PU.1 ^KI/KI mice were provided as a kind gift by Dr. Dan Tenen (Harvard Stem Cell Institute). PU.1 ^KI/KI mice were bred to C57BL/6J mice (strain #000664) from The Jackson Laboratory to generate PU.1 ^+/+ littermate controls for the study. 6–12-week-old animals were used for experiments, and animals of both sexes were used in these studies. Mice were euthanized by CO₂ inhalation followed by cervical dislocation. All animal experiments and euthanasia procedures were conducted in accordance with approved procedures reviewed by the Institutional Animal Care and Use Committee (IACUC) at University of Colorado Anschutz Medical Campus (protocol number 00091).

0.5 μg of IL-1β (Peprotech) suspended in sterile D-PBS/0.2% BSA, or D-PBS/0.2% BSA alone as a -IL-1β control, was injected intraperitoneally (i.p.) in a 100 μL bolus daily for 20 days, as previously described in the literature. In vivo puromycin incorporation assays were performed by injecting 500 μg puromycin in a 100 μL bolus intraperitoneally 1 h prior to euthanasia, following previously published protocols.

For analysis of BM cell populations, we used an identical protocol as our previously published work. BM was flushed from the four long bones (femurs + tibiae) of mice using a syringe and 21G needle filled with staining media (SM: 2% heat-inactivated FBS in HBSS without Ca²⁺ or Mg²⁺). Cells were subsequently pelleted at 500 x g and resuspended in 1x ACK (ACK 150 mM NH₄Cl/10 mM KHCO₃) on ice for 3–5 min to deplete red blood cells prior to washing with SM and filtering through a 70 micron nylon mesh to remove debris. Total cell numbers were determined using a ViCell automated counter (Beckman-Coulter) and 1 × 10⁷ BM cells were used for staining. To identify HSPC populations, BM cells were stained for 30 min on ice with PE-Cy5-conjugated anti-CD3, CD4, CD5, CD8, Gr-1 and Ter119 to exclude mature lineage + cells, plus Flk2-biotin, Mac-1-PE/Cy7, FcγR-APC, CD48-A700, and cKit-APC/Cy7. Purified rat IgG was also included as a blocking agent. Following a wash step, BM cells were stained with Sca-1-BV421, CD41-BV510, CD150-BV785, and streptavidin (SA)-BV605 in SM with a 1:4 dilution of BD Brilliant Buffer for 30 min on ice. For analysis of mature myeloid cells, a staining cocktail containing Gr-1-Pacific Blue, Ly6C-BV605, Mac-1-PE/Cy7 and rat IgG in SM with a 1:4 dilution of Brilliant Buffer. Prior to analysis, BM cells were counterstained with 1 μg/mL propidium iodide (PI) and analyzed on a 3-laser, 12 channel FACSCelesta analyzer (Becton-Dickenson) or a 4-laser, 16 channel LSRII analyzer. For splenocyte analyses, spleens were minced through a 70 micron filter basket to create a single cell suspension, which was subsequently pelleted and depleted of red blood cells with 1x ACK as described above for BM cells. 1 × 10⁷ splenocytes were subsequently stained with an identical cocktail as above to read out mature hematopoietic cells.

For cell cycle analysis, 1 × 10⁷ RBC-depleted BM cells were stained with a variation of the BM HSPC cocktail, with each antibody stain performed on ice for 30 min: PE-Cy5-conjugated anti-CD3, CD4, CD5, CD8, Gr-1 and Ter119 as a lineage exclusion stain, Flk2-biotin, Sca-1-PE/Cy7, CD48-A700, and c-Kit-APC/Cy7, followed by a cocktail of 1:4 Brilliant Buffer:SM containing SA-BV605 and CD150-BV785. Cells were subsequently fixed and stained for Ki67 as described previously: After washing in SM, cells were fixed in Cytofix/Cytoperm buffer (Becton-Dickenson) for 20 min at room temperature (RT), washed in PermWash (BD) and permeablized using Perm Buffer Plus (BD) for 10 min at RT. Cells were again washed in PermWash, re-fixed in Cytofix/Cytoperm for 5 min, washed in PermWash and incubated with anti-Ki67-PerCP-Cy5.5 diluted in PermWash for 30 min at RT. Prior to analysis on an LSRII, cells were counterstained with DAPI diluted to 1 μg/mL in D-PBS.

