Mouse Lymphoid Cell Line Selected To Have High Immunoglobulin Promoter Activity

Abstract

Immunoglobulin variable region promoters are predominantly B-cell specific, but the molecular basis for this specificity has not been elucidated. To further understand how B-cell-specific immunoglobulin promoter expression is mediated, the murine lymphoid cell line 2017 was engineered to express the green fluorescent protein under the control of an immunoglobulin heavy chain promoter and selected for high activity using multiple rounds of fluorescence-activated cell sorting. Rare clones with intense and stable immunoglobulin promoter activity were isolated. Transient transfection experiments demonstrated that two different immunoglobulin promoters and two other B-cell-specific promoters have higher activities in the selected cell lines relative to the parental line and to the non-cell-type-specific histone H2B promoter. The increased immunoglobulin activity required nucleotide residues downstream of the transcription initiation site which were also important for maximal activity in B cells and which were conserved in other B-cell-specific promoters. Unlike the unselected cells, the 2017 variants also showed activation of their endogenous immunoglobulin heavy chain variable regions.

The immunoglobulin (Ig) heavy chain (IgH) locus is composed of tandem organized gene segments, V_H, D_H, and J_H, plus the constant region exons. Consecutive D-J and V-DJ recombination events during B-cell development form the mature VDJ (antigen binding) exon, and it is the combinatorial association of these segments and the imprecise joining at the junctions that accounts for a great part of the diversity of the antibody repertoire. Each V gene segment contains a promoter that will express the mature recombined Ig transcript, a small leader exon encoding the signal peptide, and the 5"-most portion of the Ig coding sequence. V gene segments in the murine IgH locus are classified into approximately 15 families, with each family defined by 80% or more coding sequence identity. Downstream of the D_H and J_H segments are the intronic enhancer (termed E_μ), the Ig constant regions, and other regulatory sequences.

Studies have shown that non-protein-coding transcripts from the V region promoters can be detected prior to recombination (1, 14, 59; reviewed in reference 49). These RNAs are termed sterile or germ line transcripts. Several studies have shown that expression of germ line transcripts roughly correlates with the occurrence of V-DJ recombination (24, 41, 59), suggesting a model in which germ line transcription promotes V-DJ recombination. However, other studies have shown that recombination efficiency does not always reflect promoter strength (1, 25). Most likely, a combination of promoter activity, proximity to the DJ gene segments, strength of the recombination signal sequence, and the timing and extent of chromatin opening accounts for the pattern of IgH recombination in B cells.

In isolation, V_H region promoter activity is specific to cells of the lymphoid lineage, especially B lymphocytes (for recent reviews, see references 16, 28, and 33). However, the molecular basis of this selective IgH promoter expression is incompletely understood. Although several regulatory motifs, including E boxes, Ets sites, C/EBP binding sites, and pyrimidine-rich sequences have been shown to contribute to promoter activity in specific V region promoters (6, 8, 15, 45), a DNA element known as the octamer motif (5"-ATGCAAAT-3") appears to be the critical determinant of Ig promoter B-cell specificity (3, 9, 27, 29). The octamer is present in most Ig promoters (34); however, this same motif often occurs in the regulatory regions of other genes, many of which are non-B-cell specific. Examples include the regulatory regions of the U1 and U6 snRNA and histone H2B genes (30, 48). Furthermore, the known trans-acting factors that interact with the octamer element also cannot account for the B-cell specificity. Three of these factors are present in B cells: Oct-1, Oct-2, and OCA-B/Bob-1/OBF-1.

Oct-1 (47, 52, 55) is a founding member of the POU domain family of transcription factors (17) that recognizes the octamer motif and related sequences. Oct-1 is conjectured to be preferentially active at Ig promoters (36, 46), yet it is ubiquitously expressed in adult cells.

Oct-2 (5, 39, 51) is another POU domain protein capable of binding the octamer motif with similar affinity and specificity. It and the coactivator OCA-B/Bob-1/OBF-1 (13, 26, 54) have a more-B-cell-restricted expression pattern, but data from mouse knockout studies indicate that neither is required for early B-cell development and Ig transcription (7, 10, 19, 32, 43). The Oct-2/OCA-B double knockout has also been reported to produce B cells that express Ig (44). Consistent with these results, data from several laboratories indicate that physiological levels of Oct-2 and OCA-B fail to activate Ig promoters and that instead they act preferentially at the IgH intronic and 3" enhancer regions (36, 53, 57, 58).

We have used an in vivo approach in order to find and characterize additional cis regions and trans factors that mediate B-cell-specific IgH promoter transcription. In this approach, changes in IgH promoter activity were measured as changes in the amount of green fluorescent protein (GFP) produced from a cDNA linked to an IgH promoter (17.2.25; see Materials and Methods). Cells with increased GFP activity were monitored, enriched, and isolated using fluorescence-activated cell sorting (FACS).

During the course of these experiments, we observed that one murine pre-T-cell line, 2017 (50), could be selected to have levels of GFP that are 100-fold higher than the baseline. Several clones from this high-GFP population have been isolated and further characterized. We also report the delineation of a new cis-acting element downstream of the transcription initiation site in IgH promoters and the promoters of several other B-cell-specific genes.

MATERIALS AND METHODS

Transfection constructs.

