Mycobacteria manipulate macrophage recruitment through coordinated use of membrane lipids

Abstract

The evolutionary survival of Mycobacterium tuberculosis, the cause of human tuberculosis, depends on its ability to invade the host, replicate, and transmit infection. At its initial peripheral infection site in the distal lung airways, M. tuberculosis infects macrophages, which transport it to deeper tissues. How mycobacteria survive in these broadly microbicidal cells is an important question. Here we show in mice and zebrafish that M. tuberculosis, and its close pathogenic relative Mycobacterium marinum, preferentially recruit and infect permissive macrophages while evading microbicidal ones. This immune evasion is accomplished by using cell-surface-associated phthiocerol dimycoceroserate (PDIM) lipids to mask underlying pathogen-associated molecular patterns (PAMPs). In the absence of PDIM, these PAMPs signal a Toll-like receptor (TLR)-dependent recruitment of macrophages that produce microbicidal reactive nitrogen species. Concordantly, the related phenolic glycolipids (PGLs) promote the recruitment of permissive macrophages through a host chemokine receptor 2 (CCR2)-mediated pathway. Thus, we have identified coordinated roles for PDIM, known to be essential for mycobacterial virulence, and PGL, which (along with CCR2) is known to be associated with human tuberculosis. Our findings also suggest an explanation for the longstanding observation that M. tuberculosis initiates infection in the relatively sterile environment of the lower respiratory tract, rather than in the upper respiratory tract, where resident microflora and inhaled environmental microbes may continually recruit microbicidal macrophages through TLR-dependent signalling.

Figure 1. PDIM mediated evasion of MyD88 dependent macrophage recruitment

a, Schematic of a 2dpf zebrafish showing the hindbrain ventricle (HBV) injection site outlined with dashed white line. Scale bar = 500 μm. b-c, Mean macrophage recruitment at 3 HPI into the HBV of wild-type or MyD88-morphant (MO) fish following infection with 150 S. aureus, 200 P. aeruginosa (b), 80 M. marinum or 85 M. smegmatis (c). Representative of three separate experiments. d, Mean bacterial burdens at 3 DPI following HBV infection of wild-type fish with 80 wild-type, ΔmmpL7, or Δerp M. marinum. Representative of three separate experiments. e, Mean macrophage recruitment at 3 HPI into the HBV of wild-type or MyD88 MO fish following infection with ΔmmpL7 or Δerp M. marinum. Representative of four separate experiments. f, Mean bacterial burdens of wild-type or MyD88 MO fish at 3 DPI following HBV infection with wild-type or ΔmmpL7 M. marinum. Representative of three separate experiments. Significance testing for all panels done using one-way ANOVA, with Bonferroni's post-test for comparisons shown, *P < 0.05; ***P < 0.001.

Figure 2. Increased iNOS-dependent microbicidal activity of macrophages recruited to PDIM-deficient mycobacteria

a, b, Representative images of wild-type (a) and ΔmmpL7 (b) M. marinum-infected fish from (c). N=13 (wild-type) and 14 (ΔmmpL7) larvae per group. Scale bar = 50μm. c, Percent of infected macrophages that were iNOS-positive, in the HBV at 3 DPI with 80 wild-type, ΔmmpL7, or Δerp M. marinum. Representative of three separate experiments. d, Mean bacterial burdens of 2dpf control (CTRL) or RNS scavenger (CPTIO) treated fish following HBV infection with 80 wild-type or ΔmmpL7 M. marinum. Representative of two separate experiments. e-h, Mean bacterial volume of red fluorescent wild-type M. marinum (infection inoculum 30-40) when co-infected with 30-40 green fluorescent wild-type, ΔmmpL7, or Δerp M. marinum at 3 DPI in wild-type (e), pu1 MO (f), myd88 MO (g), or CPTIO-treated (h) larvae. e, g, Co-infection of wild-type Mm and ΔmmpL7 M. marinum in wild-type or myd88 MO fish is representative of at least three separate experiments, and co-infection with Δerp is representative of two separate experiments. f, h, Representative of two separate experiments. a-h, Significance testing done using one-way ANOVA, with Bonferroni's post-test for comparisons shown, *P < 0.05; **P < 0.01; ***P < 0.001. i, Mean bacterial volume of red fluorescent wild-type M. marinum at 3 DPI (infecting inoculum 30-40) when co-infected with the volume equivalent of 30-40 heat-killed, crushed wild-type M. marinum. Representative of two separate experiments. Student's unpaired t test.

Figure 3. Elevated frequencies of iNOS expressing inflammatory monocytes in mice infected with PDIM-deficient M. tuberculosis

C57B6 mice were infected via the aerosol route with H37Rv or an isogenic PDIM-deficient mutant (ΔdrrA). Lung tissue was harvested on 21 DPI and iNOS (protein expression was measured via flow cytometry. Representative FACS plots (a) and graphical depiction (b) of frequencies of iNOS expressing cells within the Ly6C⁺CD11b⁺ inflammatory monocyte population. Representative of two separate experiments. Student's unpaired t test.

