Hematopoietic stem and progenitor cells improve survival from sepsis by boosting immunomodulatory cells

New therapeutic strategies to reduce sepsis-related mortality are urgently needed, as sepsis accounts for one in five deaths worldwide. Since hematopoietic stem and progenitor cells (HSPCs) are responsible for producing blood and immune cells, including in response to immunological stress, we explored their potential for treating sepsis. In a mouse model of Group A Streptococcus (GAS)-induced sepsis, severe immunological stress was associated with significant depletion of bone marrow HSPCs and mortality within approximately 5–7 days. We hypothesized that the inflammatory environment of GAS infection drives rapid HSPC differentiation and depletion that can be rescued by infusion of donor HSPCs. Indeed, infusion of 10,000 naïve HSPCs into GAS-infected mice resulted in rapid myelopoiesis and a 50–60% increase in overall survival. Surprisingly, mice receiving donor HSPCs displayed a similar pathogen load compared to untreated mice. Flow cytometric analysis revealed a significantly increased number of myeloid-derived suppressor cells in HSPC-infused mice, which correlated with reduced inflammatory cytokine levels and restored HSPC levels. These findings suggest that HSPCs play an essential immunomodulatory role that may translate into new therapeutic strategies for sepsis.


Introduction
Sepsis accounts for one in five deaths worldwide and is a common final pathway for many disease processes such as cancer, diabetes, and cardiovascular disease (1). Sepsis is an inflammatory syndrome largely driven by the activation of immune cells by pathogen associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs) (2,3). After recognizing these molecules via pattern recognition receptors (PRR), immune cells become activated and produce proinflammatory cytokines, notably interleukins IL-1, IL-6, interferons (IFNs), and tumor necrosis factor that contribute to fever, vasodilation, and multiorgan dysfunction (3)(4)(5)(6). For patients that progress to septic shock, mortality rates remain as high as 40% (7).
Leukopenia is a feature of severe sepsis that arises from apoptosis of peripheral immune cells and is an independent risk factor for death. To counteract the adverse effects of leukopenia, investigators have used immunotherapies such as GM-CSF (8) or granulocyte infusions in an attempt to restore leukocyte numbers and improve survival. These strategies have produced mixed results (9)(10)(11). The benefits of granulocyte infusions in cancer patients with fever and neutropenia are limited by the difficulty of obtaining sufficient cells and the short-lived nature of those cells (12,13).
Recent work from our group and others indicates that hematopoietic stem and progenitor cells (HSPCs) express surface receptors for cytokines, chemokines, and pathogen-associated molecular patterns (PAMPs) (14)(15)(16)(17)(18)(19)(20)(21) and respond rapidly upon direct and indirect stimulation by these signals. HSPCs, the progenitors of all blood and immune cells, are comprised of five subgroups of hematopoietic cells: hematopoietic stem cells (HSC) which have long term selfrenewal capacity, and four types of multipotent progenitors (MPP 1-4), which are defined by lower self-renewal capacity and myeloid or lymphoid differentiation biases (6,(22)(23)(24)(25)(26). Immune responses induce HSPCs in the bone marrow (BM) to produce effector immune cells via a process called emergency hematopoiesis (18,(20)(21)(22)27,28). The capacity of HSPCs to directly detect pathogen-derived molecules, cytokines, and chemokines suggests that emergency granulopoiesis can be mobilized from even the most primitive hematopoietic progenitors and that HSPCs have an active role in fighting infections. However, the extent and mechanism by which HSPC responses contribute to immunity in the acute setting remain poorly defined.
We recently showed that chronic inflammatory stress impairs HSPC quiescence and selfrenewal while promoting their activation and terminal differentiation (28). Upon direct sensing of inflammatory cytokines such as interferon gamma (IFNg), HSCs are dislodged from their normal position near quiescence-enforcing CXCL12-abundant reticular cells in the niche. Inflammatory signaling induces transcription factors such as Pu.1, CEBPb, and BATF2 (18,20,28,29) to promote myeloid differentiation, leading to the expansion of granulocyte and monocyte populations. Disruption of the homeostatic balance of self-renewal and differentiation eventually leads to depletion of the progenitor compartment (20,22,28). Collectively, these studies point toward a direct role for HSPCs in supplying the myeloid cells critical to the immune response against infection.
To test their contribution to immune responses during acute infection, we examined the role of HSPCs in a mouse model of Streptococcus pyogenes infection, also known as Group A Streptococcus (GAS). GAS is a common pathogen that causes a plethora of diseases, from mild skin infections to life-threatening necrotizing fasciitis and sepsis (6,(30)(31)(32). GAS infections can infiltrate the bloodstream and other organs, causing high systemic levels of inflammatory cytokines including IFNg, TNF, IL-1, and IL-6. As HSPCs have been shown to activate and differentiate in response to these cytokines (15,18,(20)(21)(22)(33)(34)(35), in this study we sought to determine the role of HSPCs in immune responses against infections.
Here we found that GAS infection significantly depletes HSPCs in the bone marrow. We tested the idea of infusing HSPCs to restore the hematopoietic progenitor pool. Mice treated with HSPCs showed restored HSPC numbers in the BM, increased myeloid cell production, and significantly improved overall survival. Surprisingly, HSPC infusion did not reduce pathogen burden. Instead, HSPC infusion correlated with a significant increase in the abundance of myeloid-derived suppressor cells (MDSCs) and a dampening of overall systemic inflammation. In summary, our studies indicate that HSPCs contribute to survival from sepsis by supporting the production of immunosuppressive MDSCs.

