SOX7-enforced expression promotes the expansion of adult blood progenitors and blocks B-cell development

During embryogenesis, the three SOXF transcription factors, SOX7, SOX17 and SOX18, regulate the specification of the cardiovascular system and are also involved in the development of haematopoiesis. The ectopic expression of SOX17 in both embryonic and adult blood cells enhances self-renewal. Likewise, the enforced expression of SOX7 during embryonic development promotes the proliferation of early blood progenitors and blocks lineage commitment. However, whether SOX7 expression can also affect the self-renewal of adult blood progenitors has never been explored. In this study, we demonstrate using an inducible transgenic mouse model that the enforced expression of Sox7 ex vivo in bone marrow/stroma cell co-culture promotes the proliferation of blood progenitors which retain multi-lineage short-term engrafting capacity. Furthermore, SOX7 expression induces a profound block in the generation of B lymphocytes. Correspondingly, the ectopic expression of SOX7 in vivo results in dramatic alterations of the haematopoietic system, inducing the proliferation of blood progenitors in the bone marrow while blocking B lymphopoiesis. In addition, SOX7 expression induces extra-medullary haematopoiesis in the spleen and liver. Together, these data demonstrate that the uncontrolled expression of the transcription factor SOX7 in adult haematopoietic cells has dramatic consequences on blood homeostasis.


Introduction
The SOX family of transcription factors is highly conserved throughout evolution and in mouse comprises 20 members divided into subgroups based on sequence similarities [1]. During embryogenesis, SOX factors are essential for the regulation of many developmental processes, often implicating several SOX factors in one developmental path, such as, for example, SRY and SOX9 in sex determination or SOX1, 2 and 3 in neural tube formation [2]. In addition to their critical roles during embryogenesis, SOX factors are also often implicated in the maintenance and identity of stem cell populations [3]. For instance, SOX2 is critical for the expansion and maintenance of neuronal stem progenitor cells [4], while SOX9 determines mammary stem cell state [5].
SOX7 and its two close homologues, SOX17 and SOX18, belong to the SOXF subgroup, which are known to play important roles in cardiovascular development [6,7]. Compensatory mechanisms and some level of redundancy have been shown to occur between the three SOXF members during cardiovascular development and maintenance [8,9]. The SOXF factors have also been implicated in the development of the haematopoietic system, which is closely linked to endothelium development during embryogenesis [10]. SOX17 drives the expansion of fetal haematopoietic stem cells [11] while at the earliest stages of blood specification all three SOXF factors are expressed in haemogenic endothelium [12][13][14][15], a specialized subset of endothelium cells giving rise to blood progenitors upon endothelial to haematopoietic transition [16]. At E10. 5  License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original author and source are credited. [8,17]. We and others have demonstrated that the ectopic expression of SOXF factors in haematopoietic cells leads to dramatic alterations in the balance between proliferation and differentiation. The enforced expression of SOX17 in adult blood cells was shown to enhance self-renewal capacity and to confer fetal-like haematopoietic properties; long-term ectopic expression of SOX17, however, led to leukaemogenesis [18]. Similarly, in E10.5 aorta-derived haematopoietic clusters, the enforced expression of SOXF factors, and SOX17 in particular, resulted in the maintenance and expansion of blood progenitors and HSCs [17]. A role for SOX17 in promoting the myeloid specification of E10.5 blood progenitors was also suggested in follow-up studies [19]. Our group showed that Sox7 expression was upregulated in mesoderm precursors at the onset of blood specification and downregulated as differentiation progresses to committed blood lineages [13,14]. The enforced expression of SOX7 in E7.5-derived embryo cells or in in vitro differentiated embryonic stem cells was shown to promote the self-renewal of early blood progenitors harbouring endothelial-like features and to block further differentiation to committed lineages [13,14]. The enforced expression of SOX18 in these early embryonic populations led to a similar phenotype [15,17]. Given the potential of SOXF factors in maintaining the selfrenewal properties of blood progenitors, we hypothesized that the ectopic expression of SOX7 may also confer a proliferative or survival advantage to adult haematopoietic cells. Using a transgenic inducible mouse model, we explore here the consequences of SOX7 ectopic expression on adult haematopoiesis both ex vivo and in vivo. We show that SOX7-enforced expression in adult bone marrow cells leads to the dramatic expansion of blood progenitors and specifically impairs B lymphopoiesis.

Transplantation
Bone marrow cells were transplanted i.v. into sub-lethally irradiated (125 cGy) Nod Scid IL2rg-deficient mice (NSG, Charles River). After four weeks, mice were fed or not with doxycycline diet (Harlan). Mouse health was assessed by blood analysis, weight and general health monitoring.

