The Gene Encoding the Hematopoietic Stem Cell Regulator CCN3/NOV Is under Direct Cytokine Control through the Transcription Factors STAT5A/B*

Cytokines control the biology of hematopoietic stem cells (HSCs) and progenitor cells in part through the transcription factors STAT5A/B. To investigate the target genes of STAT5A/B activated by cytokines in HSCs and progenitors, we performed microarray analyses using Lineage− Sca-1+ c-Kit+ (KSL) cells in the presence and absence of STAT5A/B. Stimulation with a mixture containing IL-3, IL-6, stem cell factor, thrombopoietin, and Flt3 ligand induced Ccn3/Nov mRNA over 100-fold in WT (control) but not Stat5a/b-null KSL cells. CCN3/NOV is a positive regulator of human HSC self-renewal and development of committed blood cells. Without stimulation, the Ccn3/Nov signal level was low in control KSL cells similar to Stat5a/b-null KSL cells. To determine which cytokine activates the Ccn3/Nov gene, we analyzed Lineage− c-Kit+ (KL) and 32D cells using quantitative PCR and ChIP assays. Although stimulation with a mixture lacking IL-3 prevented the induction of Ccn3/Nov in control KL cells, IL-3 alone could induce Ccn3/Nov mRNA in control KL and 32D cells. ChIP assays using 32D cells revealed IL-3-induced binding of STAT5A/B to a γ-interferon-activated sequences site in the Ccn3/Nov gene promoter. This is the first report that Ccn3/Nov is directly induced by cytokines through STAT5A/B.

STAT5A and STAT5B (referred to as STAT5A/B) are two closely related members of the family of signal transducers and activators of transcription (STAT) proteins. STATs are activated upon binding of a ligand to the receptor by phosphorylation of a critical tyrosine residue through Janus kinases (JAK). Activated STATs bind to specific DNA sequences (TTC-nnnGAA), named ␥-interferon-activated sequences (GAS 2 sites), and initiate transcription of target genes (1). Genetic evidence links STAT5A/B to the normal physiology and pathophysiology of a wide range of cell types (1), including murine and human HSCs (2)(3)(4)(5). Notably, cell-specific ablation of the Stat5a/b locus in mice resulted in a complete loss of T and B cells (6,7), defective granulopoiesis (8), and erythropoiesis (9). In HSCs, STAT5A is activated by thrombopoietin, IL-3, and granulocyte-macrophage colony-stimulating factor (10). Activation of STAT5A/B is also required for self-renewal and quiescence of HSCs (4,11), but little is known about the underlying molecular mechanisms and the target genes downstream of STAT5A/B. To investigate STAT5A/B target genes upon stimulation with cytokines in HSCs and progenitor cells, we sorted Lineage Ϫ Sca-1 ϩ c-Kit ϩ (KSL) cells from wild type (control) and Stat5a/b-null mice and performed microarray analyses.
CCN3/Nov, originally characterized as a site of proviral integration in an avian model of nephroblastoma, belongs to the CCN family of multimodular matricellular signaling proteins (12,13). The Ccn3/Nov gene encodes a secreted protein that associates with the extracellular matrix and is thought to be involved in the regulation of proliferation, differentiation, survival, adhesion, and migration through integrin-mediated signaling pathways (14). Ccn3/Nov has been reported to act as a tumor suppressor gene in solid tumors including hepatocellular carcinomas, Wilm tumors, Ewing sarcomas, gliomas, and adrenocortical carcinomas (15)(16)(17). On the other hand, CCN3/Nov has been suggested as a candidate marker for HSCs, and it has been shown to be essential for HSC and hematopoietic progenitor cell functional integrity (18,19). It has been reported that CCN3/Nov is essential for self-renewal of HSCs and progenitors and development of committed blood cells after bone marrow transplantation using human cord blood stem cells (19). Therefore, CCN3/Nov and STAT5A/B can be considered positive regulators in normal hematopoietic cells. Here we show that Ccn3/Nov is a cytokine-inducible gene and a direct target of STAT5A/B.
