C/EBPα:AP-1 Leucine Zipper Heterodimers Bind Novel DNA Elements, Activate the PU.1 Promoter, and Direct Monocyte Lineage Commitment More Potently Than C/EBPα Homodimers or AP-1

Abstract

The basic-region leucine zipper (BR-LZ or bZIP) transcription factors dimerize via their LZ domains to position the adjacent BRs for DNA-binding. Members of the C/EBP, AP-1, and CREB/ATF bZIP subfamilies form homodimeric or heterodimeric complexes with other members of the same subset and bind specific DNA motifs. Here we demonstrate that C/EBPα also zippers with AP-1 proteins and that this interaction allows contact with novel DNA elements and induction of monocyte lineage commitment in myeloid progenitors. A leucine zipper swap:gel shift assay demonstrates that C/EBPα zippers with c-Jun, JunB, or c-Fos, but not with c-Maf or MafB. To evaluate activities of specific homodimers or heterodimers we utilized LZs with acid (LZE) or basic (LZK) residues in their salt bridge positions. C/EBPαLZE:C/EBPαLZK preferentially binds a C/EBP site, c-JunLZE:c-FosLZK an AP-1 site, and C/EBPαLZE:c-JunLZK a hybrid element identified as TTGCGTCAT by oligonucleotide selection. In murine myeloid progenitors, C/EBPα:c-Jun or C/EBPα:c-Fos LZE:LZK heterodimers induce monocyte lineage commitment with markedly increased potency compared with C/EBPα or c-Jun homodimers or c-Jun:c-Fos heterodimers, demonstrating a positive functional consequence of C/EBP:AP-1 bZIP subfamily interaction. C/EBPα:cJun binds and activates the endogenous PU.1 promoter, providing one mechanism for induction of monopoiesis by this complex.

Introduction

Transcription factor dimerization provides functional diversity. C/EBP, AP-1, or CREB/ATF basic region-leucine zipper (BR-LZ or bZIP) proteins readily dimerize with other members of their subfamily. Interaction between the AP-1 and ATF or C/EBP and ATF proteins also occurs (Hai and Curran 1999; Gombart et al 2007). c-Fos or c-Jun bind the C/EBPβ bZIP domain and prevent DNA-binding by C/EBPβ even if the c-Fos or c-Jun BRs are deleted, suggesting direct interaction via their LZ domains (Hsu et al 1994). Similarly, deletion of the LZ from either C/EBPα or c-Jun disrupts their interaction, and endogenous C/EBPα and c-Jun co-immunoprecipitate (Rangatia et al 2002). These studies suggest direct zippering between c-Jun or c-Fos and C/EBPα or C/EBPβ as opposed to contact via their outer LZ surfaces. However, none of the AP-1 or Maf LZs interacts with C/EBP LZs on a zipper peptide chip (Newman and Keating 2003). We now provide evidence for zippering between C/EBP and AP-1 proteins. Due to the relative weakness of these interactions we propose that heterodimer formation results in gene activation via novel DNA elements rather than factor cross-inhibition.

To assess the function of specific bZIP homodimers or heterodimers, one group linked Jun and Fos proteins via an 18 amino acid tether (Bakiri et al 2002). Concern for pertubation of tertiary structure led us to develop an alternative strategy utilizing acidic and basic LZs (O’Shea et al 1993). We find that C/EBPα homodimers bind a C/EBP cis element, that C/EBPα:c-Jun prefers a hybrid element containing C/EBP and AP-1 half-sites, and that c-Jun:c-Fos interacts most strongly with an AP-1 site.

