repeated in tandem that fold around zinc ions and function as nucleic acid binding domains (3-6). About one-third of mammalian Cys₂-His₂ zinc finger proteins contain a conserved domain of approximately 75 amino acids called KRAB (Kruppel-associated box) (7). The KRAB domain, located at the N terminus of Cys₂-His₂ multifinger proteins, can confer strong distance-independent transcriptional repression of both activated and basal RNA polymerase II promoter activity (8-14). Since they encode a repression motif and a potential DNA binding domain, members from the large KRAB/Cys₂-His₂ protein family are presently thought to function as transcriptional regulators of gene expression.

We previously cloned ZNF74, a gene that encodes a KRAB/Cys₂-His₂ protein (15). ZNF74 lies a few kilobases proximal to a polymorphic CA repeat (D22S264) (16) that was shown to be a distal marker for 22q11.2 deletions associated with increased susceptibility to schizophrenia (17). ZNF74 is also one of the few genes found hemizygously deleted in the majority of patients with the DiGeorge syndrome, a microdeletion disorder associated with a wide variety of congenital malformations including cardiac defects, thymic hypoplasia, and hypocalcemia (18-20). To date, however, both the role of embryologically expressed ZNF74 in the DiGeorge syndrome (19) and its biochemical and cellular functions remain unclear.

Although the presence of a KRAB motif and of 12 Cys₂-His₂ zinc finger motifs suggest that ZNF74 may function as a transcription factor, we recently demonstrated that its zinc finger domain harbors an RNA binding activity in vitro (15). This result suggests that, in addition to its potential function in transcriptional regulation, ZNF74 may also be involved in RNA metabolism. We also previously reported that ZNF74 is tightly associated with the nuclear matrix, since detergent, DNase, RNase, and high salt treatments could not release it from this nuclear scaffold structure (15). In this study, we report that the ZNF74 zinc finger domain acts as a nuclear matrix targeting sequence. Furthermore, to further define ZNF74 function, we searched for proteins from the nuclear matrix that directly interact with ZNF74. We now show that ZNF74 interacts, via its multifinger domain, with a hyperphosphorylated form of the largest subunit of RNA polymerase II (pol IIo) and colocalizes with this protein in irregularly shaped subnuclear domains. Because the hyperphosphorylated form of this RNA pol II¹ subunit has been shown to colocalize (21, 22) and associate (23-25) with splicing factors and to be involved in pre-mRNA processing (26, 27), our results indicate that ZNF74 might play a role in the regulation of gene expression not only by transcriptional but also by post-transcriptional mechanisms.

EXPERIMENTAL PROCEDURES

pCGN, a cytomegalovirus enhancer-driven eukaryotic expression vector (28) that encodes an N-terminal hemagglutinin (HA) epitope was used to generate HA fusion constructs. pMAL-c vector (New England Biolabs) was used to generate maltose-binding protein (MBP) fusion constructs for expression in Escherichia coli. The following constructs were previously described in Grondin et al. (15): HA-ZNF74-(1-572), HA-ZNF74Krab-(68-572), HA-NKrab-Zn-(68-509), HA-Zn-C-(175-572), HA-Zn-(175-509), MBP-ZNF74-(1-572), MBP-ZNF74-(106-572), MBP-Zn-(175-509), and MBP--gal- (also called MBP herein). GAL4 DNA binding domain (GAL4-(1-147)) fusion proteins were obtained by introducing either the zinc finger domain (aa 175-509) or the N-terminal domain (aa 1-174) of ZNF74 in a Rous sarcoma virus enhancer-driven eukaryotic expression vector pRSVGAL4 (kindly provided by Dr. Robert Rehfuss). The zinc finger domain (aa 175-509) or the N-terminal domain (aa 1-174) of ZNF74 were PCR-amplified with oligonucleotide primers containing XbaI cloning sites. In each case, the 3-primer included an in frame stop codon. The PCR-amplified products were subcloned in frame at the 3-end of the GAL4 DNA binding domain in a unique XbaI site of the pRSVGAL4 vector.

