Lack of the ubiquitin-editing enzyme A20 results in loss of hematopoietic stem cell quiescence

To identify whether A20 is expressed during early hematopoiesis, we divided BM Lineage⁻Sca1⁺c-Kit⁺ (LSK) cells from C57BL/6 WT mice into LT-HSCs (CD150⁺CD48⁻LSK), ST-HSCs (CD34⁺Flt3⁻LSK), and multipotent progenitors (MPPs; CD34⁺Flt3⁺LSK), sorted them individually, and quantified expression levels of A20 mRNA by RT-PCR. Indeed, A20 was expressed in all three LSK subsets (i.e., LT-HSCs, ST-HSCs, and MPPs), in c-Kit⁺Lin⁻ myeloid progenitor cells, and in total BM cells (Fig. 1 A). Of note, the highest expression levels of A20 were found in ST-HSCs (Fig. 1 B). To study the functions of A20 in HSCs, we generated mice with loxP-flanked alleles of the gene encoding mouse A20 (Fig. 1 C). To analyze Cre-mediated ablation of A20 expression, we crossed A20 Floxed mice with the ubiquitous-Cγ1-Cre animals to obtain A20^Flox/Flox Cγ1-Cre (designated as A20^−/−) mice. To test whether A20 is deleted in these mice, mesenchymal embryonic fibroblasts (MEFs) were generated, from either control or A20^−/− embryos, and stimulated with TNF (as it induces A20 expression in MEFs). Western blot analysis indicated the presence and induction of A20 protein in control MEFs, but it was completely abolished in A20^−/− MEFs (not depicted). We also used antibodies that recognize the N-terminal epitope of A20, which may still present in the truncated A20 protein, and confirmed that no aberrant A20 protein was expressed in these cells (not depicted). To delete A20 in HSCs, we crossed A20^F/F mice with the Vav-Cre transgenic mice (Ogilvy et al., 1999) to generate A20^Flox/FloxVav^Cre/+ (called A20^Hem-KO here) mice and confirmed that A20 protein was predominantly absent in the hematopoietic organs, such as BM, spleen, and thymus. As expected, A20 expression was intact in nonhematopoietic organs, such as brain, kidney, and heart (Fig. 1 D).

A20^Hem-KO mice were born at Mendelian ratios; however, they exhibited ∼50% reduction of body size and weight (Fig. 1, E and F), as compared with their littermate controls. Strikingly, almost all A20^Hem-KO mice died before their weaning age (∼21 d) and had a median lifespan of 14 d (Fig. 1 G). Analysis of 14-d-old A20^Hem-KO mice indicated (Fig. 1 H) thymic atrophy (reduced thymus), splenomegaly (increased spleen size), and hepatomegaly (enlarged liver). Cell count analysis showed an ∼50% reduction in BM cellularity, ∼90% reduction in thymic cellularity, and ∼400% increase in splenic cellularity of A20^Hem-KO mice (Fig. 1 I). Histopathological analysis of spleen and liver of A20^Hem-KO mice revealed disrupted architecture, infiltration of leukocytes (Fig. 1 J), and extramedullary hematopoiesis (Fig. 1 K). Peripheral blood count analysis of A20^Hem-KO mice indicated elevated WBC counts and reduced hemoglobin and platelets counts (Fig. 1 L). Collectively, these data suggest that A20 has an indispensable role in restricting pathological hematopoiesis.

To determine whether the HSC pool of A20^Hem-KO mice was altered, we performed immunophenotyping experiments of HSCs and HSPCs by flow cytometry. Analysis of the LSK compartment, consisting of LT-HSCs, ST-HSCs, and MPPs, revealed a remarkable increase (3.2% in control and 17.9% in KO) of relative numbers in the BM of A20^Hem-KO mice (Fig. 2 A). Further investigation of LSK cells indicated a striking reduction of LT-HSCs (4.8% in control and 0.09% in KO) and MPPs (64% in control and 8% in KO) in the BM of A20^Hem-KO mice (Fig. 2 A). Interestingly, frequencies of ST-HSCs were increased (27% in control and 86% in KO) in the BM A20^Hem-KO mice (Fig. 2 A). Next, we enumerated the absolute numbers of HSPC subsets of the BM. Even though the relative frequencies of LSK cells were increased in A20^Hem-KO mice, the absolute numbers of LSK cells were comparable between controls and A20^Hem-KO mice (Fig. 2 B). Consistent with their relative numbers, the absolute numbers of LT-HSCs and MPPs were decreased, and the absolute numbers of ST-HSCs were increased in the BM of A20^Hem-KO mice (Fig. 2 B).