C/EBPα(−/−) neonatal mice lack granulocytes and have reduced monocytes and CFU-M (Zhang et al 1997; Heath et al 2004). Deletion of C/EBPα in adult mice blocks the common myeloid progenitor (CMP) to granulocyte-monoyte progenitor (GMP) transition (Zhang et al 2004). To evaluate the effect of C/EBPα on myeloid specification, we transduced C/EBPα-ER into mononuclear murine marrow cells and carried out lineage depletion after drug selection. This strategy minimizes bias due to C/EBPα-mediated cell cycle inhibition. Activation of C/EBPα-ER with estradiol (E2) increases formation of monocytes and their CFU-M progenitors, compared to granulocytes or CFU-G (Wang et al 2006). C/EBPαGZ did not induce monopoiesis. This variant contains a GCN4 LZ, restricting heterodimerization. c-Jun, c-Fos, c-Maf, and MafB each induce monocytic differentiation of myeloid cells (Lord et al 1993; Hegde et al 1999; Kelly et al 2000). We find that C/EBPα zippers with Jun or Fos but not Maf proteins and that C/EBPα:c-Jun or C/EBPα:c-Fos heterodimers induce monocyte lineage specification more potently than C/EBPα or c-Jun homodimers or c-Jun:c-Fos heterodimers. Moreover, C/EBPα:c-Jun binds and activates the PU.1 promoter, potentially accounting for induction of monopoiesis.

Results

C/EBPs zipper with c-Jun, JunB, and c-Fos but not c-Maf or MafB

Detecting a band of intermediate size on a gel shift assay provides evidence for hydrophobic surface zippering of LZ proteins with identical DNA-binding specificity (Kataoka et al 1994). To evaluate interaction between C/EBP and AP-1 proteins, which differ in their DNA specificity, a novel approach was utilized in which heterologous LZs replace the C/EBPα LZ. For example, C/EBPα(JZ) contains the c-Jun LZ in place of the C/EBPα LZ. These full-length proteins were co-expressed with a smaller bZIP domain, such as C/EBPα(bZIP), and incubated with a radiolabeled C/EBP binding site (Figure 1a). Zippering between the C/EBPα, C/EBPβ, or C/EBPδ LZs and the c-Jun LZ was readily detected (Figures 1b, 1c lanes 1–7). Note increased expression of βbZIP, compared with αbZIP or δbZIP, and consequently its heterodimer. The GCN4 LZ (GZ), linked to the C/EBPα BR in the short αGZ(bZIP) protein, did not interact with the c-Jun LZ (in αJZ) or with the C/EBPα LZ (Figure 1c lanes 8–11). C/EBPα(L12V), in which the first two leucines of the LZ are changed to valine, did not bind DNA, confirming a requirement for zippering. The c-Fos LZ (FZ) zippered with C/EBPα or C/EBPβ, but the MafB LZ (MBZ) did not (Figure 1d). As expected, c-Fos alone did not bind DNA (Figure 1d, lane 5), with the slight bands seen representing background. In contrast, c-Fos:αbZIP and c-Fos:βbZIP complexes were readily detected (lanes 6 and 7). The JunB LZ (JB) bound C/EBPα or C/EBPβ in this assay (Figure 1e, lanes 3 and 4), whereas the c-Maf LZ (CMZ) did not (Figure 1f). Promoter competition likely causes diminished expression of C/EBPαbZIP or C/EBPβbZIP when expressed with C/EBPαMBZ or C/EBPαCMZ.

Figure 1.

The C/EBP and AP-1 LZs interact to enable DNA-binding. (a) Diagram of the zipper swap:gel shift assay. (b–f) The indicated proteins were co-expressed in 293T cells, and nuclear extracts were prepared and subjected to gel shift assay using the NE-C/EBP probe. * marks bZIP proteins, ** marks full-length proteins, and arrows mark heterodimers. Abbreviations: α-C/EBPα; αbZIP, βbZIP, δbZIP - bZIP domains of C/EBPα, C/EBPβ, or C/EBPδ; αJZ, αFZ,αJB, αMBZ, αCMZ - C/EBPα with c-Jun, c-Fos, JunB, MafB, or c-Maf LZs, L12V C/EBPα mutant L317V/L324V; αGZbZIP the C/EBPα BR linked to the GCN4 LZ.

