Type I IFNs drive hematopoietic stem and progenitor cell collapse via impaired proliferation and increased RIPK1-dependent cell death during shock-like ehrlichial infection

Type I interferons (IFNα/β) regulate diverse aspects of host defense, but their impact on hematopoietic stem and progenitor cells (HSC/HSPCs) during infection remains unclear. Hematologic impairment can occur in severe infections, thus we sought to investigate the impact of type I IFNs on hematopoiesis in a tick-borne infection with a virulent ehrlichial pathogen that causes shock-like disease. During infection, IFNα/β induced severe bone marrow (BM) loss, blunted infection-induced emergency myelopoiesis, and reduced phenotypic HSPCs and HSCs. In the absence of type I IFN signaling, BM and splenic hematopoiesis were increased, and HSCs derived from Ifnar1-deficient mice were functionally superior in competitive BM transplants. Type I IFNs impaired hematopoiesis during infection by both limiting HSC/HSPC proliferation and increasing HSPC death. Using mixed BM chimeras we determined that type I IFNs restricted proliferation indirectly, whereas HSPC death occurred via direct IFNαR -mediated signaling. IFNαR-dependent signals resulted in reduced caspase 8 expression and activity, and reduced cleavage of RIPK1 and RIPK3, relative to Ifnar1-deficient mice. RIPK1 antagonism with Necrostatin-1s rescued HSPC and HSC numbers during infection. Early antibiotic treatment is required for mouse survival, however antibiotic-treated survivors had severely reduced HSPCs and HSCs. Combination therapy with antibiotics and Necrostatin-1s improved HSPC and HSC numbers in surviving mice, compared to antibiotic treatment alone. We reveal two mechanisms whereby type I IFNs drive hematopoietic collapse during severe infection: direct sensitization of HSPCs to undergo cell death and enhanced HSC quiescence. Our studies reveal a strategy to ameliorate the type I IFN-dependent loss of HSCs and HSPCs during infection, which may be relevant to other infections wherein type I IFNs cause hematopoietic dysfunction.

Introduction Acute infection induces demand-adapted hematopoiesis, characterized by increased hematopoietic stem cell and progenitor cell (HSC and HSPC) proliferation, to support production and mobilization of immune cells or platelets [1][2][3][4][5]. Infection induced 'emergency myelopoieisis' results in increased production of effector myeloid cells that promote bacterial clearance [3,6]. However, excessive proliferation of HSCs and HSPCs can lead to functional impairment and induce hematopoietic suppression [7][8][9][10], [11], though the precise mechanisms driving HSC/HSPC impairment have only recently been investigated [3,[12][13][14][15]. The Ehrlichiae are emerging tick-borne pathogens that cause an acute, febrile disease called human monocytic ehrlichiosis (HME) [16]. Ehrlichia are obligate, intracellular alpha-proteobacteria of the Anaplasmataceae family, and contain gram-negative cell wall structures but lack the genes that encode lipopolysaccharide and peptidoglycan [17,18]. HME disease severity can vary greatly, and in some cases life-threatening complications include multi-organ failure similar to septic shock syndrome [19]. Ixodes ovatus ehrlichia (IOE) is a highly virulent strain that causes shock-like illness in mice [20,21], and is therefore an ideal model to study fatal HME [22]. Vector borne diseases are increasing, and current vaccines are lacking [23], therefore, acute and chronic sequelae induced by tick-borne infections are clinically significant and represent a growing health care concern. HSCs are essential for lifelong hematopoiesis and supply all cells necessary for hemostasis, immunity, and oxygenation, thus delineating the mechanisms that impact HSC function during acute infection is important for our full understanding of infection-induced pathology.
Type I interferons (IFNα/β) are induced in response to nearly all infections. IFNα receptor (IFNαR) signaling stimulates diverse immune cell effector functions, and IFNα/β regulate HSCs directly and through the bone marrow (BM) microenvironment [24,25]. However, it is currently unclear how type I IFNs regulate HSC function during infection. Sterile IFNα/β stimulation induces HSC proliferation, caspase activation, and apoptosis [8]. HSPCs from IFNα-treated patients do not exhibit apoptotic priming [26], however, and IFNα/β promote hematopoietic precursor survival in a murine model of opportunistic lung infection [27]. Therefore, type I IFNs can have diverse impacts on HSC function. HSCs sustain immune function over the lifetime of an organism, and must be maintained through multiple rounds of inflammatory stress, thus clarifying the impact of IFNα/β on hematopoiesis during infection and disease has important implications.
Type I IFNs mediate host defense against viral infection, but are also involved in immunopathology, including BM hypocellularity and impaired immunity in a number of bacterial infections [11]. Type I IFNs drive receptor interacting serine/threonine protein kinase 3 (RIPK3)-dependent macrophage necroptosis in Salmonella typhimurium infection [35], and are required for TNFα-induced systemic inflammatory response syndrome [36], wherein RIPK3 drives pathology [37]. In a shock-like infection induced by a virulent ehrlichial pathogen, IFNα/β-induced BM failure and hematopoietic suppression via increased RIPK1 activation and diminished caspase 8 expression. HSPC depletion and cell death depended upon cellintrinsic IFNαR expression, whereas enforced HSC quiescence also depended on IFNαR, but was not due to direct signaling in HSCs. Together, impaired proliferation and increased cell death contributed to hematopoietic collapse during severe infection and reduced HSC and HSPC recovery following infection.