To independently confirm these findings through an alternative technique, we performed “side population” (SP) analysis (Goodell et al., 1996). This technique utilizes the fluorescent DNA-binding dye Hoechst 33342 to identify and purify mouse HSCs in the BM, as only HSCs, but not differentiated hematopoietic cells, will have the abilities to pump out the Hoechst 33342 dye. Our SP analysis identified a remarkable decrease (0.047% in control and 0.015% in KO) of SP cells in the BM of A20^Hem-KO mice (Fig. 2, C and D).

To further strengthen these data, we adopted the immunophenotyping strategy, developed by the Trumpp laboratory (Wilson et al., 2008), that distinguishes HSPCs of the mouse BM into five distinct stages: HSC (LSK CD150⁺CD48⁻CD34⁻Flt3⁻), MPP1 (LSK CD150⁺CD48⁻CD34⁺Flt3⁻), MPP2 (LSK CD150⁺CD48⁺CD34⁺Flt3⁻), MPP3 (LSK CD150⁻CD48⁺CD34⁺Flt3⁻), and MPP4 (LSK CD150⁻CD48⁺CD34⁺Flt3⁺). According to this study (Wilson et al., 2008), the most primitive HSC (LSK CD150⁺CD48⁻CD34⁻Flt3⁻) fraction remains largely quiescent, and the transition from MPP1 to MPP4 stage is accompanied by a gradual loss of quiescence and an increase of proliferation. As expected, we detected almost no HSCs (13% in control and 0% in KO), slightly increased MPP1 (82% in control and 100% in KO), modestly reduced MPP2 (92% in control and 63% in KO), remarkably increased MMP3 (26% in control and 89% in KO), and highly reduced MPP4 (66% in control and 6% in KO) in the BM of A20^Hem-KO mice (Fig. 2 E). These data suggested that the more quiescent HSC fraction was completely depleted and the transition from MPP3 to MPP4 was severely affected in the absence of A20-mediated functions in HSCs. Overall, the immunophenotyping experiments demonstrated that the HSC pool is remarkably reduced in the BM of A20^Hem-KO mice.

To assess the functions of A20 mutant HSCs, we performed BM transplantation (BMT) experiments. Transfer of A20-deficient BM cells (10⁶) into lethally (10 Gy) irradiated WT animals failed to provide radioprotection in ∼70% of the recipients (Fig. 3 A). Even though the remaining ∼30% of recipients (which received A20-deficient BM) survived for >3 wk, they developed severe leukopenia, anemia, and thrombocytopenia, starting at 2 wk after transplantation (Fig. 3 B). In these survivors, A20 mutant–derived chimerism was >95% at 2 wk after transplantation, but it was dropped to ∼10% at 4 wk after transplantation (Fig. 3, C and D). Transfer of BM from the survivors that showed a donor chimera of ∼10% into lethally irradiated WT secondary recipients failed to provide detectable levels (<1%) of donor chimerism at 4 wk after transplantation (Fig. 3 E). We repeated these experiments using purified LSK cells and obtained similar results (not depicted). Based on the fact that we detected A20 mutant–derived hematopoiesis at 2 wk after transplantation, we reasoned that homing defects might not be responsible for the reduced engraftment in recipients. To substantiate this, we performed homing experiments using CFSE-labeled LSK cells and found that the homing capacities were comparable between control and A20^Hem-KO HSPCs (Fig. 3 F).

Next, we tested the potential of A20-deficient BM to repopulate the hematopoietic system under competitive conditions. A20-deficient and control BM cells (CD45.2⁺) were mixed with congenic (CD45.1⁺) WT BM cells at a ratio of either 1:1 or 3:1 (donor/competitor) and were injected into lethally irradiated congenic (CD45.1⁺) recipients. Analysis of donor-derived chimera at 4 wk after transplantation indicated a significant reduction of donor-derived (CD45.2⁺) hematopoiesis (Fig. 3, G and H), as well as multilineage differentiation (Fig. 3, I and J), in recipients that received A20 mutant BM. Because no detectable levels of A20 mutant–derived chimerism were seen in recipients at 4 wk after transplantation, we increased the donor/competitor ratio to 5:1 and injected into lethally irradiated WT recipients. Analysis of donor-derived chimera indicated that even though A20 mutant–derived hematopoiesis was detectable (albeit ∼50% lower than that of the control group) in recipients at 2 wk after transplantation, it was undetectable in the BM, spleen, thymus, and peripheral blood at 9 wk after transplantation (Fig. 3, K and L). Collectively, these data indicate that A20 deficiency impairs HSC functions.

