Production and Differentiation of Myeloid Cells Driven by Proinflammatory Cytokines in Response to Acute Pneumovirus Infection in Mice

Respiratory viruses induce a variety of symptoms and pathologies, with important impacts on health. Most research has focused on characterizing the inflammatory response and disease processes at the site of infection in the airways and lung tissue, but emerging evidence suggests that this inflammatory response does not remain compartmentalized to the lung (1–3). Rather, localized viral infection can have systemic effects, including elevated circulating cytokines levels and alterations in bone marrow hematopoiesis (1–3). The systemic response to respiratory viral infections and the impact on disease outcomes remains poorly understood. In our investigation, we gain new insights into the impact of viral lung infection in vivo on systemic immune responses by assessing changes in cytokine levels and alterations in bone marrow hematopoiesis.

Hematopoiesis proceeds through a tightly regulated hierarchy of cell stages, whereby hematopoietic stem cells (HSCs) differentiate through committed multipotent progenitor (MPP) and lineage-specific progenitor stages, before differentiating into mature hematopoietic lineages. During differentiation, hematopoietic stem/progenitor cells (HSPCs) progressively lose multilineage potential as they undergo commitment to specific lineages. The regulation of HSPC populations by inflammatory signals and infection has been extensively reviewed (3–5). Recent findings suggest that, rather than acting as quiescent bystanders, HSPC populations are modulated by inflammatory cytokine stimulation (including IFN-γ [6–12] and TNF-α [13–16], which feature prominently in respiratory virus infection [17–19]). Inflammatory cytokine stimulation and/or direct interaction of HSPCs with pathogens (3–5) may modulate bone marrow homeostasis (20, 21). Thus, HSPCs respond rapidly and appropriately to distinct inflammatory signals. Although a growing body of literature suggests a role for inflammatory cytokines in modulating hematopoiesis, the majority of these studies have been conducted through direct administration of individual cytokines. Relatively few studies have assessed changes during active infection, particularly using assays that quantify HSC and downstream progenitor function. As such, the mechanisms underlying HSPC regulation remain unclear but have important implications for disease management, particularly as new therapies are being developed that target inflammatory mediators in disease settings (22).

In this study, we use pneumonia virus of mice (PVM) in an acute model of respiratory infection (23). PVM (Family Paramyxoviridae, Genus Pneumovirus) is a natural mouse pathogen related to human respiratory syncytial virus (RSV). PVM infection reproduces many of the clinical and pathological features of severe RSV infection seen in human infants, inducing impaired respiratory function and proinflammatory chemokine production (24) and myeloid cell recruitment to the lung (25, 26). In humans, early-life exposure to RSV has long-term effects, being associated with increased asthma susceptibility later in life (27, 28). Similarly, PVM infection in early life can induce an asthmatic phenotype in mice (29) or drive spontaneous asthma-like pathology in the context of TLR7 deficiency (30). The local response to PVM has been extensively studied and specifically highlights a role for CCL3 in myeloid cell recruitment, viral clearance, and clinical outcome (25). In contrast, comparatively few investigations focus on systemic or hematopoietic responses to acute PVM infection.

In this study, we demonstrate that acute pneumovirus infection results in a profound increase in myeloid cells in the lung, circulation, spleen, and bone marrow, defined by cell-surface expression of CD11b, Ly6G, and Gr-1 lineage markers. This increase occurs despite restriction of PVM replication to the lung tissue, occurs before the onset of severe symptoms, and is associated with increases in both local and systemic inflammatory cytokine levels. We also observe a dramatic increase in Sca-I expression across multiple hematopoietic lineages and increased bone marrow HSPC proportions (assessed using flow cytometry). Functional assays demonstrate this occurs through increased committed myeloid progenitor numbers, with no changes in HSC numbers. Furthermore, systemic administration of neutralizing anti–TNF-α or anti–IFN-γ Abs limited myeloid progenitor expansion in the bone marrow and interfered with viral clearance in the lung, demonstrating a central role for viral-induced inflammatory responses in promoting myeloid cell development.