DNA used for transfection was double-band CsCl purified (38). Unless otherwise noted, the murine 17.2.25 hybridoma heavy chain promoter was used in Ig promoter analyses and in the formation of stable cell lines. For the luciferase constructs, the pGL3 basic vector (Promega) was digested with BglII and HindIII, treated with alkaline phosphatase, and gel purified. The plasmid pμΔ1 (12) was used as a PCR template with PCR primers containing restriction enzyme sites on their 5" ends. pμΔ1 PCR products were digested with BglII and HindIII, gel purified, and ligated into the pGL3 backbone using T4 DNA ligase (Life Technologies). The full-length IgH promoter corresponded to residues from −154 to +35 relative to the transcription initiation site. The sequence of the IgH-154Bgl 5" primer used to amplify the promoter sequences was 5"-GGAGATCTGGATCCAGAGTTNGG-3". The sequences of the 3" primers were as follows: IgH + 35Hind, 5"-GGAAGCTTGAGATTCAGTGTCT-3"; and IgH + 1Hind, 5"-GGAACGTTATGATGTTTGGTGTG-3". For the H2B promoter constructs, two oligonucleotides containing sequences from the human H2B promoter (48) were synthesized with phosphate groups on their 5" ends. The oligonucleotides were as follows: H2B-595"Bgl, 5"-GATCTTCTTCACCTTATTTGCATAAGCGATTCTATATAAAAGCGCCTTGTCATACCCTACTCACGA-3"; and H2B-593"Hind, 5"-AGCTTCGTGAGTAGGGTATGACAAGGCGCTTTTATATAGAATCGCTTATGCAAATAAGGTGAAGAA-3". The oligonucleotides were annealed together, creating a double-stranded molecule with BglII- and HindIII-compatible ends, and ligated directly into pGL3. The final portion of the H2B promoter incorporated into the luciferase construct spanned sequences from −59 to +1 relative to the transcription initiation site. The IgH-154 + 35 construct with the E_μ intronic enhancer was made by inserting a 700-bp enhancer sequence from the EVH186.2 construct (a gift of F. Alt) into a distal pGL3 XbaI site located approximately 2 kb upstream of the promoter sequence. The IgH-154 + 1/186.2 chimeric promoter was made by inserting a double-stranded oligonucleotide with HindIII overhangs and 5" phosphate groups into the HindIII site of the IgH-154 + 1 construct. The sequences of the oligonucleotides were as follows: 186.2HinTOP, 5"-AGCTTGATCACTGTTCTCTTTACAGTTACTGAGCACACAGGA-3"; and 186.2HinBOT, 5"-AGCTTCCTGTGTGCTCAGTAACTGTAAAGAGAACAGTGATCA-3". The sequence and orientation of the insert were confirmed by DNA sequencing and PCR.

Construction of the scanning promoter mutagenesis constructs utilized the EcoRV site located at −21 relative to the transcription initiation site in the 17.2.25 promoter. The full-length −154 to +35 luciferase construct was digested with EcoRV and HindIII, treated with alkaline phosphatase (Life Technologies), and gel purified. Oligonucleotides (73-mer) containing the promoter sequence between the EcoRV and HindIII sites and containing 3-bp mutations were amplified by seven rounds of PCR using Taq DNA polymerase (Perkin-Elmer) and primers homologous to the 5" and 3" ends. The mutations were A→C, G→T, C→A, and T→G. As an example, the sequence of the oligonucleotide IgMut-15 was 5"-CGCAGCGATATCACAACCAAACATCATATGAGCCCTATCTTCTCTACAGACACTGAATCTCAAGCTTCGAGCC-3". The letters in bold indicate the positions of the mutation. The EcoRV and HindIII sites are underlined. The PCR primers were 5"-AGCCGCAGCGATATC-3" and 5"-AGCGGCTCGAAGCTT-3". The products were digested with EcoRV and HindIII and ligated into the digested −154 to +35 construct.

For GFP constructs, the pEGFP-1 plasmid (Clontech) was digested with BglII and HindIII, treated with alkaline phosphatase, and gel purified. The luciferase construct (−154 to +35) was digested with BglII and HindIII, gel purified, and ligated into the pEGFP-1 backbone creating the plasmid pEGFP-IgH-154 + 35.

Cell culture, transfections, and stable cell lines.

The cell lines 2017, BJA-B, M12, 70Z/3, and WERI-27 were grown in RPMI with HEPES (Life Technologies) plus 10% heat-inactivated fetal calf serum (IFS) (HyClone), 2 mM l-glutamine, 50 U of penicillin-streptomycin/ml, and 50 μM 2-mercaptoethanol. EL-4, p5424, and θ4b were grown in Dulbecco’s modified Eagle medium with HEPES (Life Technologies), 10% IFS, l-glutamine, penicillin-streptomycin, and 2-mercaptoethanol. When transfections were performed using electroporation, 2 × 10⁶ subconfluent cells were harvested and resuspended in 400 μl of the appropriate medium without IFS. Ten micrograms of total DNA was added (typically 2 μg of luciferase reporter, 2 μg of pCMV-βGal, and 6 μg of pUC18 DNA), and the reaction mixtures were incubated on ice for 10 min. Cells were electroporated with a Bio-Rad Gene Pulser set at 260 V, 960 μF, and 200 Ω and with cuvettes with a gap length of 4 mm. After a further 10-min incubation on ice, the cells were plated in 10-cm-diameter dishes with 10 ml of complete medium. Cells were harvested and assayed after 24 h. Lipid-mediated transfection experiments used the FuGene6 transfection reagent (Roche) according to the manufacturer's instructions. Total DNA (2.5 μg) (typically 500 ng each of the luciferase construct and pCMV-βGal and 1.5 μg of pUC18 DNA) were combined with 7.5 μl of transfection reagent in 200 μl of serum-free medium, and applied to 10⁶ cells in six-well plates containing 3 ml of complete medium in each well. Cells were assayed after 24 h.