Type I IFNs impair hematopoiesis during acute ehrlichial infection
IOE infection elicits robust production of IFNα/β, which is detected in the serum, spleen, and liver of wild type (WT) mice [28,29], as well as in the BM (Fig 1A). IOE infection induces BM hypocelluarity with a profound decrease in cellularity on day 8 post-infection, which is mitigated by Ifnar1 deletion (Fig 1B). IFNαR-dependent signals also impaired myeloid progenitor cell activity in the BM during infection (S1A Fig), and significantly enhanced monocyte differentiation was observed in cultured Lincells from Ifnar1 -/mice, relative to wild type mice (S1B Fig). HSPCs can mobilize from the BM and engraft the spleen during systemic inflammation [38], though neither WT nor Ifnar1 -/mice showed evidence of HSPC mobilization in the peripheral blood (S1C Fig), and WT mice showed little to no induction of splenic extramedullary hematopoiesis (EMH). In contrast, robust EMH was observed in the spleens of Ifnar1 -/mice (S1D-S1G Fig), corresponding to increased mature myeloid cells (S1H-S1K Fig), demonstrating IFNα/β suppress hematopoiesis systemically during IOE infection.
HSPCs are heterogeneous, proliferative and consist of lineage-biased progenitors important for daily blood production. The HSPC pool requires replenishment by HSCs, which typically reside in a dormant state in the BM. We examined HSPCs and HSCs on day 7 post-infection, when we detect type I IFNs in the BM and the day prior to the striking decrease in cellularity. At this time, both HSPCs (Lineage -, c-Kit + ) and phenotypic long-term HSCs (Lineagec-Kit + CD135 -CD150 + CD48) were depleted from the BM of WT mice during infection, but in Ifnar1 -/mice HSPCs were maintained and HSCs underwent a significant expansion as determined by both frequency and number (Fig 1C and 1D). Neutralization of IFNα/β during infection of WT mice resulted in improved BM cellularity and significant protection against infection-induced loss of HSPCs and HSCs relative to isotype control treated mice (Fig 1E), similar to observations in Ifnar1 -/mice.
Infection of Ifnar1 -/mice resulted in a striking increase in phenotypic HSCs (Fig 2A). To address whether their phenotype corresponded to function we performed transplantation assays. In the absence of infection, IFNα/β signaling did not impact HSPC repopulating activity in transplantation (S2A Fig). However, sort-purified HSCs derived from IOE-infected Ifnar1 -/mice were functionally superior to those derived from WT mice as evidenced by increased contribution to blood production and reconstitution of both lymphoid and myeloid compartments (Fig 2B and 2C and S2B Fig). HSCs derived from IOE-infected WT mice exhibited a loss of lymphoid potential, demonstrating reduced stem cell function. Thus, the decline in HSC number and function during shock-like infection was dependent on IFNα/β, and may, in part, be due to the intense demand for and loss of HSPCs.

