Critical Role for the Oct-2/OCA-B Partnership in Ig-Secreting Cells¹

Immunoglobulin-secreting cells, as represented by plasmacytomas, express Oct-2, PU-1, and OCA-B. Oct-2 and OCA-B are transcription factors largely limited to the B lymphocyte lineage, whereas PU.1 is also expressed in myeloid cells (1, 2, 3, 4, 5). All three transcription factors are expressed throughout B cell development, as represented by pre-B, B, and Ig-secreting cell lines (2, 6, 7). Gene knockout studies have shown that PU.1-deficient mice lack B lymphoid progenitors, ascribing a function to this transcription factor even before cell commitment to this lineage (8, 9). The B cell lineage appears to develop normally up to the surface Ig⁺ cell stage in the absence of either Oct-2 or OCA-B or in the absence of both factors (10, 11, 12, 13). Past this stage, however, B lymphocyte function is impaired, and OCA-B/Oct-2-deficient mice lack germinal centers, the sites where activated B lymphocytes normally proliferate and differentiate into memory and Ig-secreting cells (13).

Although these gene knockout experiments demonstrate a function for each of these transcription factors at times before the Ig-secreting cell stage, all three proteins are also produced in this terminally differentiated effector cell. We have used an alternate experimental strategy that, unlike the conventional gene knockout approach, has allowed us to explore these transcription factors’ functions in these late-stage, Ig-secreting cells. Cell fusions between a murine plasmacytoma MPC11 (Ig-secreting cells) and the murine T lymphoma BW5147 lead to silencing of all genes specifically expressed by the plasmacytoma (14, 15). The hybrid lines, instead, express genes unique to the T lymphoma (16). We have used this experimental system to explore the means by which one genetic program is chosen over another and to identify the regulatory proteins essential to each program.

As previously reported, Oct-2, OCA-B, and PU.1 are silenced in hybrids arising from plasmacytoma X T lymphoma fusions (17). We found, however, that if we sustained constitutive expression of Oct-2 over the course of cell fusion, using an Oct-2 transgene under viral promoter control, the phenotype of resulting hybrids was completely reversed. All hybrids were now plasmacytoma-like. This led us to conclude that Oct-2 plays a pivotal role in sustaining the gene expression profile of the Ig-secreting cell, and its constitutive expression shifts T lymphoma-programmed chromosomes toward expression of the plasmacytoma program. In other experiments we showed that most of these plasmacytoma-like hybrids retained a full complement of both T lymphoma and plasmacytoma-derived chromosomes (16).

In more recent experiments we have shown that the related octamer-binding transcription factor Oct-1 cannot supplant Oct-2 in this function and that the C-terminal domain of Oct-2 distinguishes it from Oct-1 in this regard (18). One question arising from these findings is whether any plasmacytoma-specific gene encoding a regulatory factor can shift the balance in favor of the plasmacyte’s genetic program if this gene is shielded from silencing at the time of cell fusion. Alternatively, Oct-2 might play a unique role, and genes encoding other tissue-specific transcription factors lie downstream of Oct-2’s dominant effect on programming.

To address this question, we tested the effect of introducing a gene encoding PU.1 or a gene encoding OCA-B into the plasmacytoma or T lymphoma parent before cell fusion. As in our previous studies with Oct-2, the transgenes were under CMV promoter control so that they were not subject to fusion-mediated silencing. As detailed below, we found that when the OCA-B transgene was introduced into the plasmacytoma before cell fusion, this transcription factor, like Oct-2, could shift the hybrid cell phenotype to that of the plasmacytoma parental line. In contrast, PU.1, a transcription factor pivotal to the early events giving rise to the B lymphocyte lineage, could not do so and so must play a subordinate role in the Ig-secreting cell. Previous work from our laboratory suggested that the function of human Oct-2 (hOct-2) in Ig-secreting cells requires a tissue-specific (plasmacytoma-specific) coactivator (17). In the present study we show that OCA-B is that cofactor and works synergistically with Oct-2 to sustain the gene expression profile of the Ig-secreting cell.

BW5147.G.1.4.OUA^R is a variant subline of an AKR/J mouse thymoma resistant to 10⁻⁴ M 6-thioguanine and to 10⁻³ ouabain (ATCC CRL 1588; American Type Culture Collection, Manassas, VA) (19). MPC11 refers to a γ2b/κ-producing cell line (45.6.2.4) derived from a BALB/c mouse tumor (20). Both cell lines were maintained in DMEM (catalogue no. 12100-061; Life Technologies, Grand Island, NY) containing 10% bovine calf serum (catalogue no. A-2151-L; HyClone Laboratories, Logan, UT), 100 U/ml penicillin, 100 μg/ml streptomycin (catalogue no. 15140-015; Life Technologies), and 0.1 mM nonessential amino acids (catalogue no. 11140-019; Life Technologies).

