Endothelial cells translate pathogen signals into G-CSF-driven emergency granulopoiesis

Abstract

Systemic bacterial infection induces a hematopoietic response program termed "emergency granulopoiesis" that is characterized by increased de novo bone marrow (BM) neutrophil production. How loss of local immune control and bacterial dissemination is sensed and subsequently translated into the switch from steady-state to emergency granulopoiesis is, however, unknown. Using tissue-specific myeloid differentiation primary response gene 88 (Myd88)-deficient mice and in vivo lipopolysaccharide (LPS) administration to model severe bacterial infection, we here show that endothelial cells (ECs) but not hematopoietic cells, hepatocytes, pericytes, or BM stromal cells, are essential cells for this process. Indeed, ECs from multiple tissues including BM express high levels of Tlr4 and Myd88 and are the primary source of granulocyte colony-stimulating factor (G-CSF), the key granulopoietic cytokine, after LPS challenge or infection with Escherichia coli. EC-intrinsic MYD88 signaling and subsequent G-CSF production by ECs is required for myeloid progenitor lineage skewing toward granulocyte-macrophage progenitors, increased colony-forming unit granulocyte activity in BM, and accelerated BM neutrophil generation after LPS stimulation. Thus, ECs catalyze the detection of systemic infection into demand-adapted granulopoiesis.

Figure 1

LPS-induced emergency granulopoiesis requires intact MYD88-mediated TLR signaling in nonhematopoietic cells. (A) Graphical scheme depicting experimental outline to induce LPS-induced emergency granulopoiesis in mice with MYD88 expression restricted to hematopoietic (WT→Myd88^−/−) or nonhematopoietic (Myd88^−/−→WT) cells, and respective control mice (WT→WT and Myd88^−/−→Myd88^−/−). (B) Frequency and absolute number of PB CD11b⁺ cells, and (C) BM cellularity after LPS stimulation in BM chimeric and control mice. (D) Representative FACS profile showing characteristic changes in BM CD11b⁺Gr1^high mature and BM CD11b⁺Gr1^low immature neutrophils after LPS stimulation. (E) Frequencies and absolute numbers of BM CD11b⁺Gr1^high mature, (F) BM CD11b⁺Gr1^low immature neutrophils, and (G) plasma G-CSF levels in reciprocal Myd88^−/− BM chimeras upon systemic LPS injection. Black squares, PBS-injected mice; red squares, LPS-injected mice; rel., relative; abs., absolute. Data from 2 independent experiments are shown. Two-tailed Student t tests were used to assess statistical significance (*P < .05, **P < .01, ***P < .001; ns, nonsignificant).

Figure 2

LPS-induced emergency granulopoiesis is abrogated in Tie2-Cre;Myd88^fl/fl mice. (A) Frequency and absolute number of PB CD11b⁺ cells after LPS stimulation in control Myd88^fl/fl, and experimental Nes-Cre;Myd88^fl/fl, Pdgfrb-Cre;Myd88^fl/fl, Alb-Cre;Myd88^fl/fl, and Tie2-Cre;Myd88^fl/fl mice. (B) BM cellularity after LPS stimulation in control Myd88^fl/fl, and experimental Nes-Cre;Myd88^fl/fl, Pdgfrb-Cre;Myd88^fl/fl, Alb-Cre;Myd88^fl/fl, and Tie2-Cre;Myd88^fl/fl mice. (C) Representative FACS profile showing characteristic LPS-induced changes in BM CD11b⁺Gr1^high mature and BM CD11b⁺Gr1^low immature neutrophils in control Myd88^fl/fl, and experimental Nes-Cre;Myd88^fl/fl, Pdgfrb-Cre;Myd88^fl/fl, Alb-Cre;Myd88^fl/fl, 0and Tie2-Cre;Myd88^fl/fl mice. (D) Frequencies and absolute numbers of BM CD11b⁺Gr1^high mature and (E) BM CD11b⁺Gr1^low immature neutrophils, and (F) plasma G-CSF levels in the different sets of tissue-specific Myd88^−/− mice after systemic LPS injection. Black squares, PBS-injected mice; red squares, LPS-injected mice; rel., relative; abs., absolute. Data from at least 3 independent experiments are shown. Two-tailed Student t tests were used to assess statistical significance (*P < .05, **P < .01, ***P < .001).