Flow cytometry
Single-cell suspensions from adult bone marrow, spleen, liver and blood or OP9 co-culture were stained and analysed with FACSCalibur or LSRII and sorted with Influx or Aria flow cytometers (all BD Biosciences). Staining for sorting was performed in IMDM with 10% FCS, whereas cell surface staining for analysis was performed in PBS with 10% FCS. Cells were incubated with primary antibodies for 30 min at 48C, then washed in PBS with 10% FCS and stained with secondary antibodies for 30 min at 48C. After the secondary staining, cells were washed in PBS with 10% FCS and re-suspended in PBS with 10% FCS for cell surface staining or IMDM with 10% FCS for sorting. All antibodies and streptavidin used for staining were purchased from eBioscience. Details are available upon request. Data were analysed using the FLOWJO software (TreeStar).

Immunohistochemistry
Reticulin staining was performed on paraffin sections using the Gordon and Sweet's stain. Sections were incubated for 5 min in a potassium permanganate (3% sulfuric acid) solution followed by washes in tap water. Next, 1.5% oxalic acid was applied until clear. After washes in tap water, sections were incubated with 2% ferric ammonium sulfate for 15 min. Multiple washes in distilled water were done before applying the ammoniacal silver solution (10% silver nitrate, ammonium hydroxide and 3% sodium hydroxide solution). Sections were washed in distilled water and fixed in 10% formalin for 5 min. After washes in tap water, slides were stained with 5% sodium thiosulfate for 2 min to remove unreduced silver, rinsed in tap water and incubated in 0.2% gold chloride for 3 min and then after washes in tap water, 0.1% neutral red was applied for 10 s. Finally, slides were dehydrated, cleared and mounted on clear glass covered by a coverslip.
For CD45R (B220) staining, sections were incubated with Dako target retrieval solution (S2369) at 1258C for 1 min, and then cooled to 908C for 10 s. Slides were brought to room temperature in running water. Endogenous peroxidase was blocked in 0.3% TBS for 10 min. After washing thoroughly, blocking solution (10% rabbit serum) was applied for 30 min. Anti-CD45R antibody or Rat IgG2b control (BD Biosciences) was applied overnight at 48C. After washing twice in TBS for 5 min, a rabbit anti-rat biotinylated antibody (Dako) was applied for 30 min at room temperature. Slides were washed twice in TBS and stained with Strep-AB-Complex-HRP (Dako). Sections were rinsed in running water and counterstained with 1X Gills Haematoxylin for 1 min, then washed in water for 1 min, dipped in alkaline water for 30 s and washed again in water for 1 min. Finally, slides were dehydrated, cleared and mounted on clear glass covered by a coverslip. Pictures were taken using a slide scanner machine (Leica SCN400).

May-Grünwald-Giemsa staining
Cells were cyto-spun onto slides, fixed with methanol for 10 min and stained with five volumes of o-dianidisine, one rsob.royalsocietypublishing.org Open Biol. 6: 160070 volume of 3% hydrogen peroxide and one volume of sodium nitroferricyanide mix for 10 min. After rinsing with water, slides were stained with May-Grü nwald solution (Merck) for 3 min, rinsed with water and stained with a 1 in 20 dilution of Giemsa's stain solution (VWR) for 20 min. Slides were rinsed with water, left to dry and mounted with a coverslip. Haematoxylin and eosin staining was performed as previously described [20].

Transcriptomic analysis
Total RNA was extracted using the RNeasy Mini kit (Qiagen) and tested for RNA quality on an Agilent Bioanalyzer 2100 (Agilent Technologies). cDNA libraries were hybridized to Affymetrix mouse exon 1.0 ST array. Background adjustment and data normalization were performed with the robust multi-array average algorithm. Probesets were filtered by Affymetrix detection above background (DABG) scores. A probeset was kept if its DABG p-value was less than 0.01 in at least two samples. After these filtering steps, 189 956 probesets (representing 22 140 unique genes) from mouse array remained for differential gene expression analysis. Annotations and cross-mappings between probesets, exons and genes were obtained from Ensembl mouse genome using the R/BIOCONDUCTOR package Annmap (BIOCONDUCTOR 2.14).

Statistical analyses
Data were analysed using Student's t-test. Significant differences are indicates with *p , 0.05 and **p , 0.01.

