Type-I-IFNs act upon hematopoietic progenitors to protect and maintain hematopoiesis during Pneumocystis lung infection in mice

Abstract

Although acquired bone marrow failure (BMF) is considered a T cell-mediated autoimmune disease, few studies have considered contributing roles of innate immune deviations following otherwise innocuous infections as a cause underlying the immune defects that lead to BMF. Type-I-IFN signaling plays an important role in protecting hematopoiesis during systemic stress responses to the opportunistic fungal pathogen Pneumocystis. During Pneumocystis lung infection, mice deficient in both lymphocytes and type-I-IFN-receptor (IFrag^−/−) develop rapidly progressing BMF associated with accelerated hematopoietic cell apoptosis. However, the communication pathway eliciting the induction of BMF in response to this strictly pulmonary infection has been unclear. We developed a conditional-null allele of Ifnar1 and used tissue-specific induction of the IFrag^−/− state and found that, following Pneumocystis lung infection, type-I-IFNs act not only in the lung to prevent systemic immune deviations, but also within the progenitor compartment of the BM to protect hematopoiesis. In addition, transfer of sterile-filtered serum from Pneumocystis-infected mice as well as intra-peritoneal injection of Pneumocystis into uninfected IFrag^−/− mice induced BMF. Although specific cytokine deviations contribute to induction of BMF, immune-suppressive treatment of infected IFrag^−/− mice ameliorated its progression but did not prevent loss of hematopoietic progenitor functions. This suggested that additional, non-cytokine factors also target and impair progenitor functions; and interestingly, fungal β-glucans were also detected in serum. In conclusion, our data demonstrates that type-1-IFN signaling protects hematopoiesis within the BM compartment from the damaging effects of pro-inflammatory cytokines elicited by Pneumocystis in the lung and possibly at extra-pulmonary sites via circulating fungal components.

Introduction

Microbial challenges induce expansion and lineage-specific differentiation of hematopoietic progenitor cells (HPC) in the bone marrow (BM) (1, 2). Communication between site of injury and BM is accomplished via systemic secretion of cytokines including IL-1, IL-6, G-CSF, type-I-IFNs and TNF-α by resident immune cells (3–11). However, hematopoietic stem cells (HSCs) and early HPCs also directly respond to circulating pathogen-associated molecular patterns (PAMPs) via TOLL-like receptor (TLR)-mediated signaling pathways (12). Under these circumstances, HPCs can be a source of pro-inflammatory cytokines to drive stress-induced hematopoiesis (13).

Consistent with this, non-disseminating lung infections with the fungus Pneumocystis cause expansion of HPCs (14). Thus, either inflammatory mediators elicited in the lung entering the blood and/or low levels of circulating fungal PAMPs, such as circulating β-glucans, likely drive BM responses (15, 16). While this response is designed to meet increased demand for inflammatory cells during infection, inflammation can also harm hematopoiesis and result in BM dysfunctions due to HSC destruction or exhaustion (5, 17); however the exact mechanisms contributing to these dysfunctions remain unclear.

BM failure (BMF) occurs in the context of inherited and acquired conditions and can manifest as aplastic anemia with severe peripheral cytopenias and acellular BM spaces (18). While most acquired aplastic anemias are thought to result from T cell-mediated autoimmune responses to unknown, likely infectious stimuli, inherited forms typically associate with gene defects affecting the viability of HSCs in response to inflammatory stimuli (19–22). BM suppression also occurs as a complication of inflammatory syndromes including sepsis, rheumatoid diseases and AIDS (23–25). Thus, while the mechanisms underlying BMF are complex, a common theme appears to be the presence of inflammatory stimuli accompanied by immune deviations.

We recently showed that type-I-IFN signaling protects on-demand hematopoiesis during infections with the strictly pulmonary pathogen Pneumocystis in a mouse model of the disease (26). Specifically, mice lacking both lymphocytes- and type-I-IFN-receptor (IFrag^−/− mice) die of BMF 16- to 21-days after Pneumocystis lung infection. This response is ameliorated in lymphocyte-competent, type-I-IFN-receptor-deficient mice (IFNAR^−/−) to a transient BM depression accompanied by profound extramedullary hematopoiesis (26, 27). While Pneumocystis does not disseminate to the BM in this model, exuberant systemic cytokine secretion, including IL-1β, IL-18, and IFN-γ, resembles the innate immune activation seen in inflammasome-activation-induced inflammatory syndromes that promote accelerated neutrophil depletion (27–29). Hematopoiesis remains unperturbed in IFNAR-competent lymphocyte-deficient (Rag1^−/−) or wild type mice and, in both cases, a robust systemic IFN-α response is observed.

Type-I-IFNs are conserved cytokines consisting of IFNα_1–14, IFNβ, IFN-ε, -κ and -ω that signal via a common receptor (IFNAR) expressed on all cells (30). IFN-α and IFN-β mediate immunity to viral, bacterial and fungal infections (31–33) by directing the activation, maturation and differentiation of macrophages, dendritic-, T-, NK- (34–36) and B-cells (37). Type-I-IFNs also induce IL-10 production (38–40), transcriptionally repress TNF-α (41), and inhibit inflammasome activation and subsequent IL-1β/IL18 processing (42). These pleiotropic functions are also evident in their diverse roles in autoimmunity and hematopoiesis (43–45). Type-I-IFNs act as neutrophil survival factors that prevent apoptosis during inflammation (46, 47) and induce HSC proliferation and differentiation in this normally dormant compartment (8). In contrast, chronic exposure to IFN-α can result in HSC exhaustion and loss of function (8, 48).