IFNα/β impair HSPC proliferation and maintain HSC quiescence during IOE infection
IFNα/β induce HSC proliferation during sterile inflammation [8,39,40]. We reasoned that loss of HSCs and HSPCs during IOE infection may be driven by enhanced and unsustainable IFNαR-dependent proliferation. Our data did not support this, however, as HSPCs and HSCs were more proliferative in Ifnar1 -/mice, exhibiting enhanced incorporation of BrdU during IOE infection, relative to WT mice ( Fig 3A). HSCs are typically in G 0 , though HSC quiescence was reduced in Ifnar1 -/mice at both steady state and upon infection (Fig 3B and 3C) indicating that HSCs proliferated more in Ifnar1 -/mice whereas WT HSCs remained quiescent. Infection caused a reduction in the proportion of G 0 HSPCs in both WT and Ifnar1 -/mice, though a more marked reduction was observed in the absence of type I IFNs. Because the inflammatory milieu varies between WT and Ifnar1 -/mice (S3 Fig), we evaluated cell cycle regulation within the same microenvironment using WT:Ifnar1 -/mixed BM chimeric mice. When in the same microenvironment, WT and Ifnar1 -/-HSPCs and HSCs showed similar cycling at both steady state and during infection ( Fig 3D). Thus, the infection-induced difference in proliferation of HSCs and HSPCs did not depend on direct IFNαR-signaling, and was due to an extrinsic or soluble factor. Our data suggest that increased HSCs and HSPCs in Ifnar1 -/mice may result from their enhanced proliferation and reduced quiescence, relative to WT mice.

IFNα/β mediated HSPC cell death
HSPCs undergo type I IFN-dependent p53-mediated apoptosis in models of sterile, polyI:C (pI:C)-driven inflammation [8,40], suggesting type I IFNs may drive infection-induced hematopoietic failure via increased cell death of HSPCs. To model IOE infection-induced IFNα/β exposure, we repeatedly administered endotoxin-free pI:C to WT and Ifnar1 -/mice, which caused transient BM hypocellularity, and an IFNαR-dependent reduction in HSCs and HSPCs (S4A- S4D Fig). Type I IFNs were also required for the pI:C-induced increase in dead and dying Lin -CD48cells (enriched for long-and short-term HSCs; S4E Fig; [41]. Consistent with prior reports that pI:C induces apoptosis, we observed increased numbers of cells that bound fluorescent probes for active caspases (caspase 3/7 and 8; S4F Fig). Thus, during sterile inflammation IFNα/β-dependent depletion of HSPCs correlates with induction of caspase activity and sensitization to apoptosis.
Similar to sterile inflammation achieved by pI:C injection, IOE infection induced an IFNαR-dependent loss in HSCs and HSPCs. To assess whether IOE infection induced apoptosis we evaluated Annexin V staining, and found increased Annexin V+ HSPCs in response to IOE infection in WT, but not Ifnar1 -/mice (S5A and S5B Fig). More striking was the nearly twenty-fold increase in the proportion of dead and dying cells (7-AAD+) among the HSPC pool in WT mice, whereas Ifnar1 -/mice exhibited no such increase at day 7 post-infection ( Fig 4A). In contrast to pI:C exposure, enhanced caspase 3/7/8-positive cells were not observed in IOE-infected WT mice ( Fig 4B). Rather, we noted a significant increase in caspase 3/7/8-positive cells in IOE infected Ifnar1 -/mice. To determine if direct IFNα/β signaling sensitized HSPCs to die, we evaluated viability and caspase activity in WT:Ifnar1 -/mixed BM chimeric mice. A striking and direct IFNα/β-dependent increase in dead (7AAD+) and apoptotic (Annexin V+) progenitors was observed 7 d.p.i. (Fig 4C). WT cells were not caspase 3/7/8-positive, whereas a significant increase in caspase 3/7/8-positive HSPCs was observed among Ifnar1 -/cells in the same mice ( Fig 4D). These data establish that IFNα/β deplete HSPCs directly during infectious inflammation, and suggest IFNα/β may sensitize progenitors to undergo caspase-independent necroptotic cell death during IOE infection.
Caspase 8 and FADD are effectors of apoptosis that can also block necroptosis by cleaving RIP1 and RIP3 [42], suggesting improved HSC/HSPCs in Ifnar1 -/mice may be due to distinct modes of cell death. In support of this, we observed enhanced RIPK1 cleavage in BM of IOEinfected Ifnar1 -/mice, relative to infected WT mice (Fig 4E and 4F). At the same time, we observe a clear RIPK3 cleavage product in the BM of Ifnar1 -/mice, and significantly reduced cleavage of RIPK3 in WT mice. Thus, in Ifnar1 -/mice increased caspase-3/7/8-positive cells correlated with enhanced cleavage of key effectors of necroptotic and apoptotic signaling.
TNFα induces pathology in IOE infection [21], and TNFα is a driver of necroptotic cell death, a form of programmed necrosis involving RIPK1, RIPK3, and the effector protein MLKL [43]. While IFNα/β can stimulate TNFR2 expression in some cellular contexts [44], we observed similar BM TNFα levels and equivalent TNFR2 expression by progenitor cells in IOE-infected WT and Ifnar1 -/mice (S6A and S6B Fig). These results suggest differences in TNF signaling do not account for the differences observed in WT and Ifnar1 -/mice. To investigate a role for type I IFNs in necroptosis we next examined RIPK1 and RIPK3 in sort-purified HSPCs derived from IOE-infected mice. In whole BM we found similar expression of RIPK3 and MLKL in WT and Ifnar1 -/mice during IOE infection (S6C and S6D Fig). Immunostaining and confocal imaging of RIPK1 and RIPK3 in sort-purified HSPCs revealed distinct IFNαR-dependent changes during IOE infection. RIPK1 intensity was greater in WT mice, relative to Ifnar1 -/mice (S7A Fig), and we noted increased colocalized RIPK3 and RIPK1 signals in the cytoplasm of HSPCs derived from infected WT mice, relative to Ifnar1 -/mice (S7B Fig), suggestive of RIPK1 and RIPK3 interaction.

