Lymphoid to Myeloid Cell Trans-Differentiation Is Determined by C/EBPβ Structure and Post-Translational Modifications

The transcription factor C/EBPβ controls differentiation, proliferation, and functionality of many cell types, including innate immune cells. A detailed molecular understanding of how C/EBPβ directs alternative cell fates remains largely elusive. A multitude of signal-dependent post-translational modifications (PTMs) differentially affect the protean C/EBPβ functions. In this study we apply an assay that converts primary mouse B lymphoid progenitors into myeloid cells in order to answer the question how C/EBPβ regulates (trans-) differentiation and determines myeloid cell fate. We found that structural alterations and various C/EBPβ PTMs determine the outcome of trans-differentiation of lymphoid into myeloid cells, including different types of monocytes/macrophages, dendritic cells, and granulocytes. The ability of C/EBPβ to recruit chromatin remodeling complexes is required for the granulocytic trans-differentiation outcome. These novel findings reveal that PTMs and structural plasticity of C/EBPβ are adaptable modular properties that integrate and rewire epigenetic functions to direct differentiation to diverse innate immune system cells, which are crucial for the organism survival.


Introduction
Understanding the molecular attributes and post-transcriptional regulation of transcription factors in cell fate determination remains a challenging task in molecular genetics and developmental biology. Ectopic expression of some key transcription factors can perturb cellular differentiation programs and install new ones, such as during lymphoid to myeloid reprogramming or trans-differentiation induced by CCAAT enhancer binding proteins (C/EBPs) [1,2]. Trans-differentiation experiments may help to determine plasticity of cell differentiation and how lineage decisions are accomplished and epigenetically fixed, providing important information for future regenerative medicine.
C/EBPs are gene regulators involved in many cell differentiation and growth control processes in different cell types, including cells from the hematopoietic system [3]. C/EBPb trans-differentiates B lymphoid cells into inflammatory macrophages, activates eosinophil genes in hematopoietic progenitors, acts as a pioneering factor during dendritic cell (DC) specification and is involved in emergency granulopoiesis [4,5,6,7,8,9,10]. C/EBPb orchestrates cell type specification in combination with other transcription factors and co-factors: C/EBPb together with c-Myb activates myeloid genes in fibroblasts, together with PU.1 evokes macrophage differentiation, and together with TAL1 and FLI1 binds to and establishes early priming of hematopoietic lineage genes [11,12,13].
To answer the emerging question about the importance of C/ EBPb structure and PTMs for determination of cell fate, here we used an assay for trans-differentiation of primary B lymphoid into myeloid cells [4,10]. We identified the essential requirement of a core trans-activating region of C/EBPb that was previously shown to interact in a regulated fashion with several transcription factors and co-factors. Distinct C/EBPb PTM site or CR mutations variegate reprogramming outcomes to yield cellular phenotypes that correspond to at least four different myeloid cell types. Interestingly, the granulocytic outcome depends on the capacity of C/EBPb to recruit chromatin remodelers. Our data demonstrate that a multitude of PTMs in connection with structural plasticity are pivotal for the fine-tuning of the epigenetic C/EBPb functions to determine cell fate in the innate immune system.