To better understand the opposing functions of type-I-IFNs on BM, we here investigated how and where type-I-IFN signaling acts to prevent BM dysfunctions following Pneumocystis lung infection. By using immune-defined mice in combination with immunosuppressive treatment and serum transfer studies in IFrag^−/− mice, we show that, following Pneumocystis lung infection, type-I-IFNs act within the progenitor compartment of the BM to protect hematopoiesis. Finally we show that, in addition to actions of pro-inflammatory cytokines, circulating fungal PAMPs contribute to BM dysfunctions when type-I-IFN-signaling is impaired. This study provides mechanistic insights into the important roles of type-I-IFN signaling in maintaining homeostasis of the BM progenitor compartment in the face of potential overstimulation by circulating cytokines and fungal PAMPs.

Material and Methods

Mice

Animals were housed at the AAALAC-approved Animal Care Center at MSU in HEPA-ventilator cages (Tecniplast) on sterile feed (Picolab 5053) and water. Protocols were approved by the MSU-IACUC. Wild type C57Bl/6 mice were obtained from Jackson Laboratories (Bar Harbor, Maine, stock JAX-000664) or Charles Rivers Laboratories (Charleston, South Carolina; stock CRL-027). C.B17 SCID mice, C57Bl/6 Rag1^−/− mice, B6.Cg-Tg(Itgax-cre)1-1Reiz/J mice, B6.129P2-Lyz2^tm1(cre)Ifo/J (“LysM^cre”) mice, and B6.129(Cg)-Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,-EGFP)Luo}/J (ROSA^mT-mG) mice were from Jackson Laboratories (stock JAX-001803, 002096, 008068, 004781, 007676, respectively).

A conditional-null allele of the Ifnar1 gene in which exon 3 is flanked by loxP sites (“floxed”, allele entitled Ifnar1^fl) (49), was generated on a purely C57Bl/6 background (Figure 1A; details are presented in Supplemental Figure S1). This Ifnar^fl mouse line will be available from The Jackson Laboratory as Stock#028256. The allele was targeted into the C57Bl/6 chromosome-16 of F1 C57BL/6 x 129SvEv hybrid ES cells (Figure 1B). A germline Ifnar1⁻ allele on a pure C57Bl/6 background was generated by breeding with deleter-Cre (50) (Figure 1C&D). Primary mouse fibroblast cultures (PMFC) were established from ear biopsies from Ifnar1^+/+, Ifnar1^fl/−, Ifnar1⁺/⁻, and Ifnar1^−/− mice (51). Analysis of IFNα-induced Mx-1 mRNA expression (52) in primary mouse fibroblasts (PMFs) of genotype Ifnar1^+/+, Ifnar1^fl/−, Ifnar1^+/−, and Ifnar1^−/− showed that only the Ifnar1^−/− PMFS failed to induce Mx-1 mRNA expression upon IFNα stimulation (Figure 1E). This confirmed that the Ifnar1⁻ allele recombined into a true null phenotype following Cre exposure (e.g., in Ifnar1^−/− PMFs, lanes 10–12) and that the Ifnar1^fl allele was functionally wild type (e.g., in Ifnar1^fl/− PMFs, lanes 4–6). Using the C57Bl/6 Ifnar1^−/− mice, we generated C57Bl/6 IFrag^−/− (“IFrag^−/−_(B6)”) by husbandry with C57Bl/6 Rag1^−/− mice and their BM responses were compared to the IFrag^−/− mice generated on a mixed C57Bl/6 x 129SvEv genetic background (“IFrag^−/−₍₁₂₉₎ mice(26)). Both IFrag^−/−₍₁₂₉₎ and IFrag^−/−_(B6) mice progressed to BMF within 16 days following Pneumocystis lung infection (Figure 1F). In both IFrag^−/− strains, this was accompanied by rapid loss of CD11b⁺Gr-1^high expressing neutrophils (Figure 1G) and increased apoptotic cell death, which was reflected in increasing Caspase-3 activity in bone marrow cells (Figure 1H). This verified that the observed bone marrow phenotype following Pneumocystis lung infection was a result of the absence of type-I-IFN-mediated pathways in lymphocyte-deficient mice. IFrag^−/−_(B6) mice were used for all subsequent experiments.

Figure 1. Generation of the Ifnar1^fl and Ifnar1⁻ alleles on a purely C57Bl/6 background demonstrates that BM phenotype in IFrag^−/− mice is not dependent on genetic background.

(A) Schematic of the Ifnar1^fl allele. Recombination between LoxP sites by Cre recombinase results in loss of exon 3, yielding the Ifnar1⁻ allele. (B) SNP analysis of Ifnar1^fl allele. Genomic DNA from F1 mouse tail biopsies (lanes 1–7) or control DNA (last three lanes) were PCR amplified with primers flanking the D16mit131 SNP, confirming that the Ifnar1^fl allele was targeted to the C57BL6/J chromosome in hybrid C57BL6/J-129X1/SvJ ES cells. (C) Ifnar1 genotype analyses with primer pair Ifnar1 intron 2-forward and Ifnar1 intron 3-reverse. DNA was isolated from tail biopsies of post-natal day 15 mice that contained wild type (+) 1012 bp, floxed (fl) 568 bp, or null (−) 288 bp combinations of Ifnar1 alleles. Genotypes are listed under numbers in lanes 1–5; lane 6, no DNA template; lane 7, DNA size marker. PCR results verify that Ifnar1^−/− animals are viable (lane 5). (D) Expression of mRNA from the Ifnar1 allele. RNA harvested from Ifnar1^+/+ liver or Ifnar1^−/− liver and thymus was used to make oligo(dT)-primed cDNA and Ifnar1 mRNA expression was analyzed using primer pair Ifnar1 exon 1-forward and Ifnar1 exon 5-reverse. The PCR product from the Ifnar1⁺ (+) allele is 689 bp and from the Ifnar1⁻ (−) allele is 510 bp. Results show that only the wild type mRNA is detectable in Ifnar1^+/+ liver (lane 2), and that accumulation of the null mRNA in Ifnar1^−/− liver and thymus (lanes 3 and 4, respectively) is detectable. (E) Analyses of type I IFN-induced Mx1 gene expression in primary mouse embryonic fibroblast cells derived from mice of the indicated genotypes. Cells that received 0, 50 or 500 units/ml of IFN-α as indicated (lanes 1–12). Lanes 13 and 14, no-DNA-template control and DNA size marker, respectively. β-actin mRNA levels served as an internal control. A comparative analysis of BM response of IFrag^−/− mice on a mixed C57Bl/6 x 129 genetic background (IFrag^−/− ₍₁₂₉₎) and the newly generated IFrag^−/− mice on C57BL/6 genetic background (IFrag^−/− _(B6)) was performed. Rag1^−/− mice served as a control to the experiment. (F) Total BM cell numbers were evaluated by hemocytometer count, (G) BM composition determined by FACS analysis and neutrophils determined as % of CD11b⁺Gr-1^hi expressing cells. (H) The rate of BM cell apoptosis was determine by FACS analysis by staining cells for the presence of activated Caspase-3 using a rhodamine-labeled Caspase-3-specific-inhibitor (CaspGLOW⁺). P values are marked as * when comparing IFrag^−/−₍₁₂₉₎ to RAG^−/− mice or ^# when comparing IFrag^−/−_(B6) to RAG^−/− mice. *^/# p< 0.05, **^/## p<0.01, ***^/### p<0.001.