Sox7-enforced expression induces the proliferation of bone marrow progenitors and impairs B-cell differentiation
To determine whether SOX7-enforced expression in adult blood cells confers a proliferative advantage as observed during embryonic development, we investigated the consequences of Sox7-enforced expression on haematopoiesis. The impact of Sox7 expression was first analysed using a murine bone marrow and OP9 stroma cell co-culture system allowing the close monitoring of haematopoietic differentiation to myeloid and lymphoid lineages [21]. To enforce Sox7 expression, we used a previously described doxycycline-inducible transgenic mouse model for SOX7 and GFP co-expression (thereafter referred to as iSox7) [14]. Bone marrow cells were extracted from iSox7 mice and serially passaged on irradiated OP9 stroma cells with or without doxycycline (figure 1a). Without doxycycline, the frequency of MAC1 þ myeloid cells progressively decreased while the frequency of B220 þ CD19 þ B cells gradually increased, revealing the dynamic of this co-culture system, which gave rise to a wave of myelopoiesis followed by a wave of B lymphopoiesis (figure 1b). When Sox7 expression was induced during the co-culture, the frequency of MAC1 þ myeloid cells remained high upon successive passages with the progressive appearance of a MAC1 low population expressing SOX7::GFP at high level (figure 1b). By contrast, the generation of B220 þ CD19 þ B cells was severely impaired and by nine passages no cells expressing B220 or CD19 were detected in the doxycycline-induced cultures. The total number of cells produced upon successive passages slowly decreased without doxycycline while a dramatic increase was observed upon Sox7-enforced expression (figure 1c). Most cells in the doxycycline-treated culture expressed CD45 and GFP (electronic supplementary material, figure S1a), and displayed increasing levels of Sox7 expression upon serial passaging (electronic supplementary material, figure S1b). In addition, the Sox7-induced cultures contained more cells with an immature blast-like morphology and fewer B lymphocytes when compared with the control cultures (electronic supplementary material, figure S1c). We next examined whether Sox7-enforced expression was essential for the maintenance of the observed phenotype. After several passages, doxycycline was either removed or maintained in the culture and cell proliferation was compared between the two conditions. The expression of Sox7 appeared critical for cell proliferation as the number of cells produced was dramatically and immediately reduced upon doxycycline removal (figure 2a). Cell cycle analysis revealed a massive decrease in the frequency of cells in the S-phase of the cell cycle upon doxycycline withdrawal (figure 2b). We next investigated the immuno-phenotype of SOX7::GFP-expressing cells that accumulated over time in these cultures. Flow cytometric analysis revealed the presence of two subpopulations based on GFP expression level (figure 2c). A small SOX7::GFP low population expressed B220 and CD19, and a predominant SOX7::GFP high population did not express these B-cell markers but expressed high levels of cKIT. To further characterize the differentiation stage of the SOX7::GFP low B220 þ population, a detailed analysis of early B-cell subsets was performed based on the expression of CD24 and BP-1, which divides the B220 þ CD43 þ population into progenitor subsets [22]. All the subsets of developing B cells were observed in the control culture, while BP-1 and CD24 expression were not detected in the SOX7::GFP low B220 þ CD43 þ population, suggesting that these cells are blocked at a pre-pro B stage or earlier (figure 2d).
Taken together, these data revealed that the enforced expression of Sox7 ex vivo in adult bone marrow cells dramatically increases proliferation but impairs the differentiation of B lymphocytes.

The cKIT þ SOX7::GFP high subpopulation contains multi-lineage progenitors
Over time, the frequency of the cKIT þ SOX7::GFP high subpopulation increased, suggesting that this subpopulation was specifically expanded by Sox7 ectopic expression (electronic supplementary material, figure S2a). To define to which extent SOX7 expression altered this population, we investigated the immuno-phenotype and biological potential of these cKIT þ SOX7::GFP high cells. The expression of VE-cadherin, a known transcriptional target of SOX7 [13], was observed on a large fraction of SOX7::GFP high cells (figure 3a), further confirming SOX7 transcriptional activity in this population. The analysis of haematopoietic stem cell marker expression revealed a large increase in the frequency of the lineage 2 SCA1 þ cKIT þ fraction (figure 3b). To define the biological potential of this population, cKIT þ SOX7::GFP high cells were sorted and plated without doxycycline in clonogenic assay to test for myeloid and erythroid potential, and on OP9 co-culture to test for lymphoid potential (figure 3c). The cKIT þ SOX7::GFP high cells gave rise to erythroid as well as all types of myeloid colonies  rsob.royalsocietypublishing.org Open Biol. 6: 160070 including megakaryocyte, granulocyte, mast and macrophage colonies (figure 3d,e); gene expression analysis for Gfi1, C/ebp1 and Irf8 further confirmed myeloid identity (figure 3f ). When plated on OP9 stroma cells, cKIT þ SOX7::GF-P high cells generated B220 þ CD19 þ B lymphocytes (figure 3g), with the expression of E2a and Pax5 further confirming B-cell identity (figure 3h). To test for in vivo engrafting potential, sorted cKIT þ SOX7::GFP high cells were injected into immuno-compromised NSG recipient mice. After four weeks, peripheral blood cell analysis revealed the presence of CD45.2 þ donor cells, indicating the success of the engraftment (electronic supplementary material, figure S2b-c). However, by 16 weeks, the frequency of CD45.2 þ cells in the peripheral blood had dropped substantially, suggesting that only shortterm engraftment capacity was maintained in the ex vivo derived cKIT þ SOX7::GFP high cell population. Altogether, these results revealed that the enforced expression of Sox7 in bone marrow cells leads to the ex vivo maintenance and expansion of a progenitor population able to give rise to myeloid, erythroid and lymphoid lineages and to provide short-term in vivo engraftment.