IFNα/β limit caspase 8-induced survival of HSPCs in IOE-infected mice
The increase in caspase-active progenitors in Ifnar1 -/mice detected by flow cytometry coincided with enhanced caspase 8 expression in the BM, including full length and the p41/43 and p18 cleavage products, all of which were less abundant in WT IOE-infected mice (Fig 5A and  5B). The striking increase in caspase 8 in Ifnar1 -/mice as well as increased FADD expression (S6E Fig), suggest protection of HSPCs in Ifnar1-/-mice may rely on caspase activation. To evaluate this we treated infected WT and Ifnar1 -/mice with the pan-caspase inhibitor zVAD-FMK (zVAD). In WT mice, zVAD did not alter HSPC or HSC depletion, consistent with our observation that WT HSPCs do not exhibit caspase activity, and confirm that HSPC depletion in WT mice is independent of caspase activity. However, zVAD significantly reduced HSPCs and HSCs in IOE-infected Ifnar1 -/mice ( Fig 5C). To test whether zVAD licensed necroptotic cell death of HSPCs, we administered a specific inhibitor of RIPK1 kinase activity, Necrostatin-1s (Nec-1s), [45]. In contrast to our prediction that caspase inhibition induced RIPK1-dependent cell death, Nec-1s did not rescue the z-VAD-induced loss of HSPCs in Ifnar1-/-mice (Fig 5D). These findings suggest a complex role for caspase activity in protection of HSPCs and HSCs, and provide evidence that in the context of global Ifnar1-deficiency caspase activity is protective to HSC/HSPCs independently of RIPK1-dependent cell death. In addition to regulating cell death, caspase 8 can induce survival and proliferation signals in lymphocytes and primitive hematopoietic progenitors [46][47][48], suggesting the profoundly increased caspase 8 expression observed in Ifnar1-/-mice may induce pro-survival and/or proliferative signals that protect and increase the pool of HSCs and HSPCs during IOE infection.