For myeloid cell-specific disruption of the Ifnar1^fl allele in CD11c^hi expressing dendritic cells, and LysM-expressing macrophages and granulocytes, colonies were generated that were either Ifnar1^fl/fl;Itgax-cre^1or2 or were Ifnar1^fl/fl;Rag1^−/−;Itgax-cre^1or2 (homozygous for the Ifnar1^fl allele and carrying the Itgax-cre transgene, either on a C57Bl/6 wild type or a C57Bl/6 Rag1^−/− lymphocyte-deficient background; in combination called “Itgax-cre;Ifnar1^fl” or “Itgax-cre;IFrag^fl”, respectively) or were LysM^cre/+;Ifnar1^fl/fl or LysM^cre/+;Rag1^−/−;Ifnar1^fl/fl (as above, but with Cre activity driven by the LysM^cre knock-in allele on chromosome 10; called “LysM^cre;Ifnar^fl” or “LysM^cre;IFrag^fl”, respectively). Similar husbandry and allelic selection were used to generate the LysM^cre;Itgax-cre;IFrag^fl mouse lines. Verification of tissue specific IFNAR-disruption was established by PCR and by using a global double-fluorescent Cre-reporter mouse system in which non-recombined cell express red-fluorescent membrane tomato (mT), while in recombined cells mT is excised to allow membrane EGFP-expression (mG) (53) (Supplemental Figure 2). All mice used in this study were between 7- and 8-weeks of age. All mouse lines used in this study are available for unrestricted non-profit research use unless specifically restricted by another party.

Verification of Ifnar1 alleles

PCR marker analysis for C57BL6/J and 129X1/SvJ used primer pair D16mit131-forward (5′-GTCACATTAACAGCAATCTTGTCT-3′) and D16mit131-reverse (5′-GTTGTCTTCTGATGTTCACATACATG-3′). Ifnar1 genotype analyses performed on mouse tail biopsy DNA used primer pair Ifnar1 intron 2-forward (5′-GTAAGTACACTGTAGCTGTCTTCAG-3′) and Ifnar1 intron 3-reverse (5′-GCACATTGACCATTACAAGAGTAG-3′). Oligo(dT)-primed cDNA samples were prepared as previously described (54). cDNA analysis of Ifnar1⁺ and Ifnar1 ⁻ alleles was performed with primer pair Ifnar1 exon 1-forward (5′-TATAGATCTCCCAAGACGATGCTCGCTGTCG-3′) and Ifnar1 exon 5-reverse (5′-TATAAGCTTATACACTGCACAGTGCTGTA-3′). Mx1 cDNA analysis used primer pair Mx1 exon 10-forward (5′-GGGTTGACTACCACTGAGATGAC-3′) and Mx1 exon 11-reverse (5′-GTTAATCGGAGAATTTGGCAAGCT-3′). β-actin cDNA analysis used primer pair β-act exon 5-forward (5′-GCTGTCTGGTGGTACCACCATGTA-3′) and β-act exon 6-reverse (5′-ATCTGCTGGAAGGTGGACAGTGAG-3′).

Pneumocystis lung infections and other treatments

Mice were intratracheally (i.t.) infected with 10⁷ Pneumocystis murina nuclei (hereafter Pneumocystis) and Pneumocystis lung-burden was assessed microscopically by enumeration of trophozoid and cyst nuclei in lung homogenates as previously described (55). The limit of detection for this technique is log₁₀ 4.2. As indicated, some mice were i.t. instilled with 100 μg Zymosan depleted of TLR stimulating properties by hot alkali treatment (Zymosan deplete; Invivogen) or received drinking water containing 8.3 mg/ml Dexamethasone. For serum transfer experiments, uninfected IFrag^−/− mice received 500 μl of pooled and sterile-filtered serum (0.2 μm filter) from Pneumocystis-infected IFrag^−/−, Rag1^−/− or SCID mice via intraperitoneal (i.p.) injection. For in vivo depletion of NK cells, IFrag^−/− mice received combined or individual treatment with 100 μg polyclonal rabbit anti-asialoGM1 (WAKO Pure Chemical Industries Ltd) and 200 μg anti-NK1.1 antibody (clone PK136, cell line obtained from ATCC) twice weekly by intraperitoneal injection (56, 57)

BM collection, differentiation, and analysis

BM cells from femur and tibia were collected, enumerated, and differential counts were performed on Diff-Quik^™ (Siemens)-stained slides (26, 58). For FACS analyses (29), antibodies were: anti-CD11b (AlexaFluor700, clone M1/70, BioLegend); anti-Ly-6G/6C (APC-Cy7, clone RB6-8C5, Pharmingen); mouse hematopoietic lineage marker mix (eFluor 450 cocktail, eBioscience), anti-ckit (APC, clone 2B8, eBioscience); and anti-Sca-1 (PE, clone D7, eBioscience). Caspase-3 activity was analyzed using CaspGLOW^™ Red Active Caspase staining reagents (Biovison) in combination with cell surface marker analysis (29). Data were acquired using an LSR FACS (Becton Dickinson) and analyzed using FlowJo Software. For analysis of proliferative responses, mice received 100 mg/kg body weight of 5-bromo-2′deoxyuridine (BrdU, Fisher Scientific) via intra-peritoneal injection two days before harvest (17) and cells were analyzed using a BrdU-Staining kit (eBioscience).

