Trauma Induces Emergency Hematopoiesis through IL-1/MyD88–Dependent Production of G-CSF

All leukocytes are generated through the process of hematopoiesis, a hierarchical progression of differentiation whereby pluripotent hematopoietic stem cells (HSC) progress through a series of more specialized progeny, ultimately culminating in the generation of terminally differentiated effector cells (1). Specialization is associated with a decreased capacity for self-renewal and the acquisition of effector cell functions (2, 3). At rest, there is a basal rate of leukocyte turnover, which is regulated by the clearance of senescent cells and the replacement of cells consumed in response to infection. At baseline, leukocytes are replaced through the proliferation of committed hematopoietic progenitors such as the granulocyte-monocyte progenitor (GrMP) cells proliferating to produce additional neutrophils or monocytes (2). In response to local infections, the proliferative activity of committed progenitors is accelerated to provide the effector leukocytes that are required to replace those consumed by senescence or in response to low-level infectious challenges (4, 5).

In contrast, the immune response to severe infections dramatically increases the demand for leukocytes, rapidly outpacing the proliferative capacity of the committed progenitor pool (4). This drives recruitment of pluripotent populations such as multipotent progenitors (MPP) and HSC into active hematopoiesis. This process is termed emergency hematopoiesis (EH) and is characterized by a broad-based activation and expansion of hematopoietic stem and progenitor cell (HSPC) populations to generate the downstream leukocyte progeny needed for an effective immune response (4, 5). Prior studies have demonstrated that severe infections such as sepsis induce EH (6, 7) and that this phenotype can be recapitulated by exogenous administration of pathogen-associated molecular patterns, such as LPS (8) or the TLR2 agonist PAM₃CSK₄ (9).

Traumatic injury alone (in the absence of infection) also creates a hematopoietic demand because of the consumption of leukocytes in the local and systemic inflammatory response to tissue injury (10–12). Sterile injury has been shown to activate committed progenitors to increase granulopoiesis and monocytopoiesis (13, 14), although plasma isolated after injury has been shown to suppress ex vivo bone marrow proliferation (15, 16). Hemorrhagic shock has been shown to increase the frequency of immature progenitors (17), but other models of sterile injury found no effect of injury on pluripotent short-term HSC (ST-HSC) in young animals (18). Taken together, these data leave unresolved the effect of sterile traumatic injury on hematopoiesis. To establish the effect of injury on hematopoiesis, we measured HSPC populations in a clinically relevant model of polytrauma. We find that trauma alone induces EH characterized by the expansion of immature hematopoietic progenitors through IL-1/MyD88–dependent production of G-CSF, resulting in a progenitor population that is skewed toward the myeloid cell production.

C57BL/6J and MyD88^−/− mice on the C57BL/6 background were obtained from The Jackson Laboratory. All studies were conducted in accordance with the institutional guidelines for the humane treatment of animals and were approved by the Washington University Animal Studies Committee.

Male C57BL/6 wild-type (WT) and MyD88^−/− mice at 10–12 wk of age were subjected to a multisystem injury consisting of bilateral lower extremity pseudofracture, limited hemorrhagic shock, and partial liver crush injury, as detailed below. Mice were maintained under general anesthesia (2% isoflurane) during the entire procedure. Pseudofracture consisted of lower extremity soft tissue crush injury, induced with a hemostat clamp, followed by the injection of a morselized bone suspension from the femurs and tibiae of a donor mouse. Limited hemorrhagic shock was induced by withdrawing 15% of the calculated total blood volume via cardiac puncture. For the liver crush injury, a hemostat clamp was used to apply six consecutive contusions over the entire area of the left liver lobe. All animals received buprenorphine (0.1 mg/kg) and fluids (1 ml saline) s.c. immediately after the procedure.