32D Cells-The IL-3-dependent mouse 32D cell line (CRL-11346, ATCC) was used in this study. Cells were grown according to the manufacturer's recommendations in the following complete culture medium: RPMI 1640 with 2 mM L-glutamine adjusted to contain 1.5 g/liter sodium bicarbonate, 4.5 g/liter glucose, 10 mM HEPES, and 1 mM sodium pyruvate supplemented with 10% heat-inactivated fetal bovine serum and 10% mouse IL-3 culture supplement (354040, BD Biosciences). Cells were starved in complete culture medium without IL-3 for 6 h and stimulated with IL-3 (100 ng/ml) for the indicated times.
Affymetrix Microarray-KSL cells derived from control and Stat5a/b-null fetal livers were stimulated with a cytokine mixture consisting of 100 ng/ml IL-3, IL-6, stem cell factor, thrombopoietin, and Flt3 ligand for 16 h or left untreated. Total RNA from four different samples (Stat5a/ b-null-unstimulated, Stat5a/b-null-stimulated, control-unstimulated, and control-stimulated) was prepared with RNeasy Plus mini kits (Qiagen) and processed/analyzed by the NIDDK Core Facility at the National Institutes of Health. Gene Chip expression 3Ј amplification reagents in a two-cycle cDNA synthesis kit (Affymetrix) were used. Affymetrix gene expression analysis arrays for the Mouse430_2 were used. The microarray signals were referred to by the Affymetrix robust multichip average algorithm. Up-and down-regulated genes were selected based on p values (Ͻ0.05), and -fold changes (Ͼ2.0 or ϽϪ2.0) were assessed by analysis of variance with the Partek Pro software (Partek). Potential STAT5A/B-regulated genes were identified by comparison of the different experimental groups (see Fig. 1, A-C, and Table 1). The microarray analysis was performed with three independent biologic sample sets. The microarray data have been uploaded to the Gene Expression Omnibus (GEO) under accession number (GSE20684).
qPCR-Sorted KL cells derived from control and Stat5a/bnull mice were stimulated with a complete cytokine mixture or an IL-3-depleted cytokine mixture for 16 h (see Fig. 2B). Control KL cells were stimulated with IL-3 for 3 or 16 h (see Fig. 2C). LT-HSCs and ST-HSCs derived from fetal liver cells or bone marrow cells were stimulated with IL-3 for 16 h (see Fig. 2D). Each cytokine was used at a 100 ng/ml concentration. RNA was isolated and subjected to qPCR. Reverse transcription of total RNA was performed using the SuperScript III protocol (Invitrogen), and TaqMan PCR reactions were done consecutively by using a 7900 HT fast real-time pcr system (Applied Biosystems). TaqMan probes for Ccn3/Nov (Mm00456855-m1) and ␤-actin (4352341E) were used (Applied Biosystems) for qPCR.
Statistical Analyses-Statistical significance was determined by Student's t test (two-tailed, unpaired/unequal variances). STAT5A/B Regulate Ccn3/Nov upon Cytokine Stimulation of KSL Cells-Total RNA was isolated after stimulation with a cytokine mixture for 16 h to investigate STAT5A/B target genes by microarray analyses. First we compared stimulated control samples with unstimulated control samples (WTϩ versus WTϪ) to determine the level of stimulation in KSL cells (Fig. 1A). 