The pCGN-OCA-B produces flu epitope-tagged hOCA-B (accession no. Z47550; provided by Dr. R. Roeder, Rockefeller University, New York, NY) (2). The vector has a CMV promoter and was cotransfected with the pSV2-hisD or pSV2-neo^r vectors (21, 22) for selection of transformants. Human and murine OCA-B share 89% amino acid identity (23).

BCMGSneo-PU.1 produces murine PU.1 under control of the CMV promoter (provided by the late Dr. M. Koshland, University of California, Berkeley, CA). BCMGSneo lacks PU.1-coding sequences and is a derivative of the expression vector BMGNeo (24).

The c-fms promoter sequence (−417 to +71) upstream of the PXP2 luciferase reporter gene was used in trans-activation studies (provided by Drs. D. Tenen and H. Radomska, Harvard Institutes of Medicine, Boston, MA). The c-fms gene encodes the receptor for M-CSF. The PXP2-luciferase vectors have been described previously (25). The c-fms promoter was excised, using BamHI-BglII, to create a promoterless construct (PXP2).

The pCGNOct-2his produces flu epitope-tagged hOct-2 under control of the CMV promoter and also expresses the selectable marker gene hisD (17). Human and murine Oct-2s have 97% amino acid identity (GenBank accession no. NM002698 and NM011138, respectively).

Transfections were performed as previously described (17). Briefly, 10⁷ cells were transfected by electroporation with 10 μg of linearized plasmid DNA (and 1 μg of selectable marker gene, where relevant).

BCMGSneo-PU.1 was linearized with PvuI, and selection was performed in 2 mg/ml G418 for MPC11 transfections (catalog no. 860-1811IJ; Life Technologies) and 3 mg/ml G418 for BW5147 transfections. The pCGN-OCA-B was linearized with DraIII, and the cotransfected vector (pSV2hisD or pSV2 neo) was linearized with EcoRI. MPC11-OCA-B clones (cotransfected with pSV2hisD) were selected in 5 mM histidinol medium. BW-OCA-B clones (cotransfected with pSV2neo) were selected in 3 mg/ml G418 medium.

BW5147 cells (4 × 10⁶) were transfected by electroporation at 280 V and 960 μF. Cells were transfected with 18 μg of reporter plasmid (c-fms-PXP2 or PXP2 luciferase), 8 μg of either BCMGSneo-PU.1 or BCMGSneo, and 2 μg of β-galactosidase expression vector. The latter was added to normalize for transfection efficiency. Cells were harvested 21 h post-transfection and assayed for β-galactosidase and luciferase activities with commercially available kits (luciferase assay, catalogue no. E1501 (Promega, Madison WI); β-galactosidase assay, catalogue no. E2000 (Promega)).

Electrofusion of somatic cells has been described previously (26). In brief, 5 × 10⁶ cells of each parental line were combined and electrofused. Approximately 48 h postfusion, hybrids were selectively grown in complete DMEM containing 10⁻⁴ M hypoxanthine, 4 × 10⁻⁶ M aminopterin, and 1.6 × 10⁻⁴ M thymidine (HAT)⁴ and 10⁻³ M ouabain. MPC11 cells die in ouabain, whereas BW5147 cells die in HAT. In all fusions, growing hybrids were recovered in <30% of the wells and, therefore, generally represented single fusion events.

Octamer-binding proteins were identified in nuclear extracts prepared as previously described (27). Binding reactions and gel electrophoresis were performed as described previously (28). In brief, 10 μg of nuclear extracts were incubated at room temperature with 10⁴ cpm of end-labeled 51-bp fragment from the IgH enhancer (51-bp probe sequence: TCAGC AAAAC ACCAC CTGGG TAATT TGCAT TTCTA AAATA AGTTG AGGATT (octamer underlined)).

For PU.1 EMSAs, 10 μg of nuclear extracts were incubated on ice with 2.5 × 10⁴ cpm of end-labeled PU.1 box probe (PU.1 probe sequence: GATCCTGAAAGAGGAACTTGGTA (the underlined sequence is the core PU.1-binding sequence); mutant PU.1 probe sequence used in competition experiments: GATCCTGAAAGACCAACTTGGTA (italicized bases are those mutated)). For supershifts, 1 μl of an Ab to the N-terminal domain of PU.1 (1297 rabbit antisera, provided by Dr. R. Maki) was added to the nuclear extracts before addition of probe.

For competition studies, 140 ng of labeled PU.1 probe (sp. act., 1.0 × 10⁷ cpm/ng) was included in each reaction tube, and competitor was added in 50-, 100-, and 500 fold-molar excess immediately before radiolabeled probe.

Total cytoplasmic RNA was isolated by the TRIzol protocol (catalogue no. 15596; Life Technologies). Approximately 20 μg of RNA was denatured in formamide and size-fractionated in 1% agarose-formaldehyde gels (29) (DNA probes: J chain (1.2-kb cDNA, Jc21) (30) and γ2b (6.8-kb γ2b gene fragment) (31)). To normalize for RNA levels, blots were stripped and rehybridized to murine GAPDH (catalogue no. 7330; Ambion, Austin, TX).