C57BL/6 male mice (8–12 wk of age) were received from the University of Newcastle Animal Services Unit, and experiments were performed in the Hunter Medical Research Institute animal facility, under specific pathogen-free conditions, following review and approval from the local animal care and ethics committee. Mice were infected by intranasal instillation of 100 PFU PVM strain J3666 in DMEM + 10% FCS, as previously described (31). This inoculum induces lethal disease, which would require euthanization by approximately days 8–10. Control animals were sham administered DMEM + 10% FCS. Postinfection, mice were monitored daily for weight loss and clinical symptoms were scored as follows: 1 = no signs of illness; 2 = consistently ruffled fur; 3 = piloerection, deeper breathing, and decreased alertness. Animals were euthanized on days 3, 6, or 8, as indicated in this article.

For competitive bone marrow reconstitution experiments, CD45.1 recipient animals were purchased from the Animal Resources Centre (Perth, WA, Australia) and housed at Australian BioResources (Moss Vale, NSW, Australia). C57BL/6 (CD45.2) and CD45.1 × CD57Bl/6 (CD45.1/2) donor mice were infected by intranasal instillation of 100 PFU PVM strain J3666. Animals were then euthanized on day 8, bone marrow cells collected, and RBC lysis performed using ammonium chloride solution. Cells from PVM-infected donors were mixed 1:1 with sham-infected donors of reciprocal genotypes (i.e., PVM-infected CD45.2 with sham-infected CD45.1/2 and vice versa) and injected into lethally irradiated CD45.1 recipient animals (1100 rad; RS2000 X-ray Irradiator; RadSource, Suwanee, GA). Donor cell mixtures and blood samples then were assessed by flow cytometry at the time points indicated.

To assess the role of TNF-α or IFN-γ, we injected PVM-infected mice i.p. on days 3 and 6 with 200 μg (in 200 μl PBS) of either anti–TNF-α (clone XT3.11; BioXCell, West Lebanon, NH), anti–IFN-γ (clone R4-6A2; BioXCell), or isotype rat IgG1 controls (clone HRPN; BioXCell), respectively, and euthanized them on day 8 postinfection.

Lung tissue, blood, spleen, and bone marrow samples were collected and processed to single-cell suspensions before staining. Lung tissue was digested in HEPES buffer containing collagenase D (Sigma-Aldrich, St. Louis, MO) and DNAse for 1 h, then forced through a 70-μm strainer. Cardiac puncture blood was collected in EDTA-coated microvette collection tubes (Sarstedt, Numbrecht, Germany). Spleen samples were forced through 70-μm strainers. Bone marrow cells were isolated by flushing femurs with PBS/2% FCS.

After isolation, RBC lysis was performed using ammonium chloride solution. All cells were treated with anti-FcγRIII/II (Fc block) in PBS/2% FCS for 20 min before staining with combinations of the following fluorochrome-conjugated Abs as indicated in the text (all Abs from BD Biosciences unless otherwise indicated; San Jose, CA): allophycocyanin-conjugated lineage mixture (containing CD3e [145-2C11]; CD11b [M1/70]; B220 [RA3-6B2]; Ly-76 [TER-119]; Gr-1 [RB6-8C5]), PE/Cy7-conjugated Sca-1 (D7), PerCP-Cy5.5–conjugated c-kit (2B8), PE-conjugated CD150 (TC15-12F12.2; BioLegend, San Diego, CA), FITC-conjugated CD48 (HM48-1; BioLegend), FITC-conjugated Ly6G (1A8), PE-conjugated F4/80 (BM8; BioLegend), PerCP-Cy5.5–conjugated CD11b (M1/70), allophycocyanin-conjugated Gr-1 (Ly6C/G; RB6-8C5), FITC-conjugated CD45 (30-F11), PE-conjugated CD3e (145-2C11), FITC-conjugated B220 (RA3-6B2), PerCP-conjugated CD8a (56-6.7), and allophycocyanin-conjugated CD4 (RM4-5). Competitive reconstitution samples were also stained using FITC-conjugated CD45.1 (A20) and PE-conjugated CD45.2 (104). For flow cytometry assessment, samples were fixed overnight in PBS/2% FCS + 0.1% PFA and collected on a BD FACSCanto II flow cytometer, then analyzed using FACSDiva software (BD Biosciences).

