Itaconate promotes the differentiation of murine stress erythroid progenitors by increasing Nrf2 activity

Steady-state erythropoiesis constantly produces new erythrocytes at a rate of 2.5 × 10⁶ per second.¹ Senescent or damaged erythrocytes are removed from circulation by macrophages in the spleen and liver at a similar rate, which maintains erythroid homeostasis.² In contrast, proinflammatory signals alter the kinetics and routes of hematopoietic differentiation, leading to increased production of myeloid cells at the expense of steady-state erythropoiesis.^3-8 Additionally, inflammation increases erythrophagocytosis, shortening the life span of erythrocytes and increases iron sequestration, which limits hemoglobin synthesis.^9-12 These demand-adapted changes in the blood system underpin a protective immune response to inflammatory insults, but they come with a cost because steady-state erythropoiesis is compromised. Given that effective oxygen transport requires adequate levels of erythrocytes, a compensatory stress erythropoiesis response is activated that uses inflammatory signals to produce erythrocytes and maintain homeostasis during inflammation.¹³ 

Stress erythropoiesis uses a different strategy to steady-state erythropoiesis. Inflammation induces the migration of short-term hematopoietic stem cells (CD34⁺Kit⁺Sca1⁺) and monocytes to the spleen.^14-16 Proinflammatory signals, such as tumor necrosis factor α (TNF-α), interleukin-1β (IL-1β), and interferon gamma, in combination with stem cell factor, canonical Wnt signaling, hedgehog, and growth and differentiation factor 15 (GDF15), promote the proliferation of a transient amplifying population of stress erythroid progenitors (SEPs).^17-22 These signals also establish a stress erythropoiesis niche.^15,18 The expansion of this population of immature SEPs and the development of the niche marks the initial stage of stress erythropoiesis. The transition of this population of SEPs from proliferating SEPs to SEPs committed to erythroid differentiation is driven be erythropoietin (Epo)–dependent changes in niche signals.¹⁸ Proinflammatory and proliferation-promoting signals are turned off, and prodifferentiation signals, such as prostaglandin E2, promote the transition to SEPs committed to differentiation. Overall, this transition relies on a switch from proinflammatory signals to proresolving signals. This change in signals is similar to that observed when macrophages promote the resolution of inflammation by increasing the expression of IL-4, IL-10, and the anti-inflammatory metabolite itaconate.^23,24 

The tight spatiotemporal coordination between niche cells and SEPs suggest that they may use the same immunomodulatory molecules to regulate their codevelopment. Although the role of these molecules in macrophages is extensively studied, it remains undetermined how immunoregulatory metabolites and cytokines act on SEPs to affect their development. In this study, we applied integrated analysis of transcriptional and metabolic profiling on SEPs at different developmental stages. We show that transition of SEPs from expansion to differentiation is dependent on a switch in progenitor cell signaling from one dominated by inflammatory signals to one dominated by resolving signals. In the expansion stage, nitric oxide (NO) synthase 2 (Nos2)–dependent NO production drives the proliferation of immature SEPs.²⁵ In contrast, the transition to SEP differentiation is marked by decreased inflammatory signals and increased resolving molecules such as itaconate. The anti-inflammatory effects of itaconate is mediated by nuclear factor erythroid 2–related factor 2 (Nfe2l2 or Nrf2). Activation of Nrf2 resolves inflammatory signals in both SEPs and the niche, which reduces NO levels and alleviates NO-mediated erythroid inhibition. These data provide a mechanistic basis for how SEP cell-fate transition is governed by immunomodulatory molecules to ensure effective erythroid regeneration.

Wild-type (WT) C57BL/6J, B6.SJL-Ptprca Pepcb/BoyJ (CD45.1; JAX stock number 002014), B6.129X1-Nfe2l2tm1Ywk/J (Nrf2^–/–; JAX stock number 017009),²⁶ B6.129P2-Il10^tm1Cgn/J (IL-10^–/–; JAX stock number 002251),²⁷ and C57BL/6N-Acod1em1(IMPC)J/J (Irg1^–/–; JAX stock number 029340) mice were purchased from The Jackson Laboratory. Mice of both sexes, aged 8 to 16 weeks, were used throughout this study. All experiments were approved by the institutional animal care and use committee at the Pennsylvania State University.