Figure 4. Macrophage recruitment and subsequent infectivity is mediated by mycobacterial PGL and host CCR2

a, Mean macrophage recruitment at 3 HPI into the HBV of wild-type or CCR2 MO fish following infection with 80 wild-type or ΔmmpL7 M. marinum. Representative of three independent experiments. One-way ANOVA, with Bonferroni's post-test for comparisons shown, *P < 0.05 b, CCL2 mRNA levels (mean +/- SEM of four biological replicates) induced at 3 hours post caudal vein infection of 2 DPF larvae with 250-300 wild-type, Δpks15, or ΔmmpL7 M. marinum. One-way ANOVA with Tukey's post-test, *P < 0.05 c, Mean macrophage recruitment at 3 HPI into the HBV following infection with 80 wild-type or Δpks15 M. marinum. Representative of three separate experiments. One-way ANOVA with Bonferroni's post-test for comparisons shown, **P < 0.01; ***P < 0.001. d, Wild-type and CCR2 MO fish, with or without the addition of μg/mL CCL2, were infected in the HBV with 1-3 wild-type or Δpks15 M. marinum. Graph shows the percent of fish that were infected (black) or uninfected (gray) after five days. n=number of larvae per group. Representative of two separate experiments. Significance was evaluated using Fisher's exact test for each comparison, **P < 0.01; ***P < 0.001.

Figure 5. MyD88-dependent macrophage recruitment elicited by other bacterial pathogens and commensals attenuates pathogenic mycobacteria

a, Mean bacterial volume of red fluorescent M. marinum (infecting inoculum 30-40) following co-infection with either 30-40 green fluorescent M. marinum or 300 P. aeruginosa at 1 and 3 DPI. Representative of three separate experiments. Significance assessed using Student's t test. b-c, Mean bacterial volume of 30-40 red fluorescent wild-type M. marinum (infecting inoculum 30-40) following co-infection with either 30-40 green fluorescent wild-type M. marinum or 300 P. aeruginosa (b) or 300 S. aureus (c) at 3 DPI, in wild-type or MyD88 MO larvae. Significance tested by one-way ANOVA with Bonferroni's post–test for comparisons shown, *P < 0.05.

Extended Data Fig. 1

Coordinate use of PDIM-mediated immune evasion and PGL-mediated recruitment by pathogenic mycobacteria. Model for infection with wild-type (WT) and PDIM-deficient mycobacteria are shown in the context of the relatively sterile lower airway versus the upper airway, with its higher levels of resident microflora and inhaled environmental organisms.

Extended Data Fig. 2

ΔmmpL7 bacteria are attenuated in zebrafish larvae. a, Kaplan-Meier graph showing daily survival of larvae infected via caudal vein injection with medium (mock), 29 wild-type or 70 ΔmmpL7 M. marinum. N=25 (mock), 31 (wild-type), or 29 (ΔmmpL7) larvae per group. Mean time to Death (days): Mock (11), wild-type (7.6) and ΔmmpL7 (11.2). Survival was compared by log-rank test: wild-type vs. mock and wild-type vs. ΔmmpL7, p<0.0001; mock vs. ΔmmpL7, p=0.5601. b, c, Larvae were infected via caudal vein injection 1 DPF with 550 wild-type, 650 ΔmmpL7, or 700Δerp, fluorescent M. marinum. b, Infection burdens were measured by Fluorescent Pixel Count (FPC, mean +/- SEM). c, Representative images at 7 DPI. N= 29 (wild-type and ΔmmpL7) or 30 (Δerp) larvae per group. Scale bar = 500 μm. At 3, 5 and 7 DPI, Log₁₀ FPC was compared by ANOVA, with Dunnett's post-test. ***, p<0.001. d,e, Representative images from wild-type (d) and ΔmmpL7 (e) M. marinum HBV infections quantified in Figure 1d. N=18 (wild-type) or 16 (ΔmmpL7) larvae per group. HBVs are outlined with a dashed white line. Scale bar = 100μm.

Extended Data Fig. 3

Knockdown of MyD88 results in a late, dose dependent hypersusceptibility to M. marinum systemic infection. a, RT-PCR for actin (upper panel) and myd88 (lower panel) demonstrating that that the majority of myd88 transcripts at 7 DPF are abnormal in MyD88 morphants. Lanes marked ‘b’ and ‘c’ correspond to morphants from the same experiments depicted in panels b and c respectively. The abnormal larger transcript (indicated by *) results from the inclusion of intron 2 in the final transcript, incorporating a premature stop codon that truncates the protein prior to the TIR (Toll/Interleukin Receptor) domain. (b, c) Caudal vein infection of MyD88 morphants with 141 (b) or 325 (c) CFU M. marinum/larva. Bacterial burden was assessed by FPC, values plotted represent the mean +/- SEM. Time points were compared by one-way ANOVA and Bonferroni's post-tests, *** p<0.001. d, Representative images of larvae at 5 DPI from experiment in (c), N=30 control, 15 Myd88 MO.