For G-CSF blockade experiments, mice were injected i.p. with 25 μg anti-mouse G-CSF Ab (MAB414; R&D Systems) or rat IgG1 isotype control Ab (MAB005; R&D Systems) at 15 h before the polytrauma procedure. A second dose of 25 μg Ab was given s.c. immediately before polytrauma. Bone marrow progenitor populations were analyzed at 24 h after polytrauma. IL-1α and IL-1β blockade experiments were performed by injection (at 15 h prior to polytrauma and at time of injury) of 150 μg anti–IL-1α Ab, clone ALF-161, and/or 150 μg anti–IL-1β Ab clone B122 (both from Bio X Cell). Control mice received polyclonal hamster IgG (Bio X Cell). For IL-1R blockade experiments, mice were treated with 3 mg rIL-1RA (anakinra, Sobi) injected s.c. 1 h before polytrauma; control animals received an equivalent volume of normal saline. In some experiments, three doses of IL-1RA were administered, at times −16, 0, and 8 h after the polytrauma procedure.

For plasma isolation, blood was taken via cardiac puncture into heparinized syringes and centrifuged at 10,000 rpm (9400 × g) for 10 min at 4°C. The plasma was then stored at −80°C until analysis. Bone marrow was isolated from the femurs of mice by centrifugation of the bones at 6000 rpm (3380 × g) for 3 min, followed by RBC lysis (Hybri-Max; Sigma-Aldrich). Splenocyte suspensions were prepared by dissociating spleens over 70-μm meshes in HBSS buffer containing 10% bovine calf serum, followed by centrifugation (300 × g for 10 min) and RBC lysis. Peripheral blood leukocytes were prepared from 500 μl whole blood (taken by cardiac puncture) by adding 5 ml RBC lysis buffer and incubating for 7 min at room temperature, followed by two washes with PBS. Peritoneal lavage was performed in euthanized mice by injecting 1 ml complete media (RPMI 1640 with 10% FCS) into the peritoneal cavity. The abdomen was briefly massaged before the collection of the resulting cell suspension within the peritoneal cavity. Nucleated cell counts and cell viability were determined by diluting cell suspensions with a mixture of acridine orange and propidium iodide and analyzing on a K2 Cellometer (Nexcelom Bioscience). For measuring IL-1 levels in liver homogenates, sections of injured and uninjured liver lobes were homogenized in PBS containing a protease inhibitor (Pierce Biotechnology/Thermo Fisher Scientific) with a TissueRuptor homogenizer (QIAGEN). Cell lysates for IL-1 measurement were prepared by resuspending cell pellets in PBS containing a protease inhibitor and 0.5% Triton X-100. Homogenates and cell lysates were spun at 10,000 rpm × g for 10 min, and supernatants were then frozen at −80°C.

Plasma G-CSF and tissue IL-1α and IL-1β levels were measured with FlexSet cytometric bead array (CBA) kits (BD Biosciences). Lower detection limit for G-CSF in plasma samples was 25 and 10 pg/ml for IL-1α and IL-1β, respectively, in cell lysates and tissue homogenates. Samples were acquired on a FACSCanto II flow cytometer and analyzed using the FCAP software (all from BD Biosciences). IL-1 tissue levels were normalized to total protein concentration by Bradford assay (IBI Scientific) according to the manufacturer’s recommendations. For some experiments, cytokines in plasma at 3, 6, 24, and 48 h after polytrauma were measured with the MILLIPLEX Cytokine/Chemokine kit (MCYTMAG-70K-PX32; MilliporeSigma) using a Bio-Plex 200 analyzer.