Mouse stress erythropoiesis cultures were done as previously described.¹⁶ Details of the media and culture conditions are provided in the supplemental Methods.

Phenylhydrazine (PHZ) was used to induce stress erythropoiesis in the context of acute hemolytic anemia. Mice were injected intraperitoneally with a single dose (100 mg/kg body weight) of freshly prepared PHZ (Sigma-Aldrich; dissolved in phosphate-buffered saline).²⁸ Heat-killed Brucella abortus (HKBA; strain 1119-3) was used to induce anemia of inflammation according to a previously described method.²⁹ 

SEPs isolated from stress erythropoiesis differentiation media (SEDM) cultures or mouse splenocytes were counted using a hemocytometer. For each sample, 2.5 × 10⁵ cells were resuspended in 2 mL MethoCult M3334 medium (Stemcell Technologies), supplemented additionally with 50 ng/mL stem cell factor (GoldBio) and 15 ng/mL bone morphogenetic protein 4 (Thermo Fisher Scientific), and cell suspension was evenly plated into 3 wells of a 12-well plate as technical triplicates. Stress burst-forming units-erythroid (BFU-E) colonies were stained with benzidine and quantified after a 5-day culture at 37°C with 2% O₂ and 5% CO₂.²⁸ 

GraphPad Prism and R were used for statistical analysis. Statistical significance between 2 groups was determined by a 2-tailed unpaired t test, except for the human culture experiment in which a paired t test was performed. Data with >2 groups were assessed for significance using 1-way or 2-way analysis of variance (ANOVA) followed by specific post hoc test, as noted in the figure legends. Tukey test was used to make every possible pairwise comparison, whereas Dunnett correction was used to compare every group with a single control. Hematocrit levels were measured at different time points from the same cohorts of mice, and data were analyzed by 2-way repeated measures analysis of variance followed by unpaired t test. Data are presented as mean ± standard error of the mean. A P value <.05 is considered as significant difference (not significant, P > .05; ∗P < .05; ∗∗P < .01; and ∗∗∗P < .001).

Additional methods are available in the supplemental Methods.

Our previous work showed that the commitment of SEPs to erythroid differentiation is driven by changes in signals made by the niche. Proinflammatory signals, such as TNF-α, IL-1β, and Wnt factors, promote proliferation, but their expression is decreased during commitment to differentiation, and they are replaced by proresolving signals such as prostaglandin E2 and prostaglandin J2.^17,18 We hypothesized that metabolites generated by these signals could contribute to the regulation of cell proliferation and differentiation. We performed liquid chromatography–mass spectrometry analysis to profile the changes of metabolites extracted from bulk SEPs on days 1 and 3 of stress erythropoiesis expansion media (SEEM) cultures, during which proinflammatory signals drive SEP proliferation. We observed that the endogenous proresolving metabolite itaconate decreased significantly from day 1 to day 3 of SEEM culture, which corresponds to the start of SEP proliferation (Figure 1A). Further analysis of itaconate levels in SEPs on days 3 and 5 during expansion culture, compared with days 1, 2, and 3 of SEDM culture, showed that itaconate increases on day 5 of expansion, which correlates with increase in stress BFU-E and peaks on day 1 of differentiation and is maintained at that level through day 3 (Figure 1B).¹⁶ Immunoresponsive gene 1 (Irg1) encodes the enzyme that catalyzes itaconate synthesis by the decarboxylation of cis-aconitate.³⁰ Analysis of Irg1 messenger RNA (mRNA) and protein showed that in Kit⁺ SEPs, mRNA increases from SEEM day 3 to day 5 and then decreases after the cultures are placed in SEDM. In contrast, the protein levels increase on SEDM day 1 and are maintained on day 5 (Figure 1C; supplemental Figure 1A). We observed similar Irg1 mRNA and protein expression levels in the stromal cells of the culture.

