Dysfunctional expansion of hematopoietic stem cells and block of myeloid differentiation in lethal sepsis

Abstract

Severe sepsis is one of the leading causes of death worldwide. High mortality rates in sepsis are frequently associated with neutropenia. Despite the central role of neutrophils in innate immunity, the mechanisms causing neutropenia during sepsis remain elusive. Here, we show that neutropenia is caused in part by apoptosis and is sustained by a block of hematopoietic stem cell (HSC) differentiation. Using a sepsis murine model, we found that the human opportunistic bacterial pathogen Pseudomonas aeruginosa caused neutrophil depletion and expansion of the HSC pool in the bone marrow. "Septic" HSCs were significantly impaired in competitive repopulation assays and defective in generating common myeloid progenitors and granulocyte-monocyte progenitors, resulting in lower rates of myeloid differentiation in vitro and in vivo. Delayed myeloid-neutrophil differentiation was further mapped using a lysozyme-green fluorescent protein (GFP) reporter mouse. Pseudomonas's lipopolysaccharide was necessary and sufficient to induce myelosuppresion and required intact TLR4 signaling. Our results establish a previously unrecognized link between HSC regulation and host response in severe sepsis and demonstrate a novel role for TLR4.

Figure 1

Loss of bone marrow neutrophils during sepsis. CD1 mice challenged with burn and inoculation of the PA14 (PA14) or 33C7 (33C7) strain, burn only (B), and normal controls (N) were killed at the indicated time points. Bone marrow (BM) cells were harvested and stained with antibodies directed to Gr1 and Mac1 markers and analyzed by flow cytometry. (A) Dot blots in the left panel show FSC (indicative of size) and SSC (indicative of granularity) of total BM samples. The FCS^low/SSC^high population was gated (R1) and analyzed for Gr1/Mac1 expression (middle panel), and sorted for morphologic analysis (right panel). (B) Dot blots show Gr1 and Mac1 expression in a representative experiment. Numbers indicate percentage of cells within the gate. (C) Line graph summarizes 3 independent experiments. Values are the average of 6 to 10 mice and indicate percentages of BM Gr1⁺/Mac1⁺ cells. The value at time 0 is the average of 15 normal controls. Error bars indicate SD. (D) Bar graph shows percentages of annexin V expression on Gr1⁺/Mac1⁺ population in N, B, and PA14 mice. Values are averages of 3 to 4 mice. Error bars indicate SD. *P = .04. (E) Bar graph shows the average percentage of total BM Gr1⁺/Mac1⁺ cells in PA14-challenged mice; percentages are normalized to Gr1⁺/Mac1⁺ cells in normal controls to equal 100%. The light portion of the column represents the average percentage of living cells; the dark portion of the column represents average percentage of annexin V⁺ apoptotic cells (n = 4). Error bars indicate SD.

Figure 2

Sepsis induces increase of primitive cells with LSK phenotype. CD1 mice challenged with burn and inoculation of the PA14 (PA14) or 33C7 (33C7) strain, burn only (B), and normal controls (N) were killed at the indicated time points. BM cells were stained with antibodies directed to Lin⁺ markers and to Sca1 and c-Kit markers. Samples were analyzed by multicolor flow cytometric analysis. (A) Dot blots show Sca1 and c-Kit expression on gated Lin⁻ cells in N, B, and PA14 mice at 24 hours after challenge in a representative experiment. Numbers indicate percentage of cells within the Lin⁻ gate. (B) Line graph indicates percentage of LSK cells on overall BM with exclusion of the R1 region in Figure 1A (to avoid relative increase of percentage due to loss of the FCS^low/SSC^high population) in a representative experiment. Values are averages of 4 to 8 mice. Time 0 is the average of 8 normal animals. (C) Bar graph shows average fold increase in LSK absolute number at 24 hours from challenge. In each column values are the average of 10 mice compared with normal controls in 3 independent experiments. Error bars indicate SE. LPS: *P = .03; PA14: *P = .01. (D) Bar graph indicates percentages of LSK cells in the S+G₂M phase of the cell cycle, as determined by Hoechst staining. Values indicate average of 3 mice at 24 hours from challenge in a representative experiment. Error bars indicate SD. *P < .01. (E) Long-term culture with limiting dilution was performed on BM cells to quantify the frequencies of hematopoietic progenitors and stem cells. Data indicate frequency of CAFCs and are represented as means ± SD. (F) Bar graph indicates percentage of Lin⁺ c-Kit⁺ cells in N, B, and PA14 mice at 24 hours from challenge. Values indicate average of 6 mice in a representative experiment at 24 hours from challenge. Error bars indicate SD. *P < .001. (G) Measure of LSK cell output. Bar graph shows ratio of percentages of Lin⁺Kit⁺ (Kit⁺) over LSK cells (r = Kit⁺/LSK). Percentages of each population were measured on identical gates to equal total BM excluding R1, as in Figure 1A. Values indicate average of 6 mice in 2 independent experiments at 24 hours from challenge. Error bars indicate SD. *P = .016.