For the construction of stable cell lines harboring the Ig-GFP reporter, 10 μg of the plasmid pEGFP-IgH-154 + 35 was electroporated into 8 × 10⁶ 2017 cells. After 24 h, the cells were placed in 0.75 mg of active G418 (Life Technologies)/ml, diluted, and aliquoted into 20 96-well plates. Approximately 1 in 25 wells showed growth in G418. Viable clones were expanded and analyzed by flow cytometry. This protocol yielded approximately 80 founder clones per transfection.

FACS.

Flow cytometry used Becton Dickinson FACScan and FACSCalibur machines. For cell sorting, the Becton Dickinson FACSVantage high-speed cell sorter was used. All analyses used CellQuest software (Becton Dickinson).

Reverse transcription PCR (RT-PCR).

The Superscript system (Life Technologies) was used. Total RNA was prepared from 5 ml of subconfluent cells using RNAzol (Teltest). GFP-specific primers were as follows: GFP-1, 5"-TGAACCGCATCGAGCTGAAGGG-3"; GFP-2, 5"-TCCAGCAGGACCATGTGATCGC-3". V region-specific primers were as follows: J5585"L, 5"-TCCTCCTGTCAGTAACTGCAG-3"; J5583"FWIII, 5"-ATGATGGCAGTGCTGGAGGAT-3". For the GFP analysis, reactions were amplified for 24 cycles in an MJ Research PTC 100 thermal cycler. For the V region analysis, reactions were amplified for 39 cycles. After 28 cycles, parallel reaction mixtures containing actin primers (Ambion) were removed to retain the actin amplification in the linear range.

RESULTS

DNA elements downstream of the transcription initiation site are necessary for full promoter activity.

The sequence of the murine 17.2.25 IgH promoter (12) is shown in Fig. 1. Key features of this promoter include the octamer motif centered at −53 relative to the transcription initiation site, a poor consensus TATA region, and a heptamer site (8) 2 bp upstream of the octamer sequence. The heptamer site is also recognized by Oct proteins (23, 37). Interestingly, mutation of the TATA sequence to a consensus adenovirus major late promoter TATA sequence reduces promoter activity (35).

FIG. 1.

Sequence of the 17.2.25 IgH promoter. The sequence was taken from reference 12. Known regulatory sequences, including the heptamer (5"-CTCATGA-3"), the octamer (5"-ATGCAAAT-3"), and the TATA region (5"-TAATATA-3"), are shown in bold. Also shown in bold are the initiating A residue at +1 and the ATG start codon (Met) at +52. Promoter sequences used in this study were cloned to either +1 or +35. Sequences upstream of the heptamer are not shown, but they were present up to −154. Also noted in brackets is the pyrimidine (Py)-rich motif (12 of 13 pyrimidine residues) centered at +12.

In an initial analysis, we tested the activity of a 190-bp fragment of the promoter by linking it to a luciferase reporter gene and transiently transfecting it into several human and mouse B- and T-cell lines using electroporation (see Materials and Methods). The sequences used contained the octamer motif, TATA region, and 35 residues downstream of the transcription initiation site (−154 to +35). A promoter construct missing the downstream residues (−154 to +1) and the backbone plasmid were also used. pCMV-βGal was cotransfected as an internal control. The cells were harvested after 24 h and assayed for luciferase and β-galactosidase (β-Gal) activity. Averaged luciferase/β-Gal activity ratios from three experiments are shown in Fig. 2A. As expected, the luciferase/β-Gal ratios for the B-cell lines (BJA-B, M12, and 70Z/3) tended to be greater than those for the T-cell lines (θ4b, P5424, and 2017). The murine T-cell line EL-4, which has been reported to have high IgH promoter activity (36) was the sole exception. Comparing pre-B (70Z/3), mature B (M12), and activated plasma cell lines (BJA-B), we also noted an increase in promoter activity with the full-length promoter construct. These studies indicated the presence of a potential regulatory element located downstream of the transcription initiation site, as removal of the sequences downstream of +1 resulted in an approximately fourfold down-regulation of promoter activity (Fig. 2A). The deletion appeared to have a greater effect on cell lines derived from more mature B cells, such as BJA-B and M12, than with earlier stage B-cell lines such as 70Z.

FIG. 2.

