An important comorbidity of chronic inflammation is anemia, which may be related to dysregulated activity of hematopoietic stem and progenitor cells (HSPCs) in the bone marrow (BM). Among HSPCs, we found that the receptor for IL-33, ST2, is expressed preferentially and highly on erythroid progenitors. Induction of inflammatory spondyloarthritis in mice increased IL-33 in BM plasma, and IL-33 was required for inflammation-dependent suppression of erythropoiesis in BM. Conversely, administration of IL-33 in healthy mice suppressed erythropoiesis, decreased hemoglobin expression, and caused anemia. Using purified erythroid progenitors in vitro, we show that IL-33 directly inhibited terminal maturation. This effect was dependent on NF-κB activation and associated with altered signaling events downstream of the erythropoietin receptor. Accordingly, IL-33 also suppressed erythropoietin-accelerated erythropoiesis in vivo. These results reveal a role for IL-33 in pathogenesis of anemia during inflammatory disease and define a new target for its treatment.

Introduction

Chronic inflammatory diseases cause local effects in diseased organs, but owing to the systemic effects of inflammation they are frequently associated with important comorbidities in other locations, including the blood and bone marrow (BM; Argollo et al., 2019; Turesson, 2016). This principle is exemplified by patients with rheumatoid arthritis and spondyloarthritis (SpA), a complex of diseases causing chronic inflammation of the axial spine, peripheral joints, and small intestine (Ranganathan et al., 2017), who may also develop anemia of inflammatory disease (AID) or BM dysfunction (Furst et al., 2013; Smith et al., 1992).

AID is an important cause of morbidity in people, limiting physical activity and impairing quality of life in ∼1 million adults in the United States (Nemeth and Ganz, 2014). Its pathogenesis is multifactorial, but important contributors include reduced absorption of iron, decreased secretion of or responsiveness to erythropoietin (EPO), and suppressive effects of inflammatory cytokines on erythroid progenitors in BM (Nemeth and Ganz, 2014; Weiss and Goodnough, 2005).

Contrary to their traditional image as isolated cells in secluded niches, hematopoietic stem and progenitor cells (HSPCs), including erythroid progenitors, express receptors for inflammatory cytokines, rendering them highly responsive to signals emanating from inflamed sites during disease (King and Goodell, 2011; Takizawa et al., 2012). Much attention has focused on the lineage-instructive effects of these cytokines on multipotent HSPCs, with both IL-1β (Pietras et al., 2016) and TNF-α (Etzrodt et al., 2019) causing early adoption of myeloid transcriptional signatures in primitive stem cells. However, cytokines may also deliver permissive or inhibitory signals to progenitors that have already committed to one lineage, modulating hematopoietic output to meet demand. In previous studies, IL-6 (McCranor et al., 2014), TNF-α (Xiao et al., 2002), and IFN-γ (Felli et al., 2005; Libregts et al., 2011) have exerted suppressive effects on erythropoiesis in various settings, but the relative importance of these and other as-yet-unknown factors in causing AID, and the progenitor stages on which they act, remain unclear.

The pathogenesis of SpA, which has been elucidated partly by studies of the murine SKG model (Benham et al., 2014; Ruutu et al., 2012), depends on contributions from the innate and adaptive immune systems, coordinated by a network of proinflammatory cytokines, including TNF-α, IL-6, IL-17, and IL-23 (Benham et al., 2014; Hata et al., 2004). To understand the lineage-specific effects and importance of these and other cytokines in the development of AID during SpA, we compared expression of cytokine receptors between myeloid and erythroid progenitors, finding that the receptor for IL-33 is expressed preferentially among the latter. IL-33 is a member of the IL-1 family of cytokines released from fibroblasts and endothelial and epithelial cells at barrier surfaces in response to damage (Liew et al., 2016). Its receptor is composed of two components: the specific chain ST2 and the IL-1 receptor accessory protein, shared with other IL-1 family members. IL-33 activates several types of ST2+ mature hematopoietic cell but, aside from a role in eosinopoiesis (Johnston et al., 2016), its impact on HSPCs has received little attention.

In this study, we find that IL-33 has an important and novel role in suppressing erythropoiesis in a mouse model of chronic inflammatory SpA and could recreate features of AID when injected in vivo. We demonstrate that IL-33 exerts a direct effect on erythroid progenitors in vitro to inhibit differentiation into mature RBCs, contingent on activation of NF-κB, and inhibits EPO-induced signaling pathways downstream of the EPO receptor (EPO-R) in a mechanism unrelated to those of previously described suppressors of erythropoiesis.

Results

SpA perturbs hematopoiesis and causes AID

A clinical phenotype of chronic disease resembling SpA develops in SKG mice injected with the bacterial cell wall extract curdlan, a β-1,3-glucan that causes release of inflammatory cytokines from innate cells after binding to the receptor dectin-1 (Kim et al., 2016; Ruutu et al., 2012; Fig. 1 A). Over a period of 4 wk following injection of curdlan, such mice developed swelling of the paws and tarsal joints (ankles; Fig. 1, B and C), with associated weight loss (Fig. 1 D). The arthritis was associated with accumulation of mature (CD11b+Ly6G+) neutrophils in paws (Fig. 1 E), with the severity of infiltration correlating with clinical score, the extent of paw and tarsal swelling, and weight loss (Fig. 1 F and Fig. S1 A).

Patients with SpA have increased blood neutrophil counts (Mercan et al., 2016), and 15% have anemia at the time of diagnosis (Furst et al., 2013). We observed a similar reduction in blood hemoglobin (Hgb) concentration and RBC count in mice with SpA (Fig. 1 G). The Hgb concentration was not associated with changes in bodyweight (Fig. S1 B) but was unexpectedly positively correlated with change in paw size (Fig. 1 H). This latter finding could be related to development of dehydration in mice with more severe disease, spuriously increasing Hgb concentrations, or to concurrent inflammation of the small intestine (enteritis), a common feature of SpA in people (De Vos et al., 1996), which seemed to be associated with more severe anemia but milder paw swelling (Fig. 1 H).

These hematologic changes prompted us to ask whether there were associated perturbations in hematopoiesis in BM. Accordingly, when we plated BM cells in methylcellulose medium, we observed a lesser number of erythroid colonies (blast-forming units–erythroid [BFU-Es]) derived from cells of SpA mice after 10 d (Fig. 1 I) compared with healthy mice. Further characterization of BM HSPCs by flow cytometry using previously published definitions (Pietras et al., 2015; Pronk et al., 2007) revealed expansion of multipotent progenitor (MPP) and primitive hematopoietic stem cell (HSC) populations (Fig. S1 C), as we described previously (Regan-Komito et al., 2020). Downstream of these HSPCs, there was a significant decrease in the proportion of erythroid progenitors (CFU-erythroid [CFU-Es]) but a significant increase in myeloid progenitors (granulocyte macrophage progenitors [GMPs]) among the pool of lineagenegcKit+Sca1neg progenitors (Fig. 1 J) and total live BM cells (Fig. 1 K). These changes were mirrored by a significant reduction in the absolute number of CFU-Es and a significant increase in GMPs (Fig. S1 D), because the total number of BM cells was unchanged (Fig. S1 E). Changes in progenitor populations were reflected in the mature cells of the BM, with a decreased proportion expressing the mature erythroid marker Ter119 in SpA, whereas the frequency of mature neutrophils was increased (Fig. 1, L and M; and Fig. S1 F).

To investigate changes in erythropoiesis further, we calculated the number of cells in the BM at successive stages of erythroid differentiation, beginning with committed progenitors (CFU-Es) and progressing to nucleated erythroblasts defined by characteristic surface markers (Fig. S1 G). When comparing these erythroid differentiation pathways for healthy mice and those with SpA, we confirmed that the rate and extent of erythroid differentiation were impaired in mice with chronic inflammation across early and late stages (Fig. 1 N).

With development of SpA, we also observed accumulation of erythroid progenitors in the spleen of inflamed mice (Fig. S1 H), but this was insufficient to prevent development of anemia (Fig. 1 G), indicating the functional output of this population did not compensate for loss of erythroid progenitors in the BM.

Taken together, these observations show that disease in SKG mice is associated with increased myelopoiesis and decreased erythropoiesis, resulting in similar hematologic changes to those described in people with SpA. To ensure these observations were not attributable to the mutation in Zap70 that predisposes SKG mice to autoimmune responses, we confirmed that inflammation induced by intraperitoneal (IP) injection of curdlan in WT C57BL/6 mice produces the same hematopoietic phenotype after 1 wk (i.e., increased GMP and decreased CFU-E percentages and absolute numbers; Fig. S1, I and J) but does not induce clinical arthritis (data not shown). Whereas curdlan injection produced a similar inhibitory effect on the BM erythroid differentiation pathway (Fig. S1 K), the blood Hgb concentration was not decreased (Fig. S1 L), presumably because RBCs have a longer circulating half-life of ∼15–20 d in blood (Amireault et al., 2011; Van Putten, 1958) than the duration of these experiments in WT mice.

The receptor for IL-33, ST2, is expressed preferentially and at high levels on erythroid progenitors

AID is a syndrome of multifactorial etiology that accompanies several diseases, including SpA, rheumatoid arthritis, and inflammatory bowel disease (Nemeth and Ganz, 2014). Inflammatory cytokines may contribute to this phenomenon by altering lineage commitment of multipotent stem cells. However, we hypothesized that cytokines may have differential inhibitory effects on terminal differentiation of lineage-committed progenitors, limiting the output of some progenitors in the context of disease. To investigate this, we compared the surface expression of a panel of inflammatory cytokine receptors between committed myeloid GMP and erythroid CFU-E progenitors (Fig. 2 A). The receptor for the prototypical myeloid cytokine GM-CSF was preferentially expressed on GMP, but we were surprised to find the receptor for IL-33, ST2, at much higher levels on CFU-E (Fig. 2 A). Differential expression of receptor proteins was reflected in the transcriptome, with the Il1rl1 gene (encoding ST2) being the only major proinflammatory cytokine receptor gene expressed at higher levels in CFU-Es compared with GMPs (Fig. 2 B and Fig. S2, A and B; Choi et al., 2019). We excluded the possibility of nonspecific staining in mice lacking the Il1rl1 gene (ST2−/−; Fig. 2 C) and further showed that ST2 expression is highest on erythroid progenitors immediately after commitment to the lineage (i.e., pre–CFU-E and CFU-E) but decreases with progressive maturation (Fig. 2 D and Fig. S2 C), suggesting any effect of IL-33 in the erythroid lineage must be exerted while cells are differentiating. Importantly, the receptor was not expressed on primitive HSCs or MPPs upstream of the erythroid lineage (Fig. S2 D), reducing the likelihood of a lineage-instructive effect.