Our analysis of A20^Hem-KO mice indicated a striking reduction of LT-HSCs in the BM (Fig. 2 A). Loss of LT-HSCs in A20 mutant BM could be explained by either increased apoptosis of LT-HSCs or increased proliferation in conjunction with decreased self-renewal of LT-HSCs. To identify the reason or reasons for the loss of the LT-HSC pool in A20 mutant mice, we first performed apoptosis experiments using Annexin V and 7-AAD. Data of these experiments indicated comparable frequencies of apoptosis between control and A20^Hem-KO mice (not depicted). These data were further validated using TUNEL assays, and we obtained consistent results (not depicted). Next, we performed proliferation experiments using BrdU (Hock et al., 2004). Data of these experiments indicated that the frequencies of BrdU⁺ cells and that the intensities of incorporated BrdU were augmented both in total LSK and in CD150⁺LSK cells from 14-d-old A20^Hem-KO mice (Fig. 4, A and B). Because LT-HSCs were almost absent in 14-d-old A20^Hem-KO mice, we repeated BrdU assays with 2-d-old (P2) pups and observed that A20 mutant CD48⁻CD150⁺LSK cells proliferated more rapidly than control cells (Fig. 4 C). Consistent with this, Ki-67 expression levels were also increased in A20 mutant LSK and CD150⁺LSK cells (Fig. 4, D and E). Next, we measured HSC quiescence using Hoechst and Ki-67. As expected, the frequencies of LSK, CD150⁺LSK, and LT-HSC fractions at G0 phase were reduced in the BM of A20^Hem-KO mice (Fig. 4 F). Finally, we performed Pyronin Y staining to distinguish quiescent HSPCs from proliferating HSPCs (Cheng et al., 2000). Pyronin Y analysis of A20 mutant BM revealed a striking reduction (40% in control and 3.5% in KO) of LSK cells in the G0 (Pyronin Y⁻Hoechst⁻) phase of the cell cycle (Fig. 4, G and H). Collectively, these data indicated that A20 deficiency might cause loss of HSC quiescence and premature depletion of the HSC pool.

A20 is a key negative regulator of NF-κB signaling, and loss of A20 results in augmented NF-κB signal transduction (Vereecke et al., 2009; Hymowitz and Wertz, 2010; Ma and Malynn, 2012). To test whether NF-κB signaling was increased in A20^Hem-KO mice, we measured IκBα expression and activation. Hematopoietic cells from control and A20 mutant mice were stimulated in vitro with LPS and assessed for protein expression. Western blot data indicated a baseline reduction of total IκBα protein in unstimulated cells of A20 mutant mice. However, in response to LPS, total IκBα levels were strikingly reduced and phospho- IκBα protein levels were increased in the hematopoietic cells of A20 mutant mice (Fig. 5, A and B). Decrease of total IκBα levels and increase of phospho-IκBα suggested that NF-κB signaling was elevated in A20-deficient cells. Consistent with increased NF-κB signaling, expression levels of several NF-κB targets, such as Relb, NF-κB2, TNF, BclX, Hif1α, and Sfpi1, were elevated in the hematopoietic cells of A20^Hem-KO mice (Fig. 5 C). These data support the previous findings (Vereecke et al., 2009; Hymowitz and Wertz, 2010; Ma and Malynn, 2012) that NF-κB signaling is augmented in A20-deficient hematopoietic cells.

