Adaptive response to inflammation contributes to sustained myelopoiesis and confers a competitive advantage in myelodysplastic syndrome HSCs

Abstract

Despite evidence of chronic inflammation in myelodysplastic syndrome (MDS) and cell-intrinsic dysregulation of Toll-like receptor (TLR) signaling in MDS hematopoietic stem and progenitor cells (HSPCs), the mechanisms responsible for the competitive advantage of MDS HSPCs in an inflammatory milieu over normal HSPCs remain poorly defined. Here, we found that chronic inflammation was a determinant for the competitive advantage of MDS HSPCs and for disease progression. The cell-intrinsic response of MDS HSPCs, which involves signaling through the noncanonical NF-κB pathway, protected these cells from chronic inflammation as compared to normal HSPCs. In response to inflammation, MDS HSPCs switched from canonical to noncanonical NF-κB signaling, a process that was dependent on TLR-TRAF6-mediated activation of A20. The competitive advantage of TLR-TRAF6-primed HSPCs could be restored by deletion of A20 or inhibition of the noncanonical NF-κB pathway. These findings uncover the mechanistic basis for the clonal dominance of MDS HSPCs and indicate that interfering with noncanonical NF-κB signaling could prevent MDS progression.

Extended Data Fig. 1 |. Inflammatory and immune pathway activation in MDS cells.

a, Survival analysis of MDS patients based on TRAF6 expression (probe: 227264_at) in CD34₊ cells (GSE58831). Patients were stratified based on TRAF6 mRNA expression (top 20%, n = 28; bottom 20%, n = 26). Log-rank (Mantel-Cox) test. b, Overview of experimental design to examine inflammatory states in MDS and human TLR-TRAF6 primed HSPC. c, Human miR-146a deficient (miR146a^−/−) and control (WT) CD34⁺ BM cells generated from healthy CD34⁺ BM using CRISPR-Cas9 gene editing or Vav-TRAF6 and WT LSK BM cells were stimulated in vitro for 90 min with 1μg/mL of LPS (or PBS) (n = 3 each per group) and then examined for differential gene expression by RNA-sequencing. The inflammatory state for each group was determined using the GSEA. NES, normalized enrichment score. d, Expression of miR-146a in miR-146a deficient (miR146a^−/−) and control (sg-CTL) CD34⁺ BM cells gene edited using CRISPR-Cas9. Results are presented as mean ± s.e.m. for n = 3 technical replicate samples. e, Immunoblot analysis of TRAF6 and IRAK1, two miR-146a targets, in miR146a^−/− and control (sg-CTL) CD34⁺ BM cells gene edited using CRISPR-Cas9. Shown is an immunoblot from a single biological replicate.

Extended Data Fig. 2 |. Evaluation of Vav-TRAF6-YFP and WT-YFP versus WT BM cell competition with LD-LPS.

a, Overview of experimental design to directly measure hematopoietic cell competition in the presence of low-dose chronic inflammation. Vav-TRAF6 CD45.2 BM cells (co-expressing a YFP transgene referred to as Vav-TRAF6-YFP) and WT CD45.2 BM cells were transplanted in equal proportions into lethally irradiated recipient mice. Two months after transplantation, chimeric mice were treated with LD-LPS (1 μg/g) or vehicle twice a week for 30 days and then examined for hematopoietic contribution of Vav-TRAF6-YFP and WT cells in the PB and BM. After the last LPS treatment, BM cells were serially transplanted into lethally irradiated recipient mice. b, Overview of experimental design to directly measure hematopoietic cell competition in the presence of low-dose chronic inflammation. Wild-type (WT) CD45.2 BM cells (co-expressing a YFP transgene) and WT CD45.2 BM cells were transplanted in equal proportions into lethally irradiated recipient mice. Two months after transplantation, chimeric mice were treated with LD-LPS (1 μg/g) or vehicle twice a week for 4 weeks and then examined for hematopoietic contribution of WT-YFP and WT cells in the BM. c, Representative flow cytometric profiles and gating strategy of YFP⁺ (WT-YFP) cells in the BM of chimeric mice after LPS or vehicle (PBS) treatment. d, The proportion of YFP⁺ (WT-YFP) cells in LSK populations 4 weeks after treatment with LPS or vehicle (PBS). Data represent the mean ± s.e.m., n = 6 mice per group.