Cytokine analysis and β-glucan assays

Serum levels of IFNγ, IL-1β, IL-18, IL-12p70, and IL-10 were measured by Bioplex (Bio-Rad) analysis system using multiplex cytokine assay plates from Bio-Rad Life Sciences. In some experiments, IL-1β and IFNγ protein concentrations were measured by ELISA using Duoset reagents (R&D Systems). Serum IFN-α levels were measured by ELISA (PBL Interferon Source; R&D systems). Serum β-glucan levels were evaluated using the Fungitell^R assay kit (Cap Cod Clinical Diagnostic).

Colony-forming cell assay for BM cells

Hematopoietic precursor cell activity in BM was assessed by hematopoietic colony forming counts (CFC-U) in methylcellulose media as previously described (29). For this, 10⁵ BM cells per animal at each time point were plated in MethCult^R GF M3534 media (StemCell Technologies). according to the manufacturer’s protocols. Colony type recognition (GM-, G-, M- forming colonies) and enumeration was performed after 7 days.

Preparation of lung cryosections and immunohistochemistry

Lungs were removed, filled with O.C.T. compound (Tissue-Tek), and frozen in liquid nitrogen. Sections (5 μm) were cut on a Cryostat (Leica Biosystems) and fixed in 10% neutral buffered formalin for 30 min. Sections were incubated in 100 mM glycine followed by washes in PBS; equilibrated to Tris-buffered saline, pH 7.5 (TBS); blocked with Rodent Block M (BioCare); and incubated with anti-ISG54 (IFIT2) antibody (polyclonal rabbit anti-human/mouse; Pierce; PA3-845) at 1:50 dilution. Detection used Rabbit on Rodent AP-Polymer (BioCare) according to the manufacture’s protocols and staining was visualized using the Vulcan Fast Red Chromogen (BioCare). Sections were evaluated using a Nikon Eclipse 80i microscope at 200X magnification and images were captured digitally.

Statistical analyses

All experimental groups consisted of four mice and each experiment was repeated at least twice. Data are presented as means ± SEM and each graph show results from representative experiments. Data were plotted using Prism Graph pad Software, Inc. and statistical analysis was performed using either a one-way or two-way ANOVA analysis of variance, followed by a Tukey or Bonferroni post hoc test, respectively. For pairwise comparison a Students-t-test was performed.

Results

Pneumocystis lung infection-induced BMF in IFrag^−/− mice is independent of the genetic background associated with diminished in vivo HPC cell proliferation

Previously we showed that complete BMF develops within 16 to 21 days following Pneumocystis infection of IFrag^−/− but not Rag1^−/− mice on mixed 129SvEv x C57Bl/6 backgrounds (26–29, 59). To investigate the mechanisms by which type-I-IFN signaling protected BM from Pneumocystis-induced failure, we developed a floxed Ifnar1 allele (Ifnar1^fl) on a purely C57Bl/6 background (Figure 1A–E) & Material and Methods) and verified that the null version of this phenocopied the previous mixed-background Ifnar1⁻ allele. Pneumocystis infected IFrag^−/− mice compared to RAG^−/− mice show total BM cell number depletion (Figure 1F & 2A) due to apoptosis-induced loss of mature and maturing Ly6G/C (Gr-1^hi) expressing neutrophils (Figure 1G,H & 2B) starting at day 7 post-infection and lack of cellular replenishment from HPC/HSC as evidenced by diminishing in vitro colony forming activity (CFU count) (Figure 2C) of plated BM cells. Decreased in vitro colony forming activity of IFrag^−/− BM cells is also associated with decreased in vivo proliferation of Lin⁻Sca-1⁺cKit⁺ (LSK) HPC progenitors by day-10 post-infection (Figure 2D,E) suggesting an infection-induced suppressive effect within the progenitor compartment.

Figure 2. BMF in IFrag^−/− mice following Pneumocystis lung infection is associated with reduced in vivo proliferation within the LSK compartment.

(A). Shown are comparisons of total BM cell numbers between Rag1^−/− and IFrag^−/− mice over a 16 day time course after Pneumocystis lung infection. (B.) Using fluorescently labeled antibody staining, neutrophils are characterized by high Gr-1-expression (Gr-1^hi). Shown is an example of comparative Gr-1 staining patterns on BM cells of IFrag^−/− and RAG^−/− mice at day 0 and 16 post-infection displayed as histograms. (C). Colony forming activity (CFU counts) in BM cells was evaluated as a measure of hematopoietic progenitor activity. Shown are comparative CFU counts of BM cells form IFrag^−/− and Rag1^−/− mice over a 16 day time course of Pneumocystis lung infection. (D, E). In vivo proliferation of the hematopoietic progenitor compartment in IFrag^−/− and Rag1^−/− mice in response to Pneumocystis lung infection was evaluated by in vivo BrdU incorporation analysis into LSK cells at day 10 post-infection using FACS analysis. (D). Shown is the gating strategy applied for this experiment on cells from uninfected and not BrdU-injected wild type animals. (E). Shown are respective BrdU-staining patterns of representative BM analysis from both Rag1^−/− and IFrag^−/− mice at day 10 post infection Pneumocytsis lung infection and data are summarized in a bar graph. (F). Pneumocystis lung burden was evaluated microscopically over the course of the experiments. Nuclei counts are given in LOG₁₀. The cutoff for microscopic detection is: LOG₁₀ 4.2. (G). Pneumocystis lung infection elicits elevated serum IFN-α levels in IFNAR-competent RAG^−/− mice to which IFrag^−/− mice cannot respond. Statistical P values are marked as * p< 0.05, ** p<0.01, *** p<0.001.