Whole cell lysates were obtained by the freeze/thaw method. Briefly, ∼10⁷ cells were washed in 1× PBS and resuspended in extraction buffer (20 mM HEPES (pH 7.9), 20% glycerol, 400 mM KCl, 0.5 mM EDTA, 0.5 mM EGTA, 0.025% Nonidet P-40, and 0.5 mM DTT). Cells were freeze-thawed six times in liquid nitrogen. Extracts corresponding to 50 μg of total protein (Bradford assay, catalogue no. 500-0006; Bio-Rad, Hercules, CA) were size-fractionated by electrophoresis (8% SDS-PAGE) and proteins electrophoretically transferred to nitrocellulose membranes (Pure-Nitrocellulose transfer membrane WP2HYA0010; Bio-Rad Transblot apparatus) as previously described (29).

Flu-tagged proteins (OCA-B and Oct-2), were detected with anti-flu epitope mouse mAb, HA.11 (Covance MMS-101P/lot 142030001; Berkeley Antibodies, Richmond, CA) and HRP-conjugated rabbit anti-mouse IgG1 (61-0120; Xymed, South San Francisco, CA). Rabbit anti-Oct-2 (C-20; SC-233) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA), and HRP-conjugated donkey anti-rabbit whole Ig (NA93) was purchased from Amersham Pharmacia Biotech (Piscataway, NJ). Rabbit anti-Oct-1 (C-21; SC-232; Santa Cruz Biotechnology) and HRP-conjugated donkey anti-rabbit whole Ig were used. Rabbit anti-OCA-B (supplied by R. Roeder) and HRP-conjugated donkey anti-rabbit whole Ig were used. The IgH and IgL chains were detected with HRP-conjugated rabbit anti-mouse whole IgG.

Blots were developed with Super Signal chemiluminescent substrate (Pierce, Rockford, IL), and chemiluminescence was visualized by various exposures to Kodak (Rochester, NY) X-OMAT film.

PU.1 has been implicated in the regulation of IgH and IgL chain genes and of the gene encoding the Ig-polymerizing protein, J chain (32). PU.1-deficient mice and mice reconstituted with PU.1^−/− bone marrow lack the lymphoid and myeloid lineages, however, so that whereas these gene knockout experiments clearly show that PU.1 is required for early commitment to the lymphoid and myeloid lineages, they provide no information regarding PU.1’s function at subsequent stages of B cell development (8, 9, 33). As PU.1 is expressed throughout B cell development, it is likely to serve a function in several, if not all, developmental stages of this lineage.

PU.1, like Oct-2, is expressed in the Ig-secreting plasmacytoma MPC11, but not in the T lymphoma BW5147. In the context of this experimental system, therefore, we refer to both these transcription factors as plasmacytoma-specific. We asked whether PU.1, like Oct-2, could sustain the plasmacytoma-specific program in hybrids between these two cell lines. Alternatively, PU.1 might act downstream of Oct-2, being unable to rescue Oct-2 gene activity, but being capable of rescuing a subset of the genes rescued by Oct-2. Another possibility was that PU.1, acting downstream of Oct-2, would not be able to rescue the expression of any other tissue-specific genes. This would follow if these genes required both PU.1 and Oct-2 (or an Oct-2-induced factor) for their activity.

A PU.1 expression vector, BCMGS-Neo-PU.1, was used to stably transform the PU.1-negative BW5147 (T lymphoma) cell line and the MPC11 plasmacytoma line (see Materials and Methods). EMSAs identified functional PU.1 protein by virtue of its ability to bind a site derived from the SV40 enhancer (5, 34). Nuclear extracts from the plasmacytoma cell line MPC11, before transfection with the PU.1 expression vector, confirmed that murine PU.1 produced from the endogenous locus was able to bind specifically to this PU.1 box (Fig. 1,A). Unlabeled, wild-type oligonucleotide successfully competed with the labeled oligonucleotide for PU.1 binding, whereas a mutant version of the same oligonucleotide did not (Fig. 1,A). Nuclear extract from a BW5147 transformant carrying the BCMGS-Neo-PU.1 gene (BWPU.1#1) revealed the same shifted complex, whereas extract from the parental BW5147 cell line (BW) did not (Fig. 1,B). An Ab specific for the N terminus of PU.1 was used to confirm that this shifted complex contained PU.1 (see supershifted complex, PU.1/Ab, Fig. 1 B).

The plasmid-encoded PU.1 protein was indistinguishable from that produced in MPC11 from the endogenous gene. As a result, EMSAs could not be used to confirm expression of the transgene in MPC11 cells (Fig. 2 C, MP-PU.1 transf.). Instead, neo^r transformants were further tested for the presence of the transgene by genomic Southern and by a PCR that distinguished the endogenous and cloned genes for PU.1 (data not shown).