For progenitor cell analysis, cells isolated from bone marrow, spleen, and blood were stained with Abs diluted in PBS with 2% bovine calf serum and 0.05% sodium azide (FACS buffer). Cells were stained on ice for 20 min, followed by incubation with a dead cell dye (Zombie Aqua or Zombie NIR, both from BioLegend). Abs used for progenitor analyses were as follows: lineage mixture consisting of PerCP-Cy5.5–labeled Abs to CD3, CD8, CD19, CD11b, Gr-1 (all from BioLegend), CD4, and Ter-119 (BD Biosciences); and Abs to progenitor markers Sca-1–FITC, CD117-BV421, CD135-PE, CD16/32-biotin, CD127-biotin, CD150-PE-Dazzle 594 (BioLegend), CD34-A647, and CD48-BV510 (BD Biosciences). Biotin Abs were detected with BV605-conjugated streptavidin (BD Biosciences). Early progenitor cells were gated as live, lineage-negative CD117⁺Sca-1⁺ (KSL) cells with the following phenotypes: CD150⁺CD48^– long-term HSCs (LT-HSC), CD34⁺CD135^– (ST-HSC), and CD34⁺CD135⁺ (MPP). MPP subsets were gated as KSL cells with the following surface phenotypes: MPP2: CD135^–CD150⁺CD48⁺; MPP3: CD135^–CD150^–CD48⁺; MPP4: CD135⁺CD150^–CD48⁺. Common lymphoid progenitors (CLP) were gated as live, lineage-negative CD117^lowSca-1^lowCD127⁺CD135⁺ cells. Myeloid progenitors were gated as live, lineage-negative CD117⁺Sca-1^– cells with the following phenotypes: CD34⁺CD16/32^lo, common myeloid progenitors, CD34⁺CD16/32^hi (GrMP); CD34^–CD16/32^lo megakaryocyte–erythroid progenitor (MEP). Representative FACS plots and gating strategy shown in Supplemental Fig. 2.

For leukocyte analysis, cells were stained with fluorochrome-conjugated Abs to CD45, CD3, CD4, CD8, CD11b, CD11c, CD19, CD115, CCR2, NK1.1, Ly6C, Ly6G, TCRγδ (all from BioLegend), Alexa Fluor 700–conjugated CCR2 (R&D Systems), and PE-CF594–conjugated Siglec F (BD Biosciences), followed by Zombie Aqua dye staining. Gating on individual immune cell subsets is shown in Supplemental Fig. 3.

Following staining, cells were fixed with Cytofix buffer (BD Biosciences). Samples were acquired on an LSRFortessa equipped with four lasers (488, 405, 640, and 552 nm) using the DIVA software (BD Biosciences), and data were analyzed with the FlowJo software (TreeStar).

To evaluate the hematopoietic potential of HSC from naive and injured animals, 10,000 bone marrow cells were mixed with methylcellulose medium containing growth factors (MethoCult GF M3434 medium; STEMCELL Technologies) and plated in 35-mm tissue culture plates. The number of colonies per plate were counted on day 7 of culture.

To define the role of sterile injury on the hematopoietic system, we first established a clinically relevant model of sterile injury by modifying our published model of polytrauma (19). Young mice were subjected to a bilateral lower extremity pseudofracture injury and a laparotomy followed by 15% blood volume hemorrhage and a blunt liver contusion. This injury resulted in <5% mortality but did cause an ∼15% loss of body weight by postinjury day 2, slowly returning to baseline weight over the following 7 d (Supplemental Fig. 1A). To characterize the magnitude of systemic inflammation induced by this injury, we measured plasma cytokines at 24 h after injury as compared with sham-manipulated animals. This polytrauma injury induced a low-magnitude systemic inflammatory response that was almost completely resolved by 24 h (Supplemental Fig. 1B). This model recapitulates the clinical situation of a complex nonlethal multisystem trauma associated with a transient acute inflammatory phase (20).