For cell enrichment experiments, lung tissues were digested, treated with anti-FcγRIII/II (Fc block), and stained with FITC-conjugated Ly6G (1A8) and allophycocyanin-conjugated Gr-1 (Ly6C/G; RB6-8C5) Abs, as described. Cells were then sequentially magnetically enriched using first anti-FITC (Ly6G⁺) microbeads, followed by anti-allophycocyanin microbeads (Gr-1⁺Ly6G⁻), according to manufacturer’s specifications (Miltenyi). Enrichment was confirmed by flow cytometry on a BD FACSCanto II flow cytometer. Cells were prepared by cytospin and stained by modified Wright-Giemsa stain, visualized on an Olympus BX51 microscope, and imaged on a DP73 digital camera (Olympus) or resuspended in TRIzol Reagent (Invitrogen) for RNA isolations.

Lung tissue was disrupted in radioimmunoprecipitation assay buffer (Sigma-Aldrich) with protease/phosphatase inhibitor mixture (Cell Signaling Technology, Danvers, MA), on a Tissuelyser LT tissue disruptor (Qiagen, Valencia, CA) at 50 Hz for 5 min and stored at −80°C. Serum was collected by cardiac puncture and centrifugation, after allowing blood to fully clot on ice. Cytokine levels for IL-6, IL-10, CCL2 (MCP-1), IFN-γ, TNF-α, and IL-12p70 were assessed using the mouse inflammation cytometric bead array kit (BD Biosciences), according to manufacturer’s specifications on a BD FACS Canto II flow cytometer and analyzed using FCAP Array software (BD Biosciences).

Lung tissue was placed in RNALater (Invitrogen, Carlsbad, CA) and stored at −80°C. Unflushed femurs were disrupted in TRIzol Reagent (Invitrogen), on a Tissuelyser LT tissue disruptor (Qiagen) at 50 Hz for 5 min and stored at −80°C. Total RNA was isolated by phenol-chloroform separation and isopropanol precipitation, and quantified on a Nanodrop 1000 spectrophotometer (Nanodrop, Wilmington, DE). cDNA was prepared by RT-PCR using random hexamer primers (Invitrogen) and MMLV reverse transcriptase (Invitrogen), on a T100 thermal cycler (Bio-Rad, Hercules, CA). Relative quantitative RT-PCR quantification was performed on a ViiA7 real-time PCR machine (Life Technologies, Carlsbad, CA), using SYBR reagents. Measured cDNA levels were normalized to the housekeeping gene HPRT. Primer sets (Table I) were designed across exon boundaries to specifically amplify mRNA products.

Bone marrows were initially processed to single-cell suspensions as described earlier. Cells were then plated in duplicate at 2 × 10⁴ cells/plate, in Methocult GF3534 (Stem Cell Technologies, Vancouver, BC), containing FBS, BSA, rh-Insulin, human transferrin, 2-ME, rm-SCF, and rm-IL-6. Plates were cultured for 7 d, and myeloid colonies (CFU-granulocyte [CFU-G]/CFU-monocyte [CFU-M]/CFU-granulocyte/macrophage [CFU-GM]) were counted by light microscopy, according to manufacturer’s instructions.

The p values were calculated using unpaired two-way Student t test.

Inoculation of C57BL/6 mice with PVM (100 PFU) resulted in rapid and significant weight loss (detected at day 6 postinoculation) and the onset of clinical symptoms immediately before sacrifice on day 8 (Fig. 1A, 1B). Virus was detected in lung tissue as early as day 3 postinoculation, with a peak viral load at day 6 coinciding with the onset of weight loss at this inoculum (Fig. 1C). Importantly, PVM virus was not detected in spleen or bone marrow at any time point by quantitative PCR (qPCR; data not shown). Leukocyte populations in lung, blood, spleen, and bone marrow were evaluated at days 6 and 8 postinoculation and compared with sham-inoculated controls.

