Novel interaction between the transcription factor CHOP (GADD153) and the ribosomal protein FTE/S3a modulates erythropoiesis.

The transcription factor CHOP (GADD153) heterodimerizes with other C/EBP family members, especially C/EBPbeta, thus preventing their homodimerization and binding to DNA sequences specific for the homodimers. Some CHOP-C/EBP heterodimers apparently bind to alternative DNA sequence and thereby regulate the transcription of other genes. Recently, we demonstrated that CHOP is up-regulated during certain stages of erythroid differentiation and that ectopic overexpression of CHOP enhances this process (Coutts, M., Cui, K., Davis, K. L., Keutzer, J. C., and Sytkowski, A. J. (1999) Blood 93, 3369-3378). In the present study, we report that CHOP also interacts with another non-C/EBP protein designated v-fos transformation effector (FTE) (Kho, C. J., and Zarbl, H. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 2200-2204), which is identical to ribosomal protein S3a (Metspalu, A., Rebane, A., Hoth, S., Pooga, M., Stahl, J. , and Kruppa, J. (1992) Gene (Amst.) 119, 313-316). Bacterially expressed His-CHOP and in vitro translated (35)S-labeled FTE/S3a-Gal4 fusion protein co-immunoprecipitated using anti-CHOP antibodies, and both anti-CHOP and anti-FTE/S3a antibodies co-immunoprecipitated CHOP and FTE/S3a from lysates of Rauscher murine erythroleukemia cells overexpressing both proteins. The in vivo interaction of CHOP and FTE/S3a was also demonstrated in cells overexpressing FTE/S3a but with endogenous expression levels of CHOP. Western blot analysis demonstrated co-localization of CHOP and FTE/S3a in both the cytosol and the nuclei of non-transfected cells. Overexpression of FTE/S3a inhibited differentiation of Rauscher cells induced either by erythropoietin or by dimethyl sulfoxide. This inhibition was reversed partially by simultaneous overexpression of CHOP or of antisense fte/S3a. FTE/S3a appears to be a bifunctional ribosomal protein that regulates CHOP and, hence, C/EBP function during erythropoiesis.

A role in erythroid growth and development has been shown or suggested for several regulatory proteins including Myc, Myb, GATA-1, and NF-E2 (4, 7, 8, 14 -19). Recently, we found that Epo up-regulates the expression of CHOP (gadd153) (20 -24), a member of the C/EBP family of transcription factors (25). Gain-of-function studies indicated that increasing CHOP expression enhances hemoglobinization of erythroid cells, indicating a functional role for CHOP in erythroid differentiation. Additionally, we obtained evidence that CHOP protein can bind to several nuclear proteins from erythroid cells, potentially some that are not C/EBP family members.
We have now used the yeast two-hybrid system (26) to screen an erythroid cell cDNA library for proteins that interact with CHOP. We report that CHOP interacts with a non-C/EBP protein designated v-fos transformation effector (FTE) (27) which is identical to ribosomal protein S3a (28). Our results indicate that this interaction inhibits the ability of CHOP to enhance erythroid differentiation.

EXPERIMENTAL PROCEDURES
Rauscher Murine Erythroleukemia Cell cDNA Library Construction and Yeast Two-hybrid Screen-A library from Rauscher murine erythroleukemia cells (29,30) was constructed into the pAD-Gal4 vector. Briefly, 5 g of poly(A) ϩ RNA was converted to double-stranded DNA using Stratascript RNase H Ϫ reverse transcriptase (Stratagene). First strand synthesis was primed with an oligonucleotide containing poly(dT) and an XhoI site (31). Second strand synthesis was carried out using Escherichia coli DNA polymerase I and RNase H (32). The cDNA ends were filled with T4 DNA polymerase, and the internal EcoRI sites of the cDNA were methylated with EcoRI methylase. Following EcoRI linker addition and size fractionation in a 1% agarose gel, doublestranded cDNA over 800 base pairs was cleaved with EcoRI and XhoI. The fragments purified from agarose gel were ligated into the Hybri-ZAP vector in unidirection (Stratagene). After in vitro packaging, 10 6 recombinant phages were plated, and the Hybrizap Lambda library was converted into the phagemid library pAD-GAL4 after in vitro mass excision.