Extended Data Fig. 4

Characteristics of macrophages recruited to wild-type and PDIM-deficient bacteria. a,. Mean mpeg1 positive macrophages recruited at 3 HPI into the HBV of wild-type fish following infection with 80 wild-type or ΔmmpL7 M. marinum. b, Data from Figure 2C expressed as mean numbers of total infected macrophages and iNOS expressing infected macrophages following HBV infection with 80 wild-type, ΔmmpL7, or Δerp M. marinum. c, bacterial burdens after L-NAME treatment. Mean bacterial burdens of 2 DPF control (CTRL) or iNOS inhibitor (L-NAME) treated fish following HBV infection with 80 wild-type or ΔmmpL7 M. marinum.

Extended Data Fig. 5

Wild-type bacterial burdens after co-infection with wild-type or ΔmmpL7 bacteria. Representative images from the HBV co-infections quantified in Figure 2e. a-b, Red fluorescent wild-type (WT) M. marinum co-infected with green fluorescent wild-type (a) or ΔmmpL7 (b) M. marinum. N=18 (wild-type) and 19 (ΔmmpL7) larvae per group. Scale bar = 50 μm. c, Wild-type bacterial burdens after co-infection with wild-type or ΔmmpL7 M. marinum with and without L-NAME treatment. Significance tested by one-way ANOVA with Bonferroni's post–test for comparisons shown.

Extended Data Fig. 6

MyD88-dependent macrophage recruitment occurs in response to PDIM deficiency rather than due to a loss of another MmpL7 exported product. a, Mean macrophage recruitment at 3 HPI into the HBV of wild-type or MyD88 MO larvae following infection with 80 Δmas M. marinum. Student's unpaired t test. b, Mean surviving bacterial volume of red fluorescent wild-type M. marinum (initial infection dose of 30-40 CFU) when co-infected with 30-40 green fluorescent wild-type, ΔmmpL7 or Δmas M. marinum at 3 DPI. Representative of two separate experiments. Significance tested by one-way ANOVA with Tukey's post-test.

Extended Data Fig. 7

Gating Strategy and Isotype Controls for iNOS Staining of Mouse Lung. a, Representative gating strategy for isolation of inflammatory monocytes. A dump channel containing anti-CD4, CD8, and CD11c was plotted against a channel exhibiting autofluorescence and also containing anti-Ly6G. Using these markers, T cell, dendritic cell, alveolar macrophage, and neutrophil cell populations were excluded from the double negative gate. Inflammatory monocytes were identified within the double negative population by their co-expression of Ly6C and CD11b. These cells were then evaluated for intracellular iNOS expression, N=4 per group (Figure 3a,b) or b, with isotype control antibodies, N=4 per group.

Extended Data Fig. 8

Specificity of CCL2-mediated macrophage recruitment in wild-type and CCR2 morphant larvae a, Mean macrophage recruitment at 3 HPI into the HBV of control (ctrl), or CCR2 MO (ccr2) larvae following injection of vehicle control (“mock”; 0.1% BSA in PBS), human CCL2 (hCCL2), human CCL4 (hCCL4), or human CCL5 (hCCL5) b-c, Mean macrophage (b) and neutrophil (c) recruitment at 3HPI into the HBV of control (ctrl), CCR2 MO (ccr2), or MyD88 MO (myd88) larvae following injection of vehicle control (mock), murine CCL2 (mCCL2), human CCL2 (hCCL2), human interleukin-8 (hIL-8), or human leukotriene B4 (hLTB₄). Representative of three separate experiments. Significance assessed by one-way ANOVA with Bonferroni's post-test for the comparisons shown, *P < 0.05; ***P < 0.001.

Extended Data Fig. 9

Identification of zebrafish CCL2 orthologue. a, mRNA levels of potential CCL2 orthologues (mean +/- SEM of four biological replicates) induced at 3 hours post caudal vein infection of 2dpf larvae with 250-300 wild-type M. marinum. These assays were performed on the same cDNA pools as the data presented in Figure 4b. b, Mean macrophage recruitment at 3 HPI into the HBV of wild-type or CCL2 MO fish following infection with 80 M. marinum. Representative of two separate experiments.

Extended Data Fig. 10

Infectivity Assay. a-b, Representative 5 HPI images from Figure 4d following HBV infection with (a) one or (b) three M. marinum. Scale bar = 100 μm. Ns for fish represented in (a) and (b) (i.e. those found to be infected with one to three bacteria) are presented in Figure 4d (18, 22, 28, 28, 28, 28, 22, 22 for the respective conditions as specified in the figure). c, Mean bacterial burdens 5 HPI HBV infection with 1-3 wild-type (WT), ΔmmpL7, or Δpks15 M. marinum.