Itaconate is an anti-inflammatory mediator,³¹ and the levels of itaconate were lower when SEPs were proliferating and increased during the transition to differentiation. NO-dependent signaling promotes the proliferation of SEPs and inhibits their differentiation (Ruan et al²⁵). We hypothesized that treating cells with a cell permeable form, 4-octyl-itaconate (OI),^30,32 could decrease the proinflammatory signaling that drives the expansion of SEP progenitor populations. OI impaired SEP expansion (Figure 2A). In fact, if OI was added at the start of culture, SEPs failed to proliferate. OI-treated SEEM cultures displayed fewer more rapidly proliferating late-stage Kit⁺Sca1⁺CD34^–CD133⁺ and Kit⁺Sca1⁺CD34^–CD133^– SEPs, whereas the numbers of the most immature Kit⁺Sca1⁺CD34⁺CD133⁺ progenitors were not affected (Figure 2B-C; supplemental Figure 1B). Treatment with OI decreased the mean fluorescent intensity of NO in SEPs, including immature Kit⁺Sca1⁺ CD34⁺CD133⁺ progenitors whose numbers were not decreased by OI treatment, which suggests that different progenitors have different requirements for NO (Figure 2D). Our analysis showed that OI reduced the levels of Nos2 mRNA in SEPs, but the levels of Nos2⁺ SEPs, as measured by flow cytometry, was not significantly affected (Figure 2E; supplemental Figure 1C). However, Nos2⁺ stromal cells were decreased by OI treatment (Figure 2F). Conversely, the defects in SEP proliferation induced by OI treatment were rescued by treatment with the NO donor S-nitroso-N-acetyl-DL-penicillamine (SNAP)³³ at either 10 or 50 μM, indicating that itaconate impairs proliferation by decreasing NO levels (Figure 2G). We observed the opposite when we cultured Irg1^–/– cells, because mutation of Irg1 accelerated the transition to more mature Kit⁺Sca1⁺CD34^–CD133^– cells, whereas Nos2 mRNA expression increased (Figure 2H). As shown above, Nos2 is expressed in both SEPs and stromal cells. Nos2 function is not cell autonomous because NO generated by stromal cells rescues Nos2^–/– SEPs (Ruan et al²⁵). In contrast, Irg1 function is cell autonomous because Irg1^–/– SEPs exhibit a defect even when cocultured with WT stroma (supplemental Figure 1D). These data support a specific role for itaconate in SEPs.

Itaconate alkylates Keap1 and acts as a potent inducer of Nrf2 activity, which is a central regulator of the response to oxidative stress.³² Nrf2 is also involved in anti-inflammatory response, and this function is essential for the immunomodulatory role of itaconate in activated macrophages.³² We hypothesized that itaconate inhibits SEP proliferation by promoting Nrf2 activation. Similar to what we observed with Irg1^–/– mutations, Nrf2^–/– SEPs cultured on WT stroma exhibited a defect in differentiation that was as severe as Nrf2^–/– cultures. However, consistent with the work of Gbotosho et al, which showed a requirement for Nrf2 signaling in macrophages during stress erythropoiesis, we also observed a smaller defect in differentiation when control SEPs were grown on Nrf2^–/– stroma.³⁴ These data demonstrate a cell-autonomous role for Nrf2 in SEPs and a smaller autonomous role in the stroma (supplemental Figure 2A-D). The expression of Nrf2 protein increased in SEPs and the stroma when cells were shifted to SEDM, which was similar to what we observed for Irg1 (Figure 3A; supplemental Figure 2E). Nrf2 transcriptional activity, as measured by mRNA expression of NAD(P)H:quinone oxidoreductase 1 (Nqo1), a direct Nrf2 target, showed low levels of expression on day 5 of SEEM culture but increased further when cells were moved to SEDM (supplemental Figure 2F).