IL-33 inhibits differentiation of erythroid progenitors

With the receptor ST2 expressed on erythroid progenitors, we hypothesized that IL-33 would exert a direct effect on these cells that might promote development of AID. To investigate this, we optimized a culture system by sorting progenitors at the earliest stage after commitment to the erythroid lineage (pre–CFU-E) by FACS from adult BM. We cultured these cells in proerythroid medium supplemented with stem cell factor and EPO, with or without IL-33 or TNF-α, for 4 d before assessing cells according to their size and expression of Ter119. With progressive differentiation, these pre–CFU-Es become smaller, up-regulate and then down-regulate the transferrin receptor CD71, lose cKit and CD105, and express Ter119 (Fig. S3 A; Chen et al., 2009; Pop et al., 2010; Tusi et al., 2018). In control conditions and with TNF-α, the majority of cells were small and had little granularity, but, in the presence of IL-33, a greater proportion of cells kept a larger, blast-like profile (Fig. 3 A), as observed in primary BM pre–CFU-Es (Fig. S3 A). Similarly, IL-33 caused a striking reduction in the proportion of cells expressing Ter119 and, even among the less mature CD71+Ter119 cells, IL-33 also prevented down-regulation of the surface receptor cKit, which is normally lost with maturation (Fig. 3, A and B; and Fig. S3 A). In the terminal stages of maturation, the nucleus is extruded from progenitors to form reticulocytes. In our culture system, addition of IL-33 decreased the number of these enucleated cells observed cytologically (Fig. 3 C) and by staining for nuclear material in Ter119+ cells by flow cytometry (Fig. 3 D).

HSPCs can secrete cytokines themselves in response to pathogen derivatives and cytokines (Zhao et al., 2014), possibly superseding the effects of the original stimulus by autocrine signaling. To evaluate whether the observed effects of IL-33 in vitro were dependent on other factors, we cultured pre–CFU-Es from mice expressing GFP under the control of the ubiquitin c promoter (Ub-GFP), which is still expressed in maturing RBCs, and from ST2−/− mice, alone or in a 1:1 mixture. As expected, IL-33 had no effect on the proportion of Ter119+ cells derived from ST2−/− cells when cultured alone (Fig. 3 E). Importantly however, the frequency of Ter119+ cells was also unchanged in ST2−/− cells in mixed cultures with Ub-GFP cells that were affected by IL-33, indicating this was a direct effect mediated through the ST2 receptor (Fig. 3 E).

Addition of IL-33 did not alter total cell yield from culture (Fig. S3 B) or the proportion of cells expressing markers of proliferation (Ki67; Fig. S3 C) or apoptosis (annexin V; Fig. S3 D), leading us to suspect it was primarily suppressing the differentiation of erythroid progenitors, as also suggested by the difference in erythroid differentiation pathways (Fig. S3 E). To test this, we separated the effects of IL-33 on proliferation and differentiation by sorting single pre–CFU-E cells and culturing them for 10 d before assessing the output by flow cytometry. We found the total cell yield was similar, but the proportion of Ter119+ cells was decreased by addition of IL-33, as it was in bulk cultures, confirming a primary effect on erythroid differentiation (Fig. 3 F). Functionally, this effect was recapitulated by decreased formation of CFU-E colonies when FACS-sorted pre–CFU-Es were plated in methylcellulose medium with IL-33 (Fig. 3 G).

Finally, we asked whether this effect of IL-33 on erythroid progenitors was common to other species by isolating CD34+CD71+ progenitors (Mori et al., 2015) from human blood leukoreduction cones. When culturing these cells for 8–9 d in similar conditions, inclusion of IL-33 decreased the frequency of mature erythroid cells expressing glycophorin A (CD235a, similar to murine Ter119; Kina et al., 2000) by approximately twofold (Fig. 3 H).

Collectively, these observations show that IL-33 suppresses differentiation of erythroid progenitors in a direct and cell-intrinsic manner, in a mechanism that appears to be common to mice and humans.

IL-33 is increased in the BM niche with inflammation

Having detected expression of its receptor at high levels on erythroid progenitors, we asked whether IL-33 had a role in erythropoiesis in healthy mice. However, mice lacking the Il33 gene had normal Hgb concentrations compared with WT counterparts (Fig. 4 A) and similar proportions of lineage-committed progenitors (Fig. 4 B). To establish whether IL-33 is produced in the BM niche, we sorted major populations of hematopoietic and stromal cells by FACS using published definitions (Chow et al., 2013; Pinho et al., 2013; Schepers et al., 2013) and found the Il33 gene, unlike the gene encoding the related cytokine IL-1β, was preferentially expressed in CD169+F4/80+ macrophages, previously described as erythroblastic island macrophages that maintain close physical contact with erythroid progenitors in the BM and contribute to their retention and differentiation (Fig. 4 C; Chow et al., 2013). Using mice expressing a citrine reporter in place of the Il33 gene, we found these CD169+ macrophages were the most abundant IL-33 producers in the BM (Fig. 4 D), exceeding the numbers of IL-33+ Ter119+ and CD45 stromal cells described previously (Kenswil et al., 2018; Wei et al., 2015).

Collectively, these observations show that IL-33 is produced in healthy mice within the BM niche inhabited by erythroid progenitors but is dispensable for maintenance of a normal erythroid compartment. However, after injecting curdlan to trigger SpA, the concentration of IL-33 measured in BM plasma increased rapidly (Fig. 4 E), whereas its expression in CD169+ macrophages decreased in mice with SpA (Fig. 4 F). This suggests increased IL-33 production in inflamed tissues, which we described previously in the SKG model (Regan-Komito et al., 2020), may supersede local production in the BM niche, leading us to hypothesize that IL-33 could be a factor regulating erythropoiesis after curdlan injection.

IL-33 is required for curdlan-induced suppression of erythropoiesis

IL-33 suppressed differentiation of erythroid progenitors in vitro and was increased in BM after induction of inflammation. Against this backdrop, we investigated the functional role of IL-33 in AID by injecting curdlan in mice lacking the Il33 gene (IL-33−/−) and matched WT C57BL/6 controls (Fig. 5 A). Whereas the frequency of erythroid progenitors decreased significantly in WT mice with this treatment, there was no significant reduction in CFU-Es in IL-33−/− mice receiving either PBS or curdlan when results from individual mice were pooled or compared across independent experiments (Fig. 5 B), with a less striking difference in the number of cells (Fig. S4 A). Conversely, the frequency of myeloid GMPs increased in both strains with induction of inflammation (Fig. 5 B and Fig. S4 A), indicating the effect of IL-33 was specific to the erythroid lineage. As in SKG mice, curdlan caused expansion of the erythroid progenitor pool in the spleens of mice of both strains (Fig. S4 B), but we could not determine the functional impact of this extramedullary hematopoiesis because the blood Hgb concentration was unaltered in either strain (Fig. S4 C), presumably because the length of the experiment was again shorter than the RBC half-life.

Having shown it was necessary for curdlan-induced suppression of BM erythropoiesis, we next asked whether administration of IL-33 alone would be sufficient to produce the same hematopoietic changes as curdlan. We therefore injected SKG mice with IL-33 over 1 wk (Fig. 5 C); this treatment caused marked reductions in the frequency and absolute number of erythroid progenitors to a similar extent as observed with curdlan injection (Fig. 5 D and Fig. S4 D). Notably, divergence between erythroid differentiation pathways of healthy and IL-33–injected mice caused a more marked deficiency in more mature Ter119+ cells (Fig. 5 E), as was apparent in cultures of primary cells exposed to IL-33 (Fig. 3 A and Fig. S3 E). Changes in BM cell numbers were also reflected in functional differences in colony formation, with an increased number of myeloid (CFU-granulocyte macrophage) and decreased number of erythroid (BFU-E) colonies observed with addition of IL-33 to methylcellulose medium (Fig. 5 F).

In mast cells and eosinophils, IL-33 causes production of other proinflammatory cytokines that may suppress erythropoiesis, including TNF-α and IL-6 (Lunderius-Andersson et al., 2012). Indeed, we found the concentration of TNF-α, but not IL-6, was increased in BM plasma of mice injected with IL-33 (Fig. S4 E). To determine whether these cytokines were required, we injected IL-33 in mice lacking IL-6 or the receptors for TNF-α (TNFRI and TNFRII), but, in all three strains, IL-33 produced similar suppression of erythropoiesis (Fig. S4, F–H), indicating they were not required to produce those effects.

At a cellular level, there was a striking reduction in expression of the genes encoding the adult globin chains (Hbb-b1 and Hbb-a1), forming the proteinaceous component of Hgb, in FACS-sorted pre–CFU-Es from mice exposed to IL-33 for 1 wk compared with controls (Fig. 5 G). There were similar reductions in expression of other genes required for production of mature RBCs, including those encoding the heme enzyme aminolevulinic acid synthase (Alas2) and the major surface proteins glycophorin A (Gypa), erythroid band protein 4.1 (Epb41), and aquaporin 1 (Aqp1; Fig. 5 G). With prolonged exposure to IL-33 over 4 wk (Fig. 5 H), these changes in cellular activity resulted in a decreased frequency of Ter119+ erythroid cells in BM (Fig. 5 I) and decreased Hgb concentrations and RBC counts in peripheral blood (Fig. 5 J). These changes were dependent on inhibition of BM erythropoiesis, because there was no significant change in the frequency of erythroid progenitors in the spleen (Fig. S4 I).

Taken together, these observations demonstrate that IL-33 is both necessary and sufficient for the full extent of curdlan-induced suppression of erythropoiesis. Remarkably, injection of IL-33 alone decreased expression of globin and heme synthesis enzyme genes, resulting in decreased blood Hgb with prolonged exposure.