For the expression of CHOP as a bait in the yeast two-hybrid system, two restriction enzyme cleavage sites were introduced into the two ends of CHOP cDNA by the polymerase chain reaction (PCR) using the following primers, 5Ј-TTCGGACGACCTAGTGCAAGCCGA-3Ј and 5Ј-TTAAGGAATTCGCAGCAGCTGAGTCCTGCCTT-3Ј. The first primer introduced a SalI second site at the 5Ј end of the cDNA, and the second primer added an EcoRI site at the 3Ј end of cDNA. The 5Ј end nucleotide sequence of CHOP was also adjusted by primer design so that the CHOP cDNA was in the right reading frame of the Gal-4 DNA binding domain. After cleavage, the PCR product was cloned into the SalI and EcoRI sites of the pBD-Gal4 yeast vector. The library was used to screen for CHOP binding partners following the manufacturer's protocol (Stratagene). Positive yeast clones were selected by prototrophy for histidine and expression of ␤-galactosidase. Yeast DNA was recovered and transformed into E. coli. Plasmids containing cDNA clones were identified by restriction mapping and were further characterized by DNA sequencing. Subsequent two-hybrid interaction was carried out with the positive and negative controls, including plasmids containing the Gal4 DNA-binding (pBD-Gal4) and Gal4 activation (pAD-Gal4) domains in Saccharomyces cerevisiae strain SFY526.
In Vitro and in Vivo Expression Constructs-For in vitro transcription and translation, fte-1 cDNA fused with Gal-4 activation domain was cleaved from the pAD-GAL4 vector with HindIII and SalI and subcloned into the same sites of the pSP72 vector (Promega). Sp6 polymerase was used for in vitro transcription and translation assays according to the manufacturer's procedure. For expression of CHOP protein in E. coli, BamHI and EcoRI restriction sites were introduced at the 5Ј and 3Ј ends of the cDNA, respectively, by PCR using the following two primers, 5Ј-TTAAGGGATCCCAGCTGAGTCCCTG-3Ј and 5Ј-TTCGGAATTCCTATGTGCAAGCCGA-3Ј. The PCR product was cleaved with BamHI and EcoRI and was cloned into the pTrc-His expression vector (Invitrogen) in the reading frame. Therefore, CHOP was expressed as a His-tagged protein. For mammalian cell expression of CHOP, an SspI/EcoRI fragment containing CHOP cDNA from the pTrc.His vector was blunted using mung bean nuclease and was subcloned into the SmaI site of pSVK3 expression vector (Amersham Pharmacia Biotech) under the transcriptional control of SV40 early promotor. The orientation of the insert was confirmed by restriction enzyme mapping and DNA sequence analysis. In order to express fte-1 in the mammalian cells, the 5Ј-untranslated region of rat fte-1 and one start codon (ATG) was added to the 5Ј end of mouse fte-1 cDNA by PCR using following primers, 5Ј-GGGAGTACTAGTGGGTCTCAGAGCGGCACC-ATGGCGGTCGGCAAGCAAGAACAACAAG-3Ј and SP6 primer 5Ј-GA-TATCACAGTGGATTTA-3Ј. The PCR product was cloned into the pZeoSV vector (Invitrogen) for eukaryotic expression under the transcriptional control of SV40 promotor and enhancer with Zeocin resistance as a selection marker.