To establish the effect of injury on hematopoiesis, we measured HSPC populations in the bone marrow after polytrauma. Mice were subjected to trauma as described above, and 24 h after injury, bone marrow was isolated and assayed by flow cytometry. HSPC populations were analyzed using previously described markers (see 2Materials and Methods and Supplemental Fig. 2) (2, 3, 21). At 24 h after injury, injured mice had significantly higher numbers of immature progenitors in the bone marrow (Fig. 1A). This was manifest in a significant increase in the heterogenous c-kit⁺/Sca-1⁺/lineage-negative (KSL) progenitor population and also in the more rigorously defined populations, including LT-HSC, short-term HSCs (ST-HSC), and MPP. We observed a significant increase in both the absolute number of cells (Fig. 1B) and also the frequency of progenitors among bone marrow cells (Fig. 1C). We also measured the number and frequency of committed progenitors, including common myeloid progenitors, CLP, GrMP, and MEP. Consistent with past studies (13), we found an increase in the frequency of GrMP and MEP, although there was no difference in the absolute numbers of committed progenitor measured.

To determine if these injury-induced changes in HSPC populations were durable over time, we repeated these studies at 10 and 21 d after trauma. At 10 d after trauma, we detected persistent increases in KSL, LT-HSC, ST-HSC, MPP, and CLP populations. By 21 d, all populations had returned to baseline, except a small but statistically significant elevation in ST-HSC frequency (Fig. 2).

The LT-HSC pool is a heterogenous population comprised of both myeloid-predominant and lymphoid-predominant HSC clones (22, 23). Myeloid-predominant HSC can be identified by the expression of the cell surface molecule CD41 (24). Twenty-four hours after injury, we measured the frequency of LT-HSC expressing CD41 by flow cytometry. We found that injury resulted in a shift toward myeloid-predominant HSC manifest as a >3-fold increase in the frequency of CD41⁺ LT-HSC (Fig. 3A). These data demonstrate that injury is acutely altering the LT-HSC cohort available for hematopoiesis. It is unclear if the increased CD41 expression reflects the expansion of the existing baseline CD41⁺ LT-HSC or induction of CD41 within the CD41⁻ LT-HSC cohort. However, the increased population of CD41⁺ LT-HSC suggests a skewing of hematopoiesis toward the production of myeloid cells.

The MPP population also contains a heterogenous population of cells distributed across a continuum of myeloid versus lymphoid potential. Several subsets of MPP can be identified on the expression of CD150, CD48, and Flk2 (CD135) (21). The MPP2 and MPP3 population predominately contribute to myelopoiesis; MPP4 are primarily lymphopoietic progenitors (21). We assayed MPP populations in the bone marrow at 1, 10, and 21 d after injury (Fig. 3B, 3C). We found increases in the frequency of all three MPP populations at 1 d after trauma (Fig. 3C). At 10 d after injury, MPP2 and MPP4 had regressed toward baseline but remained statistically significantly elevated. In contrast, the frequency of the myeloid-predominate MPP3 population remained consistently elevated 10 d after injury (Fig. 3C). At 21 d after injury, both MPP2 and MPP3 frequencies remained slightly elevated over the baseline; there was no statistically significant increase in MPP4. We then measured the absolute number of each MPP subset. Consistent with prior studies, in naive animals, MPP4 comprised >75% of the total MPP pool (Fig. 3D, 3E). One day after injury, there was a small expansion in all MPP populations. At 10 d, we found a significant expansion in the myeloid MPP population such that MPP2 and MPP3 combined to make up >40% of the total MPP populations (Fig. 3E). By day 21, the MPP4 population distribution had returned to baseline and small but statistically significant increases in the number of myeloid-predominant MPP2 and MPP3 remained detectable (Fig. 3D).

To determine if the injury-induced changes in HSPC populations were associated with functional changes in hematopoiesis, we isolated bone marrow, spleen, and blood at 10 or 21 d after injury and measured myeloid and lymphoid leukocyte numbers and frequencies by flow cytometry (Supplemental Fig. 3). In the bone marrow 10 d after injury, there was an increase in the number of Ly6G^hi (mature) neutrophils and monocytes (Fig. 4A). Eosinophil numbers, as well as T and B cell numbers, were decreased (Fig. 4A, Supplemental Fig. 4). By 21 d, we found that injury was associated with increased numbers of Ly6G^lo (immature) neutrophils and eosinophils in the injured mice (Fig. 4A). There was a trend toward increased numbers of total monocytes and a statistically significant increase in classical (CCR2^hi) monocytes (Fig. 4A, Supplemental Fig. 4). These changes were associated with a reciprocal decrease in B cell frequency but no change in absolute B cell numbers (Supplemental Fig. 4).