For puromycin staining, an identical staining panel was used as for cell cycle analysis. Following fixation with Cytofix/Cytoperm performed as with cell cycle analysis, cells were stained with anti-puromycin antibody diluted in PermWash for 1 h at RT. Cells were subsequently washed in PermWash buffer, incubated with an anti-mouse IgG2a-APC secondary antibody for 30 min at RT, washed with PermWash and resuspended in SM for analysis on an LSRII.

For analysis of cells from liquid culture, cells were stained using a similar protocol as BM, except without RBC depletion. Cells were resuspended and half the content of the well was transferred to a FACS tube and stained with Sca-1-PE/Cy7, c-Kit-APC/Cy7, Mac-1-APC, FcγR-BV711, CD18-PE, MCSFR-BV605, and Gr1-Pacific Blue using an identical approach to that described for analysis of mature BM cells. Cells were analyzed on a 3-laser, 12-channel FACSCelesta (Becton-Dickenson). A complete list of antibodies including clone information, manufacturer and dilution can be found in Supplementary Table S3.

To analyze purified SLAM HSC and MPP^GM, we harvested BM from all arm and leg bones as well as hips from mice by gently crushing bones in a mortar and pestle. BM cells were subsequently RBC depleted, passed through a 70 micron nylon mesh, and suspended on a Histopaque 1119 gradient (Sigma-Aldrich) to remove debris. BM was then enriched for c-Kit + cells by incubating on ice for 20 min with c-Kit microbeads, followed by column-based separation on an AutoMACS Pro magnetic cell separator using the Posseld2 setting. Enriched cells were subsequently stained as described above for BM HSPC analysis. Cells were subsequently sorted on a 4-laser FACSAria Fusion (Becton Dickenson) instrument equipped with a 100 micron nozzle.

Purified cells were cultured using an identical protocol as previously published. Cells were grown for 4 days in culture-treated sterile 96-well plates containing StemPro 34 serum-free medium supplemented with L-Glutamine and Anti-anti (both 100x from Gibco), in addition to the following cytokines: SCF (25 ng/mL), TPO (25 ng/mL), IL-3 (10 ng/mL), GM-CSF (20 ng/mL), Flt3L (50 ng/mL), IL-11 (50 ng/mL), EPO (4 U/mL), and ±IL-1β (25 ng/mL) at 37°C in 5% CO₂. For methylcellulose assays, 5 × 10² cultured cells were transferred to 3 cm gridded dishes containing methylcellulose medium (StemCell Technologies; M3231) supplemented with the above cytokines and without IL-1β. Colonies were counted after 1 week and 1 × 10⁴ progeny cells were re-plated into fresh methylcellulose to measure serial clonogenic activity.

Cells were sorted from mouse bone marrow as described above and analyzed for gene expression as performed previously (Reynaud et al., 2011; Pietras et al., 2015; Pietras et al., 2016; Hernandez et al., 2020; Rabe et al., 2020; Chavez et al., 2021; Ahmed et al., 2022; Chavez et al., 2022). Cells were sorted at 100 cells per well in 5 μL CellsDirect reaction buffer (Invitrogen). RNA was then reverse transcribed and preamplified with a panel of 96 DeltaGene Assay primer sets (Fluidigm) for 20 rounds with Superscript III (Invitrogen) and subsequently treated with Exonuclease I (New England Biolabs) to remove non-target genetic material. cDNA was then diluted in DNA suspension buffer and loaded onto Fluidigm 96.96 Dynamic Array IFCs along with the DeltaGene Assay primers and run on a Biomark HD (Fluidigm) using SsoFast Sybr Green (Bio-Rad) as a detector. Data were analyzed using the ΔΔCT method and normalized to Gusb expression. Hierarchical clustering and PCA analyses were performed using ClustVis. As Gusb was used to normalize data, it was not included in clustering and PCA analysis. A complete list of all Fluidigm primer sequences can be found in Supplementary Table S4. Hoxa2 and Ebf1 were excluded from all analyses due to poor primer performance in our studies.