Earlier studies identified IFN-γ as a direct target of NF-κB proteins (Sica et al., 1997), and based on the critical role of IFN-γ in the maintenance of HSCs (Baldridge et al., 2011; King and Goodell, 2011), we hypothesized that deregulated IFN-γ signaling might be potentially responsible for the HSC phenotype of A20^Hem-KO mice. To test this, we measured the expression levels of IFN-γ in A20^Hem-KO mice. In line with our hypothesis, we detected elevated levels of Ifn-γ mRNA in both BM and spleen of A20-deficient mice (Fig. 6 A). To substantiate this finding and to identify the cellular source of IFN-γ in A20 mutant mice, we crossed A20^Hem-KO mice with the IFN-γ–YFP reporter GREAT (IFN-γ reporter with endogenous polyA transcript) mice (Reinhardt et al., 2009). These transgenic mice have an IRES-eYFP reporter cassette inserted between the translational stop codon and the 3′ UTR/polyA tail of the Ifn-γ gene (Reinhardt et al., 2009). Our analysis of hematopoietic compartments from GREAT-A20^Hem-KO mice suggested an elevated expression of YFP and, therefore, IFN-γ in CD11C⁺, CD3⁺CD4⁺, and CD3⁺CD8⁺ cells of BM and spleen (Fig. 6 B). In addition, myeloid cells, such as granulocytes and monocytes, from GREAT-A20^Hem-KO mice also expressed YFP, albeit at lower levels (Fig. 6 B). Next, we tested whether Ifn-γ is a direct target of NF-κB in A20 mutant BM cells. Our analysis of mouse Ifn-γ gene sequences indicated the presence of five (AAGGACTTCCTC) NF-κB–binding sites (Wong et al., 2011) in its promoter region. Among these sites, one NF-κB–binding site (from −541 to −530) is highly conserved across species (Fig. 6, C and D), and chromatin immunoprecipitation (ChIP) data revealed an increased binding of NF-κB to the Ifn-γ promoter in A20^Hem-KO BM cells (Fig. 6 E). To understand the physiological relevance of deregulated IFN-γ expression on the HSC compartment, we investigated whether IFN-γ signaling was augmented in the HSCs from A20^Hem-KO mice. Real-time PCR data indicated that the expression levels of IFN-γ target genes, such as Irf1, Irgm1, and STAT1, were elevated in A20 mutant HSCs (Fig. 6 F). Deregulated IFN-γ signaling results in up-regulated surface expression of Sca1 in HSPCs, as Sca1 is a direct target of IFN-γ (Snapper et al., 1991; King and Goodell, 2011). Our analysis of the LSK compartment indicated increased surface expression of Sca1 in hematopoietic progenitors and in total BM cells of A20^Hem-KO mice (Fig. 6, G and H; and not depicted). Intriguingly, up-regulation of Sca1 in HSPCs has been connected with increased cell cycle activity (Essers et al., 2009).

At the molecular level, a major functional consequence of increased IFN-γ signaling is diminished HSC quiescence, probably caused by deregulated expression of cell cycle regulators, such as CcnD1 and p57, and other key transcriptional targets (de Bruin et al., 2013). Our gene expression experiments identified an increase in the expression levels of CcnD1 (a positive regulator of cell cycle) and a decrease in the expression levels of p57 (a negative regulator of cell cycle) in A20-deficient CD150⁺LSK cells (Fig. 6 I). Of note, HSCs that are deficient for D-Cyclins show reduced proliferation (Kozar et al., 2004; Pietras et al., 2011), and HSCs lacking p57 failed to reconstitute the hematopoietic system because of loss of quiescence (Matsumoto et al., 2011), a phenotype which is very similar to the phenotype of A20-deficient HSCs. Overall, these data suggest that IFN-γ expression and signaling were elevated in A20^Hem-KO mice, and as a consequence, their target genes were up-regulated in HSCs of these animals.