Extended Data Fig. 3 |. Overexpression of TRAF6 alters the response of hematopoietic progenitor cells to LD-LPS.

a, Overview of experimental design to examine the long-term effects of low-dose chronic inflammation on hematopoiesis. Vav-TRAF6 (T6) CD45.2 BM cells or WT CD45.2 BM cells were isolated from mice treated with LD-LPS (1 μg/g) or vehicle twice a week for 30 days and then transplanted with a ratio of 10:1 of CD45.1 competitor BM cells into lethally irradiated recipient mice. Three months after transplantation, BM cells were serially transplanted into lethally irradiated recipient mice. b, The proportion of donor-derived CD45.2⁺ cells in MPP2 (Flk2⁻CD150⁺CD48⁺LSK), MPP3 (Flk2⁻CD150⁻CD48⁺LSK), and MPP4 (Flk2⁺CD150⁻CD48⁺LSK) after tertiary transplantation. Results are presented as mean ± s.e.m. for n = 3 mice per group. Statistical analysis was performed by a two-tailed Studenťs t-test. *, P < 0.05.

Extended Data Fig. 4 |. TLR-TRAF6 primed HSPC exhibit expression of non-canonical NF-kB gene signatures.

a, Normalized enrichment scores (NES) and P value of gene signatures established from WT and Vav-TRAF6 LSK stimulated with LPS evaluated in constitutively active (ca) NIK expressing LSK (GSE88949). b, GSEA plots established from caNIK LSK were evaluated in miR-146a deficient (miR146a^−/−) CD34⁺ BM cells gene edited using CRISPR-Cas9 and then stimulated with 1 μg/mL of LPS (or PBS) for 90 min. The gene expression profiles are relative to unstimulated miR146a^−/− CD34⁺ BM and control (sg-CTL) (+/- LPS). (c) Capillary immunoassay of CD34⁺ cells isolated from healthy controls or MDS BM visualized by chemiluminescence using ProteinSimple. Shown is an immunoassay from a single biological replicate.

Extended Data Fig. 5 |. TET2 deficiency in hematopoietic cells results in increased myeloid-biased hematopoiesis without affecting the proportions of HSC after LD-LPS.

a, Overview of experimental design to examine the effects of low-dose chronic inflammation on hematopoiesis. Tet2^f/f VavCre CD45.2 BM cells (Tet2^−/−) or Tet2^f/f (WT) CD45.2 BM cells were isolated from mice treated with LD-LPS (1 μg/g) or vehicle twice a week for 4 weeks and then transplanted into lethally irradiated recipient mice (along with CD45.1 competitor BM cells). One month after transplantation, PB and BM cells were evaluated by flow cytometry. b, The proportion of donor-derived CD45.2⁺ myeloid (CD11b⁺) and lymphoid (B220+ and CD3⁺) cells in the PB (n = 6 mice per group). * P = 0.02. c, The proportion of donor-derived CD45.2⁺ LSK and LT-HSC in the BM of mice after treatment with LD-LPS. Results are presented as mean ± s.e.m., n = 6 for all groups; n = 3 for WT LPS treated group. * P = 0.03. Statistical analysis in b was performed by a two-tailed Studenťs t-test. Statistical analysis in c was performed by a one-tailed Studenťs t-test.

Extended Data Fig. 6 |. Generation of Traf6- and Tet2-deficient mice.

Genotyping analysis of Tet2^−/− VavCre and Traf6^−/− Tet2^−/− VavCre mice.

Extended Data Fig. 7 |. Generation of Vav-TRAF6 and A20-deficient mice.