Antibodies to Ribosomal Proteins-Polyclonal antisera recognizing FTE/S3a or ribosomal protein S26 were raised in rabbits or goats by immunization with the corresponding ribosomal proteins purified from rat liver ribosomes by a combination of carboxymethylcellulose chromatography, reversed phase liquid chromatography, and polyacrylamide gel electrophoresis. From the sera, monospecific antibodies were prepared by immunosorption to the purified ribosomal proteins immobilized to CNBr-Sepharose 4B (Amersham Pharmacia Biotech) as described (33). 35 S Labeling and Immunoprecipitation-Rabbit polyclonal antisera recognizing His-CHOP were prepared by Organon Teknika using antigen supplied by us. The polyclonal antibody was affinity purified using His-CHOP as the ligand covalently linked to Affi-Gel 15 (Bio-Rad). 35 S-Labeled proteins were generated with TNT SP6 polymerase-coupled Reticulocyte Lysate System according to the manufacturer's protocol (Promega). Five l of the 50-l translation mixture was used for immunoprecipitation. In vitro translation products were combined with 250 l of binding buffer (20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, and 0.25 mM dithiothreitol) and precleared by addition of 10 l of preimmune rabbit serum. After the mixture containing preimmune serum was incubated for 1 h at 4°C, 10 l of immunoprecipitin (Life Technologies, Inc.) were added, and the mixture was incubated for another hour. Nonspecific immune complexes were pelleted by slow speed centrifugation and discarded. The supernatant was mixed with 3 l of bacterially expressed CHOP (6.0 g), anti-CHOP antibody, and immunoprecipitin. The mixture was incubated at 4°C for 2 h. The beads were collected by brief centrifugation and washed four times with washing buffer (same as binding buffer except containing 200 mM NaCl). The pellet was resuspended in 50 l of binding buffer, and the slurry was heated to 100°C for 5 min in a water bath. After centrifugation in a microcentrifuge for 2 min, the supernatant was carefully removed, and 200 l of binding buffer was added. Another round of immunoprecipitation was carried out as described above. After extensive washing, the precipitate was fractionated on SDS-polyacrylamide gels. The gel was dried and exposed to x-ray film.
Western Blot Analysis of FTE/S3a and CHOP-Rauscher cells were harvested by centrifugation and were washed twice with ice-cold phosphate-buffered saline. A 5-fold volume excess/packed-cell pellet of Dignam buffer A (34) was added to the pellets that were gently resuspended by low speed vortexing. The mixture was incubated at 0°C for 10 min and was centrifuged at low speed in a microcentrifuge followed by vigorous vortexing. The nuclei were pelleted; the supernatant was removed, and four original cell pellet volumes of Dignam buffer C were added. The nuclei were lysed by 20 strokes of a Dounce homogenizer (type B pestle). The lysate was shaken in a microcentrifuge tube on a rotating shaker for 30 min and then centrifuged for 10 min at maximum speed in a microcentrifuge. The protein concentration was determined with a Micro BCA kit (Pierce).
For determination of the relative apparent molecular weights of FTE/S3a and CHOP, 100 g of both cytoplasmic and nuclear proteins were subjected to SDS-PAGE on 12:0.32% acrylamide/bisacrylamide polyacrylamide gels. After electrophoretic transfer to nitrocellulose, the membrane was cut in half so that each half of the membrane contained a lane of cytoplasmic protein and a lane of nuclear protein with prestained molecular weight standards on the right side of each halfmembrane. Anti-CHOP and anti-FTE/S3a antibodies were used to detect CHOP and FTE/S3a on each membrane half, respectively.
For studies to determine whether CHOP and FTE/S3a interact within the ribosomal particle, 300 g of cytoplasmic or nuclear proteins were used for immunoprecipitation by anti-CHOP antibody (Santa Cruz Biotechnology) after clearing with preimmune rabbit serum. The immunoprecipitation pellets were mixed with reducing SDS sample buffer, were boiled for 4 min, were centrifuged for 3 min, and were immediately put on ice. The samples were subjected to SDS-PAGE on a 15:0.4% acrylamide/bisacrylamide SDS gel. The proteins were transferred to nitrocellulose membrane (Millipore, 0.45 m) for 2 h in a mini-gel protein transfer apparatus (Bio-Rad) in regular transfer buffer (39 mM glycine, 48 mM Tris-HCl, and 20% methanol). The membrane was blocked with PBST (PBS ϩ 0.5% Tween) buffer containing 5% non-fat dry milk powder. After blocking, the membrane was incubated in the presence of anti-S26 antibody (20 g/ml) overnight. The membrane was washed twice for 5 min each and once for 15 min, separately. The horseradish peroxidase-conjugated secondary goat anti-rabbit antibody was added. The blot was allowed to incubate for 1 h, and the membrane was washed three times for 10 min each. Chemiluminescent detection of the bound complex was performed according to the manufacturer's instruction (Pierce).