We then evaluated the circulating and splenic leukocyte populations. In the circulation 10 d after injury, we found increased numbers of neutrophils and monocytes (Fig. 4B); concurrently, there was a decrease in B and T cell populations (See Supplemental Fig. 4). By 21 d after injury, there were no statistically significant differences in myeloid cell populations; however, absolute numbers of B and T cells were decreased (See Supplemental Fig. 4). Splenic leukocytes were also evaluated. Ten days after injury, neutrophil and monocyte numbers were increased, and B cell numbers were decreased. At 21 d, we detected significantly higher numbers of all analyzed myeloid populations; numbers of neutrophils, monocytes, eosinophils, and dendritic cells were increased in injured mice versus naive controls (Fig. 4C). This was associated with a decrease in CD4⁺ and CD8⁺ T cell frequencies, but there was no difference in absolute T cell numbers. (Supplemental Fig. 4).

Given that we measured a very limited and rapidly resolving systemic inflammatory response to trauma (Supplemental Fig. 1B), we sought to measure a broad panel of circulating mediators to determine the mechanisms by which injury induces EH. We isolated plasma from sham and injured mice 24 h after manipulation and measured soluble mediators using a multiplexed cytokine assay. We found that there were very limited changes in circulating mediator populations, with the exception of the growth factor G-CSF. We detected a 2-fold increase in G-CSF in injured as compared with sham-manipulated mice, with injured animals having >1600 pg/ml G-CSF in the plasma as compared with <800 pg/ml after sham manipulation (Table I). These data are consistent with previous data demonstrating that trauma is associated with elevated G-CSF levels both in animal model systems (25) and in injured patients (26). To better characterize the G-CSF production induced by injury, we measured G-CSF at 0, 3, 6, 24, and 48 h after trauma, and found that by 3 h after injury, elevated levels of G-CSF were detectable in the plasma, which peaked by 6 h; G-CSF levels decreased by 24 h but remained elevated above baseline at both 24 and 48 h after injury (Fig. 5A).

Prior work has demonstrated that exogenous G-CSF alone is sufficient to induce EH (12, 13). To establish the role of G-CSF in inducing EH after trauma, we assayed the effect of G-CSF Ab blockade on injury-induced changes in HSPC. Mice were treated with two doses of G-CSF–blocking Ab (or isotype-matched control Ab) before injury, and progenitor populations were analyzed by FACS as before (Fig. 5B). We found that G-CSF blockade significantly attenuated the injury-induced expansion of HSPC, blocking the expansion of KSL, LT-HSC, and ST-HSC populations (Fig. 5C). In contrast, we found no difference in committed progenitor frequency in the setting of G-CSF blockade, with an equivalent expansion of GrMP and MEP after polytrauma in mice treated with G-CSF–blocking Ab. These data suggest that the expansion of immature progenitors (i.e., HSC) is driven by G-CSF production, whereas early changes in committed progenitors are G-CSF independent.

To determine the role of G-CSF in the myeloid skewing of early hematopoietic progenitors, we measured the effect of G-CSF blockade on the MPP subset distribution 24 h after trauma. MPP populations were identified as before (Fig. 5B), and their frequencies were assessed. G-CSF blockade effectively blocked the expansion of all MPP subsets 24 h after injury (Fig. 5D). G-CSF blockade did not significantly alter the induction of CD41 on LT-HSC. These data demonstrate that G-CSF is a key driver of MPP expansion induced by trauma.