Nrf2-deficient SEPs initially grew faster than WT controls; however, there was no difference in total cell numbers on day 5 (Figure 3B). Nrf2^–/– cultures contained more late-stage Kit⁺Sca1⁺CD34^–CD133^– progenitors and fewer immature Kit⁺Sca1⁺CD34⁺CD133⁺ SEPs than WT cultures (Figure 3C). No differences in the number of F4/80⁺Vcam1⁺ macrophages in the stroma were observed (supplemental Figure 2G). The similarities in expression and phenotypes of Irg1 and Nrf2 mutant progenitors suggest a model in which itaconate levels regulate Nrf2 activity, which in turn fine-tunes SEP proliferation. To verify this mechanism, WT and Nrf2^–/– SEEM cultures were supplemented with OI to increase Nrf2 activity. This treatment increased Nqo1 mRNA expression in SEPs, which was blocked in Nrf2^–/– SEPs (Figure 3D). Furthermore, the defect in SEP proliferation caused by the addition of OI was rescued by Nrf2 mutation in Nrf2^–/– SEPs (Figure 3E). We observed a similar response when cultures were treated with dimethyl fumarate (DMF), a second known activator of Nrf2 (supplemental Figure 2H-K). These data demonstrate that itaconate increases Nrf2 activity, promoting the proliferation of SEPs.

We next examined whether Nrf2 suppresses the inflammatory signals required for SEP expansion. OI treatment decreased Nos2 mRNA expression, and this effect was compromised in the Nrf2^–/– SEPs (Figure 3F). We further confirmed the role of Nrf2 in regulating SEP proliferation because treatment of SEEM cultures with OI, DMF, or another Nrf2 activator tert-butylhydroquinone³⁵ decreased the proliferation of Kit⁺Sca1⁺ SEPs (supplemental Figure 2L). Our previous data showed that in immature SEPs, mRNA expression of Hif-1α and Pdk1 promotes glycolysis, which provides anabolic metabolites for cell proliferation.¹⁹ OI or DMF treatment decreased hypoxia inducible factor 1α (Hif-1α) and pyruvate dehydrogenase kinase 1 (Pdk1) mRNA expression in proliferating SEPs, and this decrease was reversed by Nrf2 mutation (supplemental Figure 3A-B), suggesting that activation of Nrf2 disrupts the inflammatory metabolism required for SEP proliferation.

Previously, we showed that Epo signaling in the niche promotes the transition of proliferating progenitors to erythroid differentiation.^14,16,18 Gene set enrichment analysis of RNA-sequencing (RNA-seq) data from SEPs isolated from stress erythropoiesis cultures switched from SEEM to SEDM showed an enrichment in genes in erythroid pathways in SEDM cultures, whereas SEPs from SEEM cultures showed enriched expression of genes associated with inflammatory pathways (supplemental Figure 4A). The resolution of inflammation was coupled with a profound switch of metabolism, including increased levels of itaconate, which was mirrored by increased protein expression of Irg1 in Kit⁺Sca1⁺ SEPs (Figure 1C; supplemental Figure 4B). These data suggest that the transition to differentiation increases the production of itaconate, which drives an anti-inflammatory response to promote SEP differentiation. To examine its role in differentiation, we performed SEDM cultures using control or Irg1^–/– bone marrow cells, in which itaconate production was completely impaired. We restored itaconate levels in Irg1^–/– cultures with the supplementation of OI. Compared with controls, Irg1^–/– SEPs had elevated levels of Nos2 protein and NO production, but treatment with OI decreased Nos2 protein and NO levels to levels comparable with control cells (Figure 4A-C). To test whether itaconate promotes differentiation via NO suppression, we isolated SEPs from control and Irg1^–/– SEDM cultures treated with and without the Nos2-specific inhibitor, 1400W.³⁶ Treatment of WT SEDM cultures with 1400W led to increased stress BFU-E and a superinduction of erythroid genes (Ruan et al²⁵). Irg1-deficient progenitors generated fewer stress BFU-Es and mature Kit⁺Sca1^–CD34^–CD133^– SEPs and expressed lower mRNA levels of representative erythroid genes, including Epo receptor (EpoR), Gata1, the heme biosynthetic enzyme, coproporphyrinogen oxidase (Cpox), and β-major globin (Hbb-b1), indicating that itaconate production is required for the transition to erythroid differentiation (Figure 4D-F). This defect in differentiation was rescued by treatment with 1400W, a Nos2-specific inhibitor. These data demonstrate that itaconate promotes erythroid differentiation by inhibiting Nos2-dependent NO production.