Rauscher Cell Transfections-The day before transfection, Rauscher cells were plated at the density of 10 5 cells/ml transfection in 10-cm tissue culture dishes in DMEM, 10% fetal bovine serum. Transfection was conducted by electroporation using a Gene Pulser II (Bio-Rad). In most case, cells were analyzed 48 h after transfection. For stable transfection, harvested cells were divided into four tissue culture dishes and were cultured in DMEM, 10% fetal bovine serum in the presence of 250 g/Zeocin ml for 2 weeks. During this time, fresh medium containing Zeocin was added every 2 days. Individual colonies were pipetted into 96-well plates. Transfection efficiency was analyzed by determining the galactosidase activity according to supplier's procedure (Invitrogen). The cells were collected by centrifugation and washed twice with PBS solution and once with methionine-deficient RPMI, 10% FCS medium. Cells were resuspended in methionine-deficient medium at the density of 1 ϫ 10 6 cells/ml and incubated for 30 min to deplete the excess methionine. [ 35 S]Methionine (NEN Life Science Products) was added (0.5 mCi/ml), and the cells were incubated 1 h. Cells were collected by centrifugation, washed with PBS, and resuspended directly in the lysis buffer containing 20 mM HEPES, pH 7.5, 500 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.1% Triton X-100, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml pepstatin, 100 g aprotinin, 10 g leupeptin. Cells were incubated in lysis buffer, gently rocked for 1 h at 4°C, and centrifuged at 14,000 rpm for 10 min. The supernatant was collected, aliquoted, and stored at Ϫ80°C.

RESULTS
To carry out the two-hybrid screen, CHOP was fused with the Gal-4 DNA binding domain in a pBD-Gal4 vector. This bait plasmid was co-transformed into cells with a Rauscher murine erythroleukemia cell cDNA library fused with the Gal-4 activation domain in a pAD-Gal4 vector. Ten positive clones were obtained from 3 ϫ 10 6 transformants screened.
To determine the specificity of the interaction, plasmids containing activation domain fusion proteins from the putative positive clones were co-transformed into yeast with Gal-4 binding domain CHOP and control heterologous baits. Nine clones were found to interact with the Gal-4 DNA domain protein containing CHOP but not with the DNA vector alone or with the heterologous bait. DNA sequence analysis revealed that two of the nine clones encoded the mouse homologue of rat v-fos transformation effector-1 gene (fte-1) (27), also known as ribosomal protein S3a (28). Within the coding region, the mouse and rat nucleotide sequences are 97% identical. The deduced amino acid sequences are identical. Data base searches did not identify sequences with significant similarity to the other positive clones.
We demonstrated the interaction of the CHOP and FTE/S3a proteins in vitro. The fte/S3a Gal-4 DNA binding domain clones were subcloned into a pSP72 vector containing the SP6 promotor. 35 S-Labeled protein prepared by in vitro transcription and translation was tested for association with bacterially expressed recombinant CHOP protein. Immunoprecipitation with anti-CHOP antibodies showed that CHOP bound to the GAL4-FTE/S3a fusion protein (Fig. 1).
Next, we confirmed the in vivo interaction of FTE/S3a and CHOP in erythroid cells. Rauscher cells were co-transfected transiently with fte/S3a and CHOP cDNAs and incubated with [ 35 S]methionine. Lysates were prepared and were incubated either with anti-FTE/S3a or with anti-CHOP antibodies followed by precipitation with protein A-agarose, SDS-PAGE, and autoradiography. Incubation with either antibody resulted in co-immunoprecipitation of both CHOP and FTE/S3a as a closely spaced doublet (Fig. 2).