Figure 3

Endothelial cells from various organs express high levels of Myd88 and Tlr4, respond to LPS challenge with strong upregulation of VCAM-1 and G-csf in vivo, and exogenous G-CSF administration rescues emergency granulopoiesis in Tie2-Cre;Myd88^fl/fl mice. (A) Myd88 and (B) Tlr4 expression were assessed by quantitative reverse-transcription PCR in CD45⁻Ter119⁻CD31⁺ECs isolated from BM, heart, liver, kidney, spleen, and lung (gray bars) compared with spleen DCs (black bars), ie, pooled classical dendritic cells (CD3ε⁻CD19⁻NK1.1⁻CD11c^highCD45RA⁻MHCII⁺) and plasmacytoid dendritic cells (CD3ε⁻CD19⁻NK1.1⁻CD11c⁺CD45RA⁺MHCII^high). All cells were isolated from steady-state mice. (C) Graphical scheme depicting experimental outline to assess in vivo LPS responsiveness of ECs that were flow-cytometrically sorted from BM, heart, liver, kidney, spleen, and lung of PBS- and LPS-injected mice, respectively. (D) Comparative G-csf transcript levels normalized to Actb in ECs from BM, heart, liver, kidney, spleen, and lung of LPS-injected (red bars) vs PBS-injected (black bars) wild-type mice. (E) Representative FACS analysis depicting cell-surface expression of VCAM1 (red lines) on BM ECs in steady state as well as LPS-injected control Myd88^fl/fl and Tie2-Cre;Myd88^fl/fl mice. Isotype control shown as blue line. (F) Graphical scheme showing experimental outline to assess G-CSF effects in vivo. (G) Representative FACS profile showing characteristic G-CSF–induced changes in BM CD11b⁺Gr1^high mature and BM CD11b⁺Gr1^low immature neutrophils in control Myd88^fl/fl and Tie2-Cre;Myd88^fl/fl mice. (H) Frequencies of BM CD11b⁺Gr1^high mature and (I) BM CD11b⁺Gr1^low immature neutrophils in PBS- and G-CSF–injected control Myd88^fl/fl and Tie2-Cre;Myd88^fl/fl mice. All data represent mean ± standard deviation from 2 or 3 independent experiments. Two-tailed Student t tests were used to assess statistical significance (**P < .01, ***P < .001).

Figure 4

Endothelial cells are effectively targeted in Tie2-Cre-loxP-GFP reporter mice and are the main source of G-csf after in vivo LPS stimulation. (A) Graphical scheme depicting experimental outline to induce LPS-induced emergency granulopoiesis and to assess G-csf expression in sorted CD45⁻Ter119⁻GFP⁺CD31⁺ ECs and CD45⁻Ter119⁻GFP⁺CD31⁻ non-ECs isolated from different organs of Tie2-Cre-loxP-GFP reporter mice. (B) Representative FACS profile of GFP and CD31 expression within nonhematopoietic cells (CD45⁻Ter119⁻) isolated from BM, heart, liver, kidney, spleen, and lung of Tie2-Cre-loxP-GFP reporter mice. (C) Percentages of CD31⁺ (black bars) and CD31⁻ (white bars) cells within the CD45⁻Ter119⁻GFP⁺ cell population in BM, heart, liver, kidney, spleen, and lung of Tie2-Cre-loxP-GFP reporter mice. (D) Percentages of CD31⁺ (black bars) and CD31⁻ (white bars) cells within the CD45⁻Ter119⁻GFP⁻ cell population in BM, heart, liver, kidney, spleen, and lung of Tie2-Cre-loxP-GFP reporter mice. (E) Comparative G-csf transcript levels normalized to Actb in sorted CD45⁻Ter119⁻GFP⁺CD31⁺ ECs (red bars) vs CD45⁻Ter119⁻GFP⁺CD31⁻ non-ECs (white bars) isolated from BM, heart, liver, kidney, spleen, and lung of LPS-injected Tie2-Cre-loxP-GFP reporter mice. All data represent mean ± standard deviation from 2 independent experiments (n = 5 mice). Two-tailed Student t tests were used to assess statistical significance (*P < .05, **P < .01, ***P < .001).