Results and Discussion
The B cell to Myeloid Reprogramming Potential Resides in the C/EBPb TAD To identify C/EBPb structures involved in lympho-myeloid trans-differentiation, primary B cell progenitors were purified from wild type (WT) mouse bone marrow (. S1A) and retrovirally infected with C/EBPb constructs, including the three C/EBPb isoforms (LAP*, LAP, and LIP), as well as various CR recombinants (Fig. 1, left panel). Infected cells were cultured under conditions that support both B cell and myeloid cell development [10] and surface marker expression alterations were analyzed by flow cytometry (FACS) at 6 and 9 days post-infection (dpi) to monitor reprogramming kinetics ( Fig. 1 and S2A). Both the LAP* and LAP C/EBPb isoforms up-regulated the myeloid surface marker CD11b and down-regulated the B cell marker CD19 at 6 and 9 dpi, indicating the gradual loss of the B cell phenotype and completion of lympho-myeloid trans-differentiation. In contrast, no significant change in the B cell phenotype was observed in cells infected with the LIP C/EBPb isoform, similarly to cells infected with MSCV vector or uninfected controls ( Fig. 1 and S2A). LAP* and LAP isoforms are distinguished by CR1, which determines SWI/SNF chromatin remodeling complex recruitment and differential regulation of gene subsets [20,22,23]. Omission of CR1, as in the LAP isoform or in the CR2,3,4 mutant, significantly decreased the kinetics of both acquisition of myeloid and annulation of B cell features (Fig. 1, S2A and Table S1). Deletion of CR1,2 or CR4 strongly compromised but did not entirely abolish reprogramming, whereas removal of CR3 did not affect trans-differentiation. Deletion of CR3,4 (DCR3,4) entirely abrogated both activation of CD11b and repression of CD19, however CR3,4 in combination with the bZIP was not sufficient for reprogramming but required CR2 (CR2,3,4 in Fig. 1 and S2A). The core trans-activating region of C/EBPb CR2,3,4 was previously shown to interact in a regulated fashion with several transcription factors and co-factors, including CBP/p300, CARM1/PRMT4, G9a, TBP/TFIIB, Mediator, and several other chromatin regulatory complex components [20,21,22,24,25,26,27,28]. The LIP isoform, which lacks transactivation potential and acts as a dominant negative inhibitor, not only failed to induce myeloid conversion but also failed to downregulate B cell marker expression. Thus, activation of the myeloid program and shutting down the B cell program both reside in the C/EBPb TAD. As suppression of B cell fate involves removal of Pax5 [10], one may therefore infer that inhibition of Pax5 occurs through C/EBP mediated activation of a Pax5 inhibitor, corepressor, inhibitory RNA, or proteolysis. In many cell types C/EBPb is auto-repressed and becomes activated by receptor tyrosine kinase ras/MAPK signaling, resulting in acquisition of several C/EBPb PTMs and alterations of protein interactions [14,16,21,25,29]. In fibroblasts and erythroblastoid cells deletion of the repressive RD (DCR5,6,7) enhanced myeloid gene activation by C/EBPb, whereas removal of CR6 (DCR6) represented a dominant-negative mutant [14]. Surprisingly, both RD mutants DCR5,6,7 and DCR6 displayed trans-differentiation potential similar to LAP*, suggesting that regulation of C/EBPb in B cells may differ from other cell types. The kinetics of myeloid trans-differentiation by a leucine-zipper exchange mutant (CREB LZ) was found to be similar to WT, suggesting that i) C/EBPb homodimers are able to reprogram B cells, ii) the major trans-differentiation function of C/EBPb resides in the TAD, and iii) both the bZip and the RD structures play minor roles in lineage conversion. Notably, the reprogrammed myeloid cells showed immunoglobulin gene rearrangement, confirming their B cell origin (Fig. S1B).
To exclude auto-regulatory activation of endogenous C/EBPb during lineage conversion C/EBPb deficient B cell progenitors were tested. No differences between C/EBPb isoform or mutant trans-differentiation capacity were observed between primary WT and C/EBPb 2/2 B cell progenitors ( Fig. S2B compared to Fig. 1). Likewise, no difference in the reprogramming capacity of C/EBPa p42 was detected when WT and C/EBPb deficient B cells were compared (Fig. S2C). Furthermore, the truncated C/EBPa p30 isoform, which lacks the C/EBPa TAD (equivalent to C/EBPb CR2,3,4 TAD) failed to reprogram WT B cells, suggesting that major reprogramming functions of both, C/EBPa and C/EBPb, reside within their TADs. Therefore, C/EBPa-and C/EBPbmediated reprogramming are direct effects of the ectopically expressed transcription factors.

