The lta4h locus modulates susceptibility to mycobacterial infection in zebrafish and humans

Abstract

Exposure to Mycobacterium tuberculosis produces varied early outcomes, ranging from resistance to infection to progressive disease. Here we report results from a forward genetic screen in zebrafish larvae that identify multiple mutant classes with distinct patterns of innate susceptibility to Mycobacterium marinum. A hypersusceptible mutant maps to the lta4h locus encoding leukotriene A(4) hydrolase, which catalyzes the final step in the synthesis of leukotriene B(4) (LTB(4)), a potent chemoattractant and proinflammatory eicosanoid. lta4h mutations confer hypersusceptibility independent of LTB(4) reduction, by redirecting eicosanoid substrates to anti-inflammatory lipoxins. The resultant anti-inflammatory state permits increased mycobacterial proliferation by limiting production of tumor necrosis factor. In humans, we find that protection from both tuberculosis and multibacillary leprosy is associated with heterozygosity for LTA4H polymorphisms that have previously been correlated with differential LTB(4) production. Our results suggest conserved roles for balanced eicosanoid production in vertebrate resistance to mycobacterial infection.

Figure 1. A Forward Genetic Screen Identifies Zebrafish Mutants with Altered Susceptibility to Mm Infection

(A) Diagram of zebrafish larva anatomy showing injection sites used in this study. (B) Schematic diagram of forward genetic screen showing derivation of F₂ gynogenetic diploid embryos infected at two dpf. Potential mutants were confirmed by backcrossing the corresponding F₁ female and recovering the observed mutation in the F₃ generation. (C–H) Fluorescence images of WT and mutant sibling fish at five dpi with equivalent bacterial inocula. Arrows, granulomas; white arrowheads, individual infected macrophages; yellow arrowheads, pairs of highly infected macrophages out of the focus plane that have not formed granulomas. Scale bars, 500 µm. (C) WT and (D) aggregation mutant fh212 sibling. (E) WT and (F) resistant mutant fh199 sibling. (G) WT and (H) hypersusceptible mutant fh112 sibling. (I–J) Fluorescence and Differential Interference Contrast (DIC) overlay of six dpi WT fish with granuloma (arrow) (I) and aggregation mutant fh141 sibling (J) with highly infected macrophages (arrowheads) that have not aggregated. Scale bars, 50 µm.

Figure 2. Mutations in the leukotriene A₄ hydrolase Gene Result in Decreased lta4h mRNA Expression and Increased Susceptibility to Mycobacterial Infection

(A) Fluorescence images of infection foci at five dpi with initial dose of 115 bacteria showing well-formed granuloma in WT fish (top) and profuse bacterial clusters showing cording morphology in hypersusceptible mutant fh112 sibling (bottom). Scale bars, 25 µm. (B) Proportion of fish with cording morphology at five dpi with 150–250 bacteria in five independent clutches of 50–100 animals each of fh112/+ heterozygote or WT crosses and three independent clutches of 20–25 animals each from fh112/+ x zm5961/+ heterozygote crosses. Dotted line represents the theoretical maximum for a completely penetrant recessive mutation. Error bars, SD. ***, p<0.001 (one-way ANOVA with Tukey’s post-test). (C) Genetic map of fh112 placing it between the polymorphic markers z10062 and a SNP in a PRKWNK1-related gene on zebrafish chromosome 4. Parentheses indicate the number of recombination events as a ratio of the total number of informative meioses scored. (D) Mean lta4h RNA levels measured by qRT-PCR from 3–5 biological replicates of 30 fish. Infected fh112 mRNA assessed at four dpi, infected and uninfected zm5961 RNA at three dpf (one dpi for infected zm5961 fish). Fold difference expressed relative to matched WT uninfected controls. **, P<0.01; ***, P<0.001 (ANOVA with Tukey’s post-test). Error bars, SD (E) Gene structure of zebrafish lta4h with location of the zm5961 insertion and key conserved residues in the catalytic domain indicated (Haeggstrom, 2004). (F) Fluorescent in situ hybridization (FISH) of lta4h (red) combined with mpo antibody staining for neutrophils (blue) in caudal hematopoietic tissue of two dpf uninfected fish. Yellow arrowhead, lta4h–staining, mpo-negative presumed macrophage. White arrowhead, dual staining neutrophil. Scale bar, 100 µm. Inset scale bar, 25 µm. (G) lta4h FISH in uninfected fh112/+ heterozygote (left) and zm5961/fh112 non-complementing sibling (right) at two dpf. Arrowheads point to brightly staining cells in fh112/+ heterozygote and to weakly-expressing cells in non-complementer. Scale bar, 100 µm. See also Figure S1.