The mAb CC3 (IgG_2a) was previously obtained by immunizing Balb/C mice with chick embryo proteins (29) and shown to recognize a phosphoepitope located on the carboxyl-terminal domain (CTD) of the hyperphosphorylated RNA polymerase II largest subunit (pol IIo) (23, 30). The mAb Pol3/3 recognizes a conserved region located outside of the CTD of the largest subunit of the RNA polymerase II (RNA pol II LS) and was kindly provided by E. K. F. Bautz (31). The mouse hybridoma cell line CRL 2031, secreting the mAb SC35 (IgG₁), was purchased from the American Type Culture Collection. The mAb SC35 recognizes a phosphorylated form of SC35 (32), a non-snRNP splicing protein, and possibly another structural protein that colocalizes to the same nuclear domains (33). The mAb 12CA5 (34) or the rabbit polyclonal antibody Y-11 (Santa Cruz Biotechnology Inc.) was used to detect the HA epitope tag. The GAL4 (DBD) mouse monoclonal IgG antibody (RK5C1) (Santa Cruz Biotechnology) was used to detect the DNA binding domain (aa 1-147) of GAL4 protein. A rabbit anti-MBP polyclonal antiserum (New England Biolabs) was used to detect the maltose-binding protein. The mAb 131C1 (kindly provided by Dr. Yves Raymond) was used to detect lamin A and C.

Monkey fibroblast-like COS-7 cells, mouse skin fibroblast L cells, mouse embryonal carcinoma P19 cells, human embryonal kidney HEK 293 cells, or 293T cells (35) were used to prepare the various cellular fractions. For experiments requiring transfection, cells were plated at a cell density of 4-8 × 10⁵/100-mm plate, transfected 24 h later with various plasmid constructs (25 µg of DNA) (36), and harvested 30-36 h after transfection (3-6 × 10⁶ cells obtained from a 100-mm plate). For cellular fractionation and nuclear matrix isolation, transfected or untransfected cells (3-6 × 10⁶ cells) were first submitted to a hypotonic lysis in 40 µl of RSB buffer (10 mM Tris, pH 7.4, 10 mM NaCl, 3 mM MgCl₂, 0.5 mM PMSF), and nuclei were recovered following essentially the method of Cockerill and Garrard (37) as described previously (15). Washed nuclei were subjected to subnuclear fractionation and nuclear matrix isolation using the method of He et al. (38) as previously detailed (15). In brief, the nuclei derived from a 100-mm plate were freed of the chromatin by a 50-min digestion at 30 °C with RNase-free DNase I (about 20 units/10⁶ cell nuclei) (Life Technologies, Inc.) in 40 µl of digestion buffer (10 mM Pipes, pH 6.8, 50 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, 1 mM EGTA, 0.5% (v/v) Triton X-100, 1.2 mM PMSF, 2 µg/ml aprotinin, 2 µg/ml leupeptin, and 1 µg/ml pepstatin A), and then extracted by the addition of 1 M ammonium sulfate to a final concentration of 0.25 M. The 750 × g supernatant containing digested chromatin corresponds to the soluble DNase I-treated nuclear fraction also called soluble nuclear fraction herein. The pellet was then submitted to an additional extraction with 50 µl of digestion buffer containing 2 M NaCl. The subsequent 750 × g pellet corresponding to insoluble nuclear matrix and associated ribonucleoproteins was subjected to RNase A (Quiagen) and RNase T (Boehringer Mannheim) (5 µg and 2 units, respectively) in 50 µl of digestion buffer. The resulting 750 × g pellet corresponds to the RNase-treated insoluble nuclear matrix fraction. To prepare the various nuclear fractions used in far-Western analysis, 2 mM orthovanadate was also included in the RSB buffer, and both 2 mM orthovanadate and 50 mM NaF were added to the digestion buffer. Furthermore, in some cases as indicated in the figure legends, a faster cell fractionation scheme was used to limit protein dephosphorylation by endogenous phosphatases; the centrifugation steps were reduced to 15 s at 15,000 × g, the chromatin was digested for only 25 min with twice the amount of DNase I, and the RNase A and T digestion step was omitted.