GAS infection induces trafficking of myeloid cells from the BM into circulation.
To characterize the impact of acute infections on the hematopoietic system, we inoculated mice by intramuscular injection of the hind leg with 2x10 6 colony forming units (CFU) of the model pathogen S. pyogenes strain MGAS315. To characterize differentiated hematopoietic populations during infection, we performed flow cytometric analysis of BM and peripheral blood (PB) lineage cells 24h after GAS infection ( Figure 1A) and collected serum for cytokine analyses. BM characterization of lineage cells showed a significant decrease in BM monocytes ( Figure 1C) and granulocytes ( Figure 1D) with no change in BM B-or T cells (Figure 1E & 1F). In contrast, PB lineage composition was significantly skewed towards myeloid cells with significantly higher circulating monocytes and granulocytes (Figure 1G & 1H) and lower lymphoid cells (Figure 1I &   1J). Serum cytokine characterization showed a significant increase in monocyte chemoattractant protein-1 (MCP-1; also known as CCL2) ( Figure 1B). These results suggest that myeloid cells exit the BM into circulation following an MCP-1 gradient, consistent with prior studies showing MCP-1-driven mobilization during inflammation (36,37).

GAS infection depletes bone marrow HSPCs without evidence of extramedullary hematopoiesis.
After 24h of infection, the state of HSPCs in the infected mice were analyzed by flow cytometry of bone marrow (BM) and spleen (Figure 2A) (See Table 1 for surface markers). BM cells were not gated for the common stem cell marker SCA1 (Figure 2B), since it has been previously described to be non-specifically expressed during inflammatory stress (19). The total number of HSPCs dropped significantly in just 24 hours in GAS-infected mice ( Figure 2C). More specifically, HSPC subpopulations including HSCs, multipotent progenitor 3 (MPP3s), and MPP4s were significantly lower in GAS-infected mice (Figure 2D, 2E, and 2F).
Extramedullary hematopoiesis is the proliferation and differentiation of HSCs in tissues other than the bone marrow, the canonical stem cell niche. The spleen is one of the most common sites of extramedullary hematopoiesis during infections (38). To assess whether a reciprocal increase in extramedullary hematopoiesis accompanied loss of HSPCs in the BM, we analyzed spleen tissue by flow cytometry. While there was a slight increase in total HSPCs in the spleen ( Figure 2G), there was no significant change in spleen populations that include HSCs/MPP1, MPP2s, or MPP3/4 ( Figure 2H-2J). This finding suggests that the loss of BM HSPC populations is not simply a result of migration from the BM into the spleen.