The effects of IL-33 in erythroid progenitors are dependent on NF-κB

To explore the transcriptomic events occurring after erythroid progenitors are exposed to IL-33, we sorted pre–CFU-Es from healthy mice and cultured them for 6 h in vitro before performing bulk RNA-sequencing analysis (Fig. 6 A). Control and IL-33–exposed pre–CFU-Es clustered separately based on differential expression of 482 genes, of which 388 were up-regulated and 94 down-regulated (Fig. 6 B and Fig. S5 A). Among those genes down-regulated were several with known roles in erythroid differentiation, including vimentin (Vim) required for enucleation (Xue et al., 1997) and the enzyme coproporphyrinogen III oxidase (Cpox) required for Hgb synthesis (Taketani et al., 2001), whereas CD69, which inhibits erythroid differentiation (Huang et al., 2018), was up-regulated (Fig. 6 B and Fig. S5 B).

Among the genes strikingly up-regulated by IL-33 were many members of the NF-κB pathway (Fig. 6, B and C), which transmits IL-33 signals in other cell types (Pinto et al., 2018). Members of both canonical and noncanonical pathways were up-regulated (Fig. 6 D), and we confirmed that IL-33 increased nuclear translocation of the p65 subunit (encoded by Rela) in erythroid progenitors by approximately twofold using imaging flow cytometry (Fig. 6, E and F). To define its functional role, we inhibited the canonical NF-κB pathway using a NEMO domain–binding (NBD) peptide that inhibits IκB kinase γ (Fig. 6 G). In cultures of sorted pre–CFU-Es, inclusion of this peptide inhibited the effect of IL-33, alleviating its suppression of development of Ter119+ cells by ∼50% (Fig. 6 H). Collectively, these observations confirmed that activation of NF-κB signaling, which normally decreases with progressive erythroid maturation (Zhang et al., 1998), was required for the suppressive action of IL-33.

IL-33 inhibits phosphorylation events downstream of the EPO-R, producing resistance to EPO-stimulated erythropoiesis

Analyzing the genes down-regulated by IL-33 in sorted pre–CFU-Es in more detail, we observed considerable overlap between these and a group of genes up-regulated in equivalent progenitors by injection of EPO in mice in a previous study (Fig. 7, A and B; Tusi et al., 2018). These genes were enriched in IL-33–naive pre–CFU-Es in our analysis (Fig. 7 A), leading us to hypothesize that IL-33 might interfere with the action of EPO in these cells. Exposure to IL-33 did not cause down-regulation of the Epor gene, encoding the EPO-R (Fig. 7 C), or loss of the protein from the cell surface (Fig. S5 C), nor did use of a 10-fold greater concentration of EPO rescue IL-33–induced inhibition in culture (Fig. 7 D). Collectively, this suggested any effect of IL-33 on EPO-R signaling would be downstream of the surface receptor itself.

Upon binding to EPO, the cell surface EPO-R dimerizes and causes activation of several intracellular signaling cascades, including JAK2-STAT5, PI3K-ERK1/2, p38 MAPK, and PI3K-Akt (Watowich, 2011). Of these, JAK2-STAT5 activates Bcl-xL to prevent apoptosis of erythroid progenitors (Silva et al., 1999), whereas Akt promotes differentiation (Ghaffari et al., 2006) by phosphorylating the essential erythroid transcription factor GATA1 (Zhao et al., 2006) and activating metabolic pathways via phosphorylation of the mTORC1 complex (Meng et al., 2018). To assess the effect of IL-33 on these pathways, we incubated lineageneg BM preparations with IL-33 or medium and then stimulated them with EPO before assessing phosphorylation of critical molecules in CD105+ erythroid progenitors (i.e., CFU-Es and pre–CFU-Es) by phosphoflow cytometry. In control conditions, EPO activated all four pathways downstream of EPO-R but, with prior incubation with IL-33, there was consistent suppression of STAT5, ERK1/2, and, more strikingly, Akt phosphorylation in response to EPO (Fig. 7, E and F). Among the substrates of Akt is the mTORC1 complex, which phosphorylates S6 kinase and other targets to enhance protein translation and cellular metabolism (Meng et al., 2018). Activity of the mTORC1/S6 pathway is increased with progressive erythroid differentiation (Liu et al., 2017), and, in murine erythroid progenitors, we found increased phosphorylation of mTOR and S6 kinase with EPO stimulation (Fig. S5 D). However, phosphorylation of neither molecule was affected by IL-33 exposure (Fig. S5 D), indicating IL-33–induced suppression of Akt phosphorylation did not impair EPO-induced mTORCI/S6 kinase activation.

Inflammation has been implicated as a factor causing resistance to therapeutic administration of EPO in patients with chronic anemia (Macdougall and Cooper, 2002); we hypothesized that IL-33 could contribute to this effect owing to its impact on EPO-R signaling. We therefore injected EPO in mice to produce a model of accelerated erythropoiesis (Fig. 7 G), which occurs during severe anemia to increase production of RBCs. Whereas injection of EPO alone caused marked increases in the numbers of BM erythroid progenitors (Fig. 7 H), BM Ter119+ erythroid cells (Fig. 7 I), erythroid differentiation pathway (Fig. 7 J), and blood Hgb (Fig. 7 K), concurrent administration of IL-33 largely abrogated these responses. To understand the relevance of this finding in the setting of inflammation, we injected EPO, with or without curdlan, in WT mice (Fig. S5 E), finding that curdlan clearly suppressed the EPO-induced expansion of erythroid progenitors and Ter119+ cells in BM (Fig. S5, F and G). However, this effect was partially inhibited in IL-33−/− mice, indicating IL-33 production contributed to the EPO resistance produced by curdlan (Fig. S5, F and G).

Collectively, these results show that IL-33 inhibits EPO-induced erythropoiesis in vitro and in vivo, which may be attributable to decreased activity in the Akt and, to a lesser extent, ERK1/2 and STAT5 signaling pathways. These observations provide a mechanism for the direct inhibition of differentiation of erythroid progenitors caused by IL-33 in the context of chronic inflammation.

Discussion

Our study reveals a previously unknown role for IL-33 in specific suppression of murine and human erythropoiesis in vitro and shows this suppression is necessary and sufficient to cause AID and EPO resistance in experimental models of chronic inflammation (Fig. 8). Mechanistically, we show these effects are dependent on NF-κB activation and associated with decreased responsiveness of signaling pathways downstream of EPO-R.

We found the effects of IL-33 in erythroid progenitors are dependent on NF-κB activation, and NF-κB is a common pathway used by IL-1 cytokine family members (Fields et al., 2019). However, in the context of erythropoiesis, this is important because activity of many NF-κB constituents, including p65, p50, and p52, decreases with progressive differentiation (Zhang et al., 1998). Studies of human and murine erythroid cell lines suggest high levels of NF-κB activity repress transcription at important loci required in differentiated cells, including the globin genes (Liu et al., 2003). Based on this and our own observation that NF-κB inhibition rescued cells from IL-33–mediated inhibition in vitro, we suggest IL-33 maintains inappropriately increased levels of NF-κB activity that prevent terminal differentiation of erythroid progenitors.

Building on this observation, we show that IL-33 suppresses EPO-induced Akt phosphorylation. This is important because EPO has a central and indispensable role in supporting erythroid cells during the final stages of differentiation, to the extent that EPO- and EPO-R–deficient mice suffer embryonic lethality (Wu et al., 1995). Interestingly, the onset of EPO dependence occurs shortly after formation of CFU-Es, coinciding with point at which IL-33 prevented further differentiation in our cultures. Among its pleiotropic effects on erythroid cells, EPO-induced Akt activation promotes differentiation because Akt phosphorylates and activates GATA1 (Zhao et al., 2006), providing a possible explanation for our observation that inhibition of Akt activity is associated with truncated differentiation. Interestingly, IL-33 had little or no effect on p38 MAPK, STAT5, or ERK1/2 phosphorylation, which appears to reconcile with the lack of effect of IL-33 on Ki67 and annexin V expression on pre–CFU-Es in vitro.

We confirmed IL-33 promoted myelopoiesis when administered alone (Regan-Komito et al., 2020; Talabot-Ayer et al., 2015) but was dispensable for curdlan-induced expansion of the GMP pool, suggesting it acts indirectly or alongside other cytokines to produce this effect. Teleologically, we suggest the simultaneous effects of IL-33 to permit myelopoiesis and suppress erythropoiesis could enable organisms to respond to transient damage or infection, taking advantage of the long circulating time of mature RBCs to provide a buffer against the development of anemia while BM output is diverted to meet other demands. During inflammatory diseases, this process would become detrimental if suppression of erythropoiesis were prolonged, resulting in AID as RBCs were lost but not replaced.

With curdlan injection, we show the concentration of IL-33 increased in BM plasma, an effect that could be attributable to local production in the BM niche or transport of IL-33 in the bloodstream from inflamed sites. We recently showed that development of SpA in SKG mice is associated with increased IL-33 production in the inflamed paws and small intestine (Regan-Komito et al., 2020), and others have found IL-33 is increased in the serum of patients with ankylosing spondylitis (Han et al., 2011; Li et al., 2013), the most common form of SpA in people. This suggests local production of IL-33 in the BM niche could be superseded by distant production in inflamed tissues during disease, which would reconcile with decreased expression of Il33 in CD169+ macrophages in mice developing SpA.

Completion of further experiments for this study has been limited by the pandemic of SARS-CoV-2. Although curdlan injection in WT mice produced a similar hematopoietic phenotype to that seen in SKG mice with SpA, we would ideally have injected SKG mice with a neutralizing antibody against IL-33 as they developed SpA to assess the contribution of IL-33 to AID and EPO resistance in the context of inflammatory arthritis. Additionally, we would have investigated the kinetics of IL-33 production in more detail during disease by injecting citrine reporter mice with curdlan and by measuring the concentration of the soluble ST2 receptor, which sequesters IL-33 and curtails its activity (Liew et al., 2016), in serum and BM plasma as SpA developed in SKG mice.

Collectively, our results highlight IL-33 as a novel contributor to AID through its effects on differentiation of erythroid cells. Consequently, IL-33 may represent a therapeutic target for patients with AID or EPO resistance for amelioration of an important comorbidity of chronic inflammation.