Before immunoprecipitations, the soluble DNase I-treated fractions (equivalent to 20-40 × 10⁶ cells/350 µl of digestion buffer) obtained as described under "Cellular Fractionation and Nuclear Matrix Isolation" were diluted with 2 volumes of lysis buffer (50 mM Tris, pH 7.5, 250 mM NaCl, 0.5% Triton X-100, 0.5% sodium deoxycholate, 5 mM EDTA, 2 mM orthovanadate, 50 mM NaF, 1 mM PMSF, 2 µg/ml aprotinin, 2 µg/ml leupeptin). To prepare nuclear matrix extracts for immunoprecipitations, the DNase/RNase-digested nuclear matrix pellets (from about 10-20 × 10⁶ cells) obtained as described under "Cellular Fractionation and Nuclear Matrix Isolation" were extracted with 500 µl of SDS solubilizing buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 0.8% SDS, 2 mM orthovanadate, 50 mM NaF) for 4 min at 95 °C (39). The extracts were cooled at 4 °C for 4 min before adding 500 µl of Triton buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 4% Triton X-100, 2 mM orthovanadate, 50 mM NaF) and then centrifuged at 15,000 × g for 2 min at 4 °C. To prepare total cell extracts (5-10 × 10⁶ cells/ml) for immunoprecipitations, cells were either extracted in SDS solubilizing buffer and diluted in Triton buffer as described above or extracted in lysis buffer (50 mM Tris, pH 7.5, 250 mM NaCl, 0.5% Triton X-100, 0.5% sodium deoxycholate, 5 mM EDTA, 2 mM orthovanadate, 50 mM NaF, 1 mM PMSF, 2 µg/ml aprotinin, 2 µg/ml leupeptin). The extracts were passed through a 26G1/2 needle to reduce viscosity before centrifugation at 15,000 × g for 15 min at 4 °C. Total cell, nuclear matrix, and diluted soluble DNase I-treated extracts were precleared with protein A-Sepharose (Sigma) for 1 h at 4 °C. Precleared extracts were then recovered after removal of the beads by centrifugation (1 ml/20 µl stacked protein A beads). Immunoprecipitations were then carried out for 2 h at 4 °C (or 16 h for coimmunoprecipitation experiments) using 1 ml of precleared extracts, 10-50 µl of the appropriate antibody, and 30 µl of protein A-Sepharose. For the coimmunoprecipitation experiments, 10 mM EDTA and 10 mM EGTA were added during the preclearing and the immunoprecipitation steps to stabilize the pol IIo (40). The protein A-Sepharose beads were then washed 3 times either with 1 ml of lysis buffer for immunoprecipitations or with a buffer containing 25 mM Tris, pH 7.5, 100 mM NaCl, 0.2% SDS, and 1% Triton for coimmunoprecipitations. Washed immunoprecipitates were resuspended in Laemmli buffer, and the recovered proteins were analyzed by SDS-PAGE. In some cases, the supernatants of immunoprecipitation were first concentrated using Microcon 100 filter devices (Amicon) before dilution in Laemmli buffer.

E. coli DH5 strain transformed with pMal-c-ZNF74 fusion constructs were grown to an A₆₀₀ of 0.3-0.5 (500 ml) and induced with 0.3 mM isopropyl--D-thiogalactopyranoside for 3 h. Crude soluble extracts containing the soluble MBP fusion proteins were prepared from bacterial cells resuspended in column buffer (200 mM Tris, pH 7.5, 200 mM NaCl, 10 mM -mercaptoethanol) (50 ml). MBP fusion proteins from these extracts were immobilized on amylose resin (0.5 ml) (New England Biolabs). The resin was washed with 20 volumes of column buffer and with 40 volumes of elution buffer (20 mM Hepes, pH 7.5, 100 mM NaCl). Then, the immobilized MBP fusion proteins were eluted with 1.5 ml of elution buffer containing 10 mM maltose. Depending on the preparation and on the fusion proteins, 100-1000 µg of eluted affinity-purified proteins were recovered as estimated by a Bradford protein assay (Bio-Rad) and as confirmed by Coomassie gel staining. The purified proteins were aliquoted and stored at 80 °C.

Proteins were biotinylated according to Cicchetti and Baltimore (41). In brief, the MBP-ZNF74-(106-572) fusion protein eluted from the amylose resin was dialyzed against a borate buffer (100 mM, pH 8.0). The protein was incubated for 4 h at room temperature with biotinamidocaproate N-hydroxysuccinimide ester (Sigma) at a concentration of 50 µg of the biotin derivative for 1 mg of MBP-ZNF74-(106-572) fusion protein. The reaction was stopped by adding 1 M NH₄Cl at a ratio of 4 µl for 50 µg of biotin derivative for 10 min at room temperature. The biotinylated protein was then dialyzed against a 50 mM Tris, pH 7.5, buffer containing 100 mM NaCl. Glycerol (final concentration 15%) was added to the dialyzed biotinylated protein before its storage at 20 °C. Thawed biotinylated proteins (1.5 µg/µl) were never kept longer than 2 weeks at 4 °C before use.