The ability of the transgene-encoded PU.1 to trans-activate an appropriate reporter gene was tested by transient assay. BW5147 cells were cotransfected with the PU.1 expression vector or a matched, empty vector and either a promoterless luciferase gene (PXP2) or a luciferase gene with c-fms promoter (c-fmsPXP2). The c-fms promoter (−417 to +71) has been shown previously to respond to PU.1 in HeLa cells (nonlymphoid human cell line) (35). As shown in Fig. 1 C, transgene-encoded PU.1 had no effect on the promoterless luciferase vector (PXP2 + PU.1), but augmented expression of the c-fms luciferase gene an average of 35-fold (cfmsPXP2 + PU.1). We could demonstrate, therefore, that the PU.1 protein produced from the BCMGS-Neo-PU.1 expression vector binds both DNA and behaves as a transcriptional activator when expressed in the BW5147 T lymphoma.

A T lymphoma transformant expressing PU.1, BWPU.1#1, was fused to the IgG2b-secreting plasmacytoma, MPC11. Hybrid cell lines were selected in medium containing HAT/ouabain as described in Materials and Methods and were tested for G418 resistance (evidence for the presence of the BCMGS-Neo-PU.1 gene). As in previous studies, DNA samples isolated from individual hybrids were analyzed by genomic Southern to assay for the presence of plasmacytoma and T cell-derived IgH loci (14). On the average, these intraspecific cell fusions result in the loss of about four MPC11-derived chromosomes per hybrid, although ∼45% of the hybrids lose no or only one MPC11-derived chromosome (16). As a functional IgH locus is present on only one MPC11-derived chromosome, Southern assays identified hybrid lines that remained informative with respect to Ig gene expression (retained MPC11-derived, assembled Ig genes; data not shown).

EMSA was used to confirm the continued expression of exogenous PU.1 in the BWPU.1#1 × MP hybrid lines. All confirmed hybrids carrying the PU.1 transgene expressed PU.1, albeit at varying levels (representative data, Fig. 2,B). In contrast, only one of six BW × MP hybrids (no exogenous PU.1) expressed PU.1 (Fig. 2,A, EMSA of control hybrids). In the latter hybrids, therefore, the endogenous PU.1 had been silenced, as is typical of plasmacytoma-specific genes in such cell fusions (a PCR assay confirmed the presence of the PU.1 gene in these hybrids; data not shown) (17). The one PU.1-expressing hybrid recovered is consistent with previous studies in which exceptional hybrids displaying the plasmacytoma phenotype arose at a low frequency (17). We conclude, therefore, that the PU.1 expressed in the BW-PU.1#1 × MPC11 hybrids is that encoded by the BCMGS-Neo-PU.1 transgene. Because the transgene and endogenous gene-encoded PU.1 proteins were indistinguishable, we were unable to determine whether the endogenous gene was also active (rescued from silencing) in these hybrid lines. In parallel fusions between BW5147 and MPC11 transformants expressing cloned PU.1 (MP-PU.1 transf), all resulting hybrids expressed PU.1 (Fig. 2,C), again in contrast to control hybrids (Fig. 2 A) (17).

We looked for the expression of Oct-2 and the coactivator OCA-B in control hybrids (BW × MP) and in hybrids in which the T lymphoma parent expressed PU.1 under viral promoter (CMV) control (BW-PU.1#1 × MP hybrids). As assayed by Western blot, none of the hybrids expressed OCA-B, and there was little or no Oct-2 expression in either the control or experimental hybrids (Fig. 3, A and B; faint bands comigrating with Oct-2 may indicate very low Oct-2 expression in hybrids 22 and 25 (Fig. 3,A) and in hybrids 14G and 12E (Fig. 3,B)). Ab to Oct-1 was used to normalize protein in the various cellular extracts (Fig. 3, A and B). Constitutive expression of PU.1, therefore, was unable to change the phenotype of hybrid lines with respect to these two plasmacytoma-expressed transcription factors.

As noted previously, PU.1 has been implicated in regulation of the gene encoding J chain polypeptide (34). This gene becomes activated when mature B cells contact Ag and are stimulated by T cell cytokines. As PU.1^−/− knockout mice do not produce B lymphocytes, these mice provide no means of assessing the effect of PU.1 on J chain gene expression. As shown by Northern blot in Fig. 3 C, the MPC11 parental line produced large amounts of J chain mRNA, but neither the parental cell line BWPU.1#1 nor the hybrids formed by fusion of this line with MPC11 (BWPU.1#1 × MP hybrids) produced detectable J chain mRNA. PU.1 alone, therefore, is not sufficient to either induce or sustain expression of the J chain gene.

The same Northern blots were hybridized with a probe specific for Igγ2b mRNA. As shown, the hybrids also lacked this mRNA, demonstrating that they had silenced the plasmacytoma-derived IgH gene (Fig. 3 C).

Previous somatic cell fusion experiments with the transcription factor Oct-2 showed a differential effect on the rescue of plasmacytoma-specific genes when Oct-2 was introduced into the plasmacytoma compared with the T lymphoma parental line (17). We have suggested that this difference reflects the need for another plasmacytoma-specific factor working in combination with Oct-2 (17). Reasoning that PU.1 might similarly require such a factor, we compared the phenotypes of hybrids derived from BW-PU.1 × MP fusions with those of hybrids resulting from BW × MP-PU.1 cell fusions.