C/EBPα or C/EBPα(L12V) were co-expressed with C/EBPβ, c-Jun, JunB, or c-Fos in 293T cells. C/EBPα co-immunoprecipitated C/EBPβ or the AP-1 proteins but L12V did not, demonstrating that these interactions depend on integrity of the LZ (Figure 2a). In addition, endogenous C/EBPα co-immunoprecipitated with endogenous c-Jun, JunB, or c-Fos from a myeloid cell nuclear extract (Figure 2b), and endogenous C/EBPα co-localizes with c-Jun or c-Fos by confocal microscopy in myeloid cells (Supplementary Figure 1).

Figure 2.

Interactions between C/EBPα and AP-1 proteins. (a) C/EBPα (α) or its L12V variant were co-expressed with C/EBPβ (β), c-Jun, JunB, or c-Fos in 293T cells. Nuclear extracts were subjected to immunoprecipitation (IP) with the indicated antisera followed by Western blotting with C/EBPα antibody. I – 1% of IP input extract. (b) Nuclear extracts corresponding to 4 × 10⁷ WEHI-3B D+ myeloid cells were subjected to IP with rabbit IgG (Ig) or the indicated antisera, followed by Western blotting for C/EBPα.

Acidic and basic leucine zippers direct specific heterodimerization of bZIP proteins

The LZ dimerizes as a coiled-coil in which a and d residues form a hydrophobic surface and e and g residues form salt bridges (Figure 3a). The e and g residues within the GCN4 LZ were changed to glutamic acid to generate an acidic LZ (LZE) or to lysine to generate a basic LZ (LZK) (Figure 3b). Oligonucleotides encoding LZE or LZK were ligated in place of the C-terminal C/EBPα, c-Jun, and JunB LZs or between residues 165–194 of the 380 amino acid c-Fos. Neither C/EBPαLZE nor C/EBPαLZK alone bound DNA, but when expressed together strong binding was seen (Figure 3c).

Figure 3.

Directed heterodimerization using acid and basic LZs. (a) Two α-helices dimerizing via hydrophobic a and d residues, with e and g residues forming salt bridges. (b) Diagram of C/EBPα dimerizing via acidic LZE and basic LZK. (c) Gel shift assay using 293T nuclear expressing C/EBPαLZE (αLZE), C/EBPαLZK (αLKZ), or both (E+K) with the NE-C/EBP probe. (d) Gel shift assay using extracts expressing the indicated combinations of LZE/LZK proteins with four DNA probes, NE-αα(ATTGCGCAAT), NE-αJ(tgacGCAAT), NE-JJ(TGACTCA), and NE-α-(GCAAT). The AP-1 half-site in the hybrid αJ site is shown in lower case. (e) Gel shift assay using extracts expressing the indicated LZE+LZK protein combinations, or 293T extract as a control, with the MS-JJ AP-1 site probe.

We evaluated the affinities of C/EBPα homodimers, and C/EBPα:AP-1, c-Fos:c-Jun, or c-Jun:c-Jun heterodimers for a palindromic C/EBPα site (αα), an AP-1 site (JJ), a hybrid site (αJ), or a C/EBPα half-site (α-) embedded in sequences flanking the neutrophil elastase (NE) C/EBP site (Figure 3d). C/EBPα homodimers preferentially bound the C/EBP site, whereas C/EBPα:c-Jun or C/EBPα:JunB had greater affinity for the hybrid site. These three protein combinations did not bind the AP-1 site in the context of NE flanking sequences, but neither did c-Jun:c-Fos nor c-Jun:c-Jun heterodimers (not shown). The macrosialin (MS) promoter contains an identical core AP-1 site (Li et al 1998). c-Jun:c-Fos or c-Jun:c-Jun bound the MS AP-1 site with greater potency than C/EBPα:C/EBPα, C/EBPα:c-Jun, C/EBPα:JunB, or C/EBPα:c-Fos (Figure 3e). Expression of the LZE and LZK proteins was confirmed by Western blot analysis, as was expression of LZE/K ER fusion proteins to be used for marrow transduction studies (Figure 4a). These assays indicate that C/EBP:AP-1 heterodimers bind unique DNA elements.