2171 genes were significantly induced in control KSL cells. Then we identified the genes that were significantly induced by stimulation in control but not in Stat5a/b-null cells (3163 down-regulated genes and 2423 up-regulated genes) (Fig.  1A). The expression of well known STAT5A/B target genes, such as Socs2, pim2, and IL-2R␣, was highly induced in control cells but not Stat5a/b-null cells by stimulation (in the 560-gene group shown in Fig. 1A). Genes significantly induced through STAT5A/B were identified as down-regulated in the comparison of (KOϪ versus WTϪ) and (KOϩ versus WTϩ) (Fig. 1B). 369 genes were induced through STAT5A/B upon stimulation with a cytokine mixture, only 58 genes were induced through STAT5A/B without stimulation, and 25 genes were induced through STAT5A/B unrelated to the stimulation. To select STAT5A/B target genes induced by cytokine stimulation, common genes of 560 genes and 344 genes were selected (100 genes) (Fig. 1C, Table 1). These 100 genes met the criteria of "significantly induced by stimulation in control as compared with unstimulated control KSL cells, but unchanged in Stat5a/b-null KSL cells" and "significantly induced in control as compared with Stat5a/b-null KSL cells upon stimulation." The most strongly induced gene among those 100 genes was Ccn3/Nov,

TABLE 1 mRNA levels of STAT5A/B target genes induced in KSL cells upon stimulation with cytokine mixture determined by microarray analysis
-Fold changes and p values were analyzed using three independent experimental datasets. ϩ, stimulated sample; Ϫ, unstimulated sample; WTϩ vs. WTϪ means the comparison of stimulated control samples with unstimulated control samples. The sample on the right-hand side of "vs." was used as the standard in each comparison. Genes were selected according to Fig. 1 as STAT5A

STAT5A/B Regulate Ccn3/Nov in HSCs and Multipotent Progenitors
which encodes a positive regulator of normal HSCs and progenitor cells (19). Although the Ccn3/Nov signal level in control KSL cells was similar to Stat5a/b-null KSL cells and low in the absence of cytokine mixture, treatment with cytokines resulted in a more than 100-fold elevation of Ccn3/Nov mRNA only in control but not Stat5a/b-null cells.

IL-3 Is the Key Cytokine for Induction of Ccn3
/Nov-To investigate which cytokine is responsible for the induction of Ccn3/Nov through STAT5A/B, KL cells (HSC-and multipotent progenitor-enriched population) were sorted. Phosphoflow cytometry and qPCR were performed using control KL cells. IL-3 activated STAT5a in control KL cells (Fig. 2A). The cytokine mixture induced Ccn3/Nov mRNA in control but not in Stat5a/b-null KL cells (Fig. 2B). IL-3 depletion from the cytokine mixture abolished the induction of Ccn3/Nov mRNA in control KL cells (Fig. 2B). Notably, IL-3 alone induced Ccn3/ Nov mRNA in control KL cells (Fig. 2, B and C). Although each individual cytokine in the mixture could induce Ccn3/Nov in KL cells, the strongest activator was IL-3 (supplemental Fig. 1). In addition, single cytokine depletion from the mixture demonstrated IL-3 as the stimulating cytokine of Ccn3/Nov expression (supplemental Fig. 1). The induction of Ccn3/Nov mRNA by IL-3 was also observed in LT-HSCs and ST-HSCs derived from fetal liver cells and bone marrow cells (Fig. 2D). These data demonstrate that IL-3 induced the expression of the Ccn3/Nov mRNA via STAT5A/B in HSCs and progenitors and that IL-3 is the key inducer of Ccn3/Nov among the cytokines in a mixture.