Figure 3

P aeruginosa's LPS is sufficient and necessary to induce the BM alterations observed during sepsis. Cohorts of CD1 and C57BL/6J mice were injected intraperitoneally with P aeruginosa's LPS (0.8 mg/kg) or phosphate-buffered saline (PBS) and compared with mice challenged with PA14. Mice were killed at the indicated time points, and their BM cells were analyzed. (A) BM cells harvested at 24 hours from LPS or PBS injection, or PA14 challenge, were labeled with antibodies anti-Gr1 and anti-Mac1. Bar graph shows percentage of Gr1⁺/Mac1⁺ cells in total BM. Values are average of 6 mice in 3 independent experiments. Error bars indicate SD. *P < .001. (B) BM smears from PBS- or LPS-challenged mice. Samples were stained by hematoxylin/eosin, and morphology was evaluated by light microscopy (×100 magnification). Circled area indicates the presence of neutrophils in PBS condition and of immature cells in LPS (arrows). (C) Bar graph indicates percentages of LSK cells in the S+G₂M phase of the cell cycle, as determined by Hoechst staining. Values indicate average of 2 mice at 24 hours from challenge in a representative experiment. Error bars indicate SD. (D) Each sample of sorted LSK was derived from a pool of BM of 4 to 6 mice. RNA was extracted from LSK (IL-7R⁻) cells from PBS- and LPS-challenged mice at 24 hours. Samples were analyzed for expression of SKP2 by qRT-PCR. Bar graphs represent averages of fold changes in expression in 3 independent samples from 2 independent experiments. Error bars indicate SD: LPS versus PBS, P = .04. (E) BM cells were analyzed for colony-forming ability by methylcellulose colony assay. Bar graph shows average number of myeloid colonies (CFU-GM + CFU-G + CFU-M) per 10 000 cells in 3 independent samples, each of them in quadruplicate. *P < .001.

Figure 4

P aeruginosa's LPS causes reduction of BM common myeloid progenitors and granulocyte-monocyte progenitors. Cohorts of CD1 or C57BL/6J mice were injected intraperitoneally with P aeruginosa's LPS (0.8 mg/kg) or PBS. (A) BM was harvested 24 hours after injection, stained with specific antibodies, and analyzed by multicolor flow cytometry. The dot blot on the left insight shows expression of c-Kit and Sca-1 in Lin⁻/IL-7R⁻ cells. Dot blot on the right insight shows FcγRII/III expression on gated Lin⁻/IL7-R⁻/Sca⁻/Kit⁺ cells to determine CMP, GMP, and MEP subsets. Bar graphs indicate percentages of CMPs and GMPs on the Lin⁻ population. Values are average of 10 to 12 mice and summarize 3 independent experiments. Error bars indicate SE. CMP: LPS versus PBS *P < .001; GMP: LPS versus PBS *P < .001. (B) Lys-GFP reporter mice were challenged with P aeruginosa's LPS or PBS, and the BM was harvested at 24 hours from injection. Dot plots show GFP expression on CMPs, GMPs, and granulocytes in a representative experiment. Values in the bar graph indicate percentage of cells that express no or low levels of Lys-GFP (gate A) within each specific subset and are average of 4 mice. CMP: LPS versus PBS *P = .05; GMP: LPS versus PBS *P < .001; granulocytes: LPS versus PBS *P = .035. (C) BM of C57/B6 mice challenged with LPS or PBS was analyzed for the presence of MEPs (left panel), CLPs (middle panel), and MPPs (right panel) at 24 hours from challenge. Bar graphs indicate percentages of MEPs and CLPs on the Lin⁻ population. Values are average of 10 to 12 mice and summarize 3 independent experiments. Error bars indicate SE. MEP: LPS versus PBS *P < .001; CLP: LPS versus PBS *P < .001. Short-term hematopoietic stem cells (ST-HSCs; or MPPs) were defined as Lin⁻/IL-7R⁻/Sca⁺/Kit⁺/CD34⁺/Flt3⁺ cells. Bar graph shows percentage of ST-HSCs () and long-term HSCs (LT-HSCs; ■) on total Lin⁻ population. Values are average of 12 mice and summarize 3 independent experiments. Error bars indicate SE. *P < .001. (D) In each experiment, a sample of sorted LSK was derived from a pool of BM of 4 to 6 mice. RNA was extracted from LSK (IL-7R⁻) cells from PBS- and LPS-challenged mice. Samples were analyzed for expression of C/EBPα and PU.1 by qRT-PCR. Bar graphs represent averages of fold changes in expression in 3 independent samples. Error bars indicate SD. LPS versus PBS: C/EBPα, *P = .033; PU.1 *P = .029. (E) Sepsis causes alterations in HSC differentiation (working model). During severe sepsis, bacterial LPS induces TLR4-dependent expansion of dysfunctional LSK cells displaying a defective ability to progress into the pool of myeloid progenitors: CMPs and GMPs. Reduction in CMPs and GMPs results in neutropenia.