To directly address the involvement of IFN-γ signaling in the phenotype of A20^Hem-KO mice, we crossed A20^Hem-KO mice with IFN-γ^−/− mice to obtain A20^Hem-KO IFN-γ^−/− (double KO [DKO]) mice. Strikingly, DKO mice were larger in size and showed more body weight than the A20^Hem-KO littermates (Fig. 7, A and B). However, DKO mice were slightly smaller and lighter than the A20^F/+ littermate controls (Fig. 7, A and B). Survival curve analysis indicated that postnatal lethality of A20^Hem-KO mice was rescued in the DKO mice (Fig. 7 C). Analysis of the HSPC compartment in the BM of DKO mice revealed a rescue of the LSK phenotype (Fig. 7 D). Consistently, the phenotype (relative frequencies) of LSK subsets, including LT-HSCs, ST-HSCs, and MPPs, was rescued in DKO mice (Fig. 7 D). Absolute numbers of LT-HSCs indicated a significant increase (∼300%) of LT-HSC numbers in the DKO mice when compared with their A20^Hem-KO littermates (Fig. 7 E). However, when compared with the littermate controls, the numbers of LT-HSCs were still reduced in the DKO mice (Fig. 7 E). In addition, surface expression of Sca1 was restored to WT levels in DKO progenitor cells (Fig. 7 F). Next, we performed BMT experiments to assess the functions of DKO HSCs. Transfer of total BM cells (CD45.2⁺) from control, IFN-γ^−/−, A20^Hem-KO, and DKO mice into lethally irradiated WT congenic (CD45.1⁺) recipients resulted in comparable survival among control, IFN-γ^−/−, and DKO recipient groups, whereas mice that received A20^Hem-KO BM showed reduced survival (Fig. 7 G). In these particular experiments (Fig. 7 G), we did not observe compromised survival of the recipients that received A20 mutant BM to the extent shown in Fig. 3 A. Nevertheless, analysis of survivors at 5 wk after transplantation indicated undetectable levels of donor cells in recipients that received A20^Hem-KO BM cells (Fig. 7 H). In contrast, recipients that received DKO BM cells had ∼40% donor-derived chimera at 5 wk after transplantation (Fig. 7 H), and more importantly, remarkable levels of donor-derived chimera were maintained even at 14 wk after transplantation (Fig. 7 I). To further strengthen these findings, we performed competitive repopulation experiments (donor/competitor ratio = 5:1 and 1:1, respectively) with control, A20^Hem-KO, and DKO BM cells. Analysis of recipients indicated detectable levels of donor-derived chimera in both control and DKO groups, whereas donor-derived hematopoiesis was completely absent in recipients that received A20^Hem-KO BM (Fig. 7, J and K). Nevertheless, frequencies of donor-derived hematopoiesis were lower in recipients that received DKO BM when compared with the chimera derived from littermate control BM. Donor-derived multilineage analysis indicated intact differentiation of both myeloid and lymphoid lineage cells from DKO BM, whereas it was completely absent in recipients that received A20^Hem-KO BM (Fig. 7, L and M). Overall, the data of the DKO mice suggested that the HSC phenotype of A20^Hem-KO mice was, at least in part, caused by elevated IFN-γ signaling.

In recent years, the ubiquitin-editing enzyme A20 has emerged as a central regulator of inflammation (Liu et al., 2005). Germline deficiency of A20 in mice causes persistent activation of NF-κB, leading to multiorgan inflammation, cachexia, neonatal lethality, and increased responses to TNF and LPS (Lee et al., 2000; Boone et al., 2004). To understand the functions of A20 in the immune system, the Tnfaip3 gene was deleted in selective immune compartments using tissue-specific Cre lines. Disrupting functions of A20 in B cells, using CD19 Cre, resulted in exaggerated NF-κB signaling and hyperresponsiveness to LPS/B cell receptor and CD40 ligation. In addition, A20 deficiency in B cells caused elevated levels of IL6 expression, spontaneous activation of B cells, expansion of myeloid cells, plasma cell hyperplasia, and autoimmunity at 3–4 mo of age (Tavares et al., 2010; Chu et al., 2011). DC-specific deletion of A20, using CD11c Cre mice, resulted in spontaneous activation of DCs (which results in increased secretion of proinflammatory cytokines), B cells, T cells, and myeloid cells and development of lymphadenopathy and splenomegaly in 4–6-wk-old mice (Hammer et al., 2011; Kool et al., 2011). Macrophage- and granulocyte-specific deletion of A20, using LysM Cre mice, caused spontaneous polyarthritis, increased systemic and local cytokine production, and increased osteoclast functions (Matmati et al., 2011). Although these studies have elegantly demonstrated the importance of A20-mediated NF-κB down-regulation in the immune system, almost all of these studies were focused on the role of A20 on the functions of immune cells. In the present study, we focused on the role of A20 in early hematopoiesis and found that A20 deficiency results in hyperproliferation of HSPCs and in premature depletion of the HSC pool. Thus, A20 acts as a key modulator in the maintenance of HSC quiescence, particularly under steady-state conditions.