Prior studies have demonstrated that MyD88 signaling by TLRs, including TLR4 and TLR2, is required to induce EH (8, 9, 27, 28). In contrast, studies by others have demonstrated MyD88-independent EH in the setting of experimental cecal ligation and puncture induced sepsis (7). Others have demonstrated that sterile injury can induce G-CSF production in a MyD88-dependent manner in response to sterile chemical inflammation (29). To determine the role of MyD88 in the induction of G-CSF and EH by polytrauma, we first measured G-CSF levels in WT and MyD88^−/− mice 24 h after trauma. WT and MyD88^−/− were subjected to polytrauma and both plasma G-CSF levels, and progenitor populations were assayed by flow cytometry (representative FACS plots are shown in Fig. 6A). We noted that the injury-induced G-CSF induction was completely abolished in MyD88^−/− animals (Fig. 6B).

We then evaluated the role of MyD88 on trauma-induced EH by measuring progenitor populations in WT and MyD88^−/− mice 24 h after polytrauma. Consistent with our G-CSF blockade experiments (Fig. 5), we found that in the absence of MyD88 signaling, trauma had no effect on the frequency of LT-HSC or ST-HSC populations (Fig. 6C). These data demonstrate that MyD88 signaling is required for both G-CSF production and the expansion of LT-HSC and ST-HSC after polytrauma. Inflammation can be associated with changes in the progenitor cell surface immunophenotype. To determine if the increased frequency of progenitor cells represented a true change in functional immature progenitor frequency, we measured the number of cell CFU in the bone marrow of naive and injured mice. Bone marrow cells from naive and injured mice were cultured in methylcellulose media containing growth factors, and the number of colonies formed was enumerated after 7 d. Consistent with the increase in the frequency of LT-HSC as assayed by flow cytometry, we found that the bone marrow from injured WT mice produced higher numbers of colonies as compared with that isolated from naive animals (Fig. 6D). In contrast, injury had no effect on the number of colonies produced by bone marrow from MyD88^−/− animals. We did note a trend toward an increase in the number of CFU in the bone marrow of the MyD88^−/− mice as compared with WT animals. These data are consistent with our past studies demonstrating that TLR-mediated signaling in the HSC leads to HSC quiescence and that progenitors from untreated MyD88^−/− or TLR4^−/− mice have a repopulation advantage as compared with WT cells (27).

Given that we observed MyD88-dependent production of G-CSF in the setting of sterile injury, we hypothesized that this reflected the activation of MyD88-dependent IL-1 signaling. To test this hypothesis, we treated mice with rIL-1R antagonists (IL-1RA, Anakinra, Sobi) 1 h before polytrauma. Plasma was harvested 6 h after injury and plasma G-CSF level was measured by CBA. We found that the IL-1 blockade nearly completely abrogated injury-induced elaboration of G-CSF (Fig. 7A). To determine the role of the IL-1/G-CSF axis in driving progenitor expansion, we measured the frequency of hematopoietic progenitor populations 24 h after injury in mice treated with IL-1RA. Mice were treated with three doses of IL-1RA, one dose given 16 h before injury, a second dose at the time of injury, and a third dose 8 h after injury. Bone marrow was harvested 24 h after injury, and progenitor populations were measured by flow cytometry. Progenitor populations were identified as before (Fig. 7B). Consistent with our G-CSF blockade experiments, we found that IL-1 blockade resulted in decreased frequencies of LT-HSC and ST-HSC but not the bulk MPP populations (Fig. 7C). Again consistent with G-CSF blockade, IL-1RA treatment blocked the injury-induced increase in MPP2 subpopulations, but there was only a modest (not statistically significant) trend toward a decrease in MPP3 (Fig. 7C). IL-1 blockade did not have a significant effect on the MPP4 populations (Fig. 7C).