(A) Cell lines BJA-B, M12, 70Z/3, θ4b, p5424, 2017, and EL-4 were transfected with either the pGL3 backbone (white bars) or IgH-luciferase promoter constructs containing (IgH-154 + 35) (light gray bars) or lacking (IgH-154 + 1) (dark gray bars) the sequences from the transcription initiation site to +35. pCMV-βGal was cotransfected as an internal control. Transfections were performed by electroporation. The luciferase/β-Gal activity ratio is depicted on the y axis. Averages from three replicate experiments are shown. Error bars denote standard deviations. (B) Promoter constructs containing 3-bp mutants spanning the transcription initiation site were electroporated into either the B-cell line BJA-B (white bars) or the nonlymphoid cell line WERI-27 (dark gray bars). The construct name is shown to the right of the sequence for that construct. The full-length (+35) and deletion (+1) constructs were included as controls. pCMV-βGal was cotransfected, and the luciferase/β-Gal activity ratio is shown on the x axis. The mutations were as follows: A→C, G→T, C→A, and T→G. Note that mutations beginning at +1 and +19 relative to the transcription initiation site are missing in this series. Py, pyrimidine. (C) The backbone reporter pGL3, the IgH-154 + 35 and IgH-154 + 1 constructs used for the results shown in panel A, and a reporter construct (IgH chimera) in which 37 nucleotides from the IgH 186.2 promoter were placed downstream of the IgH-154 + 1 construct (see Materials and Methods). pCMV-βGal was cotransfected as an internal control. The experiment was performed in triplicate, and error bars denote standard deviations. (D) The luciferase/β-Gal activity ratio is shown for two T-cell lines, EL-4 and 2017. pCMV-βGal was cotransfected with either the empty pGL3 vector (light gray bars) or luciferase reporter constructs containing the IgH promoter (−154 + 35) (dark gray bars) or the H2B promoter (−59 + 1) (white bars). Transfections were performed by electroporation. (E) Averages from three transfection experiments are shown. T-cell lines 2017, EL-4, and θ4b were transfected either with the pGL3 vector or with the IgH-154 + 35 reporter construct. In some cases the reporter constructs were cotransfected with expression constructs encoding human Oct-1 (pCG-Oct-1) (56) and OCA-B/Bob1 (pCATCH-Bob-1) (13). pCMV-βGal was cotransfected as an internal control. pUC18 was used to equalize the mass of the amount of DNA transfected. Transfections were performed by electroporation. The luciferase/β-Gal activity ratio is depicted on the y axis. Error bars denote standard deviations.

Deletion studies (data not shown) indicated that removal of the sequences downstream of +20 had little effect on promoter activity but that a deletion to +14 resulted in a significant reduction. We therefore undertook a more refined scanning promoter mutagenesis approach in which 3-bp mutations were introduced along the promoter DNA from −15 to +25 (Fig. 2B). These mutant constructs were cotransfected with pCMV-βGal into either the B-cell line BJA-B or the nonlymphoid cell line WERI-27. The wild-type −154 to +35 and −154 to +1 constructs were transfected as controls (Fig. 2B). These studies indicated that mutation of nucleotides +16, +17, and +18 largely eliminated the higher promoter activity of BJA-B cells relative to WERI-27 cells. These residues partially overlapped with a 13-nucleotide pyrimidine-rich sequence centered at +12 (Fig. 1). Mutating the residues immediately upstream of the transcription initiation site also largely abolished BJA-B-specific promoter activity (−3, −2, and −1) (Fig. 2B). Point mutation of the +16 to +18 region affected reporter activity more severely than complete deletion, possibly because in the deletion constructs the substituted sequences from the backbone plasmid partially complement the effect of the deletion.

To test whether the downstream sequence determinants can be complemented by sequences from another IgH promoter, a chimeric construct was made in which sequences from the V_H 186.2 promoter (see Fig. 8 and reference 20) were placed immediately downstream of the IgH-154 + 1 construct. The final construct contained sequences from the 17.2.25 promoter to +1, a six-nucleotide HindIII site, and the sequence from the 186.2 promoter downstream of the transcription start site from +6 to +42. As shown in Fig. 2C, transient transfection of this construct into BJA-B cells restored full promoter activity.

FIG. 8.

Nucleotide sequences of 11 Ig promoters and the murine B29 and mb-1 promoters are shown. The Gibbs sampling consensus is depicted underneath. For each position, the height of the letter indicates the frequency with which it is encountered. The top 25 matches to the consensus are boxed. In the case of near matches, deviations from the consensus appear in lowercase. Shown at the right is the number of correct nucleotides (n of 14) for each match. Areas of the 17.2.25 and κ-41 promoters implicated in activity by mutagenesis are underlined. In those cases in which the exact TATA region and ATG start codon had been annotated, the corresponding sequence is highlighted in bold. Annotated transcription initiation sites are shown in bold with an arrow above the nucleotide residue.

In the case of the murine pre-T-cell line 2017 (50), the IgH reporter activity from the full-length promoter was weak but measurable relative to that of B-cell lines and EL-4 (Fig. 2A). The low level of expression was promoter specific, as control promoters such as the octamer-containing but non-cell-type-specific H2B promoter were highly active in both 2017 and EL-4 (Fig. 2D). The 2017 cell line is thy-1 positive and expresses Oct-1 and Oct-2 (Oct-2 at low levels) but does not express OCA-B (unpublished data). Transient overexpression of Oct-1 or OCA-B (Fig. 2E) or stable overexpression of Oct-2 (data not shown) did not increase the level of IgH promoter activity in 2017 to that observed with B cells or EL-4. These observations provided evidence that Oct-1, Oct-2, and OCA-B cannot account for the difference in promoter expression between these cell lines.

Selection of clones for high IgH promoter-GFP expression.

The above findings were reproduced when GFP was used in place of luciferase in transient transfection assays (data not shown). Therefore, stable 2017 cell lines were constructed in which the −154 + 35 IgH promoter was linked to GFP. The stable cell lines were single-cell cloned and designated 2017-Ig-GFP. Clone no. 9 had modest GFP activity and was used in these experiments. This clone significantly up-regulates GFP activity upon polyethylene glycol-mediated fusion with the B-cell line M12 (not shown), indicating that it is poised to strongly express the stably transfected IgH promoter if trans-acting factors present in B cells are supplied. Omitting either polyethylene glycol or the M12 cell line from the fusion mixture eliminated the up-regulation.