The role of infection and inflammation in the control of hematopoiesis has gained a lot of attention in the recent years. Although it was believed that only lineage-committed progenitors, including CMP and CLP, were involved in sensing and in responding to infections (Welner et al., 2008; Baldridge et al., 2011), it is now evident that HSCs are directly involved in the primary response of both acute and chronic infections (King and Goodell, 2011). The direct response of HSCs to an immune insult might be a key determinant in the clearance of the pathogen because HSCs can dictate cell fate decisions and can preferentially differentiate into a particular hematopoietic lineage to clear the pathogen. HSCs have evolved to sense pathogens by both cell-intrinsic and -extrinsic mechanisms (Baldridge et al., 2011; King and Goodell, 2011). HSPCs have been shown to recognize pathogens in a cell-intrinsic manner, through their expression of TLRs. In contrast, the cell-extrinsic mode of recognition of an immune insult by HSCs primarily involves expression and stimulation by proinflammatory cytokines and chemokines, which includes IL6, TNF, CC-chemokine ligand 2 (CCL2), IFN-α, and IFN-γ (Baldridge et al., 2011; King and Goodell, 2011).

Indeed, long-term treatment of mice with low doses of LPS resulted in impaired self-renewal and multilineage differentiation of HSCs (Esplin et al., 2011). Exposure of HSCs to a combination of IL6, TNF, and CCL2, a condition which mimics LPS stimulation, resulted in defective long-term engraftment (Chen et al., 2010).

Recent studies identified IFNs, both type I (IFN-α) and type II (IFN-γ), as major regulators of HSC quiescence. IFNs are produced by lymphocytes, such as NK cells, NKT cells, and T cells, during bacterial and viral infections. Although in vitro stimulation of HSCs with IFN-α has antiproliferative effects, in vivo treatment of mice with IFN-α results in loss of HSC quiescence, increased phosphorylation of STAT1 and AKT1, expression of IFN-α target genes, up-regulation of surface Sca1, and increased cell cycle activity mediated by Sca1 (Essers et al., 2009). Interestingly, transient activation of HSCs by IFN-α did not affect the numbers of functional HSCs. However, long-term (chronic) activation of the IFN-α pathway in HSCs leads to decreased HSC activity and a marked competitive disadvantage in mixed BM reconstitution experiments (Essers et al., 2009). Consistently, loss of IFN regulatory factor-2 (IRF-2), a transcriptional suppressor of IFN signaling, results in reduced HSC pool, diminished HSC quiescence, and compromised HSC functions as the result of increased IFN-α signals (Sato et al., 2009). In contrast, stimulation of HSCs with IFN-γ leads to hyperproliferation of HSCs both in vitro and in vivo (Baldridge et al., 2010), and IFN-γ has been shown to be sufficient to promote proliferation of long-term repopulating HSCs, accompanied by a remarkable increase of immediate downstream progenitors (Baldridge et al., 2010). This study found a disadvantage in competitive repopulation assays after Mycobacterium avium infection (IFN-γ exposure) and proposed that baseline IFN-γ tone may influence HSC turnover and quality during normal homeostasis (Baldridge et al., 2010). Consistently, King et al. (2011) showed that dysregulated IFN-γ signaling results in reduction of the HSC pool. Overall, our data agree with the existing paradigm of IFN signaling in HSCs (Essers et al., 2009; King et al., 2011; Pietras et al., 2014). The current study indicated that IFN-γ signaling has an indispensable role in the hematopoietic phenotype of A20-deficient mice. However, analysis of A20–IFN-γ DKO mice indicated only a partial rescue of the A20^Hem-KO phenotype. These data suggest that IFN-γ is one of the several factors contributing to the HSC phenotype of A20^Hem-KO mice. Other potential factors that could influence the phenotype of A20-deficient mice may include cytokines, such as type I IFNs, IL6, and TNF. Indeed, earlier studies identified critical roles of these cytokines in the control of HSCs and hematopoiesis (Baldridge et al., 2011; King and Goodell, 2011; Pietras et al., 2011). Among these cytokines, it would be more interesting to study the role of type I IFNs, as mounting evidence documents the direct involvement of A20 in the negative regulation of two key inducers of type I IFN, IRF3 and IRF7 (Hymowitz and Wertz, 2010; Parvatiyar and Harhaj, 2011).

Further studies that aim to dissect the potential involvement of other proinflammatory cytokines in A20-deficient mice would be helpful in understanding the role of inflammation in HSC biology. We anticipate that our study contributes to a deeper understanding of hematological disorders in patients with TNFAIP3 (A20) mutations.