Molecular characterization of SOX7-mediated changes
To define at the molecular level the consequences of SOX7-mediated effect on proliferation and differentiation, transcriptome profiling analysis was performed on the two subpopulations expressing low and high levels of SOX7::GFP. Comparative analysis revealed that 1164 genes were significantly and differentially expressed, with 642 genes expressed at higher level and 522 genes expressed at lower level in SOX7::GFP high cells relative to SOX7::GFP low cells ( figure 4a,b). Gene set enrichment analysis (GSEA) performed on all genes upregulated in the SOX7::GFP high fraction demonstrated a significant enrichment for haematopoietic stem and progenitor cell genes [23,24]      rsob.royalsocietypublishing.org Open Biol. 6: 160070 (figure 5c). Immunohistochemistry analysis confirmed the dramatic decrease in B220-expressing cells in the bone marrow upon Sox7-enforced expression (figure 5d), while collagen staining revealed important fibrosis in the bone marrow (electronic supplementary material, figure S4b). In contrast to B cells, the frequencies of CD4 þ and CD8 þ T cells (figure 6a,b) as well as TER119 þ erythrocytes (figure 6c,d) were not significantly affected by Sox7-enforced expression in the bone marrow of animals fed with doxycycline when compared with control animals. However, we did observe a sharp decrease in CD41 þ megakaryocytes in the bone marrow of the treated animal (figure 6e) that was reflected by a dramatic decrease in platelet count in the blood of these animals ( figure 6f ). Finally, while neutrophil count was not significantly affected by Sox7-enforced expression, eosinophil count was decreased and basophil count was increased in the blood of the treated animals (figure 6g). It is not clear, however, if these changes were a direct consequence of SOX7-enforced expression or if they represented secondary effects due to fibrosis in the bone marrow. We next investigated whether Sox7-enforced expression also affected haematopoietic progenitors normally residing in the bone marrow. Upon clonogenic replating, the bone marrow cells of doxycycline-fed animals gave rise to two times more colonies than the control bone marrow cells ( figure 7a,b). Altogether, these findings demonstrate that in vivo the ectopic expression of SOX7 leads to B lymphopoiesis impairment and expansion of haematopoietic progenitors in the bone marrow associated with fibrosis.
The induced expression of Sox7 also led to splenomegaly and partial hepatomegaly (figure 7c,d), associated with a dramatic decrease in B220-expressing cells (figure 7e) and fibrosis (electronic supplementary material, figure S4b). The two main compartments in the spleen, white pulp and red pulp, were profoundly disorganized (figure 7f ). The red pulp is involved in filtering the blood and recycling iron from old erythrocytes, while the white pulp represents the lymphoid regions within the spleen. Upon SOX7-enforced expression, the lymphoid follicles were disrupted compared with the control samples, where a clear distinction between the follicles and red pulp was visible. The red pulp did not seem to be overtly affected by SOX7 expression. Disruption of the spleen architecture is probably due to the loss of B cells in the Sox7-expressing samples.
We next investigated the presence of extra-medullary haematopoiesis in these two organs. Clonogenic assays demonstrated a dramatic increase in the frequency of haematopoietic progenitors in both spleen and liver ( figure 7a,b). To investigate the presence of multi-potent engrafting progenitors in extra-medullary haematopoietic sites, recipient mice were engrafted with spleen cells derived from doxycycline-fed animals. Analysis at 4 and 16 weeks post-engraftment revealed the presence of CD45.2 þ donor-derived cells in the peripheral blood of recipient mice (figure 7g). At 20 weeks Together, these data reveal that Sox7-enforced expression in vivo impairs B lymphopoiesis and expands the haematopoietic progenitor compartment, resulting in splenomegaly and extra-medullary haematopoiesis.