Nuclear protein fractions or proteins immunoprecipitated from cell or nuclear extracts were resolved by SDS-PAGE on a 6% separating gel and electrophoretically transferred to 0.2-µm nitrocellulose membrane at 22 V for 12-16 h using a Mini Trans-Blot^® electrophoretic transfer cell (Bio-Rad). After transfer, the blots were rinsed in TBST buffer (10 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Tween 20) and used immediately or kept at 4 °C in the same buffer for a few days. The blots were incubated for 4-16 h in blocking buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 2% bovine serum albumin). For far-Western analysis, the blots (9-cm² strips) were probed for 12-16 h with affinity-purified MBP fusion proteins (0.5 µg/ml) or with biotinylated MBP-ZNF74-(106-572) fusion proteins (30 ng/ml) in overlay buffer (20 mM Hepes, pH 7.5, 100 mM NaCl, 1 mM EGTA, 1% Nonidet P-40, 0.5% bovine serum albumin, 0.25% gelatin) (2 ml). For competition experiments, the blots were incubated in the presence of the biotinylated MBP fusion protein and a 100-fold excess of unbiotinylated MBP fusion protein. All the above steps were performed at 4 °C. The blots were then washed in washing buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Tween 20, 0.2% gelatin) three times for 5 min at room temperature and then incubated for 1 h in TBST buffer containing 1% bovine serum albumin. Bound MBP fusion proteins were revealed by sequential incubation with a rabbit primary anti-MBP polyclonal antiserum (New England Biolabs), a secondary goat anti-rabbit horseradish peroxidase (Sigma), and a chemiluminescence reagent as described by the manufacturer (Renaissance^® kit, NEN Life Science Products). Bound biotinylated MBP-ZNF74 fusion proteins were also revealed by chemiluminescence after an incubation step with avidin-horseradish peroxidase (Sigma).

This protocol was modified from Bregman et al. (22). Proteins were immunoprecipitated as above with mAb CC3 coupled to protein A-Sepharose. For alkaline phosphatase treatment, the beads were washed with phosphatase buffer (10 mM Tris acetate, pH 7.5, 10 mM magnesium acetate, 50 mM potassium acetate) and treated with or without 5 units of calf intestinal phosphatase (Pharmacia Biotech Inc.) in 25 µl (excluding bed volume) of phosphatase buffer for 10-60 min at 37 °C. The beads were resuspended in Laemmli buffer for analysis of the eluted proteins by SDS-PAGE.

Subconfluent monkey COS-7 cells plated in 1 × 2-cm four-well Lab-Tek^TM (Nunc) were transfected with appropriate plasmid constructs (2 µg) (36). About 40 h after transfection, the cells were either fixed and processed directly for immunofluorescence when indicated or first extracted in situ as described by Bisotto et al. (42) using the method of He et al. (38). Briefly, for in situ sequential extraction, the cells were first washed in isotonic phosphate saline buffer (PBS buffer) and then extracted in cytoskeleton buffer (10 mM Pipes, pH 6.8, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, 1 mM EGTA, 0.5% Triton X-100, 1 mM PMSF), digested with DNase I (Life Technologies), and washed with 0.25 M ammonium sulfate for chromatin removal. When indicated, the transfected cells were also subjected to a high salt extraction with 2 M NaCl and an RNase A and T digestion. For immunofluorescence microscopy, untreated or in situ extracted cells were treated sequentially at room temperature with 4% formaldehyde/PBS for 1 h, with 0.2% Tween-20, 4% formaldehyde/PBS for 1 h, with 50 mM NH₄Cl/PBS for 30 min and with 0.2% cold fish gelatin (Sigma)/PBS for 30 min (15). Primary and secondary antibody incubations were then performed essentially as described (15). For single immunolabeling, the mouse mAb 12CA5 (hybridoma supernatant diluted 1:500) was used as a primary antibody to detect HA epitope-tagged ZNF74 proteins, and a mouse anti-Gal 4 antibody (1:50) (Santa Cruz Biotechnology) was used to detect GAL4 proteins. A rabbit anti-mouse IgG conjugated to fluorescein isothiocyanate (Sigma) was used as a secondary antibody. For double immunolabeling of HA-ZNF74-(1-572) and pol IIo/CC3 antigen, a rabbit polyclonal anti-HA antibody (diluted 1:200) (Santa Cruz Biotechnology) and the mouse mAb CC3 antibody (ascitic fluid diluted 1:500) were sequentially added. For immunolabeling of both HA-ZNF74-(1-572) and SC35 antigen, the rabbit polyclonal anti-HA antibody and the mouse mAb SC35 (ascitic fluid diluted 1:200) were used as primary antibodies. A goat anti-rabbit IgG conjugated to fluorescein isothiocyanate (Sigma) and a goat anti-mouse IgG conjugated to Rhodamine (Pierce) were then used as secondary antibodies. The specificity of each of the secondary antibodies was tested either by omitting one of the primary antibodies or by incubating the primary antibody individually with each of the secondary antibodies. No cross-reaction was observed between the two sets of antibodies. DNA was stained by a 5-min treatment with 2.5 µg/ml Hoechst 33258 fluorochrome (Sigma). Preparations were examined under a Leika photomicroscope equipped for epifluorescence and photographed using NEOPAN 1600 (Fuji) or EPL 400 (Kodak) films.