To generate Tnfaip3 conditional KO mice, the mouse Tnfaip3 exon 3 region and homology arms were PCR amplified from C57BL/6 mouse genomic DNA and cloned into the SalI, XhoI, and SfiI sites of the pEasy-FRTlox DT vector (provided by K. Rajewsky, Harvard Medical School, Boston, MA), which contains a neomycin resistance cassette for positive selection and a diphtheria toxin cassette for negative selection. The targeting vector was linearized by ClaI and electroporated into Bruce4 embryonic stem cells. G418-resistant clones were screened for homologous recombination by Southern blotting. Correctly targeted embryonic stem clones were injected into albino C57BL/6-cBrd/cBrd blastocysts (National Cancer Institute, Frederick, MD) and implanted into host female mice to generate chimeras. Chimeras were then bred to albino C57BL/6 mice to obtain black C57BL/6 litters, which were then screened for germline transmission of the allele. Mice were crossed to the ACTB-Flpe mouse strain (The Jackson Laboratory) to remove the neomycin cassette. VAV^Cre/+, IFN-γ^−/−, and IFN-γ reporter (GREAT) mice were purchased from The Jackson Laboratory. CD45.1 congenic animals were purchased from the National Cancer Institute. In our experiments, we used A20^+/+, A20^F/+, A20^F/F, and A20^F/+ Vav^Cre/+ mice as controls, as none of these control animals exhibited any hematopoietic phenotype. However, in each experiment, age-and gender-matched littermates were used as controls. Mice were maintained under specific pathogen–free conditions and were used according to the protocols approved by the Institutional Animal Care and Use Committee of Columbia University Medical Center.

A20^Hem-KO mice were analyzed at 14 d after birth (P14). RBCs were lysed with ammonium chloride (STEMCELL Technologies). Trypan blue (Amresco)–negative cells were counted as live cells.

Cells were analyzed by flow cytometry with either LSRFortessa or LSRII (both BD) and FACSDiva software (BD) or FlowJo software (Tree Star). Cell sorting was performed with a FACSAria (BD). The following monoclonal antibodies were used: anti-B220 (RA3-6B2), anti-CD19 (1D3), anti-CD3ε (145-2C11), anti-CD4 (GK1.5), anti-CD8 (53-6.7), anti-CD11b (M1/70), anti-CD34 (RAM34), anti-CD45.1 (A20), anti-CD45.2 (104), anti-CD48 (HM48-1), anti-CD117 (2B8), anti-Flt3 (A2F10.1), anti–Gr-1 (RB6-8C5), anti–Sca-1 (D7), and anti-Ter119 (TER119; from BD); anti-CD150 (TC15-12F12.2; from BioLegend); and anti-CD16/32 (93) and anti-CD127 (A7R34; from eBioscience). Cells incubated with biotinylated monoclonal antibodies were incubated with fluorochrome-conjugated streptavidin–peridinin chlorophyll protein–cyanine 5.5 (551419; BD) or streptavidin-allophycocyanin-Cy7 (554063; BD). In all of the FACS plots, the relative frequencies (%) of the gated fraction are indicated.

For radioprotection assays, 10⁶ of BM cells were injected into lethally irradiated (10 Gy) congenic (CD45.1⁺) recipient mice. For competitive repopulation experiments, 4 × 10⁵, 1.2 × 10⁶, or 10⁶ of BM cells from either A20 control or A20^Hem-KO mice were mixed with 4 × 10⁵, 4 × 10⁵, or 2 × 10⁵ of WT (CD45.1⁺) BM cells (to obtain a ratio of 1:1, 3:1, or 5:1, respectively) and were injected into lethally irradiated congenic WT (CD45.1⁺) recipient mice. For serial transplantation, 6 × 10⁵ BM cells from the primary transplants were injected into lethally irradiated congenic WT recipients.

Apoptotic cells were detected with the Annexin V PE Apoptosis Detection kit or Annexin V FITC Apoptosis Detection kit (BD) according to the manufacturer’s instructions. For TUNNEL assay, the In Situ Cell Death Detection kit (Roche) was used according to the manufacturer’s instructions.

For BrdU assays, either 14-d-old (P14) mice or 1-d-old (P1) pups were injected i.p. with either 1 Mg or 200 µg BrdU (BD), respectively. At 16 h after injection, mice/pups were sacrificed and BM/liver cells were stained for BrdU, according to the BrdU Flow kit (BD) manufacturer’s instructions, and detected by flow cytometry. For Ki-67 analysis, BM cells were stained for surface markers and fixed and permeabilized with FIX & PERM Cell Fixation & Cell Permeabilization kit (Life Technologies). These cells were further stained with anti–Ki-67 antibodies (B56; BD).

BM cells were incubated with 5 µg/ml Hoechst 33342 (Life Technologies) at 37°C for 90 min.