Figure 5

Endothelial cell–intrinsic MYD88 signaling is required to stimulate accelerated neutrophil production, myeloid progenitor lineage skewing toward granulocyte-macrophage progenitors (GMPs), and increased CFU-G activity in vivo. (A) Graphical scheme depicting experimental outline to induce LPS-induced emergency granulopoiesis and to assess BrdU incorporation. (B) Representative FACS profile showing BrdU incorporation in in BM CD11b⁺Gr1^high mature and BM CD11b⁺Gr1^low immature neutrophils in control Myd88^fl/fl, and Tie2-Cre;Myd88^fl/fl mice during steady-state and LPS-induced emergency granulopoiesis. (C) Frequencies of BrdU⁺ BM CD11b⁺Gr1^high mature and (D) BrdU⁺ BM CD11b⁺Gr1^low immature neutrophils in PBS- or LPS-injected Myd88^fl/fl and Tie2-Cre;Myd88^fl/fl mice. (E) Representative FACS profile showing myeloerythroid progenitors in control Myd88^fl/fl, and Tie2-Cre;Myd88^fl/fl mice during steady-state and LPS-induced emergency granulopoiesis. (F) Frequencies of Lin⁻cKit⁺Sca1⁻FcgR⁺CD34⁺ GMPs in PBS- or LPS-injected Myd88^fl/fl and Tie2-Cre;Myd88^fl/fl mice. (G) Absolute CFU numbers per 1 hind leg in control Myd88^fl/fl and Tie2-Cre;Myd88^fl/fl mice during steady-state and LPS-induced emergency granulopoiesis (CFU-G, CFU granulocyte; CFU-M, CFU macrophage; CFU-GM, CFU granulocyte/macrophage; CFU-GEMM, CFU granulocyte/erythrocyte/macrophage/megakaryocyte; BFU-E, burst-forming unit erythrocyte). Black squares, PBS-injected mice; red squares, LPS-injected mice. Data from 2 independent experiments are shown. Two-tailed Student t tests were used to assess statistical significance (*P < .05, ***P < .001).

Figure 6

Endothelial cell–intrinsic MYD88 signaling is required to efficiently stimulate emergency granulopoiesis during systemic E coli infection. (A) Frequency and absolute number of PB CD11b⁺ cells, (B) BM cellularity, (C) frequencies and absolute numbers of BM CD11b⁺Gr1^high mature and (D) BM CD11b⁺Gr1^low immature neutrophils, and (E) plasma G-CSF levels in Myd88^fl/fl mice and Tie2-Cre;Myd88^fl/fl mice in steady-state (black squares) and after infection with 4.5 × 10⁸ E coli CFU (red squares) given IP. Data from 2 independent experiments are shown. rel., relative; abs., absolute. Two-tailed Student t test was used to assess statistical significance (**P < .001, ***P < .001).

Figure 7

Model for pathogen sensing and subsequent translation into emergency granulopoiesis. (1) Gram-negative bacteria and/or their structural components that have gained access to the systemic circulation are recognized by TLR4-expressing endothelial cells (2), thereby indicating an emergency state. Upon TLR4/MYD88 signaling in endothelial cells, G-CSF is released in large quantities (3). In the bone marrow, endothelial cell–derived G-CSF acts on myeloid precursors expressing the G-CSF receptor resulting in enhanced generation, accelerated turnover, and increased neutrophil release from the bone marrow to the systemic circulation (4). These neutrophils are recruited to the site of infection (5) where they participate in clearing the pathogen (6).