GAS infection induces HSC myeloid differentiation.
Activation of HSPCs by PAMPs or cytokines promotes their proliferation and differentiation (16,18,(20)(21)(22)39). To determine the lineage fate of endogenous HSPCs following GAS infection, After five days of intraperitoneal injections of tamoxifen, mice were inoculated with GAS or saline. Since the average mammalian cell cycle takes 24h, we decided to trace the lineage of hematopoiesis 72h post GAS infection. After these 72 hours, BM and PB was harvested for flow cytometric analysis. Analysis of the BM showed that GAS infection induced the production of new HSPCs, which includes short term HSCs and MPPs (Figure 3C-3D). In addition, there was significant labeling of CD41+ HSCs, a myeloid-biased and proinflammatory subset of HSCs ( Figure 3E) (41) and new cells of the myeloid lineage ( Figure 3F-3H). While we found a significant increase in new BM monocytes ( Figure 3G), there was no significant change in the frequency of TdTomato-labeled granulocyte/monocyte progenitors (GMPs) (Figure 3I), which may simply reflect a rapid flow through this compartment to terminally differentiated populations. We also saw no statistically significant increase in BM granulocytes ( Figure 3H); however, PB analysis showed a significant increase in new myeloid cells in both monocytic and granulocytic branches ( Figure   3L-3N). While there was a significant decrease in the production of new BM T cells (Figure 3J), there was no change in BM B cells ( Figure 3K) nor PB T or B cells (Figure 3O-P). These data suggest that endogenous HSCs undergo rapid emergency myelopoiesis during GAS infection.

HSPC infusion promotes survival in GAS-infected mice.
Given that HSPCs are activated to divide and differentiate into immune effector cells upon inflammatory stimulation and we observed an acute loss of HSPCs in GAS-infected mice, we hypothesized that infusion of naïve HSPCs (Lin-Sca-1+ c-Kit+) into GAS-infected mice could improve pathogen clearance, reduce tissue damage, and prolong survival. To test this hypothesis, we infected mice with 2x10 6 CFU MGAS315 and then infused 10,000 FACS-purified HSPCs 24h later, when endogenous HSPCs are significantly decreased ( Figure 4A). This HSPC dose, equivalent to approximately 1.7x10 7 cells per m 2 body surface area, is significantly lower than the dose used for granulocyte infusions, typically between 10 8 -10 10 cells per m 2 (42). On day three post-infection, we harvested BM, limb tissue, and spleen to characterize BM populations and pathogen load ( Figure 4B).
BM characterization showed that HSPC infusion restored the relative abundance of HSPCs, HSCs, and myeloid-biased progenitors, such as MPP3s and GMPs, in the BM ( Figure   4C-4H. We observed that GAS-infected mice that received HSPCs showed lower morbidity than non-rescued mice, with improved overall body score and activity level. To assess survival, we performed Kaplan-Meier survival studies of GAS-infected mice in the absence or presence of HSPC rescue ( Figure 4K). GAS infected mice infused with HSPCs had significantly higher survival than non-rescued mice ( Figure 4L). However, to our surprise, HSPC infusion did not significantly affect the pathogen burden in the infected muscle ( Figure 4I) or pathogen spread to other tissues ( Figure J). Overall, these findings suggest that HSPC infusion is beneficial during GAS infections and promotes survival by a mechanism other than pathogen clearance.

Superinfection further depletes HSPCs in mice.
To determine the extent of the protective potential of HSPC infusion during infections, we The loss of HSPCs and HSCs was very prominent in superinfected mice, more so than in mice infected with GAS alone. Therefore, we hypothesized that an infusion of 10,000 HSPCs would also benefit mice in this model of superinfection ( Figure 5A). As expected, HSPC infusion significantly increased HSPCs and myeloid-biased progenitors in superinfected mice (Figure 5B-5G). As seen in GAS-infected mice, HSPC infusion did not promote bacterial ( Figure 5H) or viral ( Figure 5I) clearance in superinfected mice. The spread of bacteria to the spleen was also unaffected by HSPC infusion (Figure 5J).
Despite the severity of the infection, superinfected mice that received an HSPC infusion ( Figure 5K) had significantly improved survival compared to non-rescued mice ( Figure 5L). This finding suggests that the protective properties of HSPC infusion are effective even in this very severe model of infection.