Figure 4

P aeruginosa's LPS causes reduction of BM common myeloid progenitors and granulocyte-monocyte progenitors. Cohorts of CD1 or C57BL/6J mice were injected intraperitoneally with P aeruginosa's LPS (0.8 mg/kg) or PBS. (A) BM was harvested 24 hours after injection, stained with specific antibodies, and analyzed by multicolor flow cytometry. The dot blot on the left insight shows expression of c-Kit and Sca-1 in Lin⁻/IL-7R⁻ cells. Dot blot on the right insight shows FcγRII/III expression on gated Lin⁻/IL7-R⁻/Sca⁻/Kit⁺ cells to determine CMP, GMP, and MEP subsets. Bar graphs indicate percentages of CMPs and GMPs on the Lin⁻ population. Values are average of 10 to 12 mice and summarize 3 independent experiments. Error bars indicate SE. CMP: LPS versus PBS *P < .001; GMP: LPS versus PBS *P < .001. (B) Lys-GFP reporter mice were challenged with P aeruginosa's LPS or PBS, and the BM was harvested at 24 hours from injection. Dot plots show GFP expression on CMPs, GMPs, and granulocytes in a representative experiment. Values in the bar graph indicate percentage of cells that express no or low levels of Lys-GFP (gate A) within each specific subset and are average of 4 mice. CMP: LPS versus PBS *P = .05; GMP: LPS versus PBS *P < .001; granulocytes: LPS versus PBS *P = .035. (C) BM of C57/B6 mice challenged with LPS or PBS was analyzed for the presence of MEPs (left panel), CLPs (middle panel), and MPPs (right panel) at 24 hours from challenge. Bar graphs indicate percentages of MEPs and CLPs on the Lin⁻ population. Values are average of 10 to 12 mice and summarize 3 independent experiments. Error bars indicate SE. MEP: LPS versus PBS *P < .001; CLP: LPS versus PBS *P < .001. Short-term hematopoietic stem cells (ST-HSCs; or MPPs) were defined as Lin⁻/IL-7R⁻/Sca⁺/Kit⁺/CD34⁺/Flt3⁺ cells. Bar graph shows percentage of ST-HSCs () and long-term HSCs (LT-HSCs; ■) on total Lin⁻ population. Values are average of 12 mice and summarize 3 independent experiments. Error bars indicate SE. *P < .001. (D) In each experiment, a sample of sorted LSK was derived from a pool of BM of 4 to 6 mice. RNA was extracted from LSK (IL-7R⁻) cells from PBS- and LPS-challenged mice. Samples were analyzed for expression of C/EBPα and PU.1 by qRT-PCR. Bar graphs represent averages of fold changes in expression in 3 independent samples. Error bars indicate SD. LPS versus PBS: C/EBPα, *P = .033; PU.1 *P = .029. (E) Sepsis causes alterations in HSC differentiation (working model). During severe sepsis, bacterial LPS induces TLR4-dependent expansion of dysfunctional LSK cells displaying a defective ability to progress into the pool of myeloid progenitors: CMPs and GMPs. Reduction in CMPs and GMPs results in neutropenia.