Figure 4.

C/EBPα:c-Jun or C/EBPα:c-Fos induce monocytic differentiation more potently than C/EBPα or c-Jun homodimers or c-Fos:c-Jun heterodimers. (a) 293T cells were transiently transfected with retroviral plasmids expressing the indicated non-ER or ER fusion LZE or LZK proteins. Total cellular proteins harvested at 48 hrs were subjected to Western blotting with C/EBPα (lanes 1–2), c-Jun (lanes 3–4), JunB (lanes 5–6), c-Fos (lanes 7–8), or ER (lanes 9–14) antisera. Lanes were loaded with equal cell numbers except that JunB-LZE and C/EBPα-LZK were loaded 20-fold less due to their higher expression. (b) Marrow cells isolated from mice exposed to 5-FU were transduced with pBabePuro-C/EBPαLZK-ER (αLZK) and MIGR1 vectors expressing C/EBPαLZE-ER (αLZE), c-JunLZE-ER (cJLZE), or c-FosLZE-ER (FLZE). After three days of transduction and one day of puromycin selection, viable cells were isolated and subjected to lineage-depletion. The cells were then cultured +/− E2 for 2 days in IL-3/IL-6/SCF and subjected to FACS for Mac1 and Gr-1, gating on GFP⁺ cells. Representative data is shown. Cells in the upper left, Mac1⁺Gr-1⁻ quadrant are monocytes, and cells in the upper right, Mac1⁺Gr-1⁺ quadrant are granulocytes. Percent monocytes (%M) = Mac1⁺Gr-1⁻/(Mac1⁺Gr-1⁻ + Mac1⁺Gr-1⁺) cells × 100%. (c) The increase in percent monocytes due to E2 addition upon FACS analysis of marrow cells transduced with the indicated combinations of LZE-ER and LZK-ER proteins or empty vectors. (d) Increase in percent monocytes in marrow cells transduced with pBabePuro-C/EBPα-ER and MIGR1 (α), pBabePuro and MIGR1-cJun-ER (cJ), both pBabePuro-C/EBPα-ER and MIGR1-cJun-ER (α:cJ), or MIGR1-cFos-ER (cF). Data are mean and standard error from three experiments, with p values comparing +/− E2 (left). Expression of ER fusions was verified by Western analysis of transfected 293T cells (right).

C/EBPα:c-Jun or C/EBPα:-c-Fos direct monopoiesis

Marrow mononuclear cells from mice exposed to 5-FU were cultured in IL3/IL6/SCF for one day and then transduced simultaneously with pBabePuro-C/EBPαLZK-ER and either MIGR1-C/EBPαLZE-ER, MIGR1-cJunLZE-ER, or MIGR1-cFosLZE-ER. MIGR1 encodes GFP downstream of an IRES. Transduction was also carried out with the empty vectors (Pu:MIG) or with pBabePuro-cJunLZK-ER and MIGR1-cJunLZE-ER or MIGR1-cFosLZE-ER. After three days, puromycin was added, and the surviving cells were lineage-depleted to generate the D0 population. The cells were then placed in liquid culture with or without E2 and analyzed for Mac1 and Gr-1 on day 2 (D2), gating on the GFP⁺ population. Representative FACS data obtained with three of the combinations and the mean increase in monocytes upon exposure to estradiol are shown (Figures 4b, 4c). C/EBPα:c-Jun or C/EBPα:c-Fos strongly induced monocytic differentiation, C/EBPα:C/EBPα had a mild effect, and c-Jun:c-Jun, c-Jun:c-Fos, or the empty vectors had no effect. Intermediate induction was seen with C/EBPα:JunB, and C/EBPαLZK alone did not increase monocytes (not shown).