Cell lysates prepared from all but one of the BW × MP-PU.1 hybrid lines lacked both Oct-2 and OCA-B, as judged by Western blot (Fig. 4,A). Interestingly, one hybrid (#8A) produced a protein that reacted with the anti-Oct-2 Ab, but was significantly larger than the Oct-2 protein produced by MPC11 or the parental MPPU.1#1 line. This and all of the other hybrids, however, lacked both Igγ2b and Igκ chains (Fig. 4 B).

In summary, the only difference between control BW × MP hybrids and the BWPU.1#1 × MP and BW × MPPU.1#1 hybrids was that the latter two sets of hybrids carried and expressed the PU.1 transgene. Ig, J chain, and OCA-B expression were all silenced in these hybrid lines (Ig phenotypes summarized Table I). Western blots revealed faint bands comigrating with Oct-2 in two of the control hybrids and two of the BW-PU.1#1 × MP hybrids. There was a strong band that reacted with anti-Oct-2 antiserum in one BW × MPPU.1#1 hybrid. EMSAs did not confirm the expression of Oct-2 in these 5/27 hybrid lines, but even if these cell lines were expressing bona fide Oct-2, a significant effect of PU.1 on Oct-2 expression was not evident (by χ² test, p ≤ 1). From these results, we conclude that the ability to reprogram the chromosomes in a hybrid cell is not shared by all tissue-specific transcription factors.

We undertook similar experiments to determine whether an identified coactivator for Oct-2, OCA-B, would influence hybrid cell phenotype. Again, appropriate stable transformants were made from both the plasmacytoma line, MPC11, and the T lymphoma line, BW5147. Both cell lines were transfected with a CMV-regulated expression vector encoding OCA-B with an amino-terminal flu epitope tag. The tag allowed us to distinguish the transgene-encoded protein from its endogenously encoded, murine counterpart.

Two BW5147 transformants expressing the flu-tagged hOCA-B protein were isolated (BW-OCA-B#5 and BW-OCA-B#10), as were two MPC11 transformants (MPOCA-B5/2 and MPOCA-B8/6). A Western blot demonstrating flu hOCA-B expression in these cell lines is shown in Fig. 5. MP-OCA-B8/6 was fused to the T lymphoma BW5147, and hybrid lines were analyzed by Southern blot for IgH loci (data not shown). Those hybrids retaining IgH loci from both parents were further analyzed for hOCA-B expression and for expression of endogenously encoded Oct-2. Representative data for eight of 17 hybrids analyzed are shown in Fig. 6. As shown in Fig. 6,A, most of the hybrids produced both these transcription factors. Hybrid clone 34A expressed flu-tagged OCA-B at a very low level, so that its detection by Western blot was variable (no signal evident in Fig. 6,A). Endogenously encoded Oct-2 was detected by both Western blot (Fig. 6,A) and EMSA (data not shown). Some of the hybrids expressed Oct-2 at levels that approximated those of the MPC11 hOCA-B transformants (e.g., hybrids 38A and 13A), whereas others expressed Oct-2 at much lower levels (e.g., hybrids 55A and 51B). The variation in expression levels for rescued genes has been observed previously in experiments involving hOct-2 as the rescuing transcription factor (17, 18). As was seen in the latter studies, rescuing factor levels and rescued gene expression levels did not covary (e.g., compare hOCA-B and Oct-2 levels in hybrids 55A and 38A; Fig. 6 A). This is discussed in more detail below. There was one hybrid among 17 analyzed that clearly expressed hOCA-B, but did not express detectable Oct-2 (clone 21; data not shown).

EMSAs using nuclear extracts from the hOCA-B-expressing hybrids showed rescue of PU.1 in all but one hybrid (representative data; Fig. 6 B); the one exception was the hybrid that also lacked Oct-2 (clone 21; data not shown). The finding that hOCA-B could rescue PU.1 expression is consistent with previously published transient trans-activation experiments in which a cloned OCA-B gene was able to trans-activate a reporter gene with PU.1 promoter (36).

We next assayed the hybrids for expression of Ig. Sixteen of the 17 hybrids expressed both γ2b H chains and κ L chains as determined by Western blot (data not shown). Again, hybrid clone 21 that lacked Oct-2 and PU.1 also lacked these Ig proteins. The same results were obtained by Northern blot, where J chain mRNA was also detected in all but this one hybrid (representative data in Fig. 6,C; Ig expression data summarized in Table I). γ2b mRNA was barely detectable in clone 34A and was undetectable in clone 30B. γ2b protein was detectable by Western blot in clone 30B, albeit at very low levels (data not shown). Clones 30B and 34A produced low amounts of Oct-2 and PU.1 as well (Fig. 6, A and B, and data not shown). Although the expression of the rescued genes, therefore, did not covary with the rescuing protein (flu hOCA-B), the rescued genes did correlate with one another. That is, a clone that produced low amounts of Oct-2 also produced low amounts of PU.1 and Ig. Clones expressing high levels of Oct-2 also expressed high levels of PU.1 and Ig. This was also seen with respect to J chain gene expression. As shown in Fig. 6 C, J chain mRNA was clearly present in six of the clones tested, but was just detectable in clones 30B and 34A.