Genotyping analysis of Vav-TRAF6 RosaCreER and A20^−/− Vav-TRAF6 RosaCreER mice. A20 floxed allele recombination is shown after Tamoxifen treatment.

Extended Data Fig. 8 |. A20 knockdown impairs MDSL and THP1 cell function.

a, Immunoblotting of MDSL cells expressing independent shRNAs targeting A20 (shA20) or non-targeting shRNA (shControl). Shown is an immunoblot from a single biological replicate. b, Colony forming potential of MDSL cells expressing shRNAs targeting A20 (shA20) or non-targeting shRNA (shControl) in methylcellulose. Results are presented as mean ±s.e.m., for n = 3 independent biological replicates, * P = 0.006, ** P = 0.0004. c, Immunoblotting of THP1 cells expressing an shRNA targeting A20 (shA20) or non-targeting shRNA (shControl). Shown is an immunoblot from a single biological replicate. d, Colony forming potential of THP1 cells expressing shRNAs targeting A20 or non-targeting shRNA (shControl) in methylcellulose. Results are presented as mean± s.e.m., for n = 3 independent biological replicates. *, P = 0.0001. Statistical analysis in b,d was performed by a two-tailed Studenťs t-test.

Fig. 1 |. MDS HSPC are associated with inflammatory states.

a, Heatmap of all differentially expressed genes in CD34+ MDS BM cells (GSE58831,1.5-fold, P<0.05), stratified based on high (n = 28, Z=> 1.0) and low (n = 18, Z<-1.0) TRAF6 expression. Each row represents individual genes arranged by semiunsupervised clustering analysis based on the Z score (see Supplementary Table 1). Columns represent individual MDS patients. Reoccurring somatic mutations and cytogenetic alterations in the MDS patient samples are shown by black bars at the bottom. Gray bars indicate no available data. b, GSEA gene signatures from 'TRAF6 high' (TRAF6^hi) CD34+ MDS cells as compared to 'TRAF6 low' (TRAF6¹⁰) CD34⁺ MDS cells. c, GSEA gene signatures from TET2 or ASXL1 mutant CD34⁺TRAF6^hi MDS cells as compared to CD34+TRAF6¹⁰ MDS cells. d, RNA sequencing of human miR146a^−/− and WT CD34+ BM cells or Vav-TRAF6 and WT LSK BM cells were stimulated in vitro for 90 min with 1 γg ml⁻¹ of LPS (or PBS) (see Supplementary Tables 2–8). Results are presented as mean ±s.e.m., for n = 3 independently treated samples. The inflammatory state for each group was determined using the GSEA profiles established from CD34⁺TRAF6^hi MDS cells. The indicated inflammatory gene signatures are arranged by P value (log_l0).

Fig. 2 |. TLR-TRAF6-primed HSPCs outcompete WT HSPCs with LD-LPS.