Using multiple rounds of FACS, rare 2017-IgH-GFP no. 9 cells were selected that expressed GFP more strongly than the parental cell type. Upon selection, cells appeared that expressed GFP 10- to 100-fold more than the parental population (Fig. 3A). Another 2017-IgH-GFP clone could be similarly selected, indicating that the ability to up-regulate the IgH promoter is intrinsic to the 2017 cell line and not the individual clone used. In contrast, a similarly engineered B-cell line could not be selected, even after multiple rounds of FACS (data not shown). Individual clones from the high-GFP population were designated 2017-IgH-GFP no. 9 m or h depending on the level of GFP expression gated in the final round of sorting (Fig. 3A). Several single-cell clones derived from this high-GFP population were isolated and analyzed in detail. The cells had a clumped appearance (Fig. 3D and E) and slightly lower growth rates (data not shown). Western blotting revealed no change in the expression of Oct-1, Oct-2, or OCA-B among these cell lines (data not shown). The GFP activity of these cell lines was stable, as they retained their characteristics even after 4 months in continuous culture. However, upon reselection with several rounds of FACS for a low-GFP phenotype, a stable low-GFP population can reemerge (data not shown).

FIG. 3.

(A) Cell line 2017-IgH-GFP no. 9 was subjected to four rounds of high-speed FACS. At each step, approximately 100 million cells were scanned on the sorter. GFP is indicated on the x axes, and propidium iodide (PI) staining is indicated on the y axes. All scales are logarithmic. Dead (PI-positive) cells were gated out. In the final round of sorting, the cells were separated into two populations, designated m or h, based upon the degree of fluorescence. (B) Low-magnification phase-contrast microscope image of the parental cell line 2017-IgH-GFP no. 9 prior to sorting. Image taken at 100× magnification with a Zeiss Axioplan 2 epifluorescence microscope. (C) Fluorescence microscopy image taken by using a fluorescein isothiocyanate channel that induces GFP fluorescence. (D and E) Similar images taken of the high-GFP-selected clone h3 at equal densities.

To demonstrate that the GFP reporter is up-regulated at the RNA level, RT-PCR assays were performed with RNA isolated from the original cell line 2017, the parental clone 2017-IgH-GFP no. 9, and the high-GFP-selected clone h3. The primers used were predicted to amplify a 307-bp portion of the GFP cDNA. The results shown in Fig. 4 show that a band corresponding in size to the GFP RNA has been dramatically up-regulated in the sorted cell line (Fig. 4, lane 5) compared to the parental control (Fig. 4, lane 3). Actin primers were used to control for the amount of input RNA. Omission of reverse transcriptase eliminated the signal, indicating that the products were due to RNA (Fig. 4, lanes 2, 4, and 6). We conclude that at least a significant portion of the GFP up-regulation is at the RNA level.

FIG. 4.

RT-PCR assay for the presence of GFP RNA. Total RNA was prepared from cell line 2017 (lanes 1 and 2), 2017-IgH-GFP clone no. 9 (lanes 3 and 4), and the selected high-GFP clone h3 (lanes 5 and 6). Primers recognized the GFP reporter. Actin primers (Ambion) were used as a control. RT reactions were amplified using 24 cycles of PCR. Alternating lanes omitted reverse transcriptase as a control (−). GFP, amplification product corresponding to GFP; A, actin amplification product.

Increased IgH promoter activity in the selected high-GFP 2017 cells.

Transient transfection studies revealed that the activity of the full-length 17.2.25 IgH promoter had increased relative to the octamer-containing H2B promoter in a subset of the selected clones. For example clones m2, m4, and h3 (Fig. 5A) exhibited approximately fourfold-higher 17.2.25 IgH promoter activity than the parental 2017-IgH-GFP clone no. 9 cell line. In contrast, stable overexpression of human Oct-2 in this cell line did not induce increased promoter activity (data not shown). The murine pre-B-cell line 70Z/3, which displays relatively strong IgH promoter activity, was used as a positive control (Fig. 5A). Interestingly, we also found that two other B-cell-specific promoters, B29/Igβ/CD79b (−164 to +32) and mb-1/Igα/CD79a (−252 to +48) (gifts of R. Wall; for sequences, see Fig. 8) were up-regulated relative to the parental cell type (Fig. 5A). In contrast, the activity of the pre-T-cell receptor α promoter (a gift of B. Reizis) remained the same in all tested cell lines, suggesting that the increased activity is confined to B-cell-specific promoters (data not shown). These data indicate that multiple B-cell-specific promoters exhibit higher activity in the selected cell lines.

FIG. 5.

Transfection characteristics of the selected cell lines. (A) High-GFP-expressing cells from four rounds of FACS were single-cell cloned and transiently transfected with pGL3 containing four different promoters (H2B-59 + 1, IgH-154 + 35, B29-164 + 32, and mb-1-252 + 48). Cell line 2017, the parental clone 2017-IgH-GFP no. 9, and 70Z/3 were also transfected. The histone H2B promoter was used as a control. pCMV-βGal was cotransfected, and the luciferase/β-Gal ratios for the H2B construct were normalized to 1. The dark bars depict Ig promoter activity. Shaded bars are B29 and mb-1. Lipid-mediated transfection reagents were used. (B) Selected high-GFP cell lines 70Z/3 and 2017 and parent line 2017-IgH-GFP no. 9 were transiently transfected with either the 17.2.25 promoter (gray bars) or the V_H 186.2 promoter with the IgH intronic enhancer upstream (white bars). pCMV-βGal was cotransfected, and luciferase/β-Gal ratios are depicted on the y axis. Lipid-mediated transfection reagents were used. (C) Selected cell lines m4, h3, and 70Z/3, and the parental cell line 2017-IgH-GFP no. 9 were transiently transfected with either IgH-154 + 35 (gray bars) or the same construct in which the intronic enhancer had been inserted (white bars) (see Materials and Methods). Lipid-mediated transfection reagents were used. pCMV-βGal was cotransfected as an internal control. Experiments were performed in triplicate, and error bars denote standard deviations. (D) Scanning promoter mutants were transiently transfected into either the parent cell line 2017-IgH-GFP no. 9 (gray bars) or the high-GFP-selected cell line m4 (white bars). Lipid-mediated transfection reagents were used. pCMV-βGal was cotransfected, and luciferase/β-Gal ratios are depicted on the y axis.