To summarize these results, when MP-OCA-B(8/6) was fused to the T lymphoma BW5147, 16 of 17 hybrids (94%) expressed genes unique to the Ig-expressing plasmacytoma parent (Oct-2, PU.1, IgH and IgL). This ability to sustain the Ig-secreting cell’s genetic program resembles that of Oct-2. Notably, the flu-tagged hOCA-B protein sustains the expression of Oct-2 in the same way that flu-tagged hOct-2 was able to sustain OCA-B expression. We conclude that OCA-B and Oct-2 each serve a critical function in Ig-secreting cells and that they regulate one another’s expression (directly or indirectly) in a reciprocal manner at this cell stage. Moreover, as both factors rescue the expression of other regulatory and structural genes unique to the B cell program, we propose that they are at the apex of a regulatory chain responsible for the Ig-secreting cell phenotype.

We similarly analyzed the BW-OCA-B × MP hybrids for phenotype to determine whether OCA-B, like Oct-2, would demonstrate a requirement for a plasmacytoma-derived partner. If so, we expected to find a lower percentage of hybrids with the plasmacytoma phenotype than had been seen in fusions involving MP-OCA-B(8/6). As shown in Fig. 7, this was indeed what we found. The BW5147 transformant expressing higher amounts of OCA-B (BWOCA-B#10; Fig. 5) was fused to the Ig-secreting plasmacytoma MPC11. Seven hybrids were recovered that expressed the OCA-B transgene and retained Ig loci from both parental lines (Fig. 7,A, flu-OCA-B, and data not shown). Only one of these seven hybrids expressed endogenous Oct-2 (Fig. 7,A), and this was also the only hybrid to express γ2b H chain and κ L chain (Fig. 7 B). Given the relatively small number of hybrids analyzed in this experiment, it is difficult to know whether the one Ig-expressing hybrid recovered is due to a real, but weak, effect mediated by OCA-B. Alternatively, this is simply an exceptional hybrid, as occasionally arises in fusions between MPC11 and BW5147 (17). The one of seven hybrids with plasmacytoma phenotype does not represent a statistically significant shift from the zero of seven control hybrids expressing Ig (by χ² test, p ≤ 1). This is in contrast to the effect of OCA-B on hybrid cell phenotype when OCA-B was expressed in the plasmacytoma before cell fusion (by χ² test, p ≤ 0.001). Like Oct-2, therefore, OCA-B appears to require a partner(s) within the plasmacytoma nuclear environment to efficiently maintain the genetic program of the plasmacyte.

The differential effect of introducing Oct-2 or OCA-B into the plasmacytoma compared with introducing either into the T lymphoma before cell fusion suggested that both factors required a tissue-specific coregulator for their function. An obvious possibility was that they constituted a mutually dependent partnership. If this were the case, we would expect that introducing both Oct-2 and OCA-B into the parental T lymphoma line before cell fusion would uniformly yield hybrids with the plasmacytoma phenotype. In this instance, Oct-2 and OCA-B could form the necessary association to prevent silencing of any part of the plasmacytoma program when the T lymphoma and plasmacytoma were fused.

BW-OCA-B#10 was transfected with a flu-tagged Oct-2 expression vector (pCGN-Oct-2his) (17) to produce clones that were expressing both flu-tagged OCA-B and flu-tagged Oct-2. A Western blot of nuclear extracts from two such clones is shown in Fig. 5, demonstrating the expression of both flu-tagged proteins in both lines (BWOCA-B/Oct-2#1 and BWOCA-B/Oct-2#4). In additional Western blots, anti-Oct-2 and anti-OCA-B Abs were used to compare the level of each transfection factor in these BW5147 transformants with that in the MPC11 Ig-secreting cell line. In comparison with MPC11, BWOCA-B/Oct-2#1 produced 1.8× Oct-2 and 3.0× OCA-B, whereas BWOCA-B/Oct-2#4 produced 2.0× Oct-2 and 3.9× OCA-B (data not shown). The expression levels of Oct-2 and OCA-B in the BWOCA-B/Oct-2 transformants were at or below those when each was individually expressed in BW5147 (BWOCA-B (present study) and BO2 and BO6 (17)).

When clones BWOCA-B/Oct-2#1 and BWOCA-B/Oct-2#4 were fused to MPC11, four hybrids from the fusions with clone 1 and five hybrids from the fusions with clone 4 were selected for further analyses. All these hybrids retained Ig genes from both parental lines, and all continued to express the flu-tagged OCA-B and Oct-2 transgenes (data not shown). EMSAs showed that all the hybrids produced the transcription factor PU.1 (Fig. 8,A), and Northern blots showed that all produced J chain and γ2b mRNA (Fig. 8,B; data summarized in Table I). Interestingly, the level of expression of the rescued genes in the BWOCA-B/Oct2 × MPC11 hybrids very closely approximated that of the same genes in the plasmacytoma parent, MPC11 (Fig. 8 B). Unlike the MPOCA-B × BW5147 hybrids or MPOct-2 × BW5147 hybrids, these hybrids showed relatively little variation in Ig and J chain expression levels, and these expression levels were not significantly different from those in MPC11 (current study and Ref. 17).