a, Left panel, representative flow cytometric profiles of YFP⁺ (Vav-TRAF6-YFP) cells in the PB of chimeric mice before and after LPS or vehicle (PBS) treatment. Right panel, the proportion of YFP⁺ (Vav-TRAF6-YFP) cells relative to YFP⁻ (WT) cells before and after LPS or vehicle (PBS) treatment. Data represent the mean±s.e.m., n = 6 mice per group (*P< 0.05, **P< 0.0001). b, The proportion of myeloid (CD11b⁺) and lymphoid (CD3⁺ and B220⁺) YFP⁻ (WT) and YFP⁺ (Vav-TRAF6-YFP) cells in the PB after LPS or vehicle (PBS) treatment. Data represent the mean±s.e.m., n = 6 mice per group. (*P< 0.05). c, The proportion of YFP⁺ (Vav-TRAF6-YFP) cells in LK and LSK populations 30 days after treatment with LPS or vehicle (PBS). Data represent the mean±s.e.m., n = 6 mice per group (*P< 0.05, **P< 0.01). d, Pie charts indicating the relative and absolute frequency of each population (LK and LSK) in the mice treated with LD-LPS as compared to the mice treated with PBS. Data represent the mean, n = 6 mice per group. e, Chimerism of YFP⁺ (Vav-TRAF6-YFP) cells in all CD45.2⁺ (left panel), myeloid (CD11b⁺, middle panel) and lymphoid (B220⁺ and CD3⁺, right panel) in the PB at the indicated time points. Gray bar indicates in vivo LD-LPS treatment. Data represent the mean±s.e.m., n = 5 mice per group (*P = 0.05). f, Representative flow cytometric profiles of YFP⁺ (Vav-TRAF6-YFP) cells in HSC at 12 weeks after primary (upper panel) and tertiary transplantation (lower panel). g, Chimerism of YFP⁺ (Vav-TRAF6-YFP) cells in LK, LSK and HSC in the BM at the indicated time points. Gray bar indicates in vivo LPS treatment. Data represent the mean±s.e.m., n = 5 mice per group (*P< 0.05). Statistical analysis in a,c,e,g was performed by a two-tailed Studenťs t-test.

Fig. 3 |. Overexpression of TRAF6 alters the response of hematopoietic cells to LD-LPS.

a, Proportion of donor-derived CD45.2⁺ cells in HSC cells after primary, secondary and tertiary transplantation. Error bars indicate s.e.m. for n = 5 mice per group. b, Proportion of donor-derived CD45.2⁺ cells in LK and LSK cells after primary, secondary and tertiary transplantation. Data represent the mean ± s.e.m., n = 5 mice per group (*P< 0.05). c, Representative flow cytometric analysis of CD45.2 proportions in LK and LSK cells at 12 weeks after tertiary transplantation. d, Representative flow cytometric profiles of CD11b+ myeloid cells in CD45.2⁺ PB at 4weeks post transplantation. e, Proportion of myeloid (CD11b⁺) cells in the donor-derived compartment (gated on CD45.2⁺) after primary, secondary and tertiary transplantation. Data represent the mean ± s.e.m., n = 5 mice per group (*P< 0.05). Statistical analysis in b,e was performed by a one-tailed Studenťs t-test.

Fig. 4 |. LD-LPS stimulation of TLR-TRAF6-primed HSPC results in noncanonical NF-kB signaling.

a, Heatmap of differentially expressed genes in WT or Vav-TRAF6 LSK cells treated with LPS or PBS (1.5-fold, P< 0.05, n = 3 per group). b, TOPP gene analysis of the top ten enriched pathways in WT and Vav-TRAF6 LSK cells treated with LPS as compared to WT or Vav-TRAF6 control, respectively. c, Mean fluorescence intensity (MFI) of GFP in Lin⁻ BM cells from WT (WT-NF-kB-GFP) and Vav-TRAF6 (Vav-TRAF6-NF-kB-GFP) NF-kB-GFP reporter mice treated in vitro with LPS (1 μgml⁻¹). Results are presented as mean±s.e.m., for n = 3 independently treated samples (*P< 0.05). d, Immunoblotting of c-Kit⁺ BM whole cell lysates (WCL) and nuclear extracts (NEs) isolated from WT or Vav-TRAF6 mice after treatment with LPS (1 μgml⁻¹). Shown is a representative blot from two independent replicates. e, GSEA plots for noncanonical NF-kB (upper panel) and canonical NF-kB (lower panel) pathways in CD34⁺TRAF6^hi MDS BM cells as compared to CD34+TRAF6¹⁰ MDS BM cells. f, Regression analysis between RELB and TRAF6 mRNA expression in CD34⁺ MDS BM cells (R = 0.44; P=6×l0⁻¹⁰). g, Immunoblotting of nuclear extracts from THP1 cells treated with LPS (100 ng ml⁻¹) and/or NIK inhibitor (NIKi). Shown is a representative blot from two independent replicates. h, CD34⁺ del(5q) MDS BM and normal CD34⁺ BM cells treated with NIKi were evaluated for colony formation in methylcellulose. Results are presented as mean±s.e.m., for n = 3 independent technical replicates(*P<0.05, **P< 0.01). Statistical analysis in c,h was performed by a two-tailed Studenťs t-test.