Increased myeloid cell percentages were detected in lung tissue by flow cytometry using the markers Gr-1 and CD11b, compared with sham-infected controls (Fig. 1D, 1H). Specifically, the total percentage of Gr-1⁺ cells was increased in PVM-infected mice at day 6, with a further increase by day 8 (Fig. 1D, 1H). All Gr-1⁺ cells coexpressed CD11b in both sham and PVM-infected mice. Further subdividing the Gr-1⁺ population, two subpopulations (Gr⁻1⁺Ly6G⁺ [SSC^hi] and Gr-1⁺Ly6G⁻ [SSC^lo]) were evident, both of which were increased after PVM infection. A slight increase was also observed in the percentage of Gr-1⁻CD11b⁺ cells within the lung, after PVM infection. These staining profiles are characteristic of neutrophils (Gr-1⁺Ly6G⁺), infiltrating monocytes/macrophages (Gr-1⁺Ly6G⁻), and interstitial macrophages (Gr-1⁻CD11b⁺), respectively, based on previous studies (32). This increased cell proportion (displayed as % of total CD45⁺ cells) coincided with a consistent increase in total cell numbers isolated from PVM-infected lungs after processing, compared with sham controls, indicating a marked increase in neutrophil and monocyte infiltration into the lung.

Magnetic enrichment of the Gr-1⁺Ly6G⁺and Gr-1⁺Ly6G⁻ populations from lung tissue for histological assessment at day 8 postinoculation confirmed granulocytic and monocytic morphology, respectively (Fig. 2A). These surface marker profiles and morphology overlap with recently described myeloid-derived suppressor cell (MDSC) populations. Characterization of mRNA expression in total lung samples and isolated cell populations revealed increased TNF-α expression post PVM infection, as well as increased levels of Arginase 1, inducible NO synthase (iNOS), and IL-10 expression (Fig. 2B–D), which are expressed by MDSCs (33–35).

To assess systemic effects on myeloid cell levels, we evaluated peripheral blood, spleen, and bone marrow. In blood and spleen, Gr-1⁺Ly6G⁺ myeloid cells (primarily neutrophils) (36) were dramatically increased at the day 8 end point over sham-inoculated controls (Fig. 1I, 1J). In contrast, both Gr1⁺Ly6G⁻ (monocyte/macrophage) and lymphoid cell percentages were unaffected, with only a slight increase in CD3⁺CD8⁺ lymphocytes after PVM infection (Fig. 1J and data not shown). Similar changes were also seen in the blood at day 6 postinoculation, although again to a lesser extent (Fig. 1E), whereas myeloid cells percentages in the spleen were decreased at day 6 postinoculation (Fig. 1F). In bone marrow, Gr-1⁺Ly6G⁺ (neutrophil) percentages were unaffected by infection, whereas Gr1⁺Ly6G⁻ (monocyte/macrophage) percentages were increased (Fig. 1G, 1K). Total bone marrow cell counts were unaffected by PVM infection.

To identify mechanisms driving increased myeloid cell percentages during acute infection, we assessed both local and systemic inflammatory mediator levels. In the lung, PVM infection induced expression of cytokines TNF-α, IL-6, IFN-γ, and CCL2 (Fig. 3A). IL-6, IFN-γ, and CCL2 were also detected in serum postinfection (Fig. 3B), whereas serum TNF-α levels were variable, around the limit of detection in our assay (data not shown). qPCR of lung tissue confirmed increased levels of these cytokines, and likewise documented expression of CCL3, G-CSF, and osteopontin, with kinetics corresponding to the timing of peak viral load (Fig. 3A and data not shown). In total bone samples, TNF-α transcript levels progressively increased 8 d postinoculation (Fig. 3C). In separate experiments, the increase in TNF-α expression was observed in isolated bone marrow samples, as well as samples isolated from flushed bone (data not shown). Assessment of GM-CSF and IL-3 transcript levels by qPCR in lung, spleen, and bone failed to identify any changes in expression postinfection (data not shown).