HSPC infusion increases immunomodulatory MDSCs and prevents sepsis-induced cytokine exacerbation.
Production of proinflammatory cues including IL1, IL6, IL8, TNF, and MIP1a is a key driver of morbidity during sepsis. Together, these cues contribute to systemic inflammatory response syndrome (SIRS), including fever, tachypnea, vasodilation, and circulatory collapse (5,45). These cytokines are independently associated with poor outcomes and death from sepsis in humans.
Since we observed improved survival in mice receiving HSPC infusion without any changes in pathogen load, we hypothesized that HSPCs could impact immunomodulatory cell composition and the inflammatory response to severe infection. Upon analysis of PB and BM populations after HSPC infusion (Figure 6A), there were no changes in BM or PB T lymphocytes that could indicate a Treg-related activity (Supplemental Figure 3A & 3B). However, we found that HSPC infusion significantly increased PB PMN-MDSCs ( Figure 6B) and restored PB M-MDSCs levels ( Figure   6C). Similarly, HSPC infusion restored BM PMN-MDSCs and M-MDSCs populations in GASinfected mice (Figure 6D & 6E). These cells were functionally validated (Supplemental figure   3C) as immunosuppressive cells by their ability to reduce activated T cell proliferation in culture (Supplemental figure 3D). Strikingly, cytokine profiling 72 hours after GAS infection showed reduced overall levels of proinflammatory cytokines in GAS-infected mice that received HSPC infusion ( Figure 6F). These findings suggest that HSPC infusion supports the production of MDSC populations sufficient to dampen maladaptive proinflammatory cues during sepsis.

Infused HSPCs do not engraft but produce myeloid cells including MDSCs.
In order to determine whether infused HSPCs directly differentiate into MDSCs, we  Figure 7C). Collectively, these data show that infused cells differentiate towards the myeloid lineage with no sign of stem cell engraftment. Whereas MDSCs did arise directly from infused cells, their numbers were not sufficient to account for the large increase in MDSCs observed in the HSPC-rescued mice. These data suggest that HSPC infusion contributes to MDSC expansion via both direct and indirect mechanisms.

Discussion
Here, we show HSPC infusion holds therapeutic potential for bacterial sepsis. Our studies demonstrate that GAS infection induces a robust myeloid response just 24 hours after infection and significantly depletes HSPC populations in the bone marrow. After infection, endogenous HSPCs are driven to differentiate towards the myeloid lineage. However, this response is insufficient to prevent disease progression and pathogen dissemination, resulting in mortality in 5-7 days. While sepsis has been described to cause mobilization of HSPCs (46), we did not find any evidence of HSC or MPP mobilization to the spleen. Strikingly, we found infusing just 10,000 HSPCs improved survival in GAS-infected mice and mice with GAS and influenza superinfection.
This infusion was capable of increasing PB and BM hematopoietic populations of infected mice.