Discussion
Using an inducible transgenic mouse model system, we demonstrate here that the ectopic expression of SOX7 in adult bone marrow cells promotes the self-renewal of immature blood progenitors, leads to a profound block in B-cell development and globally perturbs the homeostasis of the haematopoietic system with extra-medullary haematopoiesis and extensive fibrosis in spleen and bone marrow.
The findings presented here clearly demonstrate that the ectopic expression of SOX7 has dramatic consequences for the balance between self-renewal and differentiation of blood progenitors. This is very similar to what has been observed for SOX17, for which ectopic expression was shown to alter both embryonic [17] and adult haematopoiesis [18]. Likewise, SOX18 ectopic expression was shown to promote the selfrenewal of embryonic blood progenitors [15,17]. It is currently unknown whether SOX18 mis-expression in adult blood cells   rsob.royalsocietypublishing.org Open Biol. 6: 160070 might also affect the balance between self-renewal and lineage commitment. The three SOXF factors have been shown to compensate for each other on mixed genetic background [8,9], however, when backcrossed to homogeneous background, for reasons that are still unclear, these compensatory mechanisms are lost. Additional studies have revealed a specific function for SOX18 in lymphangiogenesis [27] and various functions for SOX17 in endoderm specification [28], HSC proliferation [18] or arterial-venous specification [29] according to the cellular compartment in which this factor was deleted. To date, the specific role of SOX7 remains unknown, but SOX7 was shown to be expressed in primitive endoderm [30] and in vasculature [8]. Its deletion is linked to diaphragmatic hernia, cardiac oedema and apparent lack of yolk sac rsob.royalsocietypublishing.org Open Biol. 6: 160070 vasculature [31], but further characterization of SOX7-deficient embryos will be required to fully understand the main function of SOX7 during embryonic development. When ectopically expressed in blood progenitors, all SOXF factors are able to alter the balance between proliferation and differentiation. Given their known degree of redundancy, it is very likely that all three SOXF factors activate a similar transcriptional programme that promotes proliferation at the expense of lineage differentiation. Based on the known functions of the SOXF factors, two possible explanations can be suggested for these observations: (i) the transcriptional programme activated by all SOXF in blood cells is similar to the one activated by SOX17 in fetal HSCs, which promotes the active proliferation of these cells [11]; SOX7 and SOX18 are able to activate this programme when ectopically expressed in a blood-specific cell context. (ii) All three SOXF factors are expressed in haemogenic endothelium [12,13] and might in this cellular context induce a transcriptional programme promoting proliferation which can be induced in a blood-specific cell context upon ectopic expression. The transcriptional targets specifically activated by the SOXF factors to regulate self-renewal are still currently unknown; this will be an important avenue of investigation for future studies.
Since the ectopic expression of all three SOXF factors induced such a dramatic shift from differentiation to selfrenewal, it is surprising that the expression of these factors is not found misregulated in leukaemias. While the long-term ectopic expression of SOX17 ultimately leads to leukaemogenesis in mouse models [18], this factor has never been implicated in human haematological malignancies. Similarly, mutations in SOX18 have been linked to hypotrichosis-lymphedema-telangiectasia [32] but to date no change in SOX18 expression has been associated with blood disorders. By contrast, a role for SOX7 was recently described in acute myeloid leukaemia (AML), where this transcription factor was shown to block proliferation [33], acting in that context as a tumour suppressor. This role for SOX7 in AML is in striking contrast to the data presented here and to the block induced by SOX7enforced expression in B lymphopoiesis. This difference raises the possibility that in normal haematopoietic development SOX7 might be implicated in the commitment or proliferation of progenitors to B-lymphocytes versus myeloid lineages. During embryonic development, progenitor populations restricted to B-cell/macrophages [34] or B-cell/T-cell/macrophages [35] have been described. Given the known expression of SOX7 at the onset of blood development in the embryo [13], it is possible that SOX7 is part of the transcriptional machinery promoting the proliferation and/or commitment of these progenitors towards B cells or macrophages. The tumour suppressive role of SOX7 in AML was directly linked to the inactivation of the Wnt/b catenin pathway [33]. However, b catenin has been shown to be dispensable for B-cell development [36], suggesting that the mode of action of SOX7 in myeloid and lymphoid lineages is likely to be quite different. It will be important in future studies to elucidate how SOX7 might exert opposite effects in myeloid and lymphoid lineages. Data accessibility. The microarray data reported in this study are accessible via the NCBI Gene Expression Omnibus webpage and the accession number is GEO: GSE65552.