Primary ChIP-seq datasets from LSK/Flk2^-/CD150⁺ cells and ChIP-seq data from PU-ER, BMDM and thioglycolate-elicited primary mouse macrophages GSE21512 were analyzed as previously published (Chavez et al., 2021). Fastq files were trimmed using Cutadapt and mapped to the mm10 mouse genome using HISAT2. Peak calling was performed using HOMER in factor mode with an FDR of <0.001. Intergenic peaks nearest to a TSS were annotated as the corresponding gene. Peak data were visualized from Bigwig files using the Internet Gene Viewer (IGV) application (igv.org/app).

Statistical analyses were performed on GraphPad Prism v9.4.1. ANOVA with Tukey’s test were used for multivariate comparisons as described in the figure legends. p-values of 0.05 or less were considered statistically significant and are notated in the figures using asterisks.

To address the impact of PU.1 deficiency on the blood system under chronic inflammatory conditions, we analyzed the hematological parameters of PU.1 ^+/+ mice versus PU.1 ^KI/KI mice (Staber et al., 2013) treated for 20 days ± IL-1β (0.5 μg/day via intraperitoneal injection) (Figure 1A). Complete blood counts (CBC) showed no abnormalities in the abundance of myeloid, lymphoid, and erythroid cells in PBS-treated control PU.1 ^KI/KI mice relative to PU.1 ^+/+ controls (Figure 1B, Supplementary Figure S1A). Chronic IL-1β exposure triggered significant increases in peripheral blood neutrophils in PU.1 ^{+ I +} mice, consistent with our prior published findings (Figure 1B). Strikingly, this phenotype was exacerbated in IL-1β-treated PU.1 ^KI/KI mice. (Figure 1B). These alterations appeared confined to the myeloid lineage, as we observed no significant changes in lymphoid cell numbers and a similar degree of inflammation-induced anemia in these animals (Supplementary Figure S1A). Likewise, we observed aberrant myeloid expansion in the spleens of IL-1β-treated PU.1 ^KI/KI mice relative to IL-1β-treated PU.1 ^+/+ mice, which was also accompanied by an overall increase in spleen mass (Figures 1C, D; Supplementary Figure S1B, C). In the bone marrow (BM), myeloid cell numbers in the BM of IL-1β-treated PU.1 ^KI/KI mice were expanded with IL-1β treatment but not to an extent significantly different than their wild-type counterparts (Figure 1E), suggesting excess cells are likely being mobilized from the BM to the blood and spleen. Collectively, these data show that chronic inflammatory challenge elicits overproduction of myeloid cells in PU.1 ^KI/KI mice.