Specifically, HSPC infusion restored BM HSPCs and both PB and BM MDSC populations.
Importantly, HSPC infusion did not reduce pathogen burden, but contributed to survival via the generation of immunoregulatory cells that dampened maladaptive inflammatory signaling in infected mice.
The hematopoietic and immune systems are comprised of immune cells with antimicrobial killing capacity as well as various types of immunomodulatory cells such as MDSCs, regulatory B cells (Bregs), and regulatory T cells (Tregs) (47)(48)(49). In the short time-frame of acute sepsis, myeloid cells such as neutrophils, monocytes, and macrophages are of critical importance in rapidly recognizing and killing invading bacteria. Our data demonstrate that these cells and even the progenitors that produce them in the bone marrow can be rapidly depleted during a severe acute infection. Furthermore, lineage tracing experiments provide the first direct evidence that terminally differentiated myeloid cells are rapidly produced from the level of the HSC during an acute infection. Initially we hypothesized that replacement of HSPCs may improve outcomes from acute bacterial infection by boosting the availability of myeloid cells to kill bacteria. While myeloid cell populations were somewhat restored after HSPC infusion, this was insufficient to reduce pathogen burden.
Dysregulated inflammation is one of the main drivers of morbidity and mortality during infections (5,(50)(51)(52). For example, excessive inflammatory responses are a common result of seasonal influenza (4,50) and SARS-CoV-2 infection (51). Seasonal influenza increases the susceptibility of patients to secondary bacterial infections or superinfection (53). Superinfections exacerbate the proinflammatory environment of common viral infections and are associated with increased morbidity and mortality (53,54). To our surprise, HSPC infusion was protective in a model of influenza and GAS superinfection, suggesting that its protective effects are robust even the setting of severe inflammation. In our mouse models, infection dramatically increased cytokine levels within just three days of infection. Interestingly, HSPC infusion was accompanied by an overall decrease in serum cytokine levels and a specific dampening of cytokines involved in "cytokine storm" (5,51,52). HSPCs have been described to produce cytokines, indicating that they have the capacity to direct immune function (55). However, whether the cytokines produced by HSPCs themselves contribute to the regulation of the immune response has heretofore been unknown. Our data point toward an immunomodulatory role of HSPC infusion that could prevent immune dysregulation during sepsis.
Immunoregulatory cells in the hematopoietic system include Tregs, Bregs, and MDSCs (47,49,56). Perhaps the most recognized immunomodulatory cell known is the Treg. While Tregs have essential roles regulating immune responses to pathogens (47), we did not see differences in any lymphocyte population, including T cells, that would suggest a Treg-mediated antiinflammatory mechanism after HSPC infusion. MDSCs are immature myeloid cells that have strong anti-inflammatory roles by suppressing the responses of T-helper cells that contribute to the development of sepsis (49,57,58). PMN-MDSCs and M-MDSCs have strong antiinflammatory functions that can be beneficial or detrimental depending on the setting. In fact, some studies have shown that MDSCs contribute to clinical worsening during sepsis (49). For almost three decades, increased circulating immature myeloid cells have been a clinical marker of SIRS (59). Interestingly, increased MDSCs during sepsis have also been associated with increased development of nosocomial infections (56). However, in our GAS model of accelerated infection, the increase in MDSC populations after HSPC infusion was accompanied by lower overall cytokine levels and increased survival, suggesting that the immunomodulatory functions of MDSCs are beneficial during the early stages of systemic inflammation and could prevent sepsis-related mortality (60).
An important limitation of our study is that the lineage fate and the tissue or organ destination of the infused HSPCs at the early stages of infusion remain unknown. The small number of cells infused makes it challenging to identify them in the pool of endogenous cells of the recipient mice. While our data suggest that infused HSPCs directly and indirectly boost MDSC production by endogenous cells, further work will be required to determine the mechanisms by which HSPCs contribute to MDSC expansion. In addition, further analysis of HSPC subpopulations will be required to determine if long term HSCs or short-lived multipotent progenitors confer the greatest therapeutic potential. Identifying a short-lived hematopoietic progenitor that can signal endogenous cells to restore MDSC populations could represent a promising alternative therapeutic avenue (12,14,16) to treat sepsis while avoiding concern of possible graft versus host disease complications (61,62).
Currently, G-CSF, GM-CSF, and granulocyte transfusion (13,42,63) are used to prevent or treat sepsis in oncology patients with chemotherapy-induced fever and neutropenia. However, the clinical efficacy of granulocyte transfusion is poor (42,63). Here, we have shown infusing HSPCs is a promising alternative to granulocyte transfusion. Current granulocyte doses in humans are around 1x10 10 cells per m 2 body surface area given daily or every other day (42,64).
Our infusion model only uses a single dose of 1.7x10 7 cells per m 2 body surface area (or 10,000 HSPCs in a mouse). It is important to emphasize that 10,000 HSPCs is a relatively small number of cells to infuse into a mouse as it represents less than 0.01% of the nucleated bone marrow cells in a mouse. Collectively, the single low HSPC dose compared to multiple larger granulocyte transfusions suggests that HSPCs are more effective than granulocytes, cell for cell, in the treatment of sepsis. Our findings could lead to the development of an efficacious new therapeutic approach that could succeed where granulocyte infusions have fallen short.