We next investigated the role of Irg1 in vivo during recovery from HKBA-induced inflammatory anemia.^29,37,38 Untreated Irg1^–/– mice exhibited similar levels of SEPs in their spleens and stress BFU-E compared with WT (supplemental Figure 5A). Despite this similarity, Irg1^–/– mice treated with HKBA exhibited a significant delay in recovery over 28 days (Figure 5A). On day 8 after HKBA treatment, anemia of control mice started to improve, and over the next 8 days, their hematocrit levels significantly increased. We examined Irg1^–/– and control HKBA-treated mice during this critical period in recovery on days 8, 12, and 16. We observed that Irg1^–/– mice showed continued decreases in hemoglobin and red blood cell concentration during this time (Figure 5B). However, this defect of stress erythropoiesis is not due to a lack of SEPs in the spleen, because spleen weight and spleen cellularity were increased in Irg1^–/– mice (Figure 5C). The defect is in the differentiation of SEPs. The percentage of Kit⁺Sca1^–CD34^–CD133^– SEPs was significantly decreased in Irg1^–/– mice, whereas the percentage of Kit⁺Sca1⁺CD34^+/–CD133⁺ immature cells increased (Figure 5D-E). This decrease in mature SEPs translated to a lower frequency of stress BFU-E at each time point and fewer overall stress BFU-E on days 8 and 16 (Figure 5F). Analysis of Nos2 expression in the spleens showed that Irg1^–/– mice had increased levels of Nos2, supporting the role for itaconate synthesis in suppressing NO-dependent inhibition of erythroid differentiation (Figure 5G).

We also tested the response of Irg1^–/– mice to PHZ-induced acute hemolytic anemia. Nrf2 protein levels increased in the spleen on days 3 and 5 during recovery from PHZ-induced anemia (Figure 6A; supplemental Figure 5B). These data are consistent with in vivo metabolomics analysis on days 1 and 3 after PHZ treatment, which showed increased levels of itaconate in Kit⁺ SEPs (Figure 6B). However, Irg1^–/– mice had less Nrf2 protein in the spleen on day 3 compared with control mice (Figure 6A). Analysis of Nrf2 target gene expression showed a delay on day 1 after PHZ treatment (Figure 6C). The mRNA expression of erythroid genes was similarly delayed in Irg1^–/– mice; however, this decrease occurred only at early time points, with expression increasing at later time points during recovery (Figure 6D). The increase in stress BFU-E was also delayed, which resulted in Irg1^–/– mice reaching their lowest hematocrit levels a day earlier than control mice (Figure 6E). These defects were not caused by lower levels of Epo or a delay in increasing Epo levels in the serum (supplemental Figure 5C). Although Irg1^–/– mice exhibited defects, they were transient, and the Irg1^–/– mice eventually reached levels of erythroid gene expression, stress BFU-E, and hematocrit similar to controls (Figure 6D-E). These data suggest that an alternative mechanism for activating Nrf2 compensates during recovery from acute anemia.