Materials and methods

Mice and models of disease

All mice were bred and maintained in specific pathogen-free conditions at the University of Oxford. Experiments were conducted with appropriate licenses issued under the UK Animal (Scientific Procedures) Act 1986 and with approval from the University of Oxford animal welfare ethical review board. C57BL/6Nq-p55−/− and C57BL/6Nq-p75−/− mice were provided by the laboratory of Professor Richard Williams (University of Oxford), SKG mice were provided by Professor Shimon Sakaguchi (Osaka University, Suita, Japan), C57BL/6-Il1rl1−/− mice were provided by Professor Padraic Fallon (Trinity College Dublin, Dublin, Ireland), and BALB/C-Il33cit/+ mice were provided by Professor Andrew McKenzie (University of Cambridge, Cambridge, UK). The latter mice were backcrossed onto the C57BL/6 background until the progeny were at least 99.9% pure based on single-nucleotide polymorphism–based genotyping, and the mice were intercrossed to generate homozygous Il33cit/cit mice. WT C57BL/6, C57BL/6-Il6−/−, and C57BL/6-UbcGFP mice were purchased from The Jackson Laboratory and bred at the University of Oxford.

SKG mice (BALB/C-Zap70W163C) are predisposed to development of autoimmune responses, owing to a spontaneous mutation in the gene encoding the ZAP-70 tyrosine kinase that forms part of the T cell receptor signaling complex. These mice are healthy under specific pathogen–free conditions but develop inflammatory arthritis, enteritis, uveitis, and dermatitis 4–5 wk after IP injection of 0.3 mg of the β-1,3-glucan curdlan, derived from Alcaligenes faecalis (Wako Chemicals). Mice of a single sex were used in each experiment.

After injection of curdlan (0.3 mg IP) in SKG mice, development of arthritis was assessed by weekly measurements of the paws and tarsi using calipers and using the following clinical score: 0 = no swelling or redness; 0.5 = swelling or redness of digits and mild swelling and/or redness of carpi or tarsi; 1 = mild swelling of carpal and tarsal joints; 2 = moderate swelling of carpal and tarsal joints; and 3 = substantial swelling of carpal and tarsal joints not inhibiting normal mobility, feeding, and drinking. Scores for forepaws and hindpaws were summated for each mouse. Changes in caliper measurements of the paws and tarsi were expressed as percentage changes compared with measurements made on the day of curdlan injection.

Where indicated, male C57BL/6 (WT) and C57BL/6-Il33cit/cit (expressing a citrine reporter in place of the Il33 gene, IL-33−/−) were also injected with the same dose of curdlan and culled after 1 wk.

Where indicated, male SKG mice were injected with PBS or curdlan, followed by IP injections of 0.5 µg recombinant murine IL-33 (BioLegend) according to one of the following schedules: (1) one injection every 2 d for a total of four injections, with mice culled the day after the last injection; or (2) one injection twice weekly for 4 wk. C57BL/6-Il6−/−, C57BL/6Nq-p55−/−, and p75−/− mice were also injected with IL-33 or PBS according to schedule 1.

Where indicated, female SKG mice were injected with 0.5 µg IL-33 or PBS every 2 d for six injections. In some of these mice, EPO (BioLegend) was injected (1,875 U/injection, ∼75,000 U/kg, IP) every 2 d for five injections, starting after two injections of IL-33 had been administered. IL-33 and EPO were administered on alternate days thereafter until the mice were culled. Where indicated WT or IL-33−/− mice were injected with PBS or curdlan (0.3 mg IP) and then EPO (1,875 U/injection every 2 d for four injections), culling the day after the last dose of EPO was injected.

Isolation of cells from BM for analysis by flow cytometry

Femurs and tibiae were harvested into PBS supplemented with 0.1% BSA. BM cell suspensions were prepared by flushing the marrow from the central cavities of the femur and tibia using 5 ml PBS/0.1% BSA, 25G needles, and 1-ml syringes before breaking clumps by pipetting 10–20 times with a P1000 pipette and passing through a 70-µm filter.

Isolation of cells from spleen for analysis by flow cytometry

The whole spleen was harvested into PBS/0.1% BSA. A cell suspension was prepared by crushing the spleen through a 70-µm filter using the barrel of a 5-ml syringe and 10 ml PBS/0.1%. The suspension was centrifuged, resuspended with 2 ml RBC lysing buffer (Sigma-Aldrich) for 2 min, and then washed, gross clumps were removed, and resuspended in PBS/0.1% BSA for staining.

Isolation of cells from the soft tissue of the paw for analysis by flow cytometry

The skin was removed from one hind paw and the paw was cut into four pieces. Pieces of tissue were incubated in 1 ml RPMI 1640 supplemented with DNase I (0.1 mg/ml; Sigma-Aldrich) and Liberase (0.3 mg/ml; Roche) for 90 min at 37°C. Digested soft tissue from the paw joint was then passed through a 70-µm strainer to obtain a single-cell suspension, which was stained for flow cytometry.

Flow cytometry

Single-cell suspensions of desired tissues were obtained as described above. 1–2 × 106 cells were stained with fixable viability dye (Zombie Aqua or Zombie Green; BioLegend), unlabeled anti-CD16/32 (BioLegend) to block nonspecific staining (unless anti-CD16/32 was in the staining panel), and surface antibodies in PBS with 0.1% BSA for 30 min in the dark at 4°C. Cells were centrifuged and fixed for 15 min using fixation solution (BD Cytofix/Cytoperm buffer) at 4°C. Cells were centrifuged and resuspended in PBS with 0.1% BSA and 2 mM EDTA. The following anti-mouse antibodies were used (all from BioLegend, unless otherwise stated): anti-CD45 AF700 clone 30-F11, anti-CD11b BV510 or peridin-chlorophyll protein (PerCP)/Cy5.5 clone M1/70, anti-Ly6G BV421 clone 1A8, anti-CD3 PerCP/Cy5.5 clone 17A2, anti-CD4 PerCP/Cy5.5 clone GK1.5, anti-Gr1 PerCP/Cy5.5 clone RB6-8C5, anti-B220 PerCP/Cy5.5 clone RA3-6B2, anti-Ter119 allophycocyanin (APC) clone Ter119, anti-CD71 PE or FITC clone RI7217, anti-CD16/32 APC/Cy7 or PE/Cy7 BD clone 2.4G2, anti-CD150 BV605 clone TC15-12F12.2, anti-Conceptualization CD117 BV786 or PE BD clone 2B8, anti-CD41 PE/Cy7 clone MWReg30, anti-CD105 Pacific blue or APC clone MJ17/18, anti-Sca1 BV711 or Pacific blue clone D7, anti-IFNγR α chain biotin clone 2E2, anti-ST2 PE clone DIH9, anti-TNFRI/p55 PE clone 55R-170, anti-TNFRII/p75 AF647 AbD Serotec clone MCA351A648T, anti-GMCSFRa APC R&D Systems clone 698423, anti-CD169 APC clone 3D6.112, anti-F4/80 PE clone BM8, anti-Ly6C BV711 clone HK1.4, anti-Gr1 FITC eBioscience clone RB6-8C5, anti-CD16/32 APC clone 93, and anti-NF-κB p65 PE Santa Cruz Biotechnology clone F-6. The following anti-human antibodies were used (all from BioLegend): anti-CD3 biotin clone DKT3, anti-CD71 PE/Cy7 clone CY1G4, anti-CD19 biotin clone HIB19, anti-CD14 biotin clone 63D3, anti-CD235a APC clone HI264, anti-CD123 APC clone 6H6, and anti-CD34 FITC clone 561.

For Ki67 staining, cells were stained as above then fixed and permeabilized using the FoxP3/transcription factor staining buffer (eBioscience) set according to the instructions provided. Cells were then washed and stained for Ki67 (anti-Ki67 PE clone 16A8, 5 µl per sample) in 1× permeabilization buffer for 45 min in the dark at 4°C before washing and resuspending in PBS/0.1% BSA and 2 mM EDTA for acquisition.

For annexin V staining, cells were stained as above then resuspended in annexin V binding buffer (BioLegend) for 15 min in the dark at 4°C. Cells were washed and stained with PE-conjugated annexin V (5 µl per sample; BioLegend) for 45 min in the dark at 4°C, before washing and resuspending in binding buffer for acquisition.

Samples were acquired on an LSR II or LSR-Fortessa (BD) and analyzed using FlowJo Software.

FACS

BM suspensions were prepared as described above, and lineage-positive cells were depleted using a lineage depletion kit (eBioscience mouse hematopoietic lineage biotin panel; ThermoFisher Scientific) according to the manufacturer’s instructions. Briefly, BM derived from the femurs and tibiae of three mice was incubated with biotinylated antibodies against CD3, B220, Gr1, and Ter119 in a volume of 1 ml PBS/0.1% BSA for 10 min at 4°C. Cells were then washed and incubated with 0.4 ml streptavidin bound to magnetic beads (Magnisort beads; ThermoFisher Scientific) in a volume of 2 ml PBS/0.1% BSA in a FACS tube for 10 min at room temperature. The volume was then made up to 4 ml, and the FACS tube was placed inside a magnet for 10 min at room temperature. The fraction of cells not bound to the magnet was poured out and stained for FACS as described above.

Pre–CFU-Es and CFU-Es were sorted by FACS using a FACS ARIA III (BD) with 70-µm nozzle. For culture, cells were sorted into 0.5 ml IMDM medium with 5% FCS in a 5-ml FACS tube. For RT quantitative PCR (RT-qPCR), cells were sorted into 350 µl RLT buffer with β-mercaptoethanol in 1.5 ml tubes. The following staining panel was used for sorting: Sca-1-Pacific blue, PE-CD117, APC-CD105, PerCP Cy5.5-CD11b and Streptavidin (BioLegend), APC Cy7-CD16/32, PE Cy7-CD41, BV605-CD150.

ImageStream cytometry

BM suspensions were prepared using the lineage depletion kit, then resuspended in serum-free expansion medium (SFEM; Stem Cell Technologies). 1–2 × 106 cells were plated in SFEM and rested for 15 min at 37°C and 5% CO2. Cells were then stimulated with medium or IL-33 (50 ng/ml) for 30 min and also stained for surface markers at the same time (CD105, Pacific blue; Gr1, FITC; CD16/32, APC; cKit, PE/Cy7; BioLegend), then centrifuged and fixed with warmed 4% formalin for 15 min. Cells were washed with PBS with 0.1% Triton X-100 (Sigma-Aldrich) and 3% FCS (Sigma Aldrich), then stained with NF-κB p65-PE (5 µl per sample in the same buffer; Santa Cruz Biotechnology) for 1 h at room temperature in the dark. Cells were centrifuged and washed, then stained with 7AAD (15 µg/ml in the same buffer; BioLegend) for 1 h at room temperature. Cells were washed and resuspended in PBS/0.1% BSA for acquisition.