Differential Regulation of Key Myeloid Genes by C/EBPb WT and Mutants
To further analyze how the C/EBPb structure contributes to myeloid gene expression, several pro-inflammatory M1, antiinflammatory M2 genes, and key regulators of macrophage differentiation were examined by NanoString technology. RNA expression analyses of C/EBPb 2/2 B cell progenitors reprogrammed by WT and mutant C/EBPb showed that many M1 genes and M2 genes became up-regulated during trans-differentiation ( Fig. 2). Hierarchical gene clustering indicated no prevalence in M1 or M2 gene expression in reprogrammed cells and an overlap but also differences between C/EBPb and C/ EBPa activated genes [4]. The C/EBPb isoform LAP* and the deletion mutants DCR3 and DCR6 activated the majority of analyzed genes. Other constructs, including LAP, DCR1,2 and DCR4, showed lower or lacked trans-activation potential for several M1 and M2 genes. Both, the LAP C/EBPb isoform and the DCR1,2 mutant failed to up-regulate several macrophage polarization genes, including Mmp12, Pparg, and Chi3l3, suggesting that SWI/SNF recruitment through CR1 is a prerequisite for their activation [20,22]. Several other genes (Cxcl10, Arg1, Maf) were upregulated by LAP but not by DCR1,2, suggesting that these genes require CR2 functions that are distinct from SWI/SNF recruitment. Finally, some genes (Il1b, Cxcl10, Ccl2, Arg1, Il4ra, Maf) were more strongly activated by LAP than LAP*, in agreement with isoform-specific gene regulatory functions [30]. On the other hand, LAP* and several C/EBPb deletion mutants, but not LAP, activated the expression of Mafb, whereas LAP was the strongest activator of the Maf gene. In macrophage gene regulatory circuitry, the lysine-specific demethylase 6B Kdm6b (Jmjd3) is important for M2, but not for M1 polarization [31,32].
Interestingly, LAP and DCR1,2, which showed lower activation of Kdm6b expression (3-6-fold) as compared to LAP*, both failed to up-regulate Chi3l3, and DCR1,2 reprogrammed cells did also not express Arg1 (Fig. 2). Hence, many myeloid genes displayed designated C/EBPb CR-specific regulation, suggesting complex combinatorial, locus specific relevance of distinct C/EBPb CRs in gene regulation.
Based on myeloid surface marker expression and cell morphology, we conclude that structural alterations in C/EBPb pre-define the reprogramming outcomes into inflammatory and resident monocytes/macrophages, cDC-like cells, and granulocytes.
Mechanistically, differences between LAP* and LAP have previously been attributed to differentially regulated SWI/SNF recruitment. LAP*-specific CR1 functions and the activity of the TAD have been shown to be negatively regulated by CARM1/ PRMT4 and G9a methylation of R3 and K39, respectively [20,22,27]. Furthermore, CR1 was reported to control SUMOylation [30], thus integrating various signals to yield epigenetic consequences. Accordingly, we refined the trans-differentiation analysis using C/EBPb point mutants that affect the above mentioned modification sites. As shown in Figure 4, amino acid substitution of the G9a K39 methylation sites or the UBC9 binding/SUMOylation/methylation sites K156A/E158A, enhanced granulocytic trans-differentiation, similar to DCR6 (Fig. 3). The LAP* R3L mutant, which mimics the R3 methylated state, abrogated SWI/SNF recruitment, and failed to induce the neutrophil elastase gene [20], strongly decreased granulocytic trans-differentiation, whereas the LAP* R3A mutant, which abrogates methylation, maintained granulocytic trans-differentiation (Fig. 4). Therefore, decoration of C/EBPb with PTMs modifies its trans-differentiation capacity and, in agreement with other data [37], that recruitment of chromatin remodeling complexes through CR1 is required for granulocytic differentiation (Fig. 4E).