An intrinsic role for RIPK3 activity in infection-induced HSPC depletion
In the absence of caspase 8 activity, RIPK1-RIPK3 interactions promote RIPK3-dependent phosphorylation of the terminal necroptosis effector, MLKL [49], and can also drive cell death via NLRP3 inflammasome activation [50]. Unlike Ifnar1 -/mice, Caspase1 -/and Nlrp3 -/mice were not protected from IOE infection [28], thus we sought to further address a potential role for necroptotic cell death in driving HSPC loss during IOE infection.
Mice with a mutation in the kinase domain of RIPK3 (Ripk3 Δintron2 ), which is necessary for phosphorylation of MLKL, exhibited similar cellularity and HSPC and HSC numbers as IOEinfected WT mice (Fig 6A-6C). RIPK3 regulates type I IFN production during influenza infection ( [51], however, we observed normal production of type I IFNs in response to IOE infection in Ripk3 Δintron2 mice (S8A Fig). RIPK3 has two functional domains: an N-terminal serinethreonine kinase, and a C-terminal RIP homotypic interaction motif (RHIM). While kinase activity is required for necroptosis, the RHIM domain has additional functions, including promoting NF-kB signaling and cytokine production [52]. We confirmed that the RHIM domain was intact in Ripk3 Δintron2 mice using an antibody against the C-terminus of RIPK3, though RIPK3 expression is reduced in Ripk3 Δintron2 mice (S8B Fig). Because RIPK3 can drive apoptosis under some conditions [53,54] we examined cell death in infected Ripk3 Δintron2 mice. Compared to WT mice, Ripk3 Δintron2 mice exhibited increased apoptosis (Annexin V+ cells; S8C  Fig), suggesting HSPC loss in the absence of RIPK3 favors apoptosis. Nec-1s treatment did not improve HSPC or HSC frequencies in Ripk3 Δintron2 mice (S8D Fig).
As type I IFNs reduced HSCs via an extrinsic factor, we reasoned that RHIM-dependent functions of RIPK3 potentially obscured a protective in Ripk3 Δintron2 mice. To address this possibility, we generated mixed BM chimeric mice with Ripk3 Δintron2 BM (WT: Ripk3 Δintron2 ; Fig 6D). Within the HSPC and HSC populations, we observed a greater proportion of Ripk3 Δintron2 cells relative to WT cells, even in the absence of infection, suggesting RIPK3 impairs HSC and HSPCs in the context of transplantation (Fig 6E). In response to infection, the proportion of Ripk3 Δintron2 HSPCs was increased relative to WT cells, indicating a cellintrinsic role for RIPK3 in the infection-induced loss of HSPCs. However, equal proportions of WT and RIPK3 HSCs were observed after infection, demonstrating intrinsic RIPK3 kinase activity does not mediate HSC loss. These data are consistent with our observations in WT: Ifnar1 -/chimeras and suggest RIPK3 activation is downstream of type I IFN signals, as observed in macrophages [55]. Our data suggest that type I IFNs induce intrinsic, RIPK3mediated death of HSPCs.

RIPK1 activity promotes HSPC death and drives HSC loss during infection
To test the impact of RIPK1 signaling on IFNα/β-dependent HSPC depletion we next treated IOE-infected WT mice Nec-1s (Fig 7A). RIPK1 antagonism rescued HSPC numbers, reduced dead/dying HSPCs, and expanded the HSC pool above that of mock-infected controls ( Fig  7B-7D). Blocking RIPK1 signaling using Nec-1s during acute IOE infection rescued HSPC and HSC depletion in IOE infection, though it did not improve survival (Fig 7E). The failure of Nec-1s to protect mice, despite the remarkable rescue of the hematopoietic compartment may be due to high bacterial burden in Nec-1s treated mice, which was similar to vehicletreated mice (Fig 7F).

Necrostatin-1s treatment improves hematopoietic recovery from shock-like infection
Early antibiotic treatment with doxycycline (doxy) is the current standard of care for patients with human monocytic ehrlichiosis, and related rickettsial infections, as delayed diagnosis is common and associated with poor outcomes [31]. Severe immune-mediated pathology can occur independently of bacterial burden [56], however, and survivors of sepsis-induced shock suffer long-term immunosuppression and other sequelae [57]. To control bacterial growth and evaluate the impact of RIPK1 on hematologic recovery from shock-like infection, we administered antibiotics (doxycycline; doxy) in conjunction with Nec-1s to WT IOE-infected mice. To mimic delayed diagnosis, common in rickettsial infections [31], we began treatment on day 6 post-infection (Fig 8A). Without antibiotics, or with Nec1s alone, IOE-infected mice succumbed to infection by day 10, whereas mice administered doxy survived (Fig 8B), and could be subsequently analyzed. IOE-infected mice administered doxy alone exhibited significantly reduced frequencies and numbers of HSPCs and HSCs at 15 d.p.i., relative to mock-infected control mice (Fig 8C and 8D). However, co-administration of Nec-1s with doxy resulted in increased HSPC and HSC frequencies 15 d.p.i., resembling those seen in uninfected mice. Furthermore, treatment with both doxy + Nec-1s resulted in HSPC numbers similar to those in mock-infected mice and a significant increase in HSCs relative to mice treated with doxy alone. Bacterial burden was similar during acute infection (7 d.p.i.) between treatment groups, however, during the recovery phase (15 d.p.i.), bacterial burden was lower in mice that received Nec-1s in addition to doxy (Fig 7E). These data demonstrate that RIPK1 activity impairs hematopoiesis during acute infection, and has lasting effects on HSC and HSPC numbers. Our findings suggest that RIPK1 activity can protract disease and RIPK1 antagonism may be therapeutically relevant for improving hematopoiesis and long-term outcomes in patients surviving acute, severe infections where type I IFNs are elevated.