Figure 5

P aeruginosa's LPS causes defective myeloid differentiation. C57BL/6J mice were injected intraperitoneally with P aeruginosa's LPS (0.8 mg/kg) or PBS. BM was harvested at 24 hours from injection, and Lin⁻ cells were sorted by FACS and grown in vitro in prodifferentiative conditions in the presence of SCF, IL-3, and G-CSF. Cells were harvested at the indicated time points and analyzed for expression of differentiation markers by flow cytometry. (A) Dot blots shows expression of c-Kit and of a combination of myeloid differentiation markers (Gr1 + Mac1 + F4/80) at day 1 and day 3 of culture in a representative experiment. (B) Line graph represents acquisition of the myeloid markers (Gr1, Mac1, F4/80) over time. Values represent average percentages of 3 samples from a representative experiment of 3 independent experiments. Error bars indicate SD. Day 3 *P < .001. (C) Line graph shows kinetic of LSK cells in culture in 3 independent experiments. Values represent average percentage of LSK cells (n = 7).

Figure 6

Septic HSPCs show functional defects in BM transplantation assays. C57BL/6J CD45.2 mice were injected intraperitoneally with LPS from P aeruginosa (0.8 mg/kg) or PBS. BM was harvested at 24 hours from injection, and Lin⁻ or LSK cells, IL-7R⁻ (to exclude contamination from CLPs), were sorted by FACS. Sorted populations were transplanted into irradiated CD45.1 donors, and engraftment was evaluated at each indicated time point by analysis of CD45.2⁺ cells in the peripheral blood. (A) A total of 2 × 10⁵ Lin⁻CD45.2⁺ sorted cells were injected into a lethally irradiated CD45.1⁺ Boy/J recipient together with 3 × 10⁵ Lin⁻CD45.1⁺ competitor cells. Line graph shows engraftment as percentage of CD45.2 cells during time. Values are average of 10 to 12 mice from 3 independent experiments. Bars indicate SE. P value is nonsignificant. (B) A total of 2500 LSK/CD45.2⁺ sorted cells were injected into a lethally irradiated CD45.1 Boy/J recipient together with 10⁵ CD45.1⁺ competitor total BM cells. Line graph shows engraftment as percentage of CD45.2 cells during time. Values are average of 10 to 12 mice from 3 independent experiments. Bars indicate SE. *P < .001. (C) Bar graph shows level of donor engraftment in the PB at 4 weeks of transplantation in mice receiving decreasing doses (2500, 1000, 500) of LSK cells derived from PBS controls or LPS-challenged mice. Values are average of 5 to 8 mice. *P < .001. (D) PB cells were collected and stained with antibodies directed to the myeloid lineage. Bar graph shows percentages of donors Gr1⁺/CD45.2⁺ cells on total GR1⁺ cells (100%) at 2, 4, and 8 weeks from transplantation. Values are average of 8 to 12 mice. Error bars indicate SE. *P < .001.

Figure 7

The effects induced by LPS on HSCs and neutrophils are abrogated in a TLR4-null phenotype. (A) C57BL/6J CD45.2 mice were injected intraperitoneally with LPS from P aeruginosa (0.8 mg/kg) or PBS. BM was harvested at 24 hours from injection, and Lin⁻ cells were purified and stained with c-Kit, Sca1, and TLR4/MD2 antibodies. Histograms show TLR4 expression on LSK cells (black line) superimposed to TLR4 expression on LK cells (Lin⁻Kit⁺ Sca⁻; gray filled curve) in PBS control (left panel) and in LPS-challenged mice (right panel). TLR4 expression on LK cells was at limit of detection and superimposed with IgGs control. (B) C3H/OuJ(TLR4^wt) and C3H/HeJ (TLR4⁻) mice were injected intraperitoneally with LPS from P aeruginosa (0.8 mg/kg) or PBS. BM was harvested at 24 hours, stained with monoclonal antibodies necessary to identify the indicated subsets, and analyzed by multicolor flow cytometric analysis. (B) Average percentage of LSK on the Lin⁻ cells. TLR4^WT: PBS versus LPS; *P = .045. (C) Average percentage of Gr1⁺/Mac1⁺ neutrophils in the total BM population. TLR4^WT: PBS versus LPS *P < .001. (D) Average percentage of CMPs in the Lin⁻ population. TLR4^WT: PBS versus LPS; *P < .001. (E) Average percentage of GMPs in the Lin⁻ population. TLR4^WT: PBS versus LPS; *P < .001. In all graphs, values are average of 6 mice from 2 independent experiments. Error bars show SD.