Fig. 5 |. Noncanonical NF-kB activation correlates with A20 expression in TLR-TRAF6 primed HSPCs.

a, Venn diagram of upregulated genes (1.5-fold, P<0.0.5) in CD34⁺TRAF6^hi MDS BM cells (relative to CD34⁺TRAF6¹⁰ MDS BM), LPS-stimulated miR146a^−/− human CD34⁺ BM (relative to miR146a^−/− human CD34⁺ BM treated with PBS and normal CD34⁺ BM±LPS) and LPS-stimulated Vav-TRAF6 LSK cells (relative to Vav-TRAF6 LSK treated with PBS and WT LSK±LPS). b, A20 mRNA expression in TRAF6^hi and TRAF6¹⁰ CD34⁺ MDS BM cells (GSE19429) (***P<0.001). c, A20 mRNA expression in WT and Vav-TRAF6 LSK cells treated with either PBS or LPS (1μgml⁻¹) for 90 min, as determined by RNA sequencing. Results are presented as mean±s.e.m., for n = 3 independently treated samples (*P< 0.05). d, A20 mRNA expression in WT and miR146a^−/− human CD34⁺ BM cells treated with PBS, Pam3CSK4 (1μgml⁻¹), IL-1β (10 ngml⁻¹) or LPS (1μgml⁻¹) for 90 min. Results are presented as mean±s.e.m., for n = 3 independent biological replicates (**P<0.01). e, Immunoblot analysis of A20 and Relb in whole cell lysates (WCL) and nuclear extracts (NEs) from WT and Vav-TRAF6 c-Kit⁺ BM cells after LPS treatment (1μgml⁻¹). f, A20 mRNA expression in c-Kit⁺ BM cells from WT and Vav-TRAF6 mice treated with LD-LPS (1μg g⁻¹) or vehicle for 16 h. Error bars indicate s.e.m. for n = 3 technical replicates from individual mice (*P<0.05). g, Immunoblot analysis of WT and Vav-TRAF6 c-Kit⁺ BM cells from WT and Vav-TRAF6 mice treated with LD-LPS (1μg g⁻¹) or vehicle twice a week for 30 days. Shown below is the relative expression of the indicated proteins. The immunoblot is from three pooled mice per group performed as a single replicate. h, Immunoblotting of THP1 cells expressing TRAF6 or empty vector treated with LPS (100 ng ml⁻¹). Shown is a representative blot from two independent replicates. Statistical analysis in b-d,f was performed by a two-tailed Studenťs t-test. NS, nonsignificant.

Fig. 6 |. A20 expression and noncanonical NF-kB activation occurs in TET2-deficient HSPC.