We also confirmed the identities of the co-immunoprecipitated proteins and demonstrated both cytosolic and nuclear co-localization of FTE/S3a and CHOP. Non-transfected Rauscher cells were lysed, and cytosolic and nuclear protein fractions were prepared and subjected to SDS-PAGE and Western blot analysis (Fig. 3). CHOP was detected in both the cytosolic and, in a significantly higher amount, in the nuclear fraction. It migrated as a single, well resolved species aligned precisely with the 29-kDa molecular mass standard. A trace amount of a higher molecular weight cross-reacting protein was also seen. FTE/S3a was also detected in both the cytosolic and the nuclear fractions but in approximately equal amounts. It also migrated as a single, well resolved species that, in contrast to CHOP, was slightly (ϳ2 mm) ahead of the 29-kDa molecular mass standard, thus identifying CHOP and FTE/S3a as the upper and lower members, respectively, of the closely spaced doublet seen in Fig. 2.
In another experiment we established a Rauscher cell line stably overexpressing fte/S3a. Cells were lysed and the cytosolic fraction was incubated with anti-CHOP antibody followed by immunoprecipitation, SDS-PAGE, and Western blot analysis with anti-FTE/S3a antibody (Fig. 4). Cytosolic fractions were also subjected to SDS-PAGE and Western blot. Anti-FTE/ S3a antibodies identified FTE/S3a in the cytosolic fractions and in the immunoprecipitate obtained with the anti-CHOP antibodies again running slightly ahead of the 29-kDa standard.
The evidence for co-localization of CHOP and FTE/S3a in both the nucleus and the cytosol prompted us to ask whether the interaction might also occur within the ribosome itself. Immunoelectron microscopy studies have mapped FTE/S3a to the surface of the small subunit and the 80 S ribosome (33). Therefore, we reasoned that if CHOP binds to FTE/S3a on the surface of the ribosome, immunoprecipitation with anti-CHOP antibody should result in the appearance of other ribosomal proteins in the immunoprecipitate, including proteins such as RPS26, which is located within the ribosomal particle rather than at its surface (33,35) and is not known to interact with FTE/S3a. We prepared cytosolic and nuclear extracts of Rauscher cells and carried out an immunoprecipitation with anti-CHOP antibody (see "Experimental Procedures"). Then the immunoprecipitates, along with purified ribosomal total protein (containing rpS26) and Rauscher cell cytosolic lysate, were subjected to SDS-PAGE, electrophoretic transfer to nitrocellulose, and probing with anti-S26 antibody. As seen in Fig. 5, S26 protein was detected readily in the control purified ribosomal protein fraction (Fig. 5, lane 1) and in the Rauscher cell cytosolic lysate (Fig. 5, lane 2). However, S26 was not detected in the anti-CHOP cytosolic immunoprecipitate (Fig. 5, lane 3) nor in the nuclear immunoprecipitate (Fig. 5, lane 4). Similar results were obtained in several repeat experiments. Occasionally, an extremely faint band was detected in the cytosolic immunoprecipitate. However, this was not seen regularly and may have represented a small amount of ribosomal contamination of the centrifuged immunoprecipitate pellet.
Northern blot analysis showed that fte/S3a transcript is approximately 0.9 kilobase pairs, close to the size of the cDNA clones. The mRNA level was not induced by Epo or dimethyl sulfoxide (Me 2 SO) (not shown).
Since we had discovered previously that up-regulation of CHOP enhanced hemoglobinization in response to Epo or Me 2 SO (25), and since CHOP and FTE/S3a interact, we hypothesized that FTE/S3a might participate in regulating erythroid differentiation. To test this, Rauscher cells were transfected with fte/S3a. Transfection efficiency was monitored by transfection of ␤-galactosidase plasmid. Stable clones were selected by Zeocin resistance with the transfection of pZeoSV as the control. Cells from different independent transfections were incubated in the absence or presence of Epo or Me 2 SO for 48 h. Differentiation was quantified by benzidine staining for hemoglobin production (36).