The results presented in this study extend those of an earlier study in which we found that Oct-2 played a central role in sustaining the gene expression profile of an Ig-secreting plasmacytoma (MPC11) in plasmacytoma × T lymphoma hybrids. One possibility was that overexpression of any of the plasmacytoma-specific genes would lead to dominance of the Ig-secreting cell profile. We have previously shown that overexpression of the ubiquitous, octamer-binding factor Oct-1 does not affect the plasmacytoma × T lymphoma hybrid phenotype, nor does expression of a protein that differs from Oct-2 by only its C-terminal domain (Oct2.2.1) (18). In the present study we have demonstrated that PU.1, a factor critical to early events in B cell development, is also unable to affect the hybrid cell phenotype. Although expressed in Ig-secreting cells and silenced in hybrid lines, PU.1 does not play a pivotal role in sustaining the gene expression profile of these cells.

In our earlier studies we observed that the Oct-2 expression vector had different effects when introduced into the T lymphoma rather than into the plasmacytoma before cell fusion. This led us to conclude that Oct-2 was interacting with another plasmacytoma-restricted factor to achieve its effect on the hybrid cell phenotype (17, 18). We have previously hypothesized that such a factor would interact with the C-terminal domain of Oct-2. This was because we found that the related, but ubiquitous, Oct-1 transcription factor could not supplant Oct-2 function and that the C-terminal domain of Oct-2 constituted the functional difference between these two transcription factors (18). As demonstrated in this study, however, we have found that OCA-B is the missing coregulator in this experimental system. Not only did OCA-B have effects similar to those of Oct-2 when introduced into the plasmacytoma before cell fusion (94% hybrids expressed the plasmacytoma phenotype), but it also functioned synergistically with Oct-2 when both were introduced into the T lymphoma before cell fusion. OCA-B interacts with the POU (central, DNA-binding) domain of Oct-2 and can also interact with the POU domain of Oct-1 (2, 4, 37). This is a direct demonstration, therefore, that the Oct-2/OCA-B pair has functions that cannot be replicated by the Oct-1/OCA-B pair. Importantly, the assay for function involves regulation of endogenous genes (e.g., J chain, IgH, Igκ) and not artificial reporter gene constructs. The importance of the C-terminal domain of Oct-2, therefore, must not lie in its serving as a docking site for another tissue-restricted factor, but, rather, in mediating another activity that may, for example, involve interactions with the transcriptional machinery itself.

Also striking was the quantitative effect that coexpression of the Oct-2 and OCA-B transgenes had on endogenous gene expression in the Ig-rescued hybrid lines. When either OCA-B or Oct-2 was introduced individually into the plasmacytoma before cell fusion, the resulting hybrids expressed genes characteristic of the plasmacytoma parent. The level of gene expression varied among these hybrids, however, and did not directly correlate with transgene expression levels. In hybrids constitutively expressing both the OCA-B and Oct-2 transgenes, however, the levels of Ig (H and L) and J chain mRNA varied very little among the hybrids and were almost indistinguishable from those in the parental MPC11 line. This suggests that the limiting factor in the one-gene rescue experiments (OCA-B alone into MPC11 or Oct-2 alone into MPC11) was the missing partner in this functional pair. When expression was not dependent upon the endogenous loci encoding Oct-2 or OCA-B, but, rather, both proteins were available in excess from constitutively active expression vectors, the downstream targets of these regulatory factors (e.g., Ig and J chain) were expressed at wild-type levels.

What does this tell us about the function of Oct-2 and OCA-B in normal, Ig-secreting cells? The implication is that both are crucial to the specialized functions of the Ig-secreting plasmacyte. Without either factor, all specialized genes fall silent, most notably the Ig genes, which encode the critical effector molecules of these cells. This is not the case in the Ag-independent phase of B cell development, when both factors are largely dispensable (13). A change in factor function with cell stage or cell type is one of the hallmarks of the complex regulatory systems operating in vertebrates (38). Although the Oct-2/OCA-B partnership may act directly upon the Ig loci through, for example, binding sites found in the 3′IgH enhancer region, as suggested by us and others (39, 40), it is possible that they also affect Ig expression in a more indirect fashion, through a broader effect on cell function. This is true for all the cell-specific marker genes described in this study, including those encoding Oct-2 and OCA-B themselves (little is presently known about the OCA-B and Pou2f2(Oct-2) promoters and other regulatory regions associated with these genes) (41). For example, Oct-2/OCA-B’s effect on PU.1 at this stage in development could be direct, given the octamer-dependent nature of the PU.1 promoter (36), whereas their effect on J chain might be indirect, because the J chain promoter appears to be PU.1 dependent (34). In previous studies we have looked at an even larger array of genes and found that all analyzed behave as a coordinately controlled unit (15). Oct-2 and OCA-B, therefore, sit at the apex of this regulatory system, with PU.1 somewhere downstream. Interestingly, Hodgkin and Reed/Sternberg cells of Hodgkin’s lymphoma display a phenotype very similar to that of the extinguished hybrid. Although molecular studies of the Ig genes in these cells have established that they are B lymphocyte-derived, there is a decrease in or loss of expression of nearly all B-lineage-specific genes (42). It will be interesting to determine whether the defect in these cells can be attributed to modulation of the Oct-2/OCA-B central regulator.