For Pyronin Y analysis, cells were first incubated with 5 µg/ml Hoechst 33342 at 37°C for 45 min and then with 1 µg/ml Pyronin Y (Sigma-Aldrich) for an additional 45 min at 37°C. For cell cycle assays using Ki-67 and Hoechst, cells were stained with surface markers and fixed and permeabilized with Cytofix/Cytoperm fixation/permeabilization solution kit (BD). These cells were stained with Ki-67 for 30 min and subsequently stained with 20 µg/ml Hoechst 33342 for 5 min and analyzed by flow cytometry.

LSK cells were sorted from the BM and stained with CFSE (Life Technologies) for 15 min, and 5 × 10⁴ CFSE⁺LSK cells were injected i.v. into lethally irradiated (10 Gy) WT mice. 20 h after injection, mice were sacrificed, and CFSE-positive cells in the BM were analyzed by flow cytometry.

For immunoblot analyses, cells were lysed with cell lysis buffer (Cell Signaling Technology) with protease inhibitor cocktail (Complete; Roche) and 1 mM PMSF (Santa Cruz Biotechnology, Inc.). Cell lysates were boiled with sample buffer (NuPAGE; Life Technologies) containing 1% 2-Mercaptoethanol (Sigma-Aldrich). Proteins were subjected to 8–12% SDS-PAGE and transferred to PVDF membranes (Bio-Rad Laboratories). The membranes were blocked with either 5% bovine serum albumin (Life Technologies) or 5% skim milk and then treated with primary and secondary antibodies, respectively. The blots were visualized using the ProtoGlow ECL (National Diagnostics) and Image station 440 (Kodak). Antibodies used were as follows: anti-A20 (D13H3; Cell Signaling Technology), anti-IκBα (44D4; Cell Signaling Technology), anti–phospho-IκBα (5A5; Cell Signaling Technology), anti-actin (I-19; Santa Cruz Biotechnology, Inc.), HRP-conjugated anti–mouse and anti–rabbit IgG (Cell Signaling Technology), and HRP-conjugated anti–goat IgG (Santa Cruz Biotechnology, Inc.).

Splenocytes were stimulated with 1 µg/ml LPS (Sigma-Aldrich) at 37°C for 1 h, and then lysed for protein extraction. Cells with no stimulation served as controls.

ChIP assay was performed with an Agarose ChIP kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. In brief, 10⁷ of BM cells were fixed and immunoprecipitated with either anti-p65 (D14E12; Cell Signaling Technology) or rabbit IgG (Thermo Fisher Scientific) antibodies. Immunoprecipitated DNA fragments were quantified by real-time PCR with the use of the following primers, which amplify the IFN-γ promoter region containing NF-κB–binding sites: forward, 5′-ACCCTGAGTGATTTGTAGTAGGT-3′; and reverse, 5′-GTGAATTTCTCATCCACAGAGCA-3′. Fold enrichment was normalized to rabbit IgG–precipitated samples.

Total RNA was isolated with an RNeasy Mini kit (QIAGEN). cDNA was synthesized with Oligo(dT) primer and Maxima reverse transcription (Thermo Fisher Scientific). PCR was performed with T100 thermal cycler (Bio-Rad Laboratories) and TSG Taq (Lamda Biotech). Real-time PCR was performed in duplicates with a CFX-connect real-time PCR system (Bio-Rad Laboratories) and SsoAdvanced SYBR Green Supermix (Bio-Rad Laboratories) according to the manufacturer’s instructions. Relative expression was normalized to the expression levels of the internal control (housekeeping gene) Hprt.

We used the unpaired Student’s t test for all the experiments, except survival experiments. For survival curve analysis, log-rank test was used. Differences with P < 0.05 were considered statistically significant and are denoted as *, P < 0.05; **, P < 0.01; ***, P < 0.001.

We thank Riccardo Dalla-Favera for expert advice and critical reading of the manuscript and Danielle Matsushima for help with correcting the manuscript.

The authors declare no competing financial interests.

Author contributions: M.M. Nakagawa performed all of the experiments, analyzed and interpreted data, and prepared the figures. K. Thummar helped with establishing the animal colony. J. Mandelbaum and L. Pasqualucci generated the A20^Floxed mice. C.V. Rathinam conceived, designed, and directed research, interpreted data, provided advice, and wrote the manuscript. All authors discussed and reviewed the manuscript.