Samples were acquired with an Amnis ImageStream XMarkII with the following settings: magnification 60 µm, flow rate slow. Translocation of the p65 probe was assessed in erythroid progenitors (cKit+, CD16/32, CD105+) using the nuclear translocation wizard in IDEAS6.2 software (Amnis).

Phosphoflow cytometry

Single-cell suspensions of BM were obtained and lineage-positive cells were depleted as described above. The resulting suspensions of lineage-depleted BM cells were plated at 0.5–1 × 106 cells/well, resuspended in SFEM, with or without IL-33 (10 ng/ml) and incubated for 6 h at 37°C and 5% CO2. After this, cells were stained with surface antibodies and viability dye (all BioLegend unless otherwise stated: CD105-Pacific blue; FITC-CD71, PE/Cy7-CD16/32 [BD], AF700-CD11b, BV786-CD117; [BD]) and simultaneously stimulated with EPO (4 U/ml) for 20 min and then fixed with warmed BD Cytofix/Cytoperm solution for 12 min. Cells were centrifuged, washed with PBS/0.1% BSA, and permeabilized with Fixation buffer III (cooled to −20°C; BD) on ice for 30 min. Cells were then washed with PBS/0.1% BSA and stained with phospho antibodies (anti-STAT5 pY694 PE clone 47/Stat5(pY694), anti-Akt pS473 AF647 clone M89-61, anti-ERK1/2 pT202/pY204 AF647 clone 20A, anti-p38 MAPK pT180/pY182 PE clone 36/p38, anti-mTOR pS2448 PE clone MRRBY, and anti-pS6 pS235/pS236 PE clone N7-548, 5 µl/sample; all from BD) for 1 h at room temperature in the dark. Cells were then washed and acquired as described above. Erythroid progenitors were gated as CD11bcKit+CD16/32CD105+.

Complete blood cell counts

Blood was collected from mice by cardiac puncture after death was confirmed. Blood (450 µl) was collected into a 1-ml syringe containing 50 µl 4% sodium citrate (Sigma-Aldrich) and then transferred to a plain 1.5-ml tube and kept at 4°C. Within 4 h, samples were gently mixed and analyzed using a Sysmex PX21 analyzer. Results were corrected for the dilution effect of the anticoagulant (corrected value = [measured value / 9] × 10).

BM colony-forming assays

BM was prepared as described above and RBCs were lysed by mixing the cell pellet with 0.5 ml of RBC lysis buffer (Sigma Aldrich) and incubating for 2 min at room temperature. Cells were washed with PBS/0.1% BSA and counted. 5 × 104 cells were resuspended in 100 µl of IMDM medium (Gibco) with 5% FCS and then mixed with 1 ml M3434 Methocult medium (Stem Cell Technologies). Suspensions were thoroughly mixed by vortexing and shaking before plating into 35-mm Petri dishes. Where indicated, BM cells were incubated with or without recombinant murine IL-33 (10 ng/ml) in liquid IMDM medium with 5% FCS for 6 h before mixing with M3434 Methocult that also contained IL-33 (10 ng/ml). Dishes were sealed with Micropore tape, placed inside a 10-cm Petri dish with a dish of sterile water, and incubated at 37°C and 5% CO2 for 10 d. Colonies were assessed and counted manually using an inverted microscope. Images were obtained at 10× magnification using an Olympus BX51 microscope.

BM plasma

The BM of single femurs was flushed into 1.5-ml tubes using 250 µl of PBS/0.1% BSA, 1-ml syringes, and 25G needles. The suspension was vortexed briefly and then centrifuged at 13,000 g for 10 min. The supernatant was removed into a new tube and frozen at −80°C until assayed.

Concentrations of TNF-α and IL-33 were measured in BM plasma using ELISAs according to the manufacturer’s instructions (TNF, BioLegend Legend Max; IL-33, R&D Systems DuoSet) except that, for the IL-33 ELISA, samples and standards were incubated for 6 h with orbital shaking.

Bulk culture of pre–CFU-Es

Pre–CFU-Es (defined as lineageneg, Sca1neg, cKit+, CD41neg, CD16/32neg, CD105+, CD150hi) sorted by FACS were resuspended in SFEM supplemented with recombinant murine SCF (100 ng/ml; BioLegend), recombinant human EPO (2 U/ml), and, where indicated, recombinant murine IL-33 (10 ng/ml) or TNF-α (10 ng/ml; BioLegend). All cultures were completed in triplicate, with 1,000–2,000 cells in 200 µl volume per well of a 96-well tissue culture plate incubated at 37°C and 5% CO2. After 2 d, 60 µl medium was removed and replaced with fresh medium. After 4 d, plates were centrifuged and stained for flow cytometry as described above. Counting beads (BioLegend) were used to measure the number of cells in each well.

For assessment of morphology, some cells were incubated for 4 d, washed twice with plain PBS, and then spun onto plain glass slides using a Shandon Cytospin centrifuge with the following settings: medium acceleration, 800 rpm, 5 min. The slides were stained with Shandon Kwik Diff solutions (ThermoFisher Scientific) according to the manufacturer’s instructions; images were obtained using an Olympus BX51 microscope at 40× magnification.

For assessment of enucleation by flow cytometry, some cells were incubated for 4 d and then incubated with Hoechst 33342 (20 µg/ml) in SFEM for 30 min at 37°C and 5% CO2. The cells were centrifuged and stained with anti-Ter119-APC and CD71-PE (BioLegend) before acquisition as described above.

For assessment of colony formation, 1,000 sorted pre–CFU-Es were resuspended in SFEM with SCF and EPO (as described above), with or without IL-33 (10 ng/ml) and incubated for 6 h at 37°C and 5% CO2. Cells (in a volume of 300 µl) were then mixed with 0.8 ml M3236 Methocult (Stem Cell Technologies) supplemented with SCF (100 ng/ml) and EPO (2 U/ml), with or without IL-33 (10 ng/ml), and plated onto 35-mm Petri dishes. Dishes were sealed with Micropore tape and incubated for 48 h. The number of CFU-E colonies, identified as 8–32-cell clusters forming a distinct colony, was counted manually for each dish.

For assessment of the direct/indirect effects of IL-33, pre–CFU-Es were sorted as above from C57BL/6-Il1rl1−/− (ST2 knockout) and C57BL/6-UbcGFP mice (WT, expressing GFP under the control of the ubiquitin c gene). These cells were plated individually (2,000 cells per well) or mixed (1,000 cells of each type per well) and incubated and assessed as described above.

For assessment of the effect of NF-κB, NBD peptide was added (10 µM; Novus Biologicals) where indicated and left in the medium for 4 d. For assessment of the effect of a higher dose of EPO, EPO was added to the culture at a final concentration of 20 U/ml for 4 d.

Single-cell culture of pre–CFU-E

BM suspensions were prepared as described above. Single pre–CFU-Es were sorted by FACS into wells of a 96-well plate, each containing 80 µl SFEM medium supplemented with SCF (100 ng/ml), EPO (2 U/ml), and dexamethasone (10 µM; Sigma-Aldrich), with or without IL-33 (10 ng/ml). Dexamethasone was added for single-cell cultures to increase proliferation to obtain a reliable readout. The outside row/column of the plate was filled with sterile water. Plates were incubated at 37°C and 5% CO2 for 10 d. At this time, 50 µl of medium was removed and replaced with 50 µl of PBS/0.1% BSA with viability dye (Zombie Aqua; BioLegend) and surface antibodies (CD71-PE, Ter119-APC). The plate was incubated at 4°C in the dark for 1 h, and then 50 µl was aspirated from each well and replaced with 150 µl PBS/0.1% BSA for acquisition by flow cytometry. The number of cells in each well was quantified using counting beads.

Isolation and culture of human erythroid progenitors

Anonymized human leukoreduction cones were obtained from the NHS Blood and Transfusion Service, University of Oxford. Donors gave consent for the use of samples for research purposes. The cones were drained into 50-ml tubes and diluted with PBS to a final volume of 50 ml. From this, 25 ml was layered onto 15 ml Lymphoprep solution (Stem Cell Technologies) and centrifuged as follows: 600 g, 20 min, room temperature, acceleration 7, deceleration 0. The leukocyte layer was harvested with a P1000 pipette and washed three times with PBS.

From this suspension enriched for human PBMCs, 50 × 106 cells were removed and mixed with 3 µg biotinylated antibodies against CD3, CD19, and CD14 (all BioLegend) in 1 ml PBS/0.1% BSA for 10 min at 4°C. Cells were washed and mixed with 0.4 ml streptavidin conjugated with magnetic beads in a final volume of 2 ml in a FACS tube and incubated at room temperature for 10 min. The volume was then adjusted to 4 ml, and the FACS tube was placed inside a magnet for 10 min. After this, the supernatant was poured into a fresh tube and stained with the following panel: FITC-CD34, APC-CD123, PE/Cy7-CD71, PerCP/Cy5.5-Streptavidin, and 7AAD. Erythroid progenitors were sorted as live, lineageneg (i.e., PerCP/Cy5.5neg), CD34+, CD123, CD71+ by FACS as described above.

Sorted cells were resuspended in SFEM with recombinant human SCF (100 ng/ml; Peprotech), recombinant human EPO (2 U/ml; BioLegend), with or without recombinant human IL-33 (10 ng/ml; Peprotech). Cultures were performed in triplicate, with 2,000 cells in 200-µl volume per well in a 96-well tissue culture plate. Every 2 d, 60 µl medium was removed and replaced with fresh medium. After 8–9 d, plates were centrifuged and cells were prepared for flow cytometry as described above. CD235a (glycophorin A) was used as a marker of mature erythroid cells in human cultures because there is no known equivalent to the Ter119 antigen.