To assess the impact of PU.1 deficiency on the dynamics of hematopoietic stem and progenitor (HSPC) populations in response to chronic IL-1β exposure, we analyzed the abundance of phenotypic HSPC in the BM of PU.1 ^KI/KI and PU.1 ^+/+ mice treated for 20 days ± IL-1β (Figure 2A; Supplementary Figure S2A). We observed a trending increase in granulocyte/macrophage progenitors (GMP; Lin⁻/c-Kit⁺/CD41^-/CD150^-/FcγR⁺) (Pronk et al., 2007) in IL-1β-treated PU.1 ^KI/KI BM relative to PU.1 ^+I+ controls. (Figures 2B, C). Chronic IL-1β also triggered significant expansion of HSC (HSC; Lin⁻/c-Kit⁺/Sca-1⁺/Flk2^-/CD48^-/CD150⁺) (Kiel et al., 2005) and MPP populations, specifically the megakaryocyte/erythroid-biased multipotent progenitor (MPP^MkE; Lin⁻/c-Kit⁺/Sca-1⁺/Flk2^-/CD48⁺/CD150⁺; also known as MPP2) and granulocyte/macrophage-biased (MPP^GM; Lin⁻/c-Kit⁺/Sca-1⁺/Flk2^-/CD48^-/CD150^-; also known as MPP3) in PU.1 ^+/+ mice (Figures 2D, E; Supplementary Figure S2A), consistent with prior reports (Cabezas-Wallscheid et al., 2014; Pietras et al., 2015; Challen et al., 2021). Notably however, expansion of MPP^GM was significantly potentiated in PU.1 ^KI/KI mice relative to IL-1β-treated PU.1 ^+/+ controls (Figures 2D, E). Collectively, these data suggest the MPP^GM population may serve as a key axis of aberrant myeloid expansion following chronic inflammatory challenge in PU.1 ^KI/KI mice.

We previously showed that PU.1 deficiency leads to increased proliferation in HSC^LT following IL-1β stimulation, thereby driving expansion of these cells in vivo (Chavez et al., 2021). To assess whether MPP^GM expansion was likewise related to increased cell cycle activity, we analyzed cell cycle distribution via Ki67/DAPI staining in PU.1 ^KI/KI MPP^GM following treatment for 20 days ± IL-1β (Figure 3A). While MPP^GM from PU.1 ^KI/KI mice −IL-1β exhibited a higher proportion of cells in G₀ relative to PU.1 ^+/+ controls, IL-1β treatment significantly and selectively potentiated cell cycle activity in PU.1 ^KI/KI MPP^GM mice (Figure 3B; Supplementary Figure S2B). Taken together, these data indicate that IL-1β triggers increased cell cycle activity in PU.1 ^KI/KI MPP^GM, thereby contributing to the aberrant expansion of this population.

Next, we analyzed gene expression patterns in PU.1 ^+/+ and PU.1 ^KI/KI MPP^GM from mice treated for 20 days ± IL-1β using our custom Fluidigm qRT-PCR gene expression array. This approach allowed us to measure expression of 94 genes critical for HSPC function (Supplementary Table S4). As anticipated, expression of Spi1 (PU.1) was significantly reduced in PU.1 ^KI/KI MPP^GM (Supplementary Figure S3A). IL-1β was still able to induce Spi1 expression in PU.1 ^KI/KI MPP^GM, albeit at significantly reduced levels relative to PU.1 ^+/+ MPP^GM (Supplementary Figure S3A), likely reflecting the capacity of inflammation-induced signals to trigger PU.1 expression independently of the −14 kb URE PU.1 autoregulatory binding site (Ahmed et al., 2022). Furthermore, we did not notice overt defects in Il1r1 expression levels in PU.1 ^KI/KI MPP^GM (Supplementary Figure S3A). To evaluate overall differences in gene expression between PU.1 ^+/+ and PU.1 ^KI/KI MPP^GM ± IL-1β, we next performed hierarchical clustering analysis (Pearson correlation with average linkage). MPP^GM samples clustered predominantly by genotype and secondarily by treatment condition (Supplementary Figure S3B). MPP^GM samples likewise were clearly distinguished by principal component analysis (PCA), with PC1 appearing to discriminate samples based on relative PU.1 activity and PC2 distinguishing PU.1 ^+/+ controls from all other samples (Supplementary Figure S3C), collectively indicating unique gene programs present in each genotype and treatment. Given the changes in cell cycle activity triggered by IL-1β, we examined the expression of cell cycle genes in MPP^GM. Notably, genes such as Ccne1, Cdk6 and Myc were repressed in PU.1 ^+/+ MPP^GM following IL-1β exposure (Figure 3C). However PU.1 ^KI/KI MPP^GM failed to robustly repress these genes following in vivo IL-1β treatment (Figure 3C). Collectively, these data support a model in which PU.1 is required to repress expression of cell cycle genes to maintain normal MPP^GM cell cycle activity following chronic inflammatory challenge.