We observed an increase in the percentages of HSPCs, based on Lineage⁻Sca-I⁺c-kit⁺ (LSK) staining of cells from bone marrow at day 8 of infection (Fig. 4A, 4C); increased LSK percentages were not present at day 3 postinfection, and minor increases were observed at day 6 (Fig. 4B). The LSK cell gate is enriched for HSCs and is commonly used to quantify stem cell numbers, but also includes more committed MPPs (37). Further assessment using the signaling lymphocyte activation molecule markers (CD48 and CD150) demonstrate that the observed increase in the LSK population results from a minor increase in CD48⁻CD150⁺ cells (a population further enriched for HSCs) and a major increase in the percentages of CD48⁺CD150⁻ cells, which are enriched for MPP populations (Fig. 4B, 4C) (38). Further, in the spleen, the percentages of LSK cells, specifically within the CD48⁺CD150⁻ MPP gate, were also increased (Fig. 4D). Interestingly, proportions of LSK cells (specifically the MPP subset) were also increased in a low-dose infection model (12 PFU) at day 8 (Supplemental Fig. 1). These findings indicate that MPP proportions are increased post viral infection, but do not require severe disease.

To characterize this increase in HSPCs, we assessed the number of committed myeloid progenitors using a CFU assay. Post PVM infection, the total number of bone marrow myeloid CFU was increased slightly (but significantly) at day 6 and increased ∼2-fold at day 8, compared with sham-infected controls (Fig. 5A). This increase occurred for bipotent CFU-GM progenitors, as well as committed CFU-M progenitor subsets at day 6, and also for the CFU-G progenitors at day 8 (Fig. 5A). Interestingly, although increased HSPCs were also observed in the spleen by flow cytometry, no difference was observed in myeloid CFU numbers in either spleen or lung tissue (data not shown), suggesting that acute PVM infection fails to induce significant extramedullary hematopoiesis.

To assess changes in HSC number post PVM infection, we performed competitive reconstitution experiments using the CD45.1/2 system. Bone marrow cells from sham- and PVM-infected donors were mixed 1:1 and injected into lethally irradiated recipient animals. In donor cell mixtures, ratios of total cell numbers from each donor were confirmed at ∼50%, whereas consistent with our previous observations, ∼75–80% of the LSK population (in the 1:1 mixture) was from PVM-infected donors (Fig. 5B). Donor contributions to hematopoiesis were monitored in peripheral blood for 6 wk postreconstitution, and total donor reconstitution levels of >95% were achieved. Assessment of relative donor contributions revealed no difference in the relative contributions of sham- versus PVM-infected donors to reconstitution (Fig. 5C, right panel). Further, at the 7-wk end point, bone marrow exhibited a ∼50% contribution from PVM-infected and sham-infected donor cells to both total bone marrow and LSK subsets (Fig. 5B). These findings indicate that the increase in phenotypic LSK cells post acute PVM infection results from an increase in the number of committed myeloid progenitors, whereas functional HSC numbers remain unaltered.

Of interest, we also observed increased Sca-I expression in the bone marrow, lung, and spleen by flow cytometry and qPCR assessment post PVM infection. In total bone and lung samples, Sca-I mRNA levels increased progressively throughout the time course of PVM infection (Fig. 6A, 6B). Further, Sca-I expression was increased across all hematopoietic (CD45⁺) cell subsets assessed by Sca-I staining mean fluorescence intensity (MFI) in bone marrow (Fig. 6A). In the lung, Sca-I surface staining occurred on CD3⁺ T cell populations, as previously described (39), as well as mature myeloid populations, presented as the percentage of Sca-I⁺ (/CD45⁺) cells and Sca-I MFI (Fig. 6B). In the spleen, increased Sca-I mRNA levels were also seen at day 8 postinfection, as well as increased Sca-I expression assessed by staining MFI (Fig. 6C). These findings demonstrate that Sca-I expression is increased across a range of hematopoietic lineages.

To determine a mechanism underlying the viral-induced increases in myeloid cell production, we systemically (i.p.) administered a neutralizing Ab against TNF-α (or a rat IgG1 isotype control) on days 3 and 6 postinoculation into PVM-infected mice. The intervention time points were chosen based on our kinetic analysis, as TNF-α expression increased in lung tissue and bone between days 3 and 6 postinfection (Fig. 3C). Ab administration had no effect on weight loss or clinical symptoms, compared with isotype control (data not shown).

At the day 8 end point, increased virus recovery occurred from the lungs of mice receiving anti–TNF-α (Fig. 7A), although no differences were observed in myeloid cell infiltration into the lung (data not shown). Assessment of lung and serum cytokine levels demonstrate that anti–TNF-α administration partially reduced levels of TNF-α in lung homogenates but had no effect on the induction of other inflammatory mediators within the lung or serum (Fig. 7B).