RESULTS

We previously reported that the protein encoded by ZNF74-2 cDNA (EMBL accession number X92715), called ZNF74 here, is an RNA-binding protein tightly associated with the nuclear matrix (15, 19). To delimit the region required for ZNF74 association with the nuclear matrix, we assessed the subnuclear localization of various HA-tagged deletion mutants in transfected cells. Following subcellular fractionation of transfected L cells, all truncated ZNF74 proteins that included at least the zinc finger domain were found associated with the insoluble nuclear matrix fraction as assessed by immunoblot (Fig. 1A). As previously observed for the full-length protein (15), the attachment of all of these truncated ZNF74 proteins to the nuclear matrix was very tight, since none were detectable in the soluble nuclear fractions recovered after sequential DNase, high salt, and RNase treatments (not shown). Immunofluorescence studies using monolayers of transfected COS-7 cells confirmed that the multifinger region (aa 175-509) alone was targeted to the nuclear matrix, where it remained tightly bound following in situ sequential DNase/RNase extractions (Fig. 2A).

[View Larger Version of this Image (30K GIF file)]

[View Larger Version of this Image (30K GIF file)]

The zinc finger domain was also found to be sufficient to target a heterologous nucleoplasmic protein, the GAL4 DNA binding domain (GAL4-(1-147)) to the nuclear matrix. Indeed, the zinc finger domain (aa 175-509) fused to the GAL4-(1-147) was targeted to the nuclear matrix fraction, while the N-terminal domain (aa 1-174) of ZNF74 fused to GAL4-(1-147) was detected in the soluble fraction of the nucleus (Fig. 1B) as assessed by immunoblots. Similarly, a punctate signal indicating tight attachment to the matrix was clearly detected after in situ extractions of transfected cells expressing the zinc finger domain (aa 175-509) fused to GAL4 (aa 1-147) (Fig. 2B). In contrast, the GAL4-(1-147) protein, normally detected in the soluble nucleoplasm of untreated cells (Fig. 2B), was completely extractable from the nucleus, since no signal remained after in situ sequential DNase/RNase extractions (not shown). The ZNF74 zinc finger domain (previously shown to be sufficient for nuclear translocation) (15) was also found sufficient to target -galactosidase, a normally cytoplasmic protein, to the nuclear matrix (not shown).

Since DNase and RNase treatments did not affect the association of ZNF74 with the nuclear matrix, we suspected that this attachment was independent of ZNF74 nucleic acid binding properties and instead involved protein-protein interactions. To identify potential nuclear proteins associated with ZNF74, we performed far-Western blot analysis. Proteins from an insoluble nuclear matrix fraction and a soluble DNase I-treated nuclear fraction were separated by gel electrophoresis and transferred to membranes. As protein probes, we used various ZNF74 fusion proteins constructed with the MBP at their N terminus and produced in E. coli as described previously (15). We found that MBP-ZNF74 fusion proteins bound to a few high molecular weight nuclear proteins (Fig. 3) in various cell types such as monkey COS-7 cells (Fig. 3A), mouse P19 cells (Fig. 3, B-D), and human HEK 293 (not shown). A strong signal was obtained with a protein of approximately 250 kDa that partitioned between the insoluble nuclear matrix fraction and the soluble fraction of the nucleus. As seen in Fig. 3C, the zinc finger region alone was sufficient for interaction with this nuclear protein. This association appeared to be specific to the ZNF74 zinc finger domain, since no binding was observed with the MBP control protein alone (Fig. 3D).

[View Larger Version of this Image (47K GIF file)]

Using the monoclonal antibody CC3 (mAb CC3), Vincent and colleagues (42) identified a phosphoprotein of approximately 250 kDa partitioning between the nuclear matrix and the soluble nuclear fraction. Since the CC3 antigen was very recently found to associate with splicing complexes and to correspond to pol IIo, a hyperphosphorylated form of the RNA pol II LS (23, 30, 43), this antigen was envisaged as the putative 250-kDa protein interacting with ZNF74.

A comparative analysis of a far-Western blot probed with ZNF74 (Fig. 4A, left panel) and Western blots probed with mAb CC3 revealed that a major nuclear protein interacting with ZNF74 and its zinc finger domain comigrated precisely with the 250-kDa pol IIo/CC3 antigen (Fig. 4A, middle panel). In contrast, the hypophosphorylated form of the RNA pol II LS (pol IIa) was identified as a faster migrating protein (220 kDa) as revealed using the mAb Pol3/3 (31) that recognizes both the 220-kDa pol IIa and 250-kDa pol IIo forms of the polymerase (Fig. 4A, right panel).