Advancing our understanding of the importance of transcription factor regulation and PTMs in lineage decisions is instrumental to elucidate normal development and aberrant epigenetic processes in connection with disease. Previous findings have suggested that chromatin regulatory factors and epigenetic state regulation are involved in hematopoietic cell decisions [37,38]. Furthermore, it has been shown that interactions between C/EBPb and the transcriptional and epigenetic machineries are controlled by C/ EBPb PTMs [15,17,18,20,21,22,25,27] but their importance for directing differential myeloid cell differentiation is quite obscure. The B cell to myeloid lineage conversion now connects C/EBPb PTMs to alternative cell fate instruction, raising the possibility that related mechanisms control regular myelopoiesis. Although we do not imply B cell to myeloid trans-differentiation as a frequent event, it recalls the evolutionary relationship between innate and acquired immunity [39,40,41,42]. Moreover, lineage switching of B cell lymphoma to acute monoblastic leukemia or transdifferentiation of follicular lymphoma to histiocytic/DC sarcomas have been reported [43,44] and bi-phenotypic lymphoma displayed functional dependency on high C/EBPb expression [45,46,47]. These data suggest a role of lympho-myeloid plasticity in malignant transformation. It is evident that more detailed mechanistic insight in spatio-temporal modifications and co-factor recruitment requires advanced tools, such as generation of knockin mouse mutants, determination of the PTM-dependent C/EBPb interactome, PTM specific antibodies and genome wide comparison of C/EBPb mutant binding. Nevertheless, the extensive decoration with PTMs in conjunction with reprogramming data provided here suggest that C/EBPb integrates extracellular signals to accomplish alternative differentiation into diverse cells of the innate immune system.

Ethics Statement
All mice were bred and maintained in accordance with guidelines from institutional Animal Care Committee under specific pathogen-free animal facilities at the MDC/Charité. Experiments were approved by the Commission for Animal Experiments at the MDC and the Berlin Office of Health (LAGeSo), Permit Number T 0339/08. For isolation of cells, mice were sacrificed by euthanasia using carbon dioxide inhalation followed by cervical dislocation. All efforts were made to minimize animal suffering.
Cytospins GFP + CD11b + and GFP + CD19 + cells were sorted by FACS 9 days after retroviral infection and cytospins were performed. Slides were fixed in 100% methanol and stained with May-Grunwald and Giemsa (Sigma).

RNA Extraction and mRNA Expression Analysis by Nanostring Technology
Total RNA was extracted from C/EBPb 2/2 B cell progenitors 6 days after infection with C/EBPb constructs and sorting of the CD11b + reprogrammed cells or from bone marrow-derived macrophages (control, 6 days in vitro cultured) using RNeasy Micro Kit (QIAGEN) according to the manufacture's recommendations. mRNA counts were determined using Nanostring technology [50] after background subtraction and normalization to three house-keeping genes (Gapdh, Tbp, Ppia). Expression below the background level was set to value ''1''. After log 2 transformation, data were subjected to hierarchical clustering using Euclidean Distance to generate a gene and sample tree (MeV software).

Statistical Analysis
In all experiments, data are presented as mean 6 SEM (standard error of the mean). Statistical analyses were done on Prism 4.0a (GraphPad Software) applying unpaired two-tailed t test for the calculation of the P-value. The statistical significance of the P-value was defined as: P.0.05 -not significant, P = 0.01-0.05 -significant (*), P = 0.001-0.01 -very significant (**), P,0.001extremely significant (***).