Pneumocystis lung infection triggers type-I-IFNs responses in pulmonary cells of Rag1^−/− mice

During the 16-day time-course of infection used in this study, Pneumocystis lung burden remains low with little histological evidence of pneumonia in either IFrag^−/− or Rag1^−/− mice (≤ 10⁷ Pneumocystis/mouse; Figure 2F & 3). Protection from BMF in Rag1^−/− mice is associated with elevated IFN-α serum levels by day 7 post infection, indicating that Pneumocystis lung infection invokes a systemic type-I-IFN response in mouse strains resistant to Pneumocystis-induced BMF (Figure 2G). Expression of the type-I-IFN-induced protein IFIT2 (IGF54) in the lungs of day 7 Pneumocystis-infected Rag1^−/− mice was also dramatically stronger than that in lungs of infected IFrag^−/− mice (Figure 4). IFIT2 in the lungs of Rag1^−/− mice was detected in alveolar macrophages (AM), alveolar epithelium type I and type II cells (AE I & AE II) and infiltrating inflammatory cells (suspected granulocytes).

Figure 3. Pneumocystis lung infection elicits robust up-regulation type-I-IFN-mediated IFIT2 expression uniquely in lung of IFNAR-competent Rag1^−/− but not IFNAR-deficient IFrag^−/− mice.

IFIT2, a type-I-IFN-induced protein, was used to screen pulmonary cells responding to type-I-IFN-mediated signaling following Pneumocystis lung infection. Shown are comparative expression levels of IFIT2 on frozen section of lungs from day 7 Pneumocystis-infected (A) Rag1^−/− and (B) IFrag^−/− mice. Lungs were stained with primary polyclonal rabbit and human/mouse anti-IFIT2 antibody and detected with Rabbit on rodent AP-Polymer followed by Vulcan Fast Red staining. Positive cells are visualized in red: AE1= Alveolar epithelium type 1(thick arrow); AE2= Alveolar epithelium type II (asterisk); AM = Alveolar macrophage (thin arrow); infiltrating granulocytes (arrow head). Images were taken at 400X magnification.

Figure 4. Lymphocyte deficient mice with conditional deletion of Ifnar1 in either CD11c⁺ (Itgax-cre;IFrag^fl) or CD11b⁺Gr-1⁺ (LysM^cre;IFrag^fl) myeloid cell subsets display transient BM suppression following Pneumocystis lung infection.

Shown are BM cell number changes relative to uninfected litter mates in IFrag^−/− (A), Itgax-cre;IFrag^fl (B) and LysM^cre;IFrag^fl (C) mice over the course of Pneumocystis lung infection. Total BM cell numbers of the same experiment are also plotted in (E). BM cells were spun onto cytospin slides and stained with Diff-Quik^™ for microscopic morphological evaluation. (D). Shown are representative cytospin slides of BM cells from LysM^cre;IFrag^fl compared to IFrag^−/− mice at day 0, 10 and 16 after Pneumocystis lung infection. Images displayed are at 400X magnification. Serum cytokine analysis was performed for IFN-γ by sandwich ELISA in all comparison groups as an indicator cytokine to the systemic immune deviations associated with BMF in IFrag^−/− mice (F). Hematopoietic progenitor activity was evaluated by performing in vitro CFU assays by plating 10⁵ total BM cells from all comparison groups over the course of Pneumocystis lung infection (G). P values are marked as *p< 0.05, ** p<0.01, *** p<0.001.

Lymphocyte-deficient mice with myeloid-cell-specific deletion of IFNAR1 show transient BM changes following Pneumocystis lung infection

Previous chimera experiments indicated a critical role for type-I-IFN-signaling on hematopoietic cells to prevent the progression of BMF in IFrag^−/− mice (29). To test whether type-I-IFN-signaling in specific innate immune cell subsets prevented BMF in this model, BM responses of lymphocyte-deficient IFrag^−/−, LysM^cre;IFrag^fl, and Itgax-cre;IFrag^fl mice to Pneumocystis lung infection were compared. Although total BM cell numbers initially declined dramatically in all three mouse strains in response to the infection, both LysM^cre;IFrag^fl and Itgax-cre;IFrag^fl mice rebounded by 16-days of infection, and only IFrag^−/− mice progressed to complete BMF (Figure 4A–C, E). The declines in total BM cell numbers in both LysM^cre;IFrag^fl, and Itgax-cre;IFrag^fl mice were associated with subtle changes in the overall BM cell composition, including transient loss of mature neutrophils and accumulation of less mature precursor cells (Figure 4D). Although IFN-γ is a contributor to neutrophil depletion in IFrag^−/− mice (27), serum cytokine analyses demonstrated a lack of IFN-γ induction in LysM^cre;IFrag^fl or Itgax-cre;IFrag^fl (Figure 4F). Furthermore, LysM^cre; IFrag^fl or Itgax-cre;IFrag^fl mice exhibited robust BM HPC activity in CFU assays, whereas this activity rapidly declined in IFrag^−/− mice (Figure 4G).