a, Venn diagram of upregulated genes (1.5-fold, P< 0.0.5) in LPS-stimulated Tet2^−/− LSK (relative to Tet2^−/− LSK with PBS and Tet2^fl/fl LSK±LPS), and LPS-stimulated Vav-TRAF6 LSK cells (relative to Vav-TRAF6 LSK treated with PBS and WT LSK ± LPS). b, A20 mRNA expression in Tet2^fl/fl and Tet2^fl/flVavCre LSK cells treated with either PBS or LPS (1 μgml⁻¹) for 90 min was determined by RNA sequencing. Results are presented as mean±s.e.m., for n = 3 independently treated samples (*P< 0.05, **P<0.01). c, Enrichment analysis of upregulated genes from CD34⁺TRAF6^hi MDS BM cells in Tet2^fl/fl LSK (WT) or Tet2^fl/fl VavCre LSK (Tet2^−/−) treated with LPS (1.5-fold, P< 0.05). d, Enrichment analysis of upregulated genes from canonical (CD40 signaling, NES = 1.56, P = 0.02) and noncanonical (TNFR pathway) NF-kB signaling in Tet2^fl/fl LSK (WT) or Tet2^fl/fl VavCre LSK (Tet2^−/−) treated with LPS (1.5-fold. P<0.05). e, Immunoblotting of c-Kit⁺ BM cells isolated from Tet2^fl/fl (WT) or Tet2^fl/fl VavCre (Tet2^−/−) mice stimulated with LPS (1 μgml⁻¹). Shown is a representative blot from two independent replicates. Below is the relative expression of the indicated proteins. f, Immunoblotting of c-Kit⁺ BM cells isolated from Tet2^fl/flTraf6^fl/fl VavCre (Tet2^−/−Traf6^−/−) or Tet2^fl/fl VavCre (Tet2^−/−) mice stimulated with LPS (1 μgml⁻¹). Below is the relative expression of the indicated proteins. g, A20 mRNA expression in Tet2^fl/fl or Tet2^fl/flVavCre (Tet2) c-Kit⁺ BM cells after 16 h of in vivo LPS (1 μg g⁻¹) treatment. Results are presented as mean±s.e.m., for n = 3 independent replicates, **P< 0.001. h, Immunoblot analysis of A20 in Tet2^fl/fl and Tet2^fl/flVavCre c-kit⁺ BM cells after 30 days of in vivo LPS (1 μg g⁻¹) treatment. The immunoblot is from three pooled mice per group performed as a single replicate. Below is the relative expression of the indicated proteins. Statistical analysis in b was performed by a one-tailed Studenťs t-test. Statistical analysis in g was performed by a two-tailed Studenťs t-test.

Fig. 7 |. Deletion of A20 prevents noncanonical NF-kB signaling and rescues the competitive advantage of TLR-TRAF6-primed HSPCs during LD-LPS.

a, Immunoblotting of THP1 cells expressing shA20 or non-targeting shRNA (shControl) treated with LPS (100 ng ml⁻¹). Shown is a representative blot from two independent replicates. b, Immunoblotting of c-Kit⁺ BM isolated from WT, Vav-TRAF6 RosaCreER and A20^+/− Vav-TRAF6 RosaCreER mice treated with LPS (1 μgml⁻¹). Shown is a representative blot from two independent replicates. Below is the relative expression of the indicated proteins. c, The proportion of PB GFP⁻ cells in chimeric mice with Vav-TRAF6 RosaCreER or A20^+/− Vav-TRAF6 RosaCreER BM cells and WT CD45.2 BM cells (expressing a UBC-GFP transgene, hereafter WT-GFP) after 4 weeks of twice a week treatment with LD-LPS (1 μg g⁻¹) or vehicle. The data shown are after the last LD-LPS treatment as normalized values to each PBS control. Data represent the mean ± s.e.m., n = 5 mice per group, *P< 0.05. d, Percentage of myeloid (CD11b⁺) cells in donor-derived CD45.2 PB cells was determined at 4 weeks postcompetitive transplantation of CD45.2 WT, Vav-TRAF6 RosaCreER and A20^+/− Vav-TRAF6 RosaCreER BM cells isolated from mice treated with LD-LPS (1 μg g⁻¹) or vehicle twice a week for 4 weeks along with CD45.1 competitor BM cells. Data represent the mean ± s.e.m., n = 4 mice per group, *P< 0.05. e, Immunoblotting of MDSL cells expressing shA20 or non-targeting shRNA (shControl) and cDNAs expressing WT or mutant A20. Shown is an immunoblot from a single biological replicate. f, Colony forming potential of MDSL cells expressing shA20 or non-targeting shRNA (shControl) in methylcellulose. Results are presented as mean ±s.d., for n = 3 independent samples. *P< 0.05. Statistical analysis in c was performed by a one-tailed Studenťs t-test. NS, nonsignificant. Statistical analysis in d,f was performed by a two-tailed Studenťs t-test.