Having found that C/EBPα or c-Jun LZE:LZK homodimers only mildly induced monopoiesis, we confirmed that wild-type C/EBPα-ER markedly induces monopoiesis, and we also found that c-Jun-ER or c-Fos-ER potently induced monocyte formation in normal progenitors (Figure 4d). Together, our results are consistent with the idea that C/EBPα zippers with endogenous AP-1 proteins and c-Jun or c-Fos with endogenous C/EBPα to induce monopoiesis. Co-expression of C/EBPα-ER and c-Jun-ER increased monocyte formation (Figure 4d) marginally more potently than c-Jun alone (P=0.06, comparing % monocytes +E2).

In an additional set of experiments focused on commitment decisions, cells transduced with LZE-ER:LZK-ER pairs were exposed to puromycin and then subjected to flow cytometry to isolate GFP⁺ cells. After culture with or without E2 for 24 hrs, the cells were plated in methylcellulose without E2, and monocytic and granulocytic colonies (CFU-G, CFU-M) were enumerated on day 8 (Figure 5). As expected, basal monocytic development was higher in IL3/IL6/SCF than in GCSF/SCF. Nevertheless, either C/EBPα:c-Jun or C/EBPα:c-Fos significantly increased the %CFU-M in both culture conditions. In contrast, neither C/EBPα homodimers nor c-Jun:c-Fos increased the formation of CFU-M.

Figure 5.

C/EBPα:c-Jun or C/EBPα:c-Fos induce monocyte lineage specification more potently than C/EBPα homodimers or c-Fos:c-Jun. (a, b) The increase in percent CFU-M due to E2 exposure for 24 hrs just prior to culture of doubly transduced, GFP⁺ and puromycin-resistant cells in methylcellulose with IL-3/IL-6/SCF or with G-CSF/SCF is shown. %CFU-M = CFU-M/(CFU-M + CFU-G) × 100%. Data are mean and standard error from three experiments.

C/EBPα:c-Jun binds and activates the PU.1 promoter

Oligonucleotide immunoprecipitation in the presence of C/EBPαLZK and c-JunLZE, alternating use of C/EBPα and c-Jun antiserum, selected a binding site whose consensus contains both a C/EBPα and a c-Jun half-site, identical to the NE-αJ probe we had designed (Figure 6a). As increased PU.1 favors monopoiesis (Dahl et al 2003), we sought related sites in the PU.1 gene. A previously identified C/EBPα binding site at −68 in the PU.1 promoter is similar to the selected C/EBPα:c-Jun consensus; this site interacted with either C/EBPα:C/EBPα, C/EBPα:c-Jun, or C/EBPα:c-Fos in gel shift assay, albeit modestly, and only bound C/EBPα:c-Jun when its flanking sequences were changed to guanines (Figure 6b). Results with the G flank probe provide a further example of a site bound specifically by C/EBPα:c-Jun. Two strong C/EBPα-binding sites in the PU.1 –14 kb distal enhancer (Yeamans et al 2006) only bound C/EBPα homodimers (Figure 6c), and none of several other distal enhancer oligonucleotides bound C/EBPα:c-Jun (not shown). Consistent with the gel shift findings, C/EBPα, C/EBPβ, or c-Jun each bound the endogenous PU.1 promoter in 32Dcl3 myeloid cells cultured in G-CSF using chromatin immunoprecipitation (ChIP), whereas only C/EBPα or C/EBPβ bound the PU.1 distal enhancer or the NE promoter (Figure 6d).

Figure 6.