Itgax-cre- or LysM^cre-driven conditional targeting of myelocytic cells can be incomplete (60), and may be insufficient to drive as dramatic effects as observed in IFrag^−/− mice. Alternatively, innate immune cell types neither expressing Itgax-cre nor LysM^cre, such as NK-cells, may need to be deregulated to drive BMF in our model. However, antibody mediated depletion of NK and NK-like cells in IFrag^−/− mice over the course of Pneumocystis lung infection using anti-NK1.1 antibody and/or anti-asialoGM1 (56, 57) did not affect the course of BMF (data not shown). To increase the representation of IFNAR-deficient cells within the myelocytic compartment, we created combined LysM^cre;Itgax-cre;IFrag^fl mice, with which we were able to target >70% of Gr-1^hi expressing granulocytes, CD11b⁺ macrophages and CD11c⁺ dendritic cells in the BM, and > 90% of pulmonary Gr-1^hi neutrophils and CD11c^hi alveolar macrophage/dendritic cells (Supplemental Figure 2). LysM^cre;Itgax-cre;IFrag^fl mice exhibited a profound loss of mature neutrophils, demonstrating significant reduction of CD11b⁺GR-1^hi-expressing neutrophils by day 10 post Pneumocystis lung infection (Figure 5A – C) and increased serum IFN-γ levels (Figure 5D). This mirrored more completely the systemic inflammatory responses observed in IFrag^−/− mice (Figure 4F)(27, 59)). Nevertheless, BM changes remained transient and were followed by recovery of neutrophil numbers (see Discussion).

Figure 5. Lymphocyte-deficient mice with combined deletion of Ifnar1 in both CD11c⁺ and CD11b⁺Gr-1⁺ (LysM^cre;Itgax-cre;IFrag) myeloid cell subsets show systemic cytokine deviation associated with accelerated neutrophil turn over in IFrag^−/− mice.

Comparative bone marrow analysis was performed microscopically and by FACS analysis between Rag1^−/− and LysM^cre;Itgax-cre;IFrag mice over the course of Pneumocystis lung infection. (A). Shown are representative bone marrow cytospin slides from RAG^−/− and LysM^cre;Itgax-cre;IFrag mice at day 0, 10 and 16 post-infection at a 400X magnification. (B). Shown are representative histogram FACS plots demonstrating the percentage of Gr-1^hi expressing neutrophils in Rag1^−/− and LysM^cre;Itgax-cre;IFrag mice at day 0, 10 and 16 post-infection. (C). Summary of the comparison of percentage of the Gr-1^hi-expressing bone marrow cell numbers of Rag1^−/− and LysM^cre;Itgax-cre;IFrag mice at day 0, 10 and 16 post-infection. (D). Serum IFN-γ cytokine levels were measured in both comparison groups over the time course of infection by sandwich ELISA. P values are marked as *p< 0.05, ** p<0.01, *** p<0.001.

Glucocorticoid treatment ameliorates progression of BMF in IFrag^−/− mice

To test potential benefits of immune suppression on hematopoiesis, both IFrag^−/− and Rag1^−/− mice were treated with dexamethasone (Dex:1mg/kg/d) over the course of Pneumocystis lung infection and their BM responses were compared to those of untreated but infected IFrag^−/− mice. While Pneumocystis lung burden was comparable between groups (Figure 6A), total BM cell numbers remained stable over a time course of 13 days in both Dex-treated IFrag^−/− and Rag1^−/− mice, but rapidly declined in untreated IFrag^−/− mice (Figure 6B). The percentage of neutrophils (Figure 6C,D) also remained high throughout day 10 in Dex-treated IFrag^−/− mice, while it had already diminished in untreated IFrag^−/− mice. This was consistent with a suppressed IFN-γ response in Dex-treated compared to untreated IFrag^−/− mice (Figure 6E). These initial observations implied that immune suppression could protect IFrag^−/− mice from progression of BM failure. However, by day 13 post-infection neutrophil numbers in Dex-treated IFrag^−/− mice were significantly reduced compared to Dex-treated Rag1^−/− mice (Figure 6C,D). This was associated with a corresponding increase in BM eosinophila, which in this model commonly precedes BMF (26) (Figure 6F). Moreover, hematopoietic progenitor CFU-activity continued to decline rapidly at day 10 in both Dex-treated and untreated Pneumocystis-infected IFrag^−/− mice, but remained intact in Dex-treated Pneumocystis-infected Rag1^−/− mice (Figure 6G). Thus, additional factors not suppressed by dexamethasone appear to contribute to disease progression by specifically targeting progenitor functions.

Figure 6. Immune suppressive treatment with Dexamethasone ameliorates progression of BMF in IFrag^−/− mice by protecting from accelerated neutrophil turn over.

Groups of IFrag^−/− and Rag1^−/− mice received Dexamethasone (Dex:1mg/kg/d) in drinking water starting 3 days prior to Pneumocystis lung infection and the effects on BM responses were compared to untreated but infected IFrag^−/− mice over the course of 13 days post-infection. Shown are: (A) Pneumocystis lung burden in LOG₁₀ over the course of infection; (B) total BM cell numbers; (C) percentage of neutrophils in BM evaluated microscopically; (D) FACS-based neutrophil enumeration by CD11b⁺Gr-1^hi surface expression analysis; (E) serum IFN-γ levels at day 10 determined by sandwich ELISA; (F) percentage of eosinophils in BM evaluated microscopically; (G) in vitro progenitor activity in 10⁵ total BM cells determined by CFU assay. P values are marked as * when comparing untreated IFrag^−/− to Dex-treated RAG^−/− mice, ^# when comparing untreated IFrag^−/− to Dex-treated IFrag^−/− mice and ^$ when comparing Dex-treated RAG^−/− mice to Dex-treated IFrag^−/− mice **^/##/$$ p<0.01, ***^/###/$$$ p<0.001.

Circulating fungal PAMPs may contribute to BMF in IFrag^−/− mice

Pneumocystis pneumonia normally develops in lymphocyte deficient mice within 4- to 8-weeks with fungal lung burdens > 1×10⁸ organisms (61, 62). This is in contrast to the low fungal burden present in IFrag^−/− or Rag1^−/− mice within the first sixteen days of infection (< 10⁷ organisms, Figure 2F), in which IFrag^−/− mice exhibit rapidly progressing BMF.

SCID mice, used as source mice for Pneumocystis pathogen, have substantial levels of circulating fungal β-glucans at 8 weeks post-infection (Figure 7A). To evaluate if fungal β-glucans may be detectable early in the course of infection in IFrag^−/− and Rag1^−/− mice, serum was collected from both mouse strains at days 1, 3, 5 and 7 post-infection and β-glucans were analyzed. Fungal β-glucans were detectable at low levels in both mouse strains as early as day 1 post-infections (Figure 7B), while Pneumocystis lung burden steadily increased (Figure 7C).