The increased GFP activity could reflect a more global potentiation of the activity of many IgH promoters. To test this possibility, we used a construct containing a different IgH promoter (V_H 186.2, −230 to +70) (20) (for the sequence, see Fig. 8) linked to the IgH intronic enhancer (EVH186.2) (a gift of F. Alt). We observed that this promoter was also up-regulated relative to the parental cell type (Fig. 5B).

The IgH intronic enhancer appeared to remain largely inactive in the selected cell lines, as the ratio of activity of EVH186.2 to IgH-154 + 35 was approximately the same as that of the parental line (Fig. 5B). In contrast, the B-cell line 70Z/3 showed a dramatic increase in reporter activity when the construct with the enhancer was used. To more directly test this idea, we inserted the E_μ sequence from EVH186.2 into the IgH-154 + 35 construct by using a distal XbaI site (see Materials and Methods) and transfected this construct into the parental and selected 2017 cell lines. In this configuration, we observed a substantial increase in luciferase activity with the enhancer in 70Z/3 B cells, but a much more modest increase in activity with the parental 2017-IgH-GFP no. 9 cell line. While we observed an increase in reporter activity in two selected cell lines with both constructs, we did not observe any further up-regulation in the presence of the enhancer (Fig. 5C). We conclude that the high-GFP 2017 cell lines have been altered in a manner that potentiates the activity of multiple IgH promoters, but IgH intronic enhancer activity is not augmented.

To determine which regions of the IgH 17.2.25 promoter were necessary for the up-regulation in the selected cell lines, constructs lacking either the downstream regions or the octamer motif were tested in transient transfection assays. Among the constructs used, only those that lacked the sequences downstream of the transcription initiation site eliminated the higher promoter activity in the selected high-GFP clones relative to the parental cell type (data not shown). Scanning promoter mutants were used in the selected cell lines to determine which regions were important for the up-regulation (Fig. 5D). Mutations in +16, +17, and +18 and mutations in +19, +20, and +21 largely eliminated the increased promoter activity seen in the selected cells, indicating that this region was critical for the up-regulation. The mutant promoter constructs still retained significant activity over the backbone plasmid (not shown). Thus, the downstream sequences are a likely targets of the factor(s) mediating the up-regulation of promoter activity in the selected high-GFP cells.

High-density gene arrays were used to compare the mRNA expression profiles of the parental and two selected cell lines (m6 and h3), resulting in a list of 70 genes and expressed sequence tags whose expression coordinately changed by twofold or more (unpublished data). Notably absent from the list of genes with altered expression (but present on the array) were Oct-1, Oct-2, and OCA-B. Two up-regulated genes, as measured by the gene chip, were CD53 and L-selectin/MEL-14. We used commercial antibodies to stain for surface CD53 and L-selectin and analyzed the degree of expression using flow cytometry (Fig. 6). CD53 was significantly up-regulated in the selected cell line m6 compared to the 2017-IgH-GFP no. 9 parental cell line (compare Fig. 6B with 6A). L-selectin was up-regulated to a lesser extent, and in this case 2017-IgH-GFP no. 9 had a higher degree of baseline expression (compare Fig. 6F to 6E). Increased nonspecific binding of the secondary antibodies could be excluded as a factor in these results, as omitting the primary antibody eliminated the signal (Fig. 6C, D, G, and H). The GFP fluorescences of 2017-IgH-GFP no. 9 (Fig. 6I) and clone 6 (Fig. 6J) are also shown. These data suggest that the selected cell lines display an altered transcriptional profile compared to the parental cell line.

FIG.6.

Histograms of cell surface staining for CD53 or L-selectin (MEL-14). Cell line 2017-IgH-GFP no. 9 (left panels) or the high-GFP-selected clone m6 (right panels) were stained with biotinylated antibodies to CD53 (PharMingen) (A and B), L-selectin (PharMingen) (E and F), or phosphate-buffered saline as a negative control (C, D, G, and H). After being incubated and washed, the cells were incubated with streptavidin-allophycocyanin (APC) (panels A to H). Also shown in panels I and J are the GFP profiles for 2017-IgH-GFP no. 9 and h3, respectively. Note that the x axes are logarithmic.

Activation of endogenous IgH V region gene expression in selected cells.