More Materials and Methods could be found in the Materials and Methods S1. Figure S1 FACS sorting strategy, rearrangements in IgH gene loci and C/EBPb expression in the C/EBPb reprogrammed myeloid cells (related to Figure 1). A. Bone marrow single cell suspension was prepared and cells stained, as described in Materials and Methods. Lin -B220 + IgM -CD19 +/ 2 pre-pro/pro/pre B cell progenitors were sorted for the reprogramming experiments. Lin + cells were cultured in vitro for obtaining bone marrow-derived macrophages (MPh) for negative controls for IgH rearrangement PCR. Lin -B220 + IgM + bone marrow immature B cells and spleenic B220 + B cells were sorted for positive rearrangement PCR controls. B. PCR for D-J rearrangements in IgH locus. CD11b + reprogrammed myeloid cells and CD19 + MSCV-, LIP-and DCR3,4-infected B cells were sorted and PCR for D-J rearrangements in the IgH locus was performed. Controls: WT bone marrow-derived macrophages (MPh) and spleenic B cells. Data shown are representative from multiple experiments. C. Protein expression of the C/EBPb WT and deletion constructs in the virus-packaging cell line PlatE. The size of the proteins is according to the size of the deletions. D. Intracellular C/EBPb protein staining in the reprogrammed cells. The relative C/EBPb expression in the virus-infected cells was calculated as described in Materials and Methods S1. The endogenous C/EBPb expression level in WT bone marrowderived macrophages (MPh) was also assessed. The relative C/ EBPb expression values varied between the different experiments, however the tendencies were highly reproducible. (TIF) Figure S2 Reprogramming of WT and C/EBPb 2/2 B cell progenitors by C/EBPa and C/EBPb (related to Figure 1). A. Representative FACS profiles of the C/EBPb infected WT B cell progenitors at 6 and 9 dpi. FACS plots represent GFP + gated cell population, B cells -control uninfected GFP -B cell progenitors. Similar outcomes were obtained from at least two repeat experiments. B. Percentage of C/EBPb 2/2 B cell progenitors infected with C/EBPb WT and mutants expressing the B cell marker CD19 or the myeloid marker CD11b at 6 dpi. Intermediates (CD19 + CD11 + cells) are also included. Graphs represent GFP + gated cell population, B cells -control uninfected GFP -B cell progenitors. Values represent mean 6 SEM from two and more repeat experiments. C. Percentage of WT and C/ EBPb 2/2 B cell progenitors infected with WT C/EBPa p42 and p30 expressing the B cell marker CD19 or the myeloid marker CD11b at 6 dpi. Intermediates (CD19 + CD11 + cells) are also included. Graphs represent GFP + gated cell population. Values for C/EBPb 2/2 B cell progenitors represent mean 6 SEM from three repeat experiments. (TIF) Figure S3 Heterogeneity among reprogrammed myeloid cells and lack of differential apoptosis between the subpopulations of reprogrammed cells (related to Figure 3). A. Phagocytosis assay was performed after 10 days in vitro reprogramming. Red line represents cells incubated with fluorescent latex beads and the black line -the auto-fluorescence of the untreated samples. For MSCV-infected cells histograms represent GFP + CD19 + population, whereas C/EBPb-infected reprogrammed cells were gated on GFP + CD11b + cells. As positive controls for phagocytic capacity, bone marrow-derived macrophages (MPh) were used. Similar outcomes were obtained in two or more repeat experiments. B. Apoptosis assay based on AnnexinV staining and evaluated by FACS. Dead cells were excluded by DAPI staining and the apoptosis assessment was done after gating on the different GFP + cell populations (CD19 + , CD11b + Gr-1and CD11b + Gr-1 + ). na -no available cells with these surface characteristics. The graph represents data from four independent experiments. C. Expression of Ly-6C and M-CSFR myeloid cell markers on the reprogrammed cells at 6 and 9 dpi. FACS plots represent GFP + CD11b + cell population. For MSCVinfected cells FACS plots represent GFP + CD19 + cells. The myeloid cell marker staining was repeated in at least two independent experiments and similar results were obtained. (TIF) Table S1 C/EBPb WT and mutant constructs display different B-to-myeloid cell reprogramming kinetics (related to Figure 1).

(DOC)
Materials and Methods S1 Supplementary Materials and Methods (DOC)