[View Larger Version of this Image (31K GIF file)]

To confirm that the 250-kDa protein recognized on far-Western blots by ZNF74 was pol IIo, we investigated the potential interaction of ZNF74 with mAb CC3-immunoprecipitated proteins. As illustrated in Fig. 4B, the ZNF74 zinc finger region bound to the 250-kDa pol IIo immunoprecipitated from nuclear matrix and soluble DNase I-treated nuclear extracts by mAb CC3. After immunoprecipitation, such extracts were completely depleted in the 250-kDa protein interacting with ZNF74 zinc finger domain, since no signal remained in the supernatants (Fig. 4B). In contrast, the 250-kDa protein remained in the supernatants when a monoclonal antibody directed against an epitope unrelated to pol IIo was used as a control (Fig. 4B). These results showed that a single nuclear protein migrating at an apparent molecular weight of 250,000 is interacting with ZNF74 via its zinc finger domain and confirmed that this protein is pol IIo.

The specificity of ZNF74 interaction with the immunoprecipitated pol IIo was furthermore confirmed in competition experiments. As seen in Fig. 4C (right lane), the binding of a biotinylated MBP-ZNF74 fusion protein to immunoprecipitated 250-kDa pol IIo/CC3 antigen was completely abolished in the presence of an excess of unbiotinylated MBP-ZNF74 as revealed using avidin-conjugated horseradish peroxidase. Such binding, however, remained unaffected in the presence of an excess of MBP control protein (Fig. 4C, middle lane).

RNA pol II LS is subjected to cycles of phosphorylation/dephosphorylation of its CTD, a region composed of multiple tandem repeats of a conserved heptapeptide (Tyr-Ser-Pro-Thr-Ser-Pro-Ser) (44, 45). As previously discussed, two resulting main forms, the 250-kDa hyperphosphorylated pol IIo and the 220-kDa hypophosphorylated pol IIa, can be resolved by gel electrophoresis (reviewed in Refs. 44 and 45). Since ZNF74 associates only with hyperphosphorylated pol IIo, we determined whether this interaction was in fact phosphodependent. As demonstrated by far-Western blot, no ZNF74-pol IIo interaction signal could be observed following phosphatase treatment of the CC3-immunoprecipitated pol IIo (Fig. 5, left panel). In this phosphatase time course experiment, the disappearance of the ZNF74-pol IIo interaction signal was clearly coincident with the appearance of partially dephosphorylated forms of the CC3-immunoprecipitated polymerase (Fig. 5, right panel). Upon phosphatase treatment, the progressive dephosphorylation of the polymerase was indeed evidenced by its increased electrophoretic mobility as revealed on Western blot using the mAb Pol3/3 that recognizes RNA polymerase II LS independently of its phosphorylation state. This result indicated that the interaction of ZNF74 with pol IIo is dependent on the phosphorylation state of the polymerase. Consistent with this, ZNF74-pol IIo interaction was also considerably reduced when the cellular fractions used for detecting such interaction were prepared in the absence of phosphatase inhibitors (not shown).

[View Larger Version of this Image (36K GIF file)]

To determine whether ZNF74-pol IIo interaction also occurred in the context of a complex cell extract, we then tested for the presence of ZNF74-pol IIo protein complexes by performing coimmunoprecipitation of both proteins from soluble cell extracts. Since ZNF74 is very tightly associated with the nuclear matrix and not detected in the soluble nucleoplasm, SDS cell extracts renatured by dilution in Triton X-100 were used as the only mean found to recover soluble ZNF74 protein complexes. Extracts prepared from human 293T cells expressing HA epitope-tagged ZNF74 were immunoprecipitated with either the anti-HA mAb 12CA5 or anti-pol IIo mAb CC3. The immunoprecipitates were separated by SDS-PAGE and examined by immunoblot for the presence of ZNF74 and Pol IIo. These studies showed that either the antibody directed against ZNF74 (Fig. 6A) or that specific for the pol IIo (Fig. 6B) was able to coimmunoprecipitate both proteins, indicating that ZNF74-pol IIo protein complexes formed in total cell extracts. The specificity of the coimmunoprecipitation obtained with either mAb 12CA5 or mAb CC3 was demonstrated by the fact that ZNF74 and pol IIo were undetectable when nontransfected cell extracts (Fig. 6A) or a control IgG directed against the nuclear lamins (mAb 131C3) (Fig. 6B) was used.