IL-1RA blocks the activity of IL-1R1 in response to both IL-1α and IL-1β. To determine whether IL-1α or IL-1β was driving G-CSF production, we measured the effect of specific cytokine blockade on G-CSF production in response to injury. Mice were treated with blocking Abs against IL-1α, IL-1β, or both in combination; control animals were treated with isotype control Ab. Six hours after injury, plasma was harvested, and G-CSF levels were measured by CBA. We found that anti–IL-1α significantly blocked injury-induced G-CSF production (Fig. 8A). There was a trend toward decreased G-CSF in the presence of Ab-mediated blocking of IL-1β, although this was not statistically significant. Although these data and data in Fig. 7 suggest that IL-1 is driving G-CSF production and EH after injury, we did not detect any significant levels of IL-1α or IL-1β in the plasma of injured mice. To determine the cellular source of IL-1, we measured IL-1 protein levels in liver homogenates and in cell lysates prepared from bone marrow and peritoneal lavage from naive animals as well as animals at 6 h after injury. IL-1α and IL-1β were measured by CBA, and the amount of IL-1 relative to total protein was calculated to account for differences in tissue mass and/or cellular composition. We found very low levels of IL-1 within bone marrow cells; injury had no effect on the levels of IL-1α in this tissue (Fig. 8B) and was associated with a decrease in IL-1β (Fig. 8C). In contrast, we found that the peritoneal exudate cells from injured mice contained significantly higher levels of both IL-1α and IL-1β as compared with those from naive animals (Fig. 8B, 8C). There was no difference in IL-1α levels in injured versus naive liver and minimal levels of IL-1β in injured liver tissue (Fig. 8B, 8C).

We report that sterile tissue injury alone results in the expansion of immature hematopoietic progenitors, including LT-HSC and MPP populations through IL-1/MyD88–dependent production of G-CSF, leading to a sustained increase in myeloid-biased progenitors at 10 d after insult. These data demonstrate that EH can be induced independent of exogenous pathogen-associated molecular patterns and describe a novel IL-1/G-CSF access in inducing EH. These data suggest a potential mechanism for the persistent changes in immune function seen in severely injured patients (30–32).

Previous work has demonstrated that exogenous G-CSF can increase the number of phenotypic HSC in the bone marrow (27). Our data define a biologically relevant role for this pathway, with IL-1 and G-CSF driving the expansion of pluripotent hematopoietic progenitors in a clinically relevant model of moderate, nonlethal polytrauma. We find that blocking injury-induced G-CSF almost completely abrogates the expansion of HSC induced by injury. These effects of G-CSF also extend to the MPP compartment, and we find that G-CSF similarly blocks the effect of injury on the MPP2, MPP3, and MPP4 subsets. Although we did not find a statistically significant decrease in the bulk MPP population, we did observe a trend toward a decreased MPP frequency in the setting of G-CSF blockade after injury. Consistent with past literature, we have defined MPP based on the expression of CD34 and FLT3 (CD135) (33), whereas the MPP subsets are defined based on the expression of a combination of CD135/CD48 and CD150. There is incomplete overlap between the MPP2/3/4 populations and the ST-HSC and MPP populations as defined by these markers. In addition, inflammation can affect the expression of cell surface markers on the progenitor populations, and changes in cell surface marker expression may also contribute to decreased magnitude of the effect of G-CSF on the MPP population (as compared with the MPP2/3/4 populations). Interestingly, G-CSF blockade had no effect on the frequency of the committed GrMP and MEP populations, despite the well-defined role of G-CSF in granulopoiesis (4, 34, 35).

Consistent with myeloid biasing of the progenitors for up to 10 d, we do report an increase in myeloid cell populations in the bone marrow and spleen up to 3 wk after injury. We acknowledge that the differences were modest, and the functional consequences of these small differences in neutrophil and monocyte populations are unclear. Further, although we hypothesize that these data could be an expression of myeloid biasing of the progenitors by EH, there may be ongoing inflammation related to the injury that has incompletely resolved at this time. Future studies will be required to determine if the injury-induced expansion of progenitors has significant functional consequences.