RT-qPCR

To sort cells for RT-qPCR, BM was prepared as described above under Flow cytometry. For stromal populations, femurs and tibiae were dissected, cleaned, and then crushed in a Petri dish. The bone fragments were transferred to wells of a 48-well plate with 1 ml of Hanks' buffered saline solution with DNase I (15 µg/ml; Sigma-Aldrich) and collagenase I (300 U/ml; Sigma-Aldrich). The plate was incubated at 37°C with shaking at 100 rpm for 1 h. After this, bone fragments were washed and, with the digestion solution, transferred to a new tube through a 70-µm filter.

For assessment of BM IL-33 production by RT-qPCR, the following populations were defined by FACS: mesenchymal stromal/stem cells: CD45, Ter119, CD31, CD51+, PDGFRα+, Sca1; endothelium: CD45, Ter119, CD31+; osteoblastic lineage cells: CD45, Ter119, CD31, CD51+; erythroblastic island macrophages (CD169+ macrophages): CD45+, Gr1, F4/80+, SSClo, CD169+; neutrophils: CD45+, CD11b+, Ly6G+; and monocytes (Ly6C+ monocytes): CD45+, CD11b+, Ly6G, Ly6C+

Progenitors or BM cell populations were sorted by FACS directly into RLT buffer with β mercaptoethanol, vortexed briefly, and frozen at −80°C until processed. RNA was extracted using the RNeasy Micro Kit (Qiagen), and 50 ng cDNA was synthesized using the High-Capacity Reverse Transcription kit (Affymetrix eBioscience). RT-qPCR was performed using TaqMan primers/probes (Hba-a1 Mm02580841_g1, Hbb-b1 Mm01611268_g1, Alas2 Mm00802083_m1, Epb41 Mm01187466_m1, Gypa Mm00494848_m1, Epor Mm00833882_m1, Aqp1 Mm01326466_m1, Il33 Mm00505403_m1, and Il1b Mm00434228_m1; ThermoFisher Scientific) and PrecisionFast Mastermix (PrimerDesign) on a ViiA7 384-well real-time PCR system. All expression levels were normalized to an internal reference gene (Hprt) and calculated as 2−(threshold cycle [CT]Hprt− CT gene). Real-time qPCR reactions were performed in duplicate.

Immunofluorescence

BM was prepared as described above, and CFU-Es were sorted by FACS. Cells were incubated in SFEM supplemented with EPO and SCF as described above at 9 × 104 cells per well, with or without IL-33 (10 ng/ml) for 24 h. The plate was centrifuged and cells were fixed in 4% formalin for 20 min at room temperature. Cells were washed once with PBS/0.1% BSA and then stained with goat anti-mouse EPO-R antibody (15 µg/ml; R&D Systems) for 2 h at room temperature. Cells were then washed once and blocked with rabbit serum before staining with rabbit anti-goat IgG secondary AF555 antibody (1:500; Invitrogen) for 45 min at room temperature. Cells were washed once and then stained with Hoechst 33342 (Life Technologies) for 20 min. Cells were resuspended in 30 µl PBS/0.1% BSA, of which 10 µl was transferred to a plain glass slide and covered with a 1 cm diameter glass coverslip. After 10 min to allow cells to rest on the slide, cells were imaged using an Olympus BX51 microscope.

RNA sequencing

Pre–CFU-Es were sorted by FACS and resuspended in SFEM with SCF and EPO as described above. Cells were incubated with or without IL-33 (10 ng/ml) for 6 h at 37°C and 5% CO2. Plates were then centrifuged, supernatants discarded, and cells transferred to 1.5-ml tubes with 350 µl RLT buffer with β-mercaptoethanol. RNA was extracted using the RNEasy Micro kit (Qiagen), according to the manufacturer’s instructions.

RNA-sequencing analysis

RNA-sequencing data were analyzed using pipeline_scrnaseq.py (https://github.com/sansomlab/scseq/blob/master/pipelines/). Data quality was assessed using pipeline_readqc.py (https://github.com/cgat-developers/cgat-flow/). Sequence reads were aligned to the mouse genome with Hisat2 (version 2.1.0; Kim et al., 2015) using a “genome” index built from the GRCm38 release of the mouse genome and known splice sites extracted from Ensembl version 91 annotations (using the hisat2_extract_splice_sites.py tool). A two-pass mapping strategy was used to discover novel splice sites (with the additional parameters –dta and–score-min L,0.0,−0.2). The average alignment rate was 95.2% (as assessed with Picard Tools v2.10.9; https://github.com/broadinstitute/picard). Mapped reads were counted using featureCounts (Subread version 1.6.3; Ensembl version 91 annotations; with default parameters; Liao et al., 2014). Salmon v0.9.1 was used to calculate transcript per million values (Patro et al., 2017) using a quasi-index (built with Ensembl version 91 annotations and k = 31) and guanidine/cytosine (gc) bias correction (parameter “–gcBias”). Differential expression analysis of IL-33–exposed pre–CFU-E versus control samples was performed using DESeq2 (v1.24.0; Love et al., 2014). Raw data have been deposited in GEO (accession no. GSE141480).

Gene set enrichment analysis (GSEA)

GSEA was performed using fgsea R-package (version 1.10.1) for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (using for ranking the Wald statistic from DESeq2; Fig. 5 C; Sergushichev, 2016,Preprint). Enrichment of members of the NF-κB pathways and genes up-regulated by EPO in committed erythroid progenitors (CEPs) were assessed in RNA-sequencing results using GSEA software, version 4.0.2. Genes up-regulated in CEP by EPO were extracted from Supplementary Table 5 in Tusi et al. (2018). NF-κB pathway gene sets were obtained from the Pathway Interaction Database available through the GSEA software.

Analysis of Haemosphere data

Results of RNA sequencing from murine hematopoietic progenitors were obtained from the online Haemosphere database, reporting the results of the Haemopedia sequencing project (Choi et al., 2019). Specifically, gene counts (expressed as transcript per million) were downloaded for CFU-E and GMP. Differential expression of genes between these cell types was evaluated using DEseq2 (v1.24.0) and EnchancedVolcano packages in R, version 3.6.1.

Statistical analysis

Unless otherwise stated, all analysis was completed using proprietary software (GraphPad Prism). Groups were compared using unpaired t tests, Mann-Whitney U tests, one-way ANOVA, or two-way ANOVA depending on the number of groups and nature of comparisons. If significant differences were detected by ANOVA, post hoc Tukey tests (one way) or Sidak’s multiple comparison tests (two way) were completed. Correlations were assessed using linear regression analysis.

Online supplemental material

Fig. S1 presents additional data on the hematopoietic changes observed in SKG mice developing SpA and shows similar data from WT C57BL/6 mice injected with curdlan. Fig. S2 provides additional data on the expression of the IL-33 receptor ST2 on lineage-committed and MPPs. Fig. S3 provides additional data on proliferation and apoptosis markers for cultured erythroid cells. Fig. S4 provides additional data on the effect of IL-33 on hematopoiesis in mice lacking IL-6 or the receptors for TNF. Fig. S5 provides additional data on the expression of the EPO-R in cells exposed to IL-33 and on the dependence of curdlan on IL-33 to suppress EPO-accelerated erythropoiesis in vivo.

Acknowledgments

We thank Professor Claus Nerlov and Dr. Roy Drissen (University of Oxford) and Professor Nicole Horwood (University of East Anglia) for helpful discussions. We are grateful to the High-Throughput Genomics Group at the Wellcome Trust Centre for Human Genetics for sequencing data. We thank Jonathan Webber of the Kennedy Flow Cytometry Core Facility.

This study was funded by Versus Arthritis (Career Development Fellowship 20834 to T. Griseri), the Wellcome Trust (Investigator in Science Award 212240/Z/18/Z to F. Powrie), and the Kennedy Trust for Rheumatology Research. The Wellcome Trust Centre for Human Genetics is funded by Wellcome Trust grant 090532/Z/09/Z.

Author contributions: Investigation: J.W. Swann, D. Regan-Komito, and T. Griseri; Formal analysis: J.W. Swann, L.A. Koneva, and T. Griseri; Resources and supervision: S.N. Sansom, F. Powrie, and T. Griseri; Conceptualization: J.W. Swann, S.N. Sansom, F. Powrie, and T. Griseri; Data curation: J.W. Swann, L.A. Koneva, S.N. Sansom, and T. Griseri; Writing – original draft: J.W. Swann, L.A. Koneva, and T. Griseri; Writing – review & editing: J.W. Swann, D. Regan-Komito, L.A. Koneva, S.N. Sansom, F. Powrie, and T. Griseri; Funding acquisition: F. Powrie, T. Griseri; Project administration: T. Griseri.