Discussion
Demand-adapted hematopoiesis is a dynamic process sculpted by cytokines, and can vary depending upon the duration and magnitude of stimulation and the presence of secondary stimuli [1,4,58]. IOE infection causes lethal, shock-like disease accompanied by cytopenias and immune dysfunction, similar to what is observed in patients with Rocky Mountain spotted fever and severe hemorrhagic viral infections. IOE is unusual for the ehrlichia because it infects endothelial cells [59], and it elicits shock-like infection driven by TNFα and type I IFNs [20,28,29]. We previously found that blocking type I IFN signaling in both the hematopoietic and non-hematopoietic system is necessary for protection against lethal infection [29], and herein we demonstrate a profound impact of type I IFNs on hematopoiesis and HSC and HSPC function. Hematopoiesis produces mature immune cells that can impact immunity and host IFNα/β-dependent HSPC depletion in severe infection defense [3,60], thus the impact of infection and type I IFNs on HSPCs is directly relevant to recovery and survival in patients.
In sterile inflammation, type I IFNs drive proliferation and apoptosis of HSPCs and HSCs [8,39,40]. Based on our previous observation that IFNα/β promotes lethal disease in shocklike infection [29], we predicted the impact of IFNα/β on HSCs may differ between sterile and infection-dependent inflammation. A key finding from our studies is that in contrast to sterile inflammation, type I IFNs drive hematopoietic dysfunction via impaired proliferation and increased RIPK1-dependent cell death that may include both RIPK1-dependent apoptosis and necroptosis (Fig 9). Whereas IFNα/β directly reduced HSPCs, stem cell numbers were not directly responsive to IFNα/β. Rather, type I IFNs were indirectly associated with HSC loss and increased quiescence during infection. While HSC dormancy may be critical for preserving the stem cell population under inflammatory or infectious conditions, failure to proliferate in response to infectious demand was associated with impaired myelopoiesis and reduced host innate immunity. Additional studies are warranted to determine the IFNα/β dependent factors associated with impaired proliferation.
Similar to sterile inflammation, infection increased progenitor cell death. However, during infection type I IFNs mediated HSPC death by reducing caspase 8 expression and activity, which correlated with reduced RIPK1 and RIPK3 cleavage in the BM. Our conclusion that IFNα/β-dependent HSPC depletion in severe infection IFNα/β-dependent HSPC depletion in severe infection type I IFNs induce RIPK1-dependent apoptosis and/or necroptosis is based on the specific protection of HSPCs conferred by RIPK1 antagonism, the intrinsic role for RIPK3 kinase activity in HSPC non-apoptotic cell death, and the observation of limited caspase 8 expression and activity in WT mice. Our findings do not exclude a role for RIPK1-dependent apoptosis, which is also kinase-dependent and therefore sensitive Nec-1s [61]. WT mice exhibit profound rescue of HSCs and HSPCs upon RIPK1 antagonism with Nec-1s, similar to what is observed in Ifnar1 -/--deficient mice, consistent with a central role for RIPK1 kinase activity in the demise of the hematopoietic compartment in IOE infection.
Caspase 8 performs numerous and complex roles in regulating cell death and survival pathways. Caspase 8 is a critical initiator of the extrinsic pathway of apoptosis, as it cleaves effector caspases, including caspase 3, but can limit necroptosis by impairing RIPK1 and RIPK3 interactions. Caspase 8 can also form a complex with RIPK3 to activate inflammasome-independent IL-1β production [62], which may underlie the increase in IL-1β observed in the absence of IFNαR. The IFNα/β-dependent reduction in caspase activity during IOE infection is distinct from the impact of sterile inflammation on HSPCs, where increased cleaved caspase 3 was observed [8]. Despite increased caspase 3/7 activity in infected Ifnar1 -/mice we failed to detect increased apoptosis in IOE-infected Ifnar1 -/mice. The striking increase in full length pro-caspase 8, as well as active caspase 8, in Ifnar1 -/mice was apparent prior to infection, suggesting Ifnar1 -/mice may be poised to resist RIPK1-dependent cell death. Caspase 8 activity is also associated with proliferation in the hematopoietic compartment [48,63], and in lymphocytes wherein caspase-8-dependent cleavage induces more efficient recruitment of RIPK1, necessary to activate NF-kB [64]. An intriguing possibility is that increased survival and proliferation of HSCs and HSPCs in Ifnar1 -/mice is dependent on elevated caspase 8 activity consistent with our observation that the pool of HSCs expands during IOE-infection in Ifnar1 -/mice, and the extreme sensitivity of HSCs and HSPCs to zVAD.