C/EBPα:cJun binds a hybrid DNA element and the PU.1 promoter. (a) Sequences of individual DNAs selected, and their consensus, after immunoprecipitation of complexes bound to either C/EBPαLZK:cJunLZE, c-Jun, or C/EBPα. Bold nucleotide symbols match the consensus. Lower case letters indicate the second most frequent base. (b) Gel shift assay using 293T extracts expressing no exogenous protein (−), C/EBPαLZK:C/EBPαLZK (α:α), C/EBPαLZK:cJunLZE (α:cJ), or C/EBPαLZK:cFosLZE (α:cF) and either a radiolabeled probe containing the PU.1 promoter −68 bp C/EBP site with 6 bp of 5′ and 3′ flanking sequences (PU.1–68) or the same probe with the flanking sequences replaced with guanines (G flank). (c) Gel shift using these proteins and C/EBPαLZK:JunBLZE (α:JB) and probes containing either the C2 or C5 C/EBPα-binding sites in the PU.1 distal enhancer. (d) ChIP assay using 32Dcl3 cells cultured in IL-3 (top panel) or in G-CSF for one day (panels 2–6) and either rabbit Ig, C/EBPα, C/EBPβ, or c-Jun antisera. I input. PCR was done using oligonucleotides derived from the PU.1 promoter (−215 to +25), the PU.1 distal enhancer, the NE promoter (−157 to +92), the βactin gene, or −7 kb of the PU.1 gene. Results shown are representative of two independent experiments.

We generated Ba/F3 clonal lines expressing C/EBPαLZK-ER and either C/EBPαLZE-ER, c-JunLZE-ER, or c-FosLZE-ER (Figure 7a). Ba/F3 cells were utilized as they have myeloid potential and as we encountered difficulty co-expressing these proteins in 32Dcl3 cells. Co-expression of C/EBPαLZE-ER with C/EBPαLZK-ER was confirmed by RT-PCR using LZE or LZK specific primers (not shown). Their interaction with the PU.1 promoter by ChIP after induction (Figure 7b) further confirms their co-expression, as either alone cannot bind DNA. Lack of interaction −E2 indicates tightness of the regulated system. C/EBPα:c-Jun and C/EBPα:c-Fos also bound the endogenous PU.1 promoter (Figure 7b). However, only C/EBPαLZK:c-JunLZE induced endogenous PU.1 mRNA (Figure 7c).

Figure 7.

C/EBPα:cJun binds and activates the endogenous PU.1 promoter. (a) Western blot analysis of parental Ba/F3 cells or subclones expressing the indicated LZK:LZE ER fusion protein combinations, using C/EBPα, c-Jun, c-Fos, or actin antisera (top to bottom). (b) ChIP assay using the Ba/F3 lines and Ig or ER antisera. PCR was done for the PU.1 promoter, the β-actin gene, or −7 kb of the PU.1 gene. Results +E2 (bottom three panels) are representative of two independent experiments, and results −E2 (top panel) are from one experiment. (c) RNA prepared from the Ba/F3 lines cultured with or without E2 for 6 hrs was subjected to real time PCR for PU.1 mRNA, with GAPDH mRNA PCR used for normalization. Mean and standard deviation from three determinations is shown. P value for PU.1 expression with versus without E2 is from the paired student’s t test.

DISCUSSION

We have provided evidence that C/EBPα zippers with c-Jun, JunB, or c-Fos, but not with c-Maf or MafB. The co-immunoprecipitation results suggest that C/EBPα has greater affinity for C/EBPβ than for c-Jun, JunB, or c-Fos. As the C/EBP α and C/EBPβ LZs are highly conserved, we further surmise that C/EBPα has greater affinity for itself than for AP-1 proteins. A scheme for predicting interaction between LZs enumerates contacts between acidic residues and basic residues in the e and g positions (Vinson et al 1993). C/EBPα:C/EBPα, c-Fos:c-Jun, or c-Fos:JunB has four, C/EBPα:c-Jun two, and C/EBPα:JunB or C/EBPα:c-Fos three LZ salt bridges. Detection of Fos:Jun but not C/EBP:Jun or C/EBP:Fos interactions on a zipper chip may reflect weaker interaction of the latter proteins (Newman and Keating 2003). Similarly, gel shift assays using myeloid extracts and the NE-αJ probe has not detected a species super-shifted by both C/EBP and AP-1 antisera (not shown). While weakness of C/EBP:AP-1 interaction may prevent dominant inhibition between these bZIP proteins, a positive functional effect of heterodimerization via interaction with novel DNA elements remains a likely consequence. In addition, the finding that C/EBPs do not zipper with c-Maf or MafB sets a limit on the heterodimers that occur in cells.