References

References
Amireault
,
P.
,
S.
Hatia
,
E.
Bayard
,
F.
Bernex
,
C.
Collet
,
J.
Callebert
,
J.M.
Launay
,
O.
Hermine
,
E.
Schneider
,
J.
Mallet
, et al
.
2011
.
Ineffective erythropoiesis with reduced red blood cell survival in serotonin-deficient mice
.
Proc. Natl. Acad. Sci. USA
.
108
:
13141
13146
.
Argollo
,
M.
,
D.
Gilardi
,
C.
Peyrin-Biroulet
,
J.F.
Chabot
,
L.
Peyrin-Biroulet
, and
S.
Danese
.
2019
.
Comorbidities in inflammatory bowel disease: a call for action
.
Lancet Gastroenterol. Hepatol
.
4
:
643
654
.
Benham
,
H.
,
L.M.
Rehaume
,
S.Z.
Hasnain
,
J.
Velasco
,
A.C.
Baillet
,
M.
Ruutu
,
K.
Kikly
,
R.
Wang
,
H.W.
Tseng
,
G.P.
Thomas
, et al
.
2014
.
Interleukin-23 mediates the intestinal response to microbial β-1,3-glucan and the development of spondyloarthritis pathology in SKG mice
.
Arthritis Rheumatol
.
66
:
1755
1767
.
Chen
,
K.
,
J.
Liu
,
S.
Heck
,
J.A.
Chasis
,
X.
An
, and
N.
Mohandas
.
2009
.
Resolving the distinct stages in erythroid differentiation based on dynamic changes in membrane protein expression during erythropoiesis
.
Proc. Natl. Acad. Sci. USA
.
106
:
17413
17418
.
Choi
,
J.
,
T.M.
Baldwin
,
M.
Wong
,
J.E.
Bolden
,
K.A.
Fairfax
,
E.C.
Lucas
,
R.
Cole
,
C.
Biben
,
C.
Morgan
,
K.A.
Ramsay
, et al
.
2019
.
Haemopedia RNA-seq: a database of gene expression during haematopoiesis in mice and humans
.
Nucleic Acids Res
.
47
(
D1
):
D780
D785
.
Chow
,
A.
,
M.
Huggins
,
J.
Ahmed
,
D.
Hashimoto
,
D.
Lucas
,
Y.
Kunisaki
,
S.
Pinho
,
M.
Leboeuf
,
C.
Noizat
,
N.
van Rooijen
, et al
.
2013
.
CD169+ macrophages provide a niche promoting erythropoiesis under homeostasis and stress
.
Nat. Med
.
19
:
429
436
.
De Vos
,
M.
,
H.
Mielants
,
C.
Cuvelier
,
A.
Elewaut
, and
E.
Veys
.
1996
.
Long-term evolution of gut inflammation in patients with spondyloarthropathy
.
Gastroenterology
.
110
:
1696
1703
.
Etzrodt
,
M.
,
N.
Ahmed
,
P.S.
Hoppe
,
D.
Loeffler
,
S.
Skylaki
,
O.
Hilsenbeck
,
K.D.
Kokkaliaris
,
H.M.
Kaltenbach
,
J.
Stelling
,
C.
Nerlov
, et al
.
2019
.
Inflammatory signals directly instruct PU.1 in HSCs via TNF
.
Blood
.
133
:
816
819
.
Felli
,
N.
,
F.
Pedini
,
A.
Zeuner
,
E.
Petrucci
,
U.
Testa
,
C.
Conticello
,
M.
Biffoni
,
A.
Di Cataldo
,
J.A.
Winkles
,
C.
Peschle
, et al
.
2005
.
Multiple members of the TNF superfamily contribute to IFN-gamma-mediated inhibition of erythropoiesis
.
J. Immunol
.
175
:
1464
1472
.
Fields
,
J.K.
,
S.
Günther
, and
E.J.
Sundberg
.
2019
.
Structural Basis of IL-1 Family Cytokine Signaling
.
Front. Immunol
.
10
:
1412
.
Furst
,
D.E.
,
J.
Kay
,
M.C.
Wasko
,
E.
Keystone
,
A.
Kavanaugh
,
A.
Deodhar
,
F.T.
Murphy
,
J.H.
Magnus
,
E.C.
Hsia
,
B.
Hsu
, et al
.
2013
.
The effect of golimumab on haemoglobin levels in patients with rheumatoid arthritis, psoriatic arthritis or ankylosing spondylitis
.
Rheumatology (Oxford)
.
52
:
1845
1855
.
Ghaffari
,
S.
,
C.
Kitidis
,
W.
Zhao
,
D.
Marinkovic
,
M.D.
Fleming
,
B.
Luo
,
J.
Marszalek
, and
H.F.
Lodish
.
2006
.
AKT induces erythroid-cell maturation of JAK2-deficient fetal liver progenitor cells and is required for Epo regulation of erythroid-cell differentiation
.
Blood
.
107
:
1888
1891
.
Han
,
G.W.
,
L.W.
Zeng
,
C.X.
Liang
,
B.L.
Cheng
,
B.S.
Yu
,
H.M.
Li
,
F.F.
Zeng
, and
S.Y.
Liu
.
2011
.
Serum levels of IL-33 is increased in patients with ankylosing spondylitis
.
Clin. Rheumatol
.
30
:
1583
1588
.
Hata
,
H.
,
N.
Sakaguchi
,
H.
Yoshitomi
,
Y.
Iwakura
,
K.
Sekikawa
,
Y.
Azuma
,
C.
Kanai
,
E.
Moriizumi
,
T.
Nomura
,
T.
Nakamura
, et al
.
2004
.
Distinct contribution of IL-6, TNF-alpha, IL-1, and IL-10 to T cell-mediated spontaneous autoimmune arthritis in mice
.
J. Clin. Invest
.
114
:
582
588
.
Huang
,
S.Y.
,
Y.H.
Liu
,
Y.J.
Chen
,
Y.Y.
Yeh
, and
H.M.
Huang
.
2018
.
CD69 partially inhibits apoptosis and erythroid differentiation via CD24, and their knockdown increase imatinib sensitivity in BCR-ABL-positive cells
.
J. Cell. Physiol
.
233
:
7467
7479
.
Johnston
,
L.K.
,
C.L.
Hsu
,
R.A.
Krier-Burris
,
K.D.
Chhiba
,
K.B.
Chien
,
A.
McKenzie
,
S.
Berdnikovs
, and
P.J.
Bryce
.
2016
.
IL-33 Precedes IL-5 in Regulating Eosinophil Commitment and Is Required for Eosinophil Homeostasis
.
J. Immunol
.
197
:
3445
3453
.
Kenswil
,
K.J.G.
,
A.C.
Jaramillo
,
Z.
Ping
,
S.
Chen
,
R.M.
Hoogenboezem
,
M.A.
Mylona
,
M.N.
Adisty
,
E.M.J.
Bindels
,
P.K.
Bos
,
H.
Stoop
, et al
.
2018
.
Characterization of Endothelial Cells Associated with Hematopoietic Niche Formation in Humans Identifies IL-33 As an Anabolic Factor
.
Cell Rep
.
22
:
666
678
.
Kim
,
D.
,
B.
Langmead
, and
S.L.
Salzberg
.
2015
.
HISAT: a fast spliced aligner with low memory requirements
.
Nat. Methods
.
12
:
357
360
.
Kim
,
H.S.
,
K.H.
Park
,
H.K.
Lee
,
J.S.
Kim
,
Y.G.
Kim
,
J.H.
Lee
,
K.H.
Kim
,
J.
Yun
,
B.Y.
Hwang
,
J.T.
Hong
, et al
.
2016
.
Curdlan activates dendritic cells through dectin-1 and toll-like receptor 4 signaling
.
Int. Immunopharmacol
.
39
:
71
78
.
Kina
,
T.
,
K.
Ikuta
,
E.
Takayama
,
K.
Wada
,
A.S.
Majumdar
,
I.L.
Weissman
, and
Y.
Katsura
.
2000
.
The monoclonal antibody TER-119 recognizes a molecule associated with glycophorin A and specifically marks the late stages of murine erythroid lineage
.
Br. J. Haematol
.
109
:
280
287
.
King
,
K.Y.
, and
M.A.
Goodell
.
2011
.
Inflammatory modulation of HSCs: viewing the HSC as a foundation for the immune response
.
Nat. Rev. Immunol
.
11
:
685
692
.
Li
,
G.X.
,
S.
Wang
,
Z.H.
Duan
,
Z.
Zeng
, and
F.M.
Pan
.
2013
.
Serum levels of IL-33 and its receptor ST2 are elevated in patients with ankylosing spondylitis
.
Scand. J. Rheumatol
.
42
:
226
231
.
Liao
,
Y.
,
G.K.
Smyth
, and
W.
Shi
.
2014
.
featureCounts: an efficient general purpose program for assigning sequence reads to genomic features
.
Bioinformatics
.
30
:
923
930
.
Libregts
,
S.F.
,
L.
Gutiérrez
,
A.M.
de Bruin
,
F.M.
Wensveen
,
P.
Papadopoulos
,
W.
van Ijcken
,
Z.
Ozgür
,
S.
Philipsen
, and
M.A.
Nolte
.
2011
.
Chronic IFN-γ production in mice induces anemia by reducing erythrocyte life span and inhibiting erythropoiesis through an IRF-1/PU.1 axis
.
Blood
.
118
:
2578
2588
.
Liew
,
F.Y.
,
J.P.
Girard
, and
H.R.
Turnquist
.
2016
.
Interleukin-33 in health and disease
.
Nat. Rev. Immunol
.
16
:
676
689
.
Liu
,
J.J.
,
S.C.
Hou
, and
C.K.
Shen
.
2003
.
Erythroid gene suppression by NF-kappa B
.
J. Biol. Chem
.
278
:
19534
19540
.
Liu
,
X.
,
Y.
Zhang
,
M.
Ni
,
H.
Cao
,
R.A.J.
Signer
,
D.
Li
,
M.
Li
,
Z.
Gu
,
Z.
Hu
,
K.E.
Dickerson
, et al
.
2017
.
Regulation of mitochondrial biogenesis in erythropoiesis by mTORC1-mediated protein translation
.
Nat. Cell Biol
.
19
:
626
638
.
Love
,
M.I.
,
W.
Huber
, and
S.
Anders
.
2014
.
Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
.
Genome Biol
.
15
:
550
.
Lunderius-Andersson
,
C.
,
M.
Enoksson
, and
G.
Nilsson
.
2012
.
Mast Cells Respond to Cell Injury through the Recognition of IL-33
.
Front. Immunol
.
3
:
82
.
Macdougall
,
I.C.
, and
A.C.
Cooper
.
2002
.
Erythropoietin resistance: the role of inflammation and pro-inflammatory cytokines
.
Nephrol. Dial. Transplant
.
17
(
Suppl 11
):
39
43
.
McCranor
,
B.J.
,
M.J.
Kim
,
N.M.
Cruz
,
Q.L.
Xue
,
A.E.
Berger
,
J.D.
Walston
,
C.I.
Civin
, and
C.N.
Roy
.
2014
.
Interleukin-6 directly impairs the erythroid development of human TF-1 erythroleukemic cells
.
Blood Cells Mol. Dis
.
52
:
126
133
.
Meng
,
D.
,
A.R.
Frank
, and
J.L.
Jewell
.
2018
.
mTOR signaling in stem and progenitor cells
.
Development
.
145
:
145
.
Mercan
,
R.
,
B.
Bitik
,
A.
Tufan
,
U.B.
Bozbulut
,
N.
Atas