Despite increased virus recovery, mice receiving anti–TNF-α Ab responded with decreased induction of LSK cells in the bone marrow, particularly within the CD48⁺CD150⁻ MPP cell subset (Fig. 7C). Further, anti–TNF-α administration resulted in striking decreases in the numbers of myeloid CFU in the bone marrow, with numbers at the baseline levels observed in sham-infected animals (Fig. 7D). Taken together, these findings demonstrate a role for TNF-α in virus-induced bone marrow HSPC expansion, myeloid cell production, and virus clearance in the lung.

In a separate set of experiments, we also assessed the role of IFN-γ in the observed changes post acute PVM infection. IFN-γ–neutralizing Ab (or a rat IgG1 isotype control) was also administered systemically on days 3 and 6 postinfection, based on kinetics of IFN-γ production postinfection (Fig. 3C).

Similar to our findings with anti–TNF-α administration, anti–IFN-γ resulted in increased PVM viral recovery from the lungs at day 8 (Fig. 8A) but had no effect on myeloid cell infiltration into the lung (data not shown). Assessment of cytokine levels in the lung and serum revealed a striking decrease in IFN-γ protein levels after Ab treatment (Fig. 8B, 8C). In the bone marrow, anti–IFN-γ resulted in dramatic decreases in LSK cell induction and MPP cell numbers (among the LSK population; Fig. 8D). Functional assessment of CFUs in the bone marrow revealed an ablation of myeloid progenitor induction after anti–IFN-γ administration (Fig. 8E), similar to our findings with anti–TNF-α. These findings suggest that IFN-γ plays a role analogous to that played by TNF-α post lung PVM infection, driving increases in myeloid progenitors, as well as virus clearance in the lung.

Previous research has primarily focused on localized myeloid cell infiltration, inflammatory cytokine release, and tissue damage post PVM infection (24–26, 29, 39). However, numerous clinical studies have documented increased systemic inflammatory cytokine levels in virally infected patients, including increased serum IL-2, IL-4, and IFN-γ post RSV infection (1) and increased serum CCL2 levels in RSV-infected asthmatics (40) (although serum TNF-α levels remain low) (41). Further, in severe influenza infection, IL-6, IFN-γ, and IL-10 levels are increased (2, 42). Similar to these clinical findings, we observed increased serum levels of IFN-γ, CCL2, and IL-6, as well as increased TNF-α transcript in the bone, post PVM infection. Of these factors, we demonstrate an important role for TNF-α and IFN-γ in modulating hematopoiesis in response to infection.

In parallel with elevated systemic cytokines, we observed increased systemic myeloid cell percentages post PVM infection. Specifically, neutrophil proportions (Gr1⁺Ly6G⁺) were dramatically increased in the blood and spleen, whereas inflammatory monocytes/macrophages (Gr1⁺Ly6G⁻) were increased in the bone marrow. We speculate that the mild changes in the bone marrow may reflect the rapid egress of newly generated cells from the marrow niche into the circulation. Interestingly, increased systemic myeloid cell numbers (reported as CD11b⁺Gr1⁺) in the lung, blood, and bone marrow was also recently reported in severe H5N1/H1N1 influenza infections in mice (43).

Although the staining profiles of these cell populations are consistent with traditional neutrophil and monocyte/macrophage populations, they also overlap with recently described MDSC populations. MDSCs suppress T cell responses and have been best characterized in cancer models (reviewed in Ref. 44). They have also been described in infectious disease models, including lung fungal infection (45) and viral infections (46, 47). Importantly, MDSCs are nearly indistinguishable from proinflammatory myeloid cell populations by histology and surface marker staining, and controversy remains over whether they exist as distinct cells lineages or are alternative activation states of these cell populations (48). We demonstrate increased TNF-α expression post PVM infection, consistent with the presence of proinflammatory myeloid cells (Fig. 2C, 2D). Interestingly, we also detected increased levels of Arginase 1, iNOS, and IL-10 expression, which are expressed by MDSC populations (33–35). However, IL-10 protein levels were not detected in lung homogenate samples assessed by cytometric bead array. These findings demonstrate that this induced myeloid population may include a mixed population including MDSCs and inflammatory myeloid cells.