Ours is the first study to position acidic and basic LZs in place of the natural LZ in bZIP transcription factors. C/EBPα:AP-1 LZE:LZK heterodimers bound hybrid DNA elements with C/EBP and AP-1 half-sites more potently than C/EBPα LZE:LZK homodimers or c-Jun:c-Fos or JunB:c-Fos LZE:LZK heterodimers. Also, C/EBPα:c-Jun or C/EBPα:c-Fos induced monocyte lineage commitment more potently than C/EBPα or c-Jun homodimers or c-Jun:c-Fos heterodimers. In our previous study involving transduction with a single retroviral vector, C/EBPα-ER induced CFU-M whereas C/EBPα(GZ)-ER, which can only homodimerize, did not (Wang et al 2006). Together, these findings support the conclusion that C/EBPα heterodimerizes with endogenous c-Jun or c-Fos to induce monocyte lineage commitment.

Increased PU.1 levels favor monopoiesis over granulopoiesis (Dahl et al 2003), suggesting that C/EBPα:AP-1 heterodimers induce PU.1 or a protein that cooperates with PU.1 in myeloid progenitors. C/EBPα homodimers appear more important than C/EBP:AP-1 complexes for regulation of the −14 kb PU.1 distal enhancer. In contrast, C/EBPα:c-Jun binds and activates the PU.1 promoter, providing one explanation for why this heterodimer acts more potently than C/EBPα homodimers to induce monopoiesis. Although both C/EBPα:C/EBPα and C/EBPα:c-Jun bind the PU.1 promoter, the latter may uniquely attract additional coactivators in myeloid progenitors. Perhaps early during myeloid development, C/EBPα activates PU.1 to some degree, enabling the CMP to GMP transition. Subsequently, increased expression of c-Jun or its coactivators may allow further PU.1 induction and monocyte lineage commitment. Future studies will seek to identify additional genetic targets bound and activated uniquely by C/EBPα:AP-1 complexes relevant to myeloid determination or other biologic processes.

A positive functional consequence of C/EBP:AP-1 bZIP subfamily interaction had not been previously appreciated. Applying the zipper swap:gel shift assay to additional bZIP or bHLH proteins should identify the full spectrum of potentially interacting partners within these transcription factor families, and the LZE:LZK method of directed dimerization should enable assessment of the unique functions of each bZIP:bZIP or bHLH:bHLH complex.

Materials and Methods

Plasmid construction

CMV-C/EBPαbZIP, CMV-C/EBPβbZIP, or CMV-C/EBPδbZIP encode residues 272–358 of rat C/EBPα, 210–296 of murine C/EBPβ, or 183–268 of murine C/EBPδ. C/EBPαL12V contains the L317V/L324V mutations. An XhoI site was introduced upstream of C/EBPα residue T310. C/EBPα residues 310–358, containing the LZ, were then replaced with fragments encoding LZs from murine c-Jun, murine JunB, rat c-Fos, human MafB, human c-Maf, or GCN4 to generate C/EBPαJZ, C/EBPαJBZ, C/EBPαFZ, C/EBPαMBZ, C/EBPαCMZ, or C/EBPαGZ. The peptide sequence of LZE is LEDEVEELESENYHLENEVARLEKEV and that of LZK is LKDKVEELKSKNYHLKNKVARLKKKV, with e and g positions changed to E or K in the GCN4 LZ underlined. XhoI was introduced upstream of the first leucine of the c-Jun, JunB, or c-Fos LZs, and a BamHI site replaced c-Fos residues 195–196. Four overlapping oligonucleotides encoding LZE or LZK followed by a stop codon were ligated in place of the C-terminal C/EBPα, c-Jun, or JunB LZs. Similar oligonucleotides having a BamHI site instead of the stop codon were ligated in place of the internal c-Fos LZ. To generate ER fusion proteins, the CEBPα segment in C/EBPαER was replaced by segments encoding LZE or LZK proteins, c-Jun, or c-Fos. Each construct was confirmed by dideoxy-DNA sequencing.