Gene knockout experiments involving Oct-2 and OCA-B have not settled the question of when and how these factors function in cells of the B lymphocyte lineage. Early reports suggested a role for both factors in processes occurring subsequent to B cell activation (10, 11, 12, 13, 43, 44). More recently, additional defects have been detected in OCA-B^−/− mice at earlier stages of B cell development (13, 45, 46). Mice with Oct-2^−/− and/or OCA-B^−/− B-lineage cells lack germinal centers, are deficient in the non-IgM Ig isotypes, and have significantly reduced Ig responses to T-independent and T-dependent Ags (13, 44). This arises in part from reduced B cell responsiveness to Ag and Th cell-induced signals (44, 46, 47). Given that B cells are not responding normally to these signals, few cells appear to differentiate into Ig-secreting plasmacytes. In the absence of conditional knockouts of the genes encoding these factors, therefore, it has not been possible to directly examine the importance of Oct-2 and OCA-B to cells that reach this terminal, effector phase. The cell fusion system constitutes a complementary experimental approach that uniquely provides us with that opportunity, and as described in this study, our results suggest that Oct-2 and OCA-B play critical roles in controlling gene expression at this stage. This may be true of only a subset of Ig-secreting cells, however, because IgM-secreting plasmacytes have been identified in OCA-B^−/− mice by immunohistochemistry (48). Notably, surface IgM⁺ B cells are near wild-type levels in these mice, as is serum IgM, whereas the non-IgM isotypes, including IgG2b (the isotype produced by the plasmacytoma of the present study), are well below normal levels. The pronounced reduction in non-IgM serum Abs in mice reconstituted with Oct-2^−/−/OCA-B^−/− fetal liver could be due to an inability to induce isotype-switched cells to undergo differentiation to Ig-secreting cells or to an inability of such Ig-secreting cells to survive and/or function. The present study suggests that even if produced, these plasmacytes would lack normal function.

There is an additional dimension to the experimental system described in this study. Cell fusion introduces competition between the genetic programs of the two parental cell types. When the T cell program dominates, there is silencing of all tested genes characteristic of the Ig-secreting cell (15). The IgH locus of the plasmacytoma is not only transcriptionally silenced, but is de novo methylated under these circumstances (14, 15). As shown in this study, ectopic expression of Oct-2 and OCA-B in the T lymphoma before cell fusion leads to the reverse phenotype in cell hybrids. The IgH locus of the plasmacytoma (as well as all other tested plasmacytoma-specific genes) remains active, and as shown in prior studies, the IgH locus of the T lymphoma becomes demethylated in hybrids with this phenotype (15). Moreover, the genes uniquely expressed in the T lymphoma before cell fusion are silenced in Ig-expressing hybrids (16). Oct-2 and OCA-B, therefore, are the toggles in this genetic switch. It is possible that their function is not simply to sustain the Ig-secreting cell’s genetic program, but also to silence, or in other ways modify, that of the T cell lymphoma. There are reports, for example, of Ag-induced expression of Oct-2 and OCA-B in T-lineage cells that leads to downstream effects on T cell gene expression (49, 50, 51). Oct-2/OCA-B’s effect on the gene-silencing activity of the T lymphoma, therefore, may be only one part of a broader effect on T cell function. A more complete comparison of the gene expression profiles of the BW5147 T lymphoma line and its Oct-2/OCA-B-expressing derivatives may thus provide further insight into not only the mechanism for gene silencing in this cell fusion system, but also the mechanism by which natural processes, such as Ag activation, influence the gene expression profiles of T lymphocytes.

We gratefully acknowledge Drs. B. K. Birshtein (Albert Einstein College of Medicine) and M. W. Young (Rockefeller University) for critical reading of the manuscript, Dr. W. Williams (Hunter College) for advice on statistical analyses of hybrid data, Dr. R. Roeder (Rockefeller University) for both the pCGN-OCA-B expression vector and polyclonal antisera to OCA-B, Drs. D. Tenen and H. Radomska (Harvard Institutes of Medicine) for PU.1-dependent reporter plasmids, the laboratory of Dr. M. Koshland (University of California-Berkeley) for the BCMGSneo-PU.1 expression vector, and Dr. R. Maki (The Burnham Institute) for polyclonal antisera to PU.1. We thank Ryszard Stawowy, Pavel Munerman, and Steven Williams for expert technical assistance.