Figure 7. Serum transfer from Pneumocystis-infected, lymphocyte-deficient mice into IFrag^−/− mice elicits BMF and fungal β-glucan contribute to the phenotype.

Shown are β-glucan serum levels in (A) 4 and 8 week Pneumocystis-infected CB17SCID source mice; (B) 1–7 days Pneumocystis infected IFrag^−/− and Rag1^−/− mice. (C) Shown are pulmonary Pneumocystis nuclei counts for respective IFrag^−/− and Rag1^−/− mice. Serum was harvested from heavily Pneumocystis infected CB17 SCID mice, as well as from 7- to 10-day infected IFrag^−/− and Rag1^−/− mice. Serum was diluted 1:1 in PBS, sterilely filtered through a 0.2 μm syringe filter and 500 μl of serum was intra-peritoneally (i.p) injected into groups of uninfected IFrag^−/− mice. BM responses were evaluated at 14 days post-injection and compared to those of IFrag^−/− mice that had been infected with Pneumocystis pathogen or those that had received pooled serum from uninfected IFrag^−/− and Rag1^−/− mice. Shown are total BM cell numbers (D); percentage of BM neutrophils (E); and in vitro colony forming activity (CFU count) of BM progenitors (F). Comparisons were made to BM responses in uninfected, untreated IFrag^−/− mice. IFrag^−/− mice were either infected with Pneumocystis pathogen or instilled with 100 μg of LPS-depleted Zymosan and BM responses were observed at day 10 and 16 post-instillation. Shown are total BM cell numbers (G) and the percent of BM neutrophils enumerated microscopically in the comparison groups (H). Shown are total BM cell numbers in IFrag^−/− mice following either i.p instillation of whole Pneumocystis pathogen or Zymosan in comparison to i.t instillation of whole pathogen at day 14 post treatment (I). P values are marked as *p< 0.05, ** p<0.01, *** p<0.001.

To test whether circulating serum factors induce BMF in this model, we transferred sterile-filtered serum from 8-week Pneumocystis-infected SCID mice and from 7-day Pneumocystis-infected IFrag^−/− and Rag1^−/− mice into IFrag^−/− mice. BM responses of serum-transferred IFrag^−/− mice were compared after 14 days to those of Pneumocystis-lung infected IFrag^−/− mice and those transferred with pooled serum from uninfected RAG^−/− and IFrag^−/− mice. Serum from Pneumocystis-infected SCID, RAG^−/− and IFrag^−/− mice induced BMF in IFrag^−/− mice with a rapid decrease of total BM cell numbers, specific loss of neutrophils, and decreased CFU-activity in the progenitor compartment. In contrast, transfer of serum from uninfected mice had no effect (Figure 7D–F). This suggested that a serum factor common in all infected donor mouse strains promotes BMF in IFrag^−/− mice and pointed to a potential role of soluble β-glucans or another fungal PAMP.

To test whether fungal cell wall-associated PAMPs such as β-glucans alone can trigger BMF in the IFrag^−/− model, groups of IFrag^−/− mice were i.t. instilled with either Pneumocystis pathogen or hot alkali-treated Zymosan (100 μg/mouse). Although induced more profoundly in PC infected mice, both treatment groups showed significant decline of total BM cell numbers by day 16-post-instillation (Figure 7G). This was also reflected in loss of total BM neutrophils (Figure 7H) without evidence of recovery. To distinguish whether responses to circulating fungal components outside the lung can also induce BMF, we compared the effects of intraperitoneal injection (i.p) of either whole PC or Zymosan to that of the natural route of PC lung infection on BM functions. Both, i.p. and i.t instillation of whole pathogen as well as Zymosan instillation resulted in BMF in IFrag^−/− mice. However, these effects were more profound in those mice that had received whole pathogen compared to Zymosan instillation alone (Figure 7I).

Discussion

Engagement of microbial pattern recognition receptors (PRRs) on tissue-resident innate immune cells triggers a cytokine-mediated systemic acute-phase response that activates HSC within the BM to induce emergency hematopoiesis (1, 2). However, HSCs also express receptors for, and can directly respond to, PAMPs (63). While these responses are designed to maintain appropriate HSC numbers and also trigger their differentiation into lineage specific hematopoietic progenitor cells (HPC) during inflammation, excessive stimulation and proliferation can also impair HSC functions (5, 13, 17, 64).

Previously we showed that systemic IFN-α released in response to Pneumocystis lung infection critically supports on-demand hematopoiesis (26, 27, 29). This finding was unexpected because Pneumocystis infections were not previously implicated in impairing hematopoiesis and high dose or chronic type-I-IFN-exposure has been shown to suppress hematopoiesis by causing excess proliferation followed by functional exhaustion of HSC (8, 48, 65, 66). These seemingly opposing effects of type-I-IFNs on hematopoiesis emphasize the dichotomous role these cytokines can play during inflammation (45, 67).

In support of our observation that Pneumocystis lung infection also elicits BM responses, work by Shi et al. recently demonstrated a robust expansion of HPC cells in the BM within the first 2 weeks following Pneumocystis lung infection in wild type mice (14). Consistent with a potential pro-proliferative role of type-I-IFNs in these responses, our new studies in lymphocyte-deficient, IFNAR-deficient IFrag^−/− mice demonstrated that progression of BMF is associated with diminished in vivo proliferative responses of HPCs cells by day 10 after Pneumocystis lung infection when compared to IFNAR-competent Rag1^−/− mice. This is consistent with progressive lack of cellular replenishment and decreasing in vitro progenitor activity, and indicates a stem cell/early progenitor defect as a result of the infection.