Gel shift assay, co-immunoprecipitation, and Western blotting

Gel shift assay was carried out as described (Cao et al 1997). The sense strand of the NE-C/EBP oligonucleotide is 5′-TCGAGGCCAGGATGGGGCAATACAACCCG, with the C/EBP site underlined. In the αα, αJ, JJ, and α-oligonucleotides, this C/EBP site is replaced by ATTGCGCAAT, TGACGCAAT, TGACTCA, or GCAAT, respectively. The MS-C/EBP oligonucleotide is 5′-GATCCAGGTGTCTGAGCTAGGTTTGG, with the AP-1 site underlined. For co-immunoprecipitation, cell extracts were prepared from transfected 293T cells or from WEHI-3B D+ myeloid cells. Nuclear lysates were prepared using 20 mM Tris, pH 8.0, 400 mM NaCl, 0.2 mM EDTA, 20% glycerol, 1 mM DTT, and protease inhibitors and then diluted in 50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 5% glycerol, 0.25% NP40. After preclearing, 2 μg rabbit antiserum or Ig was added for 16 hr followed by A/G-Sepharose. Samples were washed and eluted in Laemmli buffer and subjected to Western blotting with mouse anti-CEBPα (Affinity Bioreagents, Golden, CO, USA) as described (Cao et al 1997). Antisera employed for immunoprecipitation were C/EBPβ (C-19) c-Jun (N), JunB (N-17), or c-Fos (4) (Santa Cruz Biotech., Santa Cruz, CA, USA). These antisera, and C/EBPα (AA14) or ERα (HC-20) (Santa Cruz), were also employed for Western analysis.

Progenitor transduction, culture, and FACS analysis

Marrow cells from C57BL/6 mice exposed to 5-fluorouracil were transduced simultaneously with pBabePuro and MIGR1 vectors for three days, selected in puromycin for one day, subjected to lineage-depletion, cultured in liquid medium for 2 days +/− E2, and subjected to FACS, gating on GFP⁺ cells (Wang et al 2006). Transduced, lineage-depleted cells were also subjected to flow cytometry to isolate GFP⁺ cells, cultured for 1 day +/− E2, and plated in methylcellulose without E2 as described (Wang et al 2006). Myeloid CFUs were enumerated on day 8. Statistical analysis was done using the student’s t test.

Oligonucleotide selection, ChIP, RT-PCR, and reporter assays

Oligonucleotide selection was done by incubation of transduced 293T nuclear extracts with 5′-GTGCTCTAGAGGATCCGACGGGGG(N16)GGGGGATCGATGGGTATA and its complement, followed by immunoprecipitation, DNA isolation, and PCR using 5′-GTGCTCTAGAGGATCCGAC and 5′-TATACCCATCGATCCCCC. After 5 cycles of selection, DNAs were digested with XbaI/ClaI, inserted into pBluescript, and sequenced. Ba/F3 cells were transduced with pBabePuro and pBabeNeo retroviral vectors as described (Wang and Friedman 2002). ChIP assay and assessment of PU.1 promoter activation in 293T cells were also done as reported previously (Kummalue and Friedman 2003). RNA was prepared and subjected to quantitative PCR for PU.1 and GAPDH as described (Wang et al 2006).