[View Larger Version of this Image (24K GIF file)]

To test whether ZNF74-pol IIo interaction may occur in vivo, we then tested whether both proteins were found in the same cell nuclear compartments. For this purpose, we performed double immunofluorescence labeling in COS-7 cells transfected with HA-tagged ZNF74. Most COS-7 cells overexpressing ZNF74 showed a clear colocalization of ZNF74 and pol IIo in discrete subnuclear domains that exhibited irregular shapes and often apparent interconnections. As illustrated, the fluorescein signal identifying ZNF74 in the nucleus of a COS-7 cell (Fig. 7A) was coincident with the rhodamine signal specific for the pol IIo/CC3 antigen (Fig. 7B) as seen in a merged image (Fig. 7C). Previous studies have shown that the pol IIo (22)/CC3 antigen (42) colocalizes in irregularly shaped nuclear speckles with the non-snRNP splicing protein SC35, a member of the SR protein family (32, 46). We thus also addressed the colocalization of ZNF74 with the splicing protein SC35 and found that the signals for these two proteins were overlapping (Fig. 7, D-F). As previously demonstrated by Vincent and co-workers (42), no change in the distribution of pol IIo/CC3 antigen and SC35 occurs after in situ extraction of cells with nonionic detergent, DNase I, and RNases to remove soluble components of the nucleus. Similarly, ZNF74 colocalization with pol IIo/CC3 antigen and SC35 splicing protein persisted following this treatment (not shown), indicating a tight association of these three proteins with the nuclear scaffold.

[View Larger Version of this Image (93K GIF file)]

DISCUSSION

We previously reported that ZNF74, a KRAB domain-containing protein with 12 tandemly repeated Cys₂-His₂ zinc finger motifs, is an RNA-binding protein tightly associated with the nuclear matrix (15). In the present study, we show that the multifinger domain of ZNF74 can function as a nuclear matrix targeting sequence and that it interacts in a phosphodependent manner with the pol IIo, a form of the RNA pol II LS functionally associated with pre-mRNA elongation and processing. Furthermore, we provide evidence that ZNF74 colocalizes in vivo with pol IIo in irregularly shaped nuclear domains enriched in splicing factors.

The results presented here indicate that the ZNF74 multifinger domain is sufficient for the tight association of ZNF74 KRAB/Cys₂-His₂ protein with the nuclear matrix and that this domain serves as a nuclear matrix targeting signal as demonstrated by its ability to target heterologous cytoplasmic or nucleoplasmic proteins to the nuclear scaffold. This exclusive interaction of ZNF74 with the nuclear matrix and its absence from the nucleoplasm contrast with the observation that several transcription factors, including the Cys₂-His₂ trifinger protein Sp1, partition between these two nuclear compartments (47). Therefore, our results provide the first evidence that a zinc finger domain can function as a nuclear matrix targeting sequence.

From our previous (15) and present results, the multifinger domain of ZNF74 appears to be plurifunctional, being responsible for ZNF74 nuclear translocation, nuclear matrix attachment, RNA binding properties, DNA binding properties,² and direct interaction with the pol IIo, a form of the RNA pol II LS hyperphosphorylated on its CTD. While both nucleic acids and the hyperphosphorylated pol IIo CTD are long negatively charged polymers, the ZNF74 zinc finger domain does not promiscuously interact with polyphosphate polymers, because it binds selectively to only some of the polyribosides previously tested (15) and because it binds solely to the hyperphosphorylated pol IIo but not to partially dephosphorylated forms of the polymerase generated by limited phosphatase treatment.

The nucleic acids and pol IIo binding properties of ZNF74 zinc finger domain suggest that ZNF74 KRAB Cys₂-His₂ protein may be involved in multiple steps of the RNA synthesis and maturation. ZNF74 interaction with the hyperphosphorylated RNA pol II LS is of particular interest in this context. The RNA pol II LS is part of a multisubunit complex responsible for the synthesis and processing of pre-mRNAs in all eukaryotic cells and exists as previously mentioned in two main forms, the hyperphosphorylated pol IIo and the hypophosphorylated pol IIa. The fact that ZNF74 interacts specifically with pol IIo in a phosphodependent manner may provide insights concerning the functional significance of this interaction. Indeed, several studies indicate that pol IIo has specific functions that differ from pol IIa. The pol IIa is more efficiently recruited in vitro to preinitiation complexes, whereas pol IIo is observed during initiation of transcription (promotor clearance) and transcription elongation (45, 48-50).