Culturing SEPs in SEEM maintains the transient amplifying population of erythroid progenitors. Switching the cultures to SEDM leads to commitment to erythroid differentiation and loss of self-renewal ability.^16,18 This switch leads to an increase in erythroid gene expression and a loss of proinflammatory gene expression (supplemental Figure 4A). In contrast, Gene set enrichment analysis of our RNA-seq data of control and Nrf2^–/– SEPs isolated on SEDM day 3 showed that Nrf2 mutant SEPs fail to upregulate erythroid genes.³⁹ Conversely, they maintain the expression of proinflammatory signals (Figure 7A-B). Furthermore, the Nrf2^–/– SEPs fail to increase the expression of genes involved in ribosome biogenesis and amino acid metabolism, suggesting a decline in translational efficiency for hemoglobin production (supplemental Figure 6A). The RNA-seq data are consistent with data showing that mutation of Nrf2 blocks the ability of exogenous OI to inhibit proliferation of SEPs cultured in SEEM (Figure 3D-F). These data suggest that the anti-inflammatory signals provided by itaconate promote SEP differentiation through Nrf2 activation. To demonstrate that Nrf2 activation drives differentiation, we cultured SEPs from WT control and Irg1^–/– mice in SEDM media. Irg1^–/– SEPs expressed significantly lower mRNA levels of representative erythroid genes, including EpoR, Gata1, and Cpox, than WT controls. However, when the cultures were treated with OI or DMF, a known Nrf2 activator, we observed significantly increased expression of EpoR, Gata1, and Cpox with OI, whereas DMF significantly increased EpoR and showed a trend toward increased Gata1 and Cpox mRNA expression (Figure 7C). Analysis of BFU-E colony-forming cells showed that mutation of Irg1 significantly decreased the frequency of BFU-E generated in the culture. Treatment with OI significantly increased the frequency BFU-E in Irg1^–/– cultures, and to a lesser extent, so did DMF (Figure 7D). These data show that increasing Nrf2 activity drives the differentiation of SEPs that lack the ability to generate itaconate and underscore the ability of itaconate to promote differentiation.

Protective immunity must balance the need to increase the production of myeloid effector cells with the need to maintain erythroid homeostasis. To accomplish these goals, proinflammatory cytokines that increase myelopoiesis also promote stress erythropoiesis to compensate for the loss of steady-state erythroid output.^13,40 Here, we present data that further underscore how changes in inflammatory signals regulate stress erythropoiesis. TNF-α and NO play key roles during the expansion stage, but the transition to differentiation is characterized by a loss of proinflammatory signals and increase in anti-inflammatory signals and a change in metabolism (Ruan et al²⁵).^17,41 Our data show that increased production of itaconate in SEPs in part catalyzes this transition. One target of itaconate is Nrf2, and our data are consistent with Nrf2 and Irg1 acting in a cell-autonomous manner. Nrf2 is well-known as a regulator of oxidative stress,^32,42 and the increased Nrf2 activity coincides with decreases in NO as SEPs transition to differentiation. Decreasing NO production is a key target of this itaconate-Nrf2 pathway. Our data showed that itaconate decreased Nos2 mRNA, but the protein levels and Nos2⁺ cells, as identified by flow cytometry, did not decline to similarly low levels. Despite the lack of a decrease in Nos2 protein, NO mean fluorescent intensity decreased in SEPs upon itaconate treatment. NO production can be regulated at multiple levels. Arginine can be used by Nos2 to make NO or generate polyamines through the action of arginase 1.⁴³ We observed that arginase 1 mRNA expression increases when SEPs were switched into differentiation media, supporting the idea that arginine metabolism changes during commitment to differentiation (data not shown).

Although we have focused on the role of itaconate in activating the Nrf2 pathway, other known itaconate targets could also play a role in stress erythropoiesis. Itaconate is known to inhibit Tet2 to dampen inflammatory responses in macrophages.⁴⁴ Recent work from Tseng et al showed that Tet2 plays a role in SEP differentiation.⁴⁵ Further work will be needed to delineate the roles of itaconate and Tet2 during the transition to differentiation. Although we have focused on events in SEPs, itaconate is a well-known anti-inflammatory metabolite in macrophages, and others have shown that mutation of Nrf2 decreases macrophage populations in erythroblastic islands in the bone marrow and spleen during recovery from phlebotomy.³⁴ These data suggest that these mediators could also affect the niche, and the extent of that contribution is not known.

Marcero et al showed that exogenous itaconate could inhibit heme biosynthesis in MEL cells.⁴⁶ Our analysis examined SEPs that are more immature than MEL cells, which correspond to poststress BFU-E and late-stage SEPs.⁴⁷ These data suggest that itaconate production by SEPs and the niche must be decreased for terminal differentiation. Our previous work showed that peroxisome proliferator activated receptor γ (PPARγ) activation in niche macrophages plays a role in promoting the transition to differentiation.¹⁸ Irg1 expression is regulated by PPARγ, so increased PPARγ activity at this time could decrease itaconate levels so as not to impede heme biosynthesis.⁴⁸ This idea is consistent with our RNA-seq data that showed that PPARγ expression increases more than threefold when SEPs commit to differentiation. Future work will be needed to address this question.