While PU.1-deficient MPP^GM exhibit increased cell cycle activity, concomitant disruption of gene programs associated with differentiation is likely required for their selective expansion under chronic inflammatory challenge. Thus, we surveyed expression of key genes associated with self-renewal in MPP^GM from mice treated for 20 days ± IL-1β (Figure 4A). Consistent with the capacity of IL-1β to trigger rapid myeloid differentiation in HSPC, expression of these genes was significantly reduced in PU.1 ^+/+ MPP^GM following IL-1β exposure (Figure 4B). Strikingly, relative to PU.1 ^+/+ MPP^GM, PU.1 ^KI/KI MPP^GM exhibited aberrantly high baseline expression of Fgd5, Ctnnal1, Egr1, Bmi1 and Hoxa9 (Figure 4B). These genes were nonetheless downregulated following IL-1β exposure, but only to levels found at steady state in PU.1 ^+/+ MPP^GM (Figure 4B). These data suggest IL-1β-mediated repression of these genes may involve other transcriptional regulators aside from PU.1 including CEBPA, which regulates many of the same genes (Koschmieder et al., 2005; Pundhir et al., 2018; Higa et al., 2021). Along these lines, we observed a significant increase in Cebpa expression in PU.1 ^KI/KI MPP^GM relative to PU.1 ^+/+ MPP^GM following IL-1β treatment (Supplementary Figure S3D), suggesting Cebpa expression may be induced to compensate for PU.1 loss in PU.1 ^KI/KI MPP^GM during chronic inflammatory challenge.

To further examine the impact of PU.1 deficiency on MPP^GM, we examined the expression of key genes regulating mechanisms that maintain stem cell activity in the hematopoietic system, specifically forkhead box O3 (Foxo3), hypoxia-inducible factor-1α (Hif1a), nuclear regulatory factor-2 (Nrf2), and N-Myc (Mycn) (Figure 4C). These transcription factors regulate numerous gene programs required for stem cell function such as autophagy, glycolytic metabolism, the antioxidant response, and apoptosis resistance. IL-1β exposure triggered robust repression of these genes in PU.1 ^+/+ MPP^GM. Interestingly, PU.1 ^KI/KI MPP^GM again exhibited increased baseline expression of these factors (Figure 4C) While IL-1β exposure likewise downregulated their expression in PU.1 ^KI/KI MPP^GM, expression was again only reduced to levels found in PU.1 ^+/+ MPP^GM at steady state (Figure 4C). Furthermore, we observed similar expression patterns in key downstream target genes of these transcription factors, including genes regulating glycolysis (glyceraldehyde-3 phosphate dehydrogenase; Gapdh), antioxidant activity (glutathione S-transferase T3; Gstt3), hypoxia response (hypoxia-inducible factor 3a; Hif3a), fatty acid oxidation (carnitine palmitoyltransferase 1a; Cpt1a), and survival (B-cell leukemia 2; Bcl2) (Figure 4D). Taken together, our data suggest PU.1-deficient HSPC may be metabolically poised to support cell cycle activity, while maintaining self-renewal activity via disruption of differentiation gene programs. These observations are broadly consistent with prior work showing that PU.1 binds to and represses key genes in glycolysis and lipid biosynthesis pathways used for production of energy and anabolic factors that support cell proliferation (Solomon et al., 2017).