The IgH locus undergoes D-J recombination and sterile transcription with relatively high frequency in T-cell lines; however, V region recombination and sterile transcription are not commonly observed and are strictly off in 2017 cells (41). Because the selected cell lines appear to have acquired a more-B-cell-like phenotype, it was possible that the endogenous Ig loci became activated. To test this possibility, we assayed for V_H region transcriptional activity by using RT-PCR. We focused on the IgH region, which is the first to become activated and rearranged in B cells, because the 2017 variants were selected for higher IgH promoter activity. We used primers homologous to the major J558 family of V_H regions. The 5" primer matched the leader sequence encoding the leader peptide, and the 3" primer was homologous to the conserved framework III region. The primers spanned an intron, allowing spliced transcripts to be distinguished from any contaminating DNA. The predicted size of the product corresponding to the spliced RNA was approximately 270 bp, whereas the signal due to contaminating DNA was predicted to be 340 bp. Actin primers were used to control the amount of RNA in the reaction mixture. The parental cell line and the mouse pre-B-cell line 70Z/3 were used as controls. As shown in Fig. 7, no PCR product corresponding in size to the spliced RNA could be detected in 2017-IgH-GFP clone no. 9 (Fig. 7, lane 1). In contrast, reaction mixtures containing RNA from two different high-GFP-expressing clones contained a product matching the size of spliced transcripts (Fig. 7, lanes 3 and 5). As expected, these same transcripts were detected in 70Z/3 cells (Fig. 7, lane 7). When reverse transcriptase was omitted from the reactions (Fig. 7, even-numbered lanes), no RNA signal was detected. These results indicate that the endogenous IgH J558 V regions have become transcriptionally active to at least a limited extent.

FIG. 7.

RT-PCR assay using total RNA prepared from cell line 2017 no. 9 (lanes 1 and 2), high-GFP clone m6 (lanes 3 and 4), h3 (lanes 5 and 6), or the B-cell line 70Z/3 (lanes 7 and 8). Primers recognized the J558 family of variable region gene segments. The 5" primer corresponded to the leader sequence, and the 3" primer corresponded to the framework III region. Actin primers (Ambion) were used as controls. Alternating lanes (−) omitted reverse transcriptase as a control. D, PCR product corresponding to amplification of contaminating DNA; R, shorter product corresponding to spliced mRNA; A, actin product.

DISCUSSION

We have identified sequences proximal to the transcription initiation site that have a role in strong cell-type-specific expression. Complete deletion of sequences between the initiation site and +35 compromises promoter activity in B and selected T cells. Interestingly, the deletion was more deleterious in cell lines derived from later-stage B cells such as BJA-B than for earlier stage lines such as 70Z/3. Scanning promoter mutagenesis pinpointed the +16 to +18 region, as well as a region immediately upstream of the transcription initiation site, as being most important for the higher activity of BJA-B cells relative to the nonlymphoid WERI-27 cells. Grosschedl and colleagues (S. Hardy and R. Grosschedl, personal communication) have analyzed the κ-41 light chain promoter using deletion and mutagenesis. They also have found that sequences near and downstream of the transcription initiation site are important for strong promoter activity as assayed by in vitro transcription (Fig. 8). Thus, analysis of two Ig promoters by using two different assay systems indicates that elements downstream of the transcription initiation site are important for controlling expression.

We have used an in vivo T-cell IgH transcription model to help elucidate new elements mediating high levels of IgH promoter expression. In this approach, 2017 cells carrying a 190-bp IgH promoter upstream of GFP were selected for high activity. This method was preferable to scoring the endogenous Ig locus because (i) the GFP activity was simple to assay and (ii) the protocol selects for higher activity of a relatively small promoter region rather than the megabase-scale endogenous IgH locus. The murine pre-T-cell line 2017 can autonomously modulate the activity of stable transfected IgH promoters, as no stimulus was given to the cells prior to sorting. In the first round of sorting, the top ∼0.1% of cells were selected. Expanding this population and assaying the cells in a second round, we observed that approximately 4% of the cells sorted into a high-GFP window. Assuming slightly slower growth for the high-GFP cells, we estimate that 1 in 20,000 cells carries the high-GFP phenotype in the original population. Although the emergence of the high-GFP variants was spontaneous, the activity was stable even after long periods of culture. Other reports have documented the ability of T-cell lines to intrinsically modulate the expression of certain genes. For example, the cell line EL-4 can spontaneously turn on or off the expression of NK-related surface markers and other markers such as CD4 and CD44 (11). In some cases, the changes were coordinate, as for example CD2 and CD3 would only change together.

The transfected 17.2.25 IgH promoter was more active in the selected cell lines. Scanning promoter mutagenesis experiments pinpointed the same region downstream of the transcription start site that was important for optimal B-cell expression in the up-regulation of promoter activity in the selected cells. This effect was investigated further by examining other promoters. In addition to two IgH promoters representing different VH family members (17.2.25 is a V_H 81X/7183 family member and V_H 186.2 is a member of the major J558 class), two other B-cell-specific promoters—B29 and mb-1—have increased activity in the selected cell lines relative to the parental cell lines and to the histone H2B promoter. In contrast, the pre-T-cell receptor α promoter remained equally active (not shown).

The difference between the levels of expression of the immunoglobulin promoters in the selected clones versus the parental cell lines strongly indicates that the activities of specific transcription factors are different in the two populations. However, the degree of up-regulation of the transiently transfected 17.2.25 IgH promoter in the selected cell lines (3- to 5-fold) was much less than the degree of up-regulation of the integrated promoter upstream of GFP (10- to 100-fold). This difference in the degree of transcriptional activation could be explained by differences in the chromatin state of the integrated and transiently transfected plasmids.

A simple explanation for the increase in expression of the IgH promoter in the selected cells would be that a rare event changed the methylation state or chromatin structure of the stable transfected reporter. Several points argue against such a local mechanism. First, multiple different stable transfected clones of 2017-IgH-GFP can be selected to have high GFP activity, indicating that the position(s) of integration does not affect the up-regulation of the IgH promoter. Second, this same promoter and other B-cell-specific promoters are up-regulated in the high-GFP cell lines when transiently transfected. In the case of the 17.2.25 IgH promoter, the up-regulation is sensitive to particular point mutations in transient transfection assays. Third, the endogenous IgH locus has apparently become activated in the selected cell lines (see below), suggesting a common function of specific trans factors at different chromosomal locations. Fourth, a gene chip indicates that a number of genes have an altered expression in the selected cell lines (Fig. 6 and unpublished data).