Ccn3/Nov Is under Direct Control through STAT5A/B upon IL-3 Stimulation-To examine the mechanism underlying the induction of Ccn3/Nov mRNA by IL-3 stimulation, STAT5A/B binding to the Ccn3/Nov gene promoter was evaluated using the IL-3-dependent lymphoblast cell line 32D. Elevated Ccn3/Nov mRNA levels were observed after 1 h, and mRNA levels were increased more than 1000-fold by 16 h (Fig.  3A). STAT5A/B consensus binding sites are located at Ϫ9560 bp (GAS1: TTCTAAGAA) and Ϫ8466 bp (GAS2: TTCT-GAGAA) upstream of the transcriptional start site of the murine Ccn3/Nov gene. To test binding of STAT5A/B to these GAS sites, ChIP was combined with qPCR (Fig. 3B). 32D cells were treated with IL-3 for 0, 1, or 6 h followed by ChIP assays. Primers for two potential STAT5A/B binding sites in the Ccn3/ Nov gene promoter, GAS1 and GAS2, were used to determine

DISCUSSION
STAT5A/B are evidently important for blood cells including HSCs. A critical role of cytokines and STAT5A/B in the self-renewal and expansion of human and mouse HSCs has been reported (2)(3)(4)(5). Decreased absolute numbers of KSL cells, lymphocytes, granulocytes, and erythrocytes were observed in the mouse model lacking STAT5A/B in blood cells. Additionally, strong activation of STAT5A/B by Flt3 mutations (internal tandem duplications) has been observed in 30% of patients with acute myelogenous leukemia (20). Constitutive STAT5 activity was demonstrated in BCR-ABL-positive chronic myelogenous leukemia (CML) cell lines, lymphoblastic cell, and hematopoietic cell lines transfected with Bcr-Abl, leading to malignant transformation (21)(22)(23). The essential role of STAT5 in BCR-ABL-induced cell growth and transformation was confirmed by several studies that suggest a critical role of STAT5A/B in the initiation and/or progression of CML (24 -26). There is no doubt that STAT5A/B play an important role in normal and malignant hematopoiesis. Notably, we and others (4,7) showed that the KSL population, but not the absolute number, of Lineage-negative liver cells in Stat5a/b-null fetuses was significantly greater than in control littermates. This suggests that in normal hematopoiesis, STAT5A/B might be more important for the maintenance of the precursor pool than the HSC and progenitor pool.
Our ChIP assays demonstrated IL-3-dependent STAT5A/B binding to the Ccn3/Nov gene promoter specifically at the GAS1 site. While the GAS1 sequence is conserved between mouse and human, located Ϫ5491 bp upstream of the transcriptional start site of the human Ccn3/Nov gene, but GAS2 is not. This suggests that CCN3/NOV might be induced via STAT5A/B in humans as well. It has been reported that CCN3/ NOV plays a role as a positive regulator of HSC self-renewal in human cord blood cells (19). In that study, CCN3/NOV was required for the development of committed lineages. In addition to the role of CCN3/NOV in normal HSCs, CCN3/NOV has been reported as a tumor suppressor in malignant disease. Decreased expression of the CCN3/NOV has been reported in BCR-ABL-positive human CML samples at diagnosis (27). It has been shown that CD34 ϩ CML cells treated with imatinib in vitro had increased intracellular and secreted CCN3/NOV protein upon entering remission and that CCN3/NOV induction reduced the colony formation ability of BCR-ABL-positive cells (27). These data suggest that CCN3/NOV is a suppressor of cell growth in CML stem cells similar to its function in solid tumors. However, the role of CCN3/NOV in CML stem cells is opposite to that in normal stem cells. Further studies should focus on elucidating the difference in CCN3/NOV signaling in malignant and normal cells.
Gene expression profiles of human umbilical cord bloodderived hematopoietic stem progenitor cells, murine embryonic stem-derived HSCs, murine HSCs, LT-HSCs, ST-HSCs, and multipotent progenitors have been published, but nei-ther CCN3/NOV nor Ccn3/Nov have been identified in these studies (28 -33). We report here that the expression of the Ccn3/Nov gene in various murine cell types, fetal liver-derived LT-HSC and ST-HSC, adult bone marrow-derived LT-HSC and ST-HSC, KSL, KL, and 32D cells, requires the inductive stimulus of cytokines that activate STAT5A/B. It is therefore likely that the absence of the Ccn3/Nov expression in the published HSC gene expression profiles is the result of a lack of cytokine stimulation in these systems. Therefore, we also report that attention should be paid to the design of gene expression profiles.
In summary, this study reveals a novel STAT5A/B target gene in HSCs and progenitor cells that is involved in the regulation of normal hematopoiesis. This is the first report that the expression of the Ccn3/Nov gene, encoding a positive regulator of hematopoiesis in HSCs, is induced after cytokine stimulation through STAT5A/B. However, the mechanism regulating CCN3/NOV expression in CML cells probably involves additional layers as CCN3/NOV levels are suppressed in CML cells despite a constitutive activation of STAT5A/B. In the future, it will be necessary to study CCN3/NOV signaling upon stimulation in adult normal HSCs and in CML cells to understand its diverse roles in normal hematopoiesis and leukemogenesis.