How type I IFNs regulate caspase 8 during infection is an important and open question. Caspase 8 can be degraded by autophagy [65], a pro-survival strategy to counter cellular stress and infection, and, of note, type I IFNs can initiate autophagy [66]. Interestingly, the Ehrlichia can also induce autophagy, and subsequently seize nutrients for their growth, and inhibition of autophagy controls ehrlichial growth [67]. As reduced bacterial burden is observed in Ifnar1 -/mice, our data are consistent with a potential role for IFNαR-dependent signaling in promoting autophagy which would support bacterial growth and, at the same time, reduce caspase 8, licensing RIPK1-dependent cell death. However, caspase 8 can also blunt autophagy by degrading components of the autophagic macherinery [68], and therefore elevated caspase 8 in Ifnar1 -/mice may prevent autophagy, thus limiting bacterial growth. Additional studies to interrogate the precise mechanisms of caspase 8 regulation may reveal important regulatory nodes critical for the control of bacterial growth and cell survival.
RIPK1 and RIPK3 can promote inflammasome activity, however, caspase-1 and Nlrp3-deficient mice were not protected from IOE infection [28]. IFNαR-and TLR9-mediated caspase-11 activation has recently been observed during IOE infection [34], and caspase-11 activity can promote pyroptotic cell death by cleaving gasdermin D [69]. Due to the robust improvement of HSPC and HSC numbers upon RIPK1 antagonism, we reason that caspase 11-mediated pyroptosis is unlikely to cause loss of HSCs and HSPCs, though pyroptotic cell death may be important for pathology in other tissues during disease, such as the liver and lung, two organs that are heavily infected [70]. In line with this possibility, Nec-1s-treated mice died with similar kinetics as vehicle treated, thus, caspase 11 activity in non-hematopoeitic cells may contribute to pathology and death during IOE infection.
Determining the in vivo relevance of various cell death pathways is challenging for a number of reasons. Depleting or blocking signaling components can render cells more susceptible to other death pathways [71], which complicates interpretations of in vivo observations. RIPK3 is a key effector in necroptotic cell death, but can also drive apoptosis or pyroptosis and regulate cytokine production. In addition, the distinct cellular context, with respect to signaling components and negative regulators of necroptosis, may regulate cell type-specific differences in cell death during infection. It is also important to consider that certain cell death pathways may be critical for control of infection [72]. Indeed, RIPK1-dependent apoptotic cell death is necessary for control of Yersinia pestis [73]. Particularly challenging is addressing cell death processes in vivo, as distinct cell types respond uniquely to stimuli, and may influence other cell types. As HSCs generate all mature cells, it is difficult to parse out the intrinsic pathways that govern cell death and survival. Using mixed BM chimeric mice we were able to define an intrinsic role for type IFNs and RIPK3 in the loss of HSPCs. Despite these complexities, delineating the functional consequences of inflammatory cell death pathways will improve the development of therapeutics to treat infection-induced hematopoietic dysfunction and suppression.
Our work highlights the potential utility of targeting RIPK1 to mitigate BM failure in patients exposed to high levels of IFNα/β. Nec-1s was not able to protect mice against death, likely due to the failure of Nec1s alone to control bacterial growth and additional type I IFNdependent effects on non-hematopoietic tissues. However, addition of Nec-1s to antibiotic treatment improved HSC and HSPC numbers, which may prove clinically relevant. Necroptosis-promoting cytokines, including IFNα/β and TNFα, are induced by numerous disease states, including infection, as well as clinical protocols, such as radiation and BM transplantation [74,75], and are a potential prognostic biomarker in patients with sepsis [76]. Therefore, therapeutic targeting of type I IFNs and/or RIPK1 and RIPK3 may prevent hematologic dysfunction in such scenarios.