As shown in Fig. 6, cultures of uninduced cells, whether transfected or non-transfected with fte/S3a, contained very few (Ͻ1%) hemoglobin-positive (Hb ϩ ) cells, as expected (Fig. 6, Ϫ/Ϫ). Non-transfected cells induced to differentiate with Epo (white bars) or Me 2 SO (black bars) were 34 and 49% Hb ϩ , respectively (Fig. 6, ϩ/Ϫ). In contrast, induced differentiation of fte/S3a-transfected cells was reduced markedly to 8 and 7%, respectively (Fig. 6, ϩ/FTE). This inhibition of differentiation in cells overexpressing fte/S3a was reversed partially by cotransfecting these overexpressing cells with an antisense fte/ S3a construct transiently (17 and 29% Hb ϩ , respectively) (Fig.  6, ϩ/FTE/AS FTE). This degree of reversal was consistent with the observed transfection efficiency of 30 -50%. Western blot analysis showed that the FTE/S3a protein concentration in the FTE/S3a-overexpressing cells was 480% that of the nontransfected cells. Transfection with an antisense fte/S3a construct reduced this to 170% that of the non-transfected cells (Fig. 7). These results suggested to us that increased formation of FTE/S3a-CHOP heterodimers in cells overexpressing fte/ S3a might prevent endogenous CHOP from interacting with other proteins, including C/EBPs, interactions of which enhance erythroid differentiation. This was supported by the results of transiently overexpressing CHOP in the fte/S3atransfected cells (Fig. 6, FTE/CHOP). Such CHOP overexpression also resulted in partial reversal of the FTE/S3a inhibition of hemoglobinization (18 and 22% Hb ϩ , respectively), similar to the results obtained with antisense fte/S3a. As observed by us previously (24), transient transfection with CHOP alone without fte/S3a overexpression enhanced hemoglobinization significantly to 53 and 62% Hb ϩ , respectively (Fig. 6, ϩ/CHOP). Transfection with antisense fte/S3a alone (Fig. 6, ϩ/AS FTE) had no significant effect. DISCUSSION Our results demonstrate a novel interaction between the transcription factor CHOP, a C/EBP family member, and FTE/ S3a, a ribosomal protein. This interaction appears to play a role in modulating erythroid differentiation. We speculate that the interaction of CHOP with one or more other proteins (especially C/EBP␤, its principle binding partner among the C/EBP family members) plays a positive role in enhancing erythroid differentiation and that its interaction with FTE/S3a interferes with this. However, it is also possible that FTE/S3a not bound to CHOP inhibits erythroid differentiation by some other mechanism and that the increased CHOP expression seen with differentiation binds to FTE/S3a and blocks this action.
FTE/S3a has been reported as a v-fos transformation effector (27) and as a ribosomal protein (28). Kho et al. (37) have attempted to reconcile this apparent dilemma and have suggested that ribosomal subunit numbers and rates of protein synthesis are important effectors of cell proliferation and neoplastic transformation. Recently, Naora et al. (38) reported that inhibiting expression of S3a could induce apoptosis in NIH 3T3 cells.
In 1992, Gordon et al. (39) found that FTE/S3a was induced by TNF-␣ in a transformed mouse embryo fibroblast cell line. Elevated expression of FTE/S3a mRNA was considered a char- acteristic feature of selected, but not all, malignancies supporting involvement of FTE/S3a in tumor development and progression. Other studies have shown that FTE/S3a expression is often higher in tumor cells than in normal cells; however, there was not a strong correlation between FTE/S3a expression levels and proliferative activities among the range of cell types examined (40,41).