Interestingly, the 17.2.25 and 186.2 IgH promoters, and the B29 and mb-1 promoters, all contain pyrimidine-rich sequences near their transcription initiation sites. Specific mutations in the 17.2.25 sequence decreased promoter activity. Linker scanner mutations in the κ-41 light chain promoter near and downstream of the transcription initiation site also compromise promoter activity (Hardy and Grosschedl, personal communication). Aligning these two promoter sequences resulted in no significant overall homology. However, the highest scoring alignments corresponded to the areas identified in the different screens (data not shown). This finding indicates that the two sequence elements may be functionally related.

We examined several other promoters from both the heavy and light chain V region genes of both mice and humans for similar sequences (Fig. 8). Many of these sequences contain obvious pyrimidine-rich sequences, typically downstream of the transcription initiation site. Although we chose promoters with minimal overall similarity to one another, all Ig V regions presumably derived from genetic duplication of an ancestral prototype, and thus aligning these promoters may highlight identity by descent. However, a Gibbs sampling algorithm (22) designed to detect conserved motifs distributed within a set of sequences produced a 14-nucleotide consensus when provided with the sequences shown in Fig. 8. The consensus is shown at the bottom of Fig. 8, and the top 25 perfect and near matches to the consensus (all of which are 9 of 14 nucleotides or better) are boxed. When the nucleotide sequences were scrambled, the algorithm did not produce a meaningful consensus. The derived sequence appears to consist of two half sites, one that is T/G rich and one that is A/C rich, with a pyrimidine-rich sequence in between them. Importantly, the nucleotide residues implicated in promoter expression by mutational analysis in the 17.2.25 heavy chain and κ-41 light chain promoters (Fig. 8) overlapped with consensus sequences. We propose that factors recognizing this consensus sequence mediate higher promoter activity in B cells and in the selected T-cell lines. The biological significance of these sequences is buttressed by the finding that downstream nucleotides from the 186.2 promoter, containing the two boxed elements shown in Fig. 8, can restore full promoter activity to the −154 to +1 segment of the 17.2.25 promoter.

The endogenous IgH locus became activated in several of the selected cell lines, as RT-PCR showed the presence of spliced J558 V region transcripts. Although it is formally possible that the transcripts are derived from a recombined mature IgH mRNA, it is unlikely given the lack of activation of the IgH intronic enhancer in the selected cell lines and the lack of activation of constant region genes present on a murine 11,000-molecular-weight gene chip (11K gene chip). We have also been unable to detect IgH recombination in DNA prepared from the selected cell lines by PCR (data not shown). More likely, these RT-PCR products represent germ line transcripts. Stable expression of the Ig intronic enhancer-binding transcription factor E47 in 2017 has been shown to activate limited recombination and sterile transcription of the D-J regions, but not the V regions, of the IgH locus (42). Therefore, high-GFP selection of a 2017-IgH-GFP cell line expressing E47 could initiate IgH production. Such a system would provide an experimental platform by which to study factors that activate the IgH locus in a T-cell line.

The octamer motif was originally identified through its high conservation in Ig promoters (9, 34). Subsequently, the octamer was found in other non-B-cell-specific genes, appearing to rule it out as a specificity element. The description of B-cell-specific complexes forming on octamer elements, later shown to be due to Oct-2 (21, 40, 47, 52), rekindled the debate over octamer specificity. More recently, B cells derived from Oct-2 homozygous null mice were shown to be capable of expressing Ig. The coactivator OCA-B was also shown to be dispensable (19, 32, 43).

One explanation for these results is that an as yet unidentified cofactor mediates B-cell specificity through the octamer, possibly in conjunction with the ubiquitous factor Oct-1. Another possibility is that the octamer motif is not the sole mediator of Ig promoter activity or B-cell specificity. There is ample evidence to support this idea. A number of Ig heavy and light chain V region promoters lack consensus octamer sites and do not interact with Oct proteins in vitro but are nevertheless efficiently expressed (2). In one such promoter, activity was shown to be mediated by a pyrimidine-rich element termed κY. Eaton and Calame also noted the presence of a pyrimidine-rich region in many Ig promoters, in this case upstream of the octamer and heptamer motifs (8). In their study, mutation of this element was as deleterious to the activity of the promoter as point mutation of the octamer. A similar sequence occurs upstream of the octamer in the 17.2.25 promoter used in these studies (centered at −105). Roeder and colleagues have used chimeric constructs containing portions of an IgH and the H2B promoter. Their data indicate that the determinants of promoter specificity lie not with the octamer sequence but with the core promoter (R. Roeder, personal communication). In support of this idea, we have found that nuclear extracts prepared from the selected high-GFP 2017 cells show an increased ability to form a specific nucleoprotein complex at the IgH core promoter relative to extracts from the unselected cell line (D. Tantin and P. A. Sharp, unpublished data).

Proteins that interact with promoter DNA proximal to and downstream of the transcription initiation site have been described previously. TFIID, for example, forms an extended footprint containing multiple downstream DNA contacts to approximately +35 on the adenovirus major late promoter (31). These contacts are especially important for promoter activity in the case of promoters with nonconsensus TATA regions (as with the IgH promoter), and they can be sequence specific (18). More recently, sequence-specific downstream promoter elements have been described for Drosophila and mammals (4). Current work is aimed at determining the nature of the factors that interact with these new Ig promoter elements.