Isolation of hematopoietic stem and progenitor cells
HSC or HSPC populations were sort-purified using a FACSARIA II (BD Biosciences) into media or directly onto poly-L-lysine-coated 12mm coverslips.

Analysis of cell death
Cellular viability was determined using 7-aminoactinomycin D or Zombie fixable viability dye (BioLegend). Caspase activity was assayed by incubating freshly isolated bone marrow cells with FAM-LETD-FMK caspase 8 and FAM-DEVD-FMK caspase 3/7 fluorescent inhibitor probes (ImmunoChemistry Technologies) according to manufacturer's instructions. Caspase activity was analyzed within viable cell populations.

ELISA
BM cell lysates for protein analysis were homogenized manually with a pestle in a buffer containing IGEPAL CA-630 and proteinase inhibitors. Total protein concentration was measured using a BCA kit (Pierce) and a 23-Plex Inflammatory Cytokine Kit (Bio-Rad). ELISA kits for IFNα and IFNβ protein were purchased from eBioscience and PBL Interferon, respectively. Protein levels were normalized to total BM protein (mg).

Immunofluorescence staining
Hematopoietic stem and progenitor cells were fixed for 10 minutes at room temperature in a 2% formaldehyde solution containing 9.8 mM NaIO 4 , and 62 mM lysine after 5 minutes of adherence to coverslips. Cells were permeabilized for 15 minutes in 0.02% saponin at room temperature and then incubated in mouse anti-RIP1 [38/RIP] from BD Biosciences and rabbit anti-RIP3 from ProSci, Inc. (clones that were previously published [55]) for 1 hour. Speciesspecific, fluorescently-conjugated IgG (H+L) F(ab') 2 fragments from Life Technologies or Cell Signaling Technology were used for detection. Coverslips were mounted with Electron Microscopy Sciences Fluorogel II containing DAPI.

Confocal microscopy and image analysis
Images were acquired on a Zeiss LSM510 META-NLO laser-scanning microscope system equipped with a Plan-Apochromat x63/1.4 numerical aperture oil objective lens, an argon laser (at 488 nm), a green HeNe diode laser (at 543 nm), and standard emission filters for Alexa488 and 555. Analysis of RIPK1 and RIPK3 signal association was performed using the colocalization module of Imaris (BitPlane, Oxford Instruments). Briefly, a colocalization channel was generated and used to calculate the Pearson's coefficient of the colocalized volume.

Quantification and statistical analysis
Flow cytometry. Flow cytometry data were analyzed using FlowJo software (TreeStar) using appropriate fluorescence minus one controls.
Densitometry. Band intensity was measured from original digital immunoblot images using Image Lab software (Bio-Rad). Briefly, band intensity of protein of interest was divided by actin band intensity for each sample and normalized values were graphed. For full length and cleaved caspase 8 analysis, the normalized cleaved caspase 8 (p18) band intensity was divided by the normalized full length caspase 8 band intensity.
Quantification of bacterial burden. 50 ng DNA was extracted from splenocyte lysates using the DNeasy Blood and Tissue kit (Qiagen), and used to quantify disulfide bond formation protein (Dsb) gene by real-time quantitative PCR, as previously described [29]. Mice with fewer than 10 3 Dsb copies/50ng DNA were excluded, as this is the limit of detection.
Statistical analysis. Data was analyzed using GraphPad Prism, version 5. Statistical analyses of data sets including multiple time points or >2 experimental groups were performed using two-or one-way ANOVA with Tukey multiple comparison post-tests. Statistical analyses of two experimental groups at a single time point were performed using two-tailed Student's t-test. All data are represented as mean ± standard error of the mean, and are representative of at least two independent experiments, unless otherwise noted. i. Briefly, 500 Lin-BM cells or 1000 Lin-spleen cells were seeded into each well of a 24-well plate containing irradiation (3000 rads) treated OP-9 cells in the presence of IL-3, IL-7, GM-CSF, SCF and Flt3L, and cultured for 10 days. Cells were analyzed to identify monocytes (CD11b+ Ly6C hi Ly6G -) and neutrophils (CD11b+ Ly6C -Ly6G + ). n = 5-7 mice/group. Ã P < 0.05, ÃÃ P < 0.001, ÃÃÃ P < 0.0001. (C) Flow cytometry plots of WT and Ifnar1 -/singlet peripheral blood cells to evaluate circulating Linc-Kit + -defined HSPCs in mock and 6, 7, and 9 d.