We also observed increased HSPCs in the bone marrow, post PVM infection, with increased committed progenitors (MPPs) assessed by flow cytometry (LSK CD48⁺CD150⁻) and colony-forming assays. Importantly, these findings occur in the absence of detectable systemic PVM virus in the spleen and bone (by qPCR), suggesting changes are not driven by direct interactions between pathogen and HSPCs (e.g., TLR stimulation of HSPCs), as has been proposed in other models (3, 4). Rather, we suggest an indirect feedback mechanism, whereby antiviral responses initiated in the lung signal via proinflammatory cytokines to the bone marrow, promoting myeloid cell production. This also fits with data from a low-dose PVM recovery model (Supplemental Fig. 1), where increases in MPP levels coincided with the kinetics of increased lung inflammatory cytokine expression.

Importantly, our initial observations of increased LSK cell percentages (as well as a minor increase in LSK CD48⁻CD150⁺) suggested the possibility of a PVM-induced increase in bone marrow HSC numbers. However, previous studies have demonstrated that inflammatory cytokines (including both TNF-α and IFN-γ) increase Sca-I expression on multiple hematopoietic cells (49), making conclusions based on phenotypic analysis alone problematic. Furthermore, recent evidence suggests that Sca-I induction on myeloid precursors plays an important functional role in driving the granulopoietic response to bacterial infection (50, 51). Increased Sca-I expression results in an apparent increase in phenotypic HSCs (if HSCs are defined based on LSK surface staining alone), emphasizing the importance of performing functional assays to quantify HSCs. Indeed, we observed striking increases in Sca-I expression across multiple hematopoietic lineages in the lung, spleen, and bone marrow. Whereas we observed increased numbers of committed myeloid progenitors (CFU-GM, CFU-G, CFU-M), no differences in bone marrow reconstitution capacity were detected. These results indicate that HSC numbers are not affected by PVM infection. Rather, hematopoiesis is altered through an increase in committed MPPs and myeloid progenitors.

Because altered cytokine levels are frequently observed in viral-infected patients (as mentioned earlier) and inflammatory cytokines have been proposed as a mechanism regulating HSPC proliferation and differentiation, we assessed levels in our model. Previous research on PVM pathogenesis identified a role for CCL3 (MIP-1α) in myeloid cell recruitment to the lungs (25). Interestingly, CCL3-induced recruitment of neutrophils was coordinated by IFN-γ (25). The macrophage chemokine CCL2 (MCP-1) is also induced by PVM infection (24, 26). Although these chemokines drive local recruitment of mature myeloid cells, it remains unclear what drives increased systemic myeloid cell numbers. GM-CSF, G-CSF, and IL-3 have well-established roles regulating myeloid progenitor cell differentiation (52–54), whereas osteopontin can induce myeloid cell migration (55, 56). IL-6 acts as a potent proinflammatory cytokine, and recent evidence suggests its production in the bone marrow environment may induce emergency myelopoiesis (57). Based on these previous reports, we assessed the levels of these cytokines in our infection model. We observed increased TNF-α, IFN-γ, CCL2, CCL3, IL-6, G-CSF, and osteopontin levels in lung tissue, as well as increased circulating levels of IFN-γ, CCL2, and IL-6, and increased TNF-α mRNA in bone. Based on the proposed roles for TNF-α and IFN-γ in regulating and inducing HSPC differentiation, we targeted these molecules using blocking Abs.