Mice
We used WT C57Bl/6 (CD45.2) and C57Bl/6.SJL (CD45.1) mice 8-10 weeks of age. Lineage tracing using KRT18-CreERT2 : Rosa26-lox-STOP-lox-TdTomato we made by crossing KRT18-CreERT2 mice obtained from Dr. Daisuke Nakada (Baylor College of Medicine) and Rosa26-lox-STOP-lox-TdTomato mice (stock # 007914) obtained from Jackson Laboratories (Bar Harbor, ME USA, www.jax.org). All mice genotypes were confirmed by polymerase chain reaction (PCR) prior to the start of the experiments. Mice were assigned to each experimental group at random. Both male and female mice were used for all the experiments except for the superinfection survival studies as it has been shown that female mice have long-lasting hyperresponsiveness to respiratory infections (65). Therefore, only male mice were used in the superinfection experiments. Individual mice were assigned to groups randomly and were age and sex matched for each independent experiment.

Pathogen inoculation and quantification
Mice were infected with Streptococcus pyogenes strain MGAS315 by intramuscular injection on the hind limb with 2x10 6 CFU. To determine the bacterial load, limb, spleen, and blood were collected from infected mice. Limb and spleen tissue were homogenized, serially diluted, and plated on blood agar plates (BAP) (BD, Franklyn Lake, New Jersey, www.bd.com). Blood was serially diluted and plated on BAP. Limb and spleen bacterial load was normalized to the grams of tissue that was homogenized.
Influenza A H1N1 PR8 strain infections were done by intranasal inoculation with 150 PFU. Viral load was quantified by collecting viral particles from lung lavage fluid using Amicon Ultra 0.5 mL (Millipore Sigma, Burlington Massachusetts. www.emdmillipore.com), and RNA was purified using the TRIZOL method followed by the quantification of viral particles by Real-Time PCR of virus-specific nucleoprotein (NP) gene. The exact quantity was calculated using a standard curve of purified viral particles with known concentration and normalized by the amount of lung tissue collected.

Flow cytometry and cell sorting
Flow cytometry analyses were done using LSR II and BD Fortessa instruments. Cells were identified by the differential expression of markers listed in Supplemental Table 1. Our cocktail of Lineage (Lin) markers include Gr1, CD11b, B220, CD4, CD8, and Ter119.
Cell sorting of HSPCs and their subpopulations were done using the SONY SH800 sorter and the BD FACS Aria Fusion using the markers listed in Supplemental Table 1. Post-sort purity test showed that sorted cells were 95-98% pure.

Cre induction
Cre activation in KRT18-CreERT2 : Rosa26-lox-STOP-lox-TdTomato was induced with Tamoxifen. Each mouse was administered tamoxifen by intraperitoneal injection at a dose of 100mg/Kg body weight for five consecutive days prior to the start of the lineage tracing experiments.

Cell infusion
All infusions were done intravenously by retroorbital injection. Rescued mice received cells resuspended in saline solution while the control mice were injected with saline solution alone.

Cytokine profiling
Serum was collected using a BD Microtainer blood collection tube (San Jose, CA. www.bdbiosciences.com). Serum levels of cytokines were analyzed through Eve Technologies company (Calgary, AB, Canada. www.evetechnologies.com).

T cell suppression assay
T cells were isolated from the spleen using anti-CD3 magnetic beads from Miltenyi Biotec (Bergisch Gladbach, Germany. www.miltenyibiotec.com) and MDSCs were sorted using the SONY SH800 sorter and the BD FACS Aria Fusion as described above. T cell were activated with anti-CD3 and anti-CD28 coated plates and supplemented with IL-2 to support proliferation and then cultured alone of with M-MDSCs or PMN-MDSCs. T cells were stained with CellTrace Violet and proliferation mas measured by dye dilution using flow cytometry.

Statistical tests
Normality was assessed using the Shapiro-Wilk test and variances were compared using F-tests.
Comparisons between two groups were made done using unpaired t-test for parametric data, Welch's t-test for parametric data without equal variances, and Mann-Whitney test for nonparametric data. Tests involving three or more comparisons we done using One Way ANOVA         Unpaired t-tests. ns=not significant, *p<0.05, **p<0.01, ***p<0.001 *****p<0.0001. Comparison of (L) survival was done using Log-rank (Mantel-Cox) test.