,
M.A.
Ozturk
,
S.
Haznedaroglu
, and
B.
Goker
.
2016
.
The Association Between Neutrophil/Lymphocyte Ratio and Disease Activity in Rheumatoid Arthritis and Ankylosing Spondylitis
.
J. Clin. Lab. Anal
.
30
:
597
601
.
Mori
,
Y.
,
J.Y.
Chen
,
J.V.
Pluvinage
,
J.
Seita
, and
I.L.
Weissman
.
2015
.
Prospective isolation of human erythroid lineage-committed progenitors
.
Proc. Natl. Acad. Sci. USA
.
112
:
9638
9643
.
Nemeth
,
E.
, and
T.
Ganz
.
2014
.
Anemia of inflammation
.
Hematol. Oncol. Clin. North Am
.
28
:
671
681: vi
.
Patro
,
R.
,
G.
Duggal
,
M.I.
Love
,
R.A.
Irizarry
, and
C.
Kingsford
.
2017
.
Salmon provides fast and bias-aware quantification of transcript expression
.
Nat. Methods
.
14
:
417
419
.
Pietras
,
E.M.
,
D.
Reynaud
,
Y.A.
Kang
,
D.
Carlin
,
F.J.
Calero-Nieto
,
A.D.
Leavitt
,
J.M.
Stuart
,
B.
Göttgens
, and
E.
Passegué
.
2015
.
Functionally Distinct Subsets of Lineage-Biased Multipotent Progenitors Control Blood Production in Normal and Regenerative Conditions
.
Cell Stem Cell
.
17
:
35
46
.
Pietras
,
E.M.
,
C.
Mirantes-Barbeito
,
S.
Fong
,
D.
Loeffler
,
L.V.
Kovtonyuk
,
S.
Zhang
,
R.
Lakshminarasimhan
,
C.P.
Chin
,
J.M.
Techner
,
B.
Will
, et al
.
2016
.
Chronic interleukin-1 exposure drives haematopoietic stem cells towards precocious myeloid differentiation at the expense of self-renewal
.
Nat. Cell Biol
.
18
:
607
618
.
Pinho
,
S.
,
J.
Lacombe
,
M.
Hanoun
,
T.
Mizoguchi
,
I.
Bruns
,
Y.
Kunisaki
, and
P.S.
Frenette
.
2013
.
PDGFRα and CD51 mark human nestin+ sphere-forming mesenchymal stem cells capable of hematopoietic progenitor cell expansion
.
J. Exp. Med
.
210
:
1351
1367
.
Pinto
,
S.M.
,
Y.
Subbannayya
,
D.A.B.
Rex
,
R.
Raju
,
O.
Chatterjee
,
J.
Advani
,
A.
Radhakrishnan
,
T.S.
Keshava Prasad
,
M.R.
Wani
, and
A.
Pandey
.
2018
.
A network map of IL-33 signaling pathway
.
J. Cell Commun. Signal
.
12
:
615
624
.
Pop
,
R.
,
J.R.
Shearstone
,
Q.
Shen
,
Y.
Liu
,
K.
Hallstrom
,
M.
Koulnis
,
J.
Gribnau
, and
M.
Socolovsky
.
2010
.
A key commitment step in erythropoiesis is synchronized with the cell cycle clock through mutual inhibition between PU.1 and S-phase progression
.
PLoS Biol
.
8
. e1000484.
Pronk
,
C.J.
,
D.J.
Rossi
,
R.
Månsson
,
J.L.
Attema
,
G.L.
Norddahl
,
C.K.
Chan
,
M.
Sigvardsson
,
I.L.
Weissman
, and
D.
Bryder
.
2007
.
Elucidation of the phenotypic, functional, and molecular topography of a myeloerythroid progenitor cell hierarchy
.
Cell Stem Cell
.
1
:
428
442
.
Ranganathan
,
V.
,
E.
Gracey
,
M.A.
Brown
,
R.D.
Inman
, and
N.
Haroon
.
2017
.
Pathogenesis of ankylosing spondylitis - recent advances and future directions
.
Nat. Rev. Rheumatol
.
13
:
359
367
.
Regan-Komito
,
D.
,
J.W.
Swann
,
P.
Demetriou
,
E.S.
Cohen
,
N.J.
Horwood
,
S.N.
Sansom
, and
T.
Griseri
.
2020
.
GM-CSF drives dysregulated hematopoietic stem cell activity and pathogenic extramedullary myelopoiesis in experimental spondyloarthritis
.
Nat. Commun
.
11
:
155
.
Ruutu
,
M.
,
G.
Thomas
,
R.
Steck
,
M.A.
Degli-Esposti
,
M.S.
Zinkernagel
,
K.
Alexander
,
J.
Velasco
,
G.
Strutton
,
A.
Tran
,
H.
Benham
, et al
.
2012
.
β-glucan triggers spondylarthritis and Crohn’s disease-like ileitis in SKG mice
.
Arthritis Rheum
.
64
:
2211
2222
.
Schepers
,
K.
,
E.M.
Pietras
,
D.
Reynaud
,
J.
Flach
,
M.
Binnewies
,
T.
Garg
,
A.J.
Wagers
,
E.C.
Hsiao
, and
E.
Passegué
.
2013
.
Myeloproliferative neoplasia remodels the endosteal bone marrow niche into a self-reinforcing leukemic niche
.
Cell Stem Cell
.
13
:
285
299
.
Sergushichev
,
A.A.
2016
.
An algorithm for fast preranked gene set enrichment analysis using cumulative statistic calculation
.
bioRxiv.
doi: 10.1101/060012 (Preprint posted June 20, 2016).
Silva
,
M.
,
A.
Benito
,
C.
Sanz
,
F.
Prosper
,
D.
Ekhterae
,
G.
Nuñez
, and
J.L.
Fernandez-Luna
.
1999
.
Erythropoietin can induce the expression of bcl-x(L) through Stat5 in erythropoietin-dependent progenitor cell lines
.
J. Biol. Chem
.
274
:
22165
22169
.
Smith
,
M.A.
,
S.M.
Knight
,
P.J.
Maddison
, and
J.G.
Smith
.
1992
.
Anaemia of chronic disease in rheumatoid arthritis: effect of the blunted response to erythropoietin and of interleukin 1 production by marrow macrophages
.
Ann. Rheum. Dis
.
51
:
753
757
.
Taketani
,
S.
,
T.
Furukawa
, and
K.
Furuyama
.
2001
.
Expression of coproporphyrinogen oxidase and synthesis of hemoglobin in human erythroleukemia K562 cells
.
Eur. J. Biochem
.
268
:
1705
1711
.
Takizawa
,
H.
,
S.
Boettcher
, and
M.G.
Manz
.
2012
.
Demand-adapted regulation of early hematopoiesis in infection and inflammation
.
Blood
.
119
:
2991
3002
.
Talabot-Ayer
,
D.
,
P.
Martin
,
C.
Vesin
,
C.A.
Seemayer
,
S.
Vigne
,
C.
Gabay
, and
G.
Palmer
.
2015
.
Severe neutrophil-dominated inflammation and enhanced myelopoiesis in IL-33-overexpressing CMV/IL33 mice
.
J. Immunol
.
194
:
750
760
.
Turesson
,
C.
.
2016
.
Comorbidity in rheumatoid arthritis
.
Swiss Med. Wkly
.
146
. w14290.
Tusi
,
B.K.
,
S.L.
Wolock
,
C.
Weinreb
,
Y.
Hwang
,
D.
Hidalgo
,
R.
Zilionis
,
A.
Waisman
,
J.R.
Huh
,
A.M.
Klein
, and
M.
Socolovsky
.
2018
.
Population snapshots predict early haematopoietic and erythroid hierarchies
.
Nature
.
555
:
54
60
.
Van Putten
,
L.M.
.
1958
.
The life span of red cells in the rat and the mouse as determined by labeling with DFP32 in vivo
.
Blood
.
13
:
789
794
.
Watowich
,
S.S.
.
2011
.
The erythropoietin receptor: molecular structure and hematopoietic signaling pathways
.
J. Investig. Med
.
59
:
1067
1072
.
Wei
,
J.
,
J.
Zhao
,
V.
Schrott
,
Y.
Zhang
,
M.
Gladwin
,
G.
Bullock
, and
Y.
Zhao
.
2015
.
Red Blood Cells Store and Release Interleukin-33
.
J. Investig. Med
.
63
:
806
810
.
Weiss
,
G.
, and
L.T.
Goodnough
.
2005
.
Anemia of chronic disease
.
N. Engl. J. Med
.
352
:
1011
1023
.
Wu
,
H.
,
X.
Liu
,
R.
Jaenisch
, and
H.F.
Lodish
.
1995
.
Generation of committed erythroid BFU-E and CFU-E progenitors does not require erythropoietin or the erythropoietin receptor
.
Cell
.
83
:
59
67
.
Xiao
,
W.
,
K.
Koizumi
,
M.
Nishio
,
T.
Endo
,
M.
Osawa
,
K.
Fujimoto
,
I.
Sato
,
T.
Sakai
,
T.
Koike
, and
K.
Sawada
.
2002
.
Tumor necrosis factor-alpha inhibits generation of glycophorin A+ cells by CD34+ cells
.
Exp. Hematol
.
30
:
1238
1247
.
Xue
,
S.P.
,
S.F.
Zhang
,
Q.
Du
,
H.
Sun
,
J.
Xin
,
S.Q.
Liu
, and
J.
Ma
.
1997
.
The role of cytoskeletal elements in the two-phase denucleation process of mammalian erythroblasts in vitro observed by laser confocal scanning microscope
.
Cell. Mol. Biol
.
43
:
851
860
.
Zhang
,
M.Y.
,
S.C.
Sun
,
L.
Bell
, and
B.A.
Miller
.
1998
.
NF-kappaB transcription factors are involved in normal erythropoiesis
.
Blood
.
91
:
4136
4144
.
Zhao
,
W.
,
C.
Kitidis
,
M.D.
Fleming
,
H.F.
Lodish
, and
S.
Ghaffari
.
2006
.
Erythropoietin stimulates phosphorylation and activation of GATA-1 via the PI3-kinase/AKT signaling pathway
.
Blood
.
107
:
907
915
.
Zhao
,
J.L.
,
C.
Ma
,
R.M.
O’Connell
,
A.
Mehta
,
R.
DiLoreto
,
J.R.
Heath
, and
D.
Baltimore
.
2014
.
Conversion of danger signals into cytokine signals by hematopoietic stem and progenitor cells for regulation of stress-induced hematopoiesis
.
Cell Stem Cell
.
14
:
445
459
.

Competing Interests

Disclosures: F. Powrie reported personal fees from GSK, grants from Janssen, personal fees from Genentech, and personal fees from Kintai Therapeutics outside the submitted work. No other disclosures were reported.

This article is available under a Creative Commons License (Attribution 4.0 International, as described at https://creativecommons.org/licenses/by/4.0/).