We previously proposed a two-pronged inflammatory pathway in determining pathogenesis in this model: One that drives accelerated neutrophil turn-over and another that blocks replenishment from early progenitors via TLR- and inflammasome-triggered cytokine release at the site of infection into the serum. Supporting this model, neutralization of IFN-γ, IL-1β, TRAIL, reactive oxygen species or TNF-α either ameliorated accelerated turnover of neutrophils or protected progenitor functions, but could not completely block all effects on hematopoiesis (27–29). However, it remained unclear whether type-I-IFNs primarily acted on HPCs to protect their functions during Pneumocystis lung infection or critically modulated immune responses in the lung that would otherwise secondarily inhibit hematopoiesis.

Pneumocystis rarely disseminates and is considered a strictly pulmonary fungal pathogen (68–70) and profound systemic immune deviations precede BMF in IFrag^−/− mice. This strongly suggests an important immune modulatory role of type-I-IFNs on classical immune sentinels in the lung to protect from cytokine responses causing BMF. The use of mice with conditional deletion of IFNAR1 in subsets of mature phagocytic cells allowed us to evaluate how lack of type-I-IFN-signaling in classical sentinel immune cells, also present in the lung, affected BM functions. Surprisingly, deletion of IFNAR1 in either CD11c⁺ dendritic cells and alveolar macrophages (Itgax-cre;IFrag), or in the CD11b⁺Gr-1⁺ granulocytic/monocytic compartment (LysM^cre;IFrag^fl), or combined IFNAR-deletion in both CD11c⁺ and Gr-1⁺cells (LysM^cre;Itgax-cre;IFrag^fl) resulted in only transient BM suppression primarily affecting neutrophil numbers. Neutrophil depletion, however, was more profound in LysM^cre;Itgax-cre;IFrag^fl mice when compared to Itgax-cre;IFrag^fl or LysM^cre;IFrag^fl mice and, like in Pneumocystis-infected IFrag^−/− mice was associated with increased IFN-γ serum levels (27). However, HPC-activity remained intact and hematopoiesis rebounded by day 16 post-infection. While this observation supported the hypothesis of a two-pronged pathway in the pathogenesis of BMF, it was unclear how deviated innate immune activation initiated in the lung of IFrag^−/− mice contributed to the “second-hit” that impaired HPC functions. It also implicated the need for type-I-IFN-signaling outside the lung, likely on early hematopoietic progenitors, to protect against BMF in response to Pneumocystis lung infection.

Although Pneumocystis does not detectably disseminate to the BM at these early time points, fungal β-glucans could be detected in the serum of infected IFrag^−/− mice as early as day 1–6 post-infection. Thus, three possible triggers for the BMF exist: 1) lung-induced immune deviations; 2) systemic release of pathogen-derived soluble fungal PAMPs such as β-glucans; or 3) a combination of both.

Glucocorticoids are powerful immunosuppressants (19, 71, 72) and dexamethasone-treatment of Pneumocystis-infected IFrag^−/− mice suppressed normally elevated IFN-γ serum levels that we had previously identified to accelerate neutrophil loss (27). However, while it abrogated neutrophil depletion, HPC activity remained severely impaired, eventually resulting in BM failure. Transfer of sterile-filtered serum from various lymphocyte-deficient Pneumocystis-infected mouse strains into previously healthy IFrag^−/− mice also markedly reduced BM numbers and triggered loss of progenitor functions. Importantly, all sera capable of inducing BMF in IFrag^−/− mice had detectable β-glucan levels. Furthermore, i.t. delivery of sterile hot alkali-treated Zymosan, in which 1–3 β-glucans among other fungal cell wall components are present, induced a profound reduction of total BM cell numbers in IFrag^−/− mice within 16 days. Although this observation is inconsistent with the previously reported need of microbe-bound β-glucans to trigger dectin-1-mediated responses (73), purified β-glucans have been shown to mediate human dendritic cell activation (74) and enhance IL-1β transcription and processing following Dectin-1 binding (75).

While these data pointed to a role of fungal PAMPs in triggering BMF in our model, it did not allow us to discern whether BM-harming responses were induced in the lung and spilled into the serum or whether such responses could also be elicited outside the lung environment in response to circulating fungal PAMPs. Indeed, intraperitoneal (i.p) administration of either whole Pneumocystis or Zymosan also showed profound effects on hematopoiesis. These effects were however ameliorated in those mice that had received Zymosan i.p. when compared to i.p injection of the whole pathogen. Thus, while cytokine-mediated responses elicited in the lung appear to drive accelerated neutrophil loss, soluble fungal PAMPs may, either directly or indirectly, contribute to the proposed “second hit” on the hematopoietic progenitor compartment that is instrumental in the pathogenesis of BMF in this model. Notably, type-I-IFN-signaling, likely in the progenitor compartment, is critically important to modulate these effects.

Although the exact mechanisms of how type-I-IFNs protect hematopoiesis in our model still remains unclear, we have observed differential expression of regulatory miRNAs in the BM of IFrag^−/− when compared to Rag1^−/− mice involved in modulating cell proliferation and apoptotic pathways. Studies further evaluating this exciting new association are currently under way.

Importantly, our studies highlight how distal fungal infections can, through the release of inflammatory mediators in form of cytokines and possibly PAMPs, have profound systemic effects that impair hematopoiesis. Furthermore, they demonstrate the protective role that type-I-IFNs have in preventing these outcomes. This emphasizes the intricate actions of cytokine networks during inflammation, which might have profound clinical implications. Immune suppressive treatment can improve BM functions in individuals with aplastic anemia, however, relapses are observed (76–78). Conversely, immunosuppressive treatment can predispose to Pneumocystis lung infection and inadvertently also suppress type-I-IFN activity. This could potentially be a contributing factor to relapses in the disease. Among the ever-increasing population of immune compromised individuals, either due to HIV infection or due to immune suppressive treatments, BM suppression is a common often unexplained finding (79–82). Like in the mouse model presented here, this could be driven by low grade Pneumocystis lung infections combined with impaired type-I-IFN-responses. This understanding might allow for the development of novel preventative or therapeutic strategies.