We next assessed whether the genes investigated above possess PU.1 binding sites. Thus, we queried our previously published PU.1 ChIP-seq datasets in which we assessed PU.1 binding in LSK/Flk2⁻/CD150⁺ HSPC. We also compared these results with data from three other publicly available PU.1 ChIP-seq datasets (Heinz et al., 2010) in bone marrow-derived macrophages (BMDM), thioglycolate-elicited macrophages, and the PU-ER cell line (Walsh et al., 2002), which is derived from PU.1-deficient fetal HSPC expressing a tamoxifen-inducible PU.1 transgene. Notably, our ChiP-seq dataset identified PU.1 peaks associated with a majority of the genes identified in Figure 4 in primary HSPC (Supplementary Figure S4A–D; Supplementary Table S1). Furthermore, these PU.1 peaks were present in at least two of the other publicly available datasets. Collectively, these data support a model in which PU.1 negatively regulates expression of numerous self-renewal genes, with PU.1-deficient MPP^GM consequently retaining high expression levels of these factors. These data are strongly reminiscent of Cebpa-deficient MPP^GM, which likewise retained high expression levels of stem cell genes including Foxo3, Mycn, Bmi1 and Bcl2 following chronic exposure to IL-1β (Higa et al., 2021).

Given our gene expression data, we hypothesized that PU.1 ^KI/KI MPP^GM would exhibit impaired capacity to differentiate in response to IL-1β. As PU.1 ^KI/KI HSPC fail to engraft in transplantation assays, and to minimize potential impacts of BM niche signals altered by IL-1β in vivo, we used a well-defined in vitro liquid culture system to study the impact of PU.1 deficiency on the differentiation kinetics of purified MPP^GM in response to in vitro IL-1β stimulation (Figure 5A) (Pietras et al., 2016; Chavez et al., 2021; Higa et al., 2021). After 4 days of culture, we analyzed the frequency and number of immature (c-Kit⁺/Sca-1⁺) and myeloid-committed (FcγR⁺/Mac-1⁺) cells in the cultures. Strikingly, we observed a significant increase in the proportion of c-Kit⁺/Sca-1⁺ immature cells in PU.1 ^KI/KI MPP^GM cultures relative to PU.1 ^+/+ control cells regardless of IL-1β stimulation (Figures 5B–D). We subsequently confirmed the increased proportion of immature cells in the day 4 PU.1 ^KI/KI MPP^GM cultures via serial clonogenic assay (Supplementary Figure S4A, B). Conversely, we observed a lower proportion and number of myeloid-committed PU.1 ^KI/KI MPP^GM during the culture period (Figures 5E–G), including a lower overall proportion of myeloid-committed PU.1 ^KI/KI MPP^GM in the -IL-1β cultures, in line with our gene expression and c-Kit⁺/Sca-1⁺ culture data. To better understand the impact of PU.1 deficiency on myeloid surface marker expression, we analyzed surface expression of a broader panel of myeloid markers in our cultures, including CD18, MCSFR and Gr-1. Unstimulated PU.1 ^KI/KI MPP^GM exhibited significantly lower surface expression of each of these markers after 4 days culture (Figure 5H). In line with these findings, we identified multiple PU.1 peaks associated with MCSFR (Csfr1), FcγR (Fcgr2b), CD18 (Itgb2), Gr-1 (Ly6c/Ly6g) and Mac-1 (Itgam, as previously reported in (Chavez et al., 2021)) in our ChIP-seq datasets, consistent with the well-known role of PU.1 in directly binding and transducing these genes (Supplementary Figure S5C; Supplementary Table S2) (DeKoter et al., 1998; Singh et al., 1999; DeKoter and Singh, 2000; DeKoter et al., 2007). IL-1β nonetheless accelerated expression of all five myeloid surface markers in PU.1 ^KI/KI MPP^GM, again to levels roughly equivalent to unstimulated WT cells (Figure 5H). Hence, reduced PU.1 expression in PU.1 ^KI/KI MPP^GM delays the activation of myeloid differentiation programs, leading to retention of an immature phenotype following IL-1β stimulation.