In mouse thymus, dexamethasone induced FTE/S3a expression and caused apoptosis within 3-6 h (40). Apoptosis was also triggered by actinomycin in the murine HL60 cell line (41) in which high levels of FTE/S3a are constitutively expressed. In cell lines with low constitutive levels of FTE/S3a expression, however, actinomycin D did not induce apoptosis. In the human promyelocytic leukemia cell line HL-60, antisense sequences of the FTE/S3a gene have been reported to induce apoptosis (42). In feline thymic tumor, FTE/S3a steady state mRNA levels were 2.7-fold higher than in normal thymus (43). Overexpression of the FTE/S3a gene was also found in surgical specimens of both primary tumors and metastatic foci of medullary thyroid carcinomas and in a panel of recurrent and metastatic medullary thyroid carcinoma, when compared with normal thyroid tissue and normal lymphatic tissue. FTE/S3a was proposed to play an important role in the progression and metastatic spread of this carcinoma (44). Interestingly, FTE/S3a was also among the up-regulated genes involved in hyperfiltration and hypertrophy of the kidney (45) as detected by representational difference analysis of cDNA of 5/6 nephrectomized and sham-operated mice suggesting that FTE/S3a maybe involved in the pathophysiological process of initial glomerular hyperfiltration and subsequent development of glomerulosclerosis of the remnant kidney. In Drosophila melanogaster, antisense suppression of the FTE/S3a gene has been reported to disrupt ovarian development (see Ref. 46).
Numerous studies have demonstrated either an up-regulation or down-regulation of FTE/S3a expression during differentiation of several cell types, implying a regulatory function for FTE/S3a. In the case of erythroid differentiation reported here, the question arises whether CHOP interacts with FTE/ S3a separately from or integrated into the ribosomal structure. During ribosome biogenesis within the nucleolus, FTE/S3a joins preribosomal particles in an initial stage; it belongs to the group of "early binding proteins" (47). In cytoplasmic ribosomes, FTE/S3a is at least partly exposed at the surface. Topologically, FTE/S3a is located at the so-called protuberance of the 40 S ribosomal subunit (33), a region that is accessible to anti-FTE/S3a antibodies, and on the surface of the 80 S ribosome. FTE/S3a has been cross-linked in polysomes with the 3ЈOH end of 18 S RNA (48) and shown to be directly involved in formation of functional ribosomal sites using cross-linking, affinity labeling, and immunoelectron microscopy. FTE/S3a has been characterized as a ribosomal protein involved in interactions of the 40 S subunit with initiation factors eIF-2␣ and eIF-2␤ (49,50), initiator-tRNA (51), initiation factor eIF-3 (52)(53)(54), mRNA (52,(55)(56)(57), and of the 80 S ribosome with elongation factor EF-2 (58,59), each of which has regulatory implications. Whether CHOP may take part in this process and influence ribosome maturation remains an unanswered question. The results presented here favor the interaction of CHOP with FTE/S3a as an isolated protein, either in the nucleus or the cytoplasm, and not with FTE/S3a as an integral part of preribosomal or ribosomal particles. However, it is still possible that CHOP also binds to 40 S or 80 S ribosomes. Additional experiments will be necessary to analyze possible binding sites and/or direct influences on ribosome activity.
In the adipocytic lineage, CHOP is related to inhibition of differentiation. The inhibition of adipocytic differentiation appears to depend on the ability of CHOP to antagonize the activities of C/EBP proteins at several levels (60). In our yeast two-hybrid screen we did not detect any C/EBP family members. However, this does not preclude the possibility of association of CHOP with C/EBP proteins in erythroid cells. Indeed, our previous work showed low levels of C/EBP␣, -␤, and -␦ mRNA in Rauscher cells, with C/EBP␤ expression increasing along with that of CHOP (25). Initially, it may seem paradoxical that CHOP enhances differentiation in erythroid cells and inhibits it in adipocytes. However, we have now shown that CHOP has a novel alternative non-C/EBP binding partner, FTE/S3a. Thus, the effects of CHOP on cell differentiation may also depend upon these alternative interactions and may be quite lineage-specific.