Irg1^–/– mice exhibit defects in recovery from HKBA and, to a lesser extent, from PHZ-induced anemia. The mutant mice eventually recover from PHZ-induced anemia, and the expression of Nrf2 target genes during recovery is only delayed in Irg1^–/– mice. These data suggest that a signal other than itaconate increases Nrf2 activity. Our preliminary data suggest that IL-10 plays a role in maintaining Nrf2 activity during differentiation of SEPs. Its expression increases both in vitro and in vivo when SEPs commit to differentiation. Exogenous IL-10 increases stress BFU-E formation (data not shown). IL-10 has not been previously implicated in stress erythropoiesis. In fact, transgenic overexpression of IL-10 increases myelopoiesis and causes anemia, suggesting that the levels, duration, and site of IL-10 expression may affect the response in the erythroid lineage.⁴⁹ Similarly, other nonerythroid cytokines may play a role in stress erythropoiesis. IL-33 inhibits bone marrow erythropoiesis and can cause anemia, but it also induces the formation of iron-recycling macrophages and increases the production itaconate and anti-inflammatory factors, including IL-10.^7,50 These data suggest that other proinflammatory or alarmin signals could play roles in stress erythropoiesis that are distinct from their effects on steady-state erythropoiesis.

In summary, our data show that inflammatory signals that induce NO production are used by stress erythropoiesis to expand a population of immature progenitors, which is followed by a regulated resolution of inflammation to ensure a successful transition of early progenitors into mature erythrocytes to restore homeostasis.

The authors thank the members of the Paulson Laboratory for suggestions on the work, especially Yuting Bai for her help with mice blood collection. Sougat Misra is also thanked for helpful discussion about metabolism. These studies were greatly helped by Rajeswaran Mani and the Huck Institutes of the Life Sciences Flow Cytometry Facility (RRID:SCR_024460) and Ashley Shay, Philip Smith, and Justin Munro and the Huck Institutes of the Life Sciences Metabolomics Facility (RRIS:SCR_023864).

This work was supported by National Heart, Lung and Blood Institute, National Institutes of Health (grant HL146528 [R.F.P.]) and National Institute of Diabetes, Digestive and Kidney Diseases. National Institutes of Health (grant DK138865 [R.F.P.]), the United States Department of Agriculture, National Instititute of Food and Agriculture Hatch Projects (PEN04960, accession number 7006577 [R.F.P.]; PEN04932, accession number 7006585 [K.S.P.]; PEN04275, accession number 1018544 [M.A.H.]; and PEN04917, accession number 7006412 [A.D.P.]), National Cancer Institute grant R01CA239256 (M.A.H.), startup funds from the College of Agricultural Sciences, Pennsylvania State University (M.A.H.), the Dr Frances Keesler Graham Early Career Professorship from the Social Science Research Institute, Pennsylvania State University (M.A.H.), and Tobacco Commonwealth Universal Research Enhancement funds from the Pennsylvania Department of Health (A.D.P. and I.K.). S.T. was supported by National Institute of Diabetes, Digestive and Kidney Diseases grant T32DK120509, and A.S. was supported by National Institute of General Medical Sciences grant T32GM108563.

Contribution: B.R., K.S.P., and R.F.P. conceived and designed the study; B.R., S.T., M.C., A.S., H.G., and Y.C. performed the experiments; B.R., J.M., and M.A.H. analyzed RNA-sequencing data; B.R., I.K., J.C., and A.D.P. performed metabolomics analysis; B.R. and R.F.P. wrote the initial draft of the manuscript; and all authors were involved in review and editing.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Robert F. Paulson, Department of Veterinary and Biomedical Sciences, Pennsylvania State University, 203 AVBS Building, 439 Shortlidge Rd, University Park, PA 16802-3500; email: rfp5@psu.edu.