Our results also define a new role for IL-1/MyD88 in EH after injury. Scumpia et al. (7) demonstrated that in the setting of endotoxemia or bacterial sepsis, LT-HSC expansion is independent of MyD88 and IL-1 activity. In contrast, we find a complete block in both G-CSF production and expansion of the LT-HSC after trauma in the MyD88 knockout mice and that the IL-1 blockade with rIL-1RA also almost completely blocked G-CSF production, largely recapitulating the effects of the G-CSF blockade on the changes in hematopoietic progenitors induced by injury. These data are consistent with previous work demonstrating that G-CSF production in response to chemical peritonitis in driven by IL-1α and that IL-1α/IL-1R signaling is required to initiate the inflammatory response to necrotic cell death (36, 37). Although we find that abrogating IL-1/MyD88 signaling effectively prevented both G-CSF production and EH, we did not detect elevated levels of IL-1α in the plasma at 3, 6, or 24 h after injury (data not shown). However, we were able to detect a significant increase in IL-1 in the peritoneal exudate cells after injury. Liver injury is associated with a significant peritoneal infiltrate, and the increased IL-1 we detect likely reflects infiltrating neutrophils and the activation of peritoneal resident macrophages and neutrophils. These data are consistent with prior work demonstrating that in the setting of chemical peritonitis, IL-1α is produced by peritoneal neutrophils and drives G-CSF production, regulated by NADPH oxidase (29). Alternatively, Boettcher et al. (34) have demonstrated that in the setting of endotoxemia, G-CSF is produced by endothelial cells in a MyD88-dependent pathway. These authors presume that MyD88 was acting downstream of the TLR in response to LPS. IL-1 is well known to be produced in response to LPS (38), and our results suggest that endothelial cell G-CSF production in the setting of LPS may be mediated by IL-1, acting as a secondary messenger produced in response to LPS. Future studies will be required to parse the sources of G-CSF and IL-1.

Prior studies have evaluated changes in early hematopoietic progenitors after sterile injury (14, 17, 18). Schneider et al. (17) reported that hemorrhagic shock caused an increase in the frequency of the mixed population of progenitors, whereas Silva et al. (14) and Nacionales et al. (18) found no change in pluripotent progenitor populations in young animals after burn injury or polytrauma, although Silva et al. did report an increase in myeloid-committed progenitors based on ex vivo culture assays. These studies use a heterogenous collection of cell surface phenotypes to define the progenitor population. We describe a complete differential of progenitors in the bone marrow, and using rigorous definitions, we find that sterile injury results in the expansion of both pluripotent progenitor populations and myeloid-committed progenitors. We further find that the effects of injury on the hematopoietic compartment are sustained for at least 10 d after injury. The persistent changes in the bone marrow progenitor cohort suggest that changes in progenitor function could underlie the persistent immune function defects observed in critically injured patients. Injured patients develop nosocomial infections at 5–10 d after their trauma (39). Most myeloid cells have a circulation half-life of 8–24 h under baseline conditions (40, 41), and systemic insults such as critical illness are associated with the margination of circulating cells with the rapid replacement of those cells by new bone marrow emigrants (42). The leukocytes in circulation by 48 h after injury are likely derived from progenitor cells in the bone marrow after the insult, and therefore any injury-induced changes in function may be encoded by the progenitor cells during differentiation. This hypothesis is consistent with recent data demonstrating that IL-1–mediated “immunologic training” of hematopoietic progenitors can lead to persistent changes in innate immune function (43, 44).

In conclusion, we find that sterile polytrauma injury induces the expansion of the pluripotent HSPC populations. Progenitor expansion is mediated by G-CSF, produced in response to IL-1 through a MyD88-dependent pathway. Progenitor expansion is associated with an increase in myeloid-biased progenitors, including CD41⁺ LT-HSC and MPP2/MPP3 subpopulations, with the increase in myeloid-biased progenitors persisting for up to 21 d after injury. These results suggest that the persistent defects in immune function observed after injury may result from changes in the progenitor cohort and suggest that the modulation of progenitor function may be a potential pathway to modulate injury-induced immune dysfunction.

The authors have no financial conflicts of interest.