TNF-α induces HSC proliferation, although limiting long-term engraftment potential (13, 14), and plays a role in the maintenance of baseline homeostasis in the absence of infection (15, 16). In Ab-blocking experiments, anti–TNF-α Ab administration decreased both MPP induction (LSK CD48⁺ CD150⁻) and myeloid CFU induction. This decrease in myeloid HSPC induction occurred despite increased PVM viral load within the lung, highlighting a role for TNF-α in infection-driven myeloid cell production and a disconnect between viral levels and HSPC induction. Our results differ slightly from other observations on the effects of anti–TNF-α interventions in mouse models of RSV infection. In one study, mice were primed with recombinant vaccinia virus and subsequently infected with RSV (or influenza), causing rapid weight loss over the first 4 d postinfection (58). Systemic blockade of TNF-α resulted in decreased weight loss, reduced lung infiltration and pathology, and decreased IFN-γ production from CD4⁺ T cells, but had no effect on viral load (58), which is quite different from our observations. Another study assessed RSV infection in unprimed mice, which demonstrated minimal weight loss and pathology (59). In this model, TNF-α depletion resulted in very minor decreases in peak weight loss and slightly reduced cell recruitment to the lung, while also limiting the number of RSV-specific T cells, resulting in increased viral load (59). Further, previous work demonstrated that anti–TNF-α treatment in RSV-infected mice resulted in increased weight loss and delayed recovery (17). Importantly, none of these studies assessed the impact of anti–TNF-α on bone marrow hematopoiesis or myeloid progenitor numbers. Taken together, these findings suggest that the effect of anti–TNF-α may differ depending on the timing of treatment or infection and the severity of disease.

IFN-γ also has an impact on HSC proliferation and survival, with varying effects reported, including apoptosis induction (10, 11), impaired proliferation (9), and promotion of proliferation and differentiation (6–8, 12). As such, it remains unclear what effect IFN-γ has on HSPC function and whether effects vary based on disease-specific conditions. Further, IFN-γ exposure induces Sca-I expression on isolated Lineage⁻c-kit⁺ cells (60). Ab-mediated blocking of IFN-γ decreased LSK cell induction (which may, in part, be explained by decreased IFN-γ–mediated Sca-I induction) and blocked induction of myeloid CFUs, demonstrating a role for IFN-γ in driving increased myeloid progenitor numbers. As with our observations using anti–TNF-α, anti–IFN-γ also resulted in increased PVM viral load in the lung, demonstrating a requirement for IFN-γ in the appropriate clearance of lung viral infection. Taken together, these experiments suggest that TNF-α and IFN-γ may function through a similar pathway to induce increased myeloid cell progenitors in the bone marrow post viral infection.

Although both anti–TNF-α and anti–IFN-γ treatment resulted in reduced MPP proportions and increased viral recovery, our study cannot make conclusions over whether a direct link exists between these observations. TNF-α and IFN-γ are each important mediators of antiviral defense, and the observed increase in viral levels may reflect localized effects within the lung. Further, although the interventions impacted on myeloid cell production in the marrow, no change was observed in myeloid cell recruitment to the lung. This recruitment is likely driven by chemokine production in the lung, which was unaltered by either anti–TNF-α or anti–IFN-γ treatment. Thus, although our observations clearly demonstrate a role for both TNF-α and IFN-γ in the upregulation of myeloid cell production post lung infection, the direct impact of this increase on antiviral immunity remains unclear.

Intriguingly, recent evidence in long-term autoimmune colitis models (52) revealed very similar alterations to those observed in our model. Induction of colitis resulted in increased HSC populations in the spleen and bone marrow (assessed by flow cytometry), through an IFN-γ–dependent mechanism (52). Of note, increased HSC numbers was not confirmed functionally. Induced HSC populations were skewed toward the myeloid lineage, and GM-CSF induced extramedullary hematopoiesis within the spleen and colon (52) (which was not observed in PVM infection). The similarity between this report and our current findings highlight the key role for inflammatory cytokines in regulating hematopoiesis and myeloid cell production, and demonstrate a conserved mechanism underlying myeloid cell production across a range of inflammatory conditions. Further, differences between disease models may highlight tailored responses, which are modulated by the duration, cause, and dynamics of disease.

In summary, our study provides evidence that localized pneumovirus infection drives systemic inflammatory responses, promoting increased myeloid cell production and differentiation in the bone marrow, and increased systemic myeloid cell percentages. Increased myeloid cell production occurs in the absence of viremia and is associated with responses to the cytokines TNF-α and IFN-γ. This paradigm may provide a more general mechanism whereby the innate immune system recognizes localized infection and upregulates myeloid cell differentiation, to respond appropriately and augment pathogen clearance.

We thank the Analytical Biomolecular Research Facility at the University of Newcastle for flow cytometry support.

The authors have no financial conflicts of interest.