While the role of pol IIo in transcription remains unclear (Refs. 51 and 52 and references therein), several groups recently provided independent evidence that the RNA pol II LS, most probably via its phosphorylated CTD, may participate in RNA processing (53). These include experiments showing that the pol IIo, but not the pol IIa, is associated with splicing complexes isolated in vitro (23, 24, 43) and is coimmunoprecipitated with snRNPs and proteins of the non-snRNP SR family (24, 25). Furthermore, in contrast to pol IIa that interacts with proteins associated with the preinitiation transcription complex (such as TATA-binding protein, TFIIE, and TFIIF) (Refs. 25 and 26 and references therein), the only proteins shown to date to directly interact with pol IIo, namely members of the SR family (54, 55) and CstF/CPSF cleavage-polyadenylation factors (27), are from two protein classes involved in RNA processing. Taking advantage of the fact that all known phosphorylation sites of the RNA pol II LS have been located to its heptapeptide (YSPTSPS) repeats within the CTD (44, 45), several very recent studies have further documented the importance of the pol IIo phosphorylation state in RNA processing. Indeed, these studies showed (i) the inhibition of in vitro splicing in the presence of anti-pol IIo antibodies recognizing specifically the phosphorylated CTD (43) or in the presence of CTD heptapeptides (peptides with mutated phosphorylation sites are not inhibitory) (54), (ii) the dispersal of snRNPs and SR splicing factors from discrete subnuclear domains with a speckled pattern as well as the in vivo inhibition of splicing in transfected cells overexpressing an in vivo phosphorylated RNA pol II LS CTD domain (26), and (iii) the requirement of the CTD for efficient in vivo splicing, 3-end cleavage, and polyadenylation of mRNA precursors (26, 27).

Given that all proteins shown to interact with pol IIo are contributing to RNA processing and considering the emerging role of pol IIo in this process, it is tempting to speculate that ZNF74, due to its phosphodependent interaction with pol IIo, represents a new class of proteins involved in the regulation of RNA maturation. Supporting this proposal, ZNF74 was also found to colocalize with Pol IIo in irregularly shaped subdomains of the nucleus enriched in the splicing SC35 protein and previously reported to be also enriched in other proteins of the SR family as well as in factors involved in the cleavage and ligation steps essential to generate mature mRNA (21, 22, 24, 42). While there is some controversy concerning the functional involvement of such nuclear subdomains exhibiting a speckled pattern in splicing, since some studies suggested that they may represent sites of storage for splicing factors rather than the sites of active splicing (56, 57), recent results indicate that pre-mRNA processing can take place within such domains (33, 46, 58-62). By examining the nuclear localization of transiently and stably expressed nascent RNA transcripts containing or lacking introns as well as the localization of splicing factors, Huang and Spector (58) furthermore demonstrated the intron-dependent recruitment of pre-mRNA splicing factors to sites of transcription seen as nuclear subdomains ranging from small dots to larger clusters. Huang and Spector (58) also provided a reconciling view by proposing that, depending on the relative rate of splicing and transcription of a studied template and on its copy number at one locus, the splicing factors may or may not be in sufficient abundance at the transcription sites to be detected by the light microscope, whereas they are easily detectable in storage sites where they are present in high amounts.

Because ZNF74 is an RNA-binding protein (15) that (i) colocalizes with the pol IIo and the SC35 splicing factor in subnuclear domains, (ii) remains tightly associated with splicing factors in the nuclear matrix, (iii) coimmunoprecipitates with pol IIo complexes, and (iv) interacts directly with pol IIo in vitro as a few proteins involved in mRNA processing, we suggest that this KRAB/Cys₂-His₂ protein, the expression of which is restricted to fetal tissues (19), plays a role in the developmental regulation of pre-mRNA maturation. The multifunctional properties of the ZNF74 zinc finger domain raise the intriguing possibility that ZNF74 KRAB Cys₂-His₂ protein may be involved in transcriptional regulation via its DNA binding activity and in RNA processing as a result of its interaction with RNA and pol IIo. A dual transcriptional and post-transcriptional role has been previously proposed for the Wilm's tumor protein, a Cys₂-His₂ tetrafinger protein that also behaves as a DNA- and an RNA-binding protein, whereas it associates with speckled subdomains of the nucleus in an RNA-dependent fashion (63, 64). Considering that ZNF74 interacts with pol IIo like few other described pre-mRNA maturating factors, different modes of action where ZNF74 may facilitate or compete against (65) the recruitment of splicing or maturating factors to pre-mRNA may be envisaged. It will be interesting to determine whether the multifunctional properties of the ZNF74 multifinger domain, and in particular its ability to interact with the pol IIo within specialized nuclear subdomains, may be a general trait of the large KRAB/Cys₂-His₂ family.