Figure 3. lta4h MO Knockdown Results in Increased Susceptibility to Mm

(A) Representative fluorescence images of control (top) and lta4h morphant (bottom) fish five dpi with 173 bacteria. Scale bar, 200 µm. (B) Mean bacterial loads per embryo for control (wt) and lta4h morphant embryos at 1.5 and six dpi with 150 bacteria (n=4 groups of 5 animals for each timepoint). Error bars, SD. (C) Survival of control and lta4h morphant fish mock-injected or injected with 177 bacteria (n=25 per group). Hazard Ratio for death of infected morphants = 9.0, P<0.0001 (Kaplan Meier method with log-rank [Mantel-Cox] test). Data representative of three independent experiments. (D) Fluorescence image of non-cording bacteria within a granuloma in control animals (top) and cording bacteria in lta4h morphant (bottom) at five dpi with infection dose of 150 bacteria. Scale bars, 20 µm. (E) Mean proportion of animals with cording in three independent groups of 15–40 animals four dpi with 173 bacteria. P=0.003 (Student’s unpaired t-test). Error bars, SEM. (F) Bacterial burden at five dpi as determined by fluorescence pixel counts (FPC) for vehicle-treated or 100 µM bestatin-treated animals after injection with ~150 CFU Mm. P=0.03 (Student’s unpaired t-test). (G) Fluorescence image of non-cording bacteria within a granuloma in control animals (top) and cording bacteria in animals treated with 100 µM bestatin (bottom) at five dpi as described in (F). Scale bars, 10 µm. (H) Quantitation of cording at four dpi in five independent groups of 10–30 animals treated with vehicle or 100 µM bestatin after injection with 100–200 CFU Mm. P=0.007 (Student’s unpaired t-test). Error bars, SEM.

Figure 4. lta4h Deficiency Compromises TNF Induction and Phenotypically Resembles TNF Signaling Deficiency

(A) DIC image of HBV of 30 hpf-embryo six hours post-injection with vehicle (left) or 1.5×10⁻¹⁴ mol LTB₄ (right). Arrows, macrophages. (B) Fluorescence and DIC overlay image of HBV of 30 hpf mutant fh112 embryo (obtained from incross of fh112/+ heterozygotes) six hours post-injection with GFP-expressing Mm. Arrows, infected macrophages. After scoring for macrophage recruitment, genotype of embryos was identified based on cording phenotype at four dpi. Scale bar, 20 µm. (C) lta4h FISH showing staining of an early-stage granuloma two dpi (left); dual FISH showing tnf expression in macrophages of four dpi granuloma macrophages by demonstrating co-localization of the macrophage marker fms (green) and tnf (red). Scale bar, 20 µm, (D) Relative TNF mRNA levels (relative to a β-actin standard) assessed by qRT-PCR at one dpi following infection of 30 WT or zm5961 larvae with 93 WT bacteria. Error bars, SEM. Representative of two independent experiments. (E) Representative fluorescence images of control (top) or lta4h morphant (bottom) animals three dpi with 130 erp mutant bacteria. (n=10 for each condition.) Scale bar 20 µm. (F) Quantification of bacterial burdens by fluorescence pixel counts (FPC) of all animals from (D) at six dpi. (G) Proportion of single macrophages containing >10 bacteria in 15 control and 12 lta4h morphant animals at three dpi with 130 erp mutant bacteria. Error bars, SD. (H–J) Serial assessment of granuloma formation in control (dark bars) and lta4h morphants (light bars) injected with 230 bacteria (n=15 each) by fluorescence and DIC microscopy for four dpi. (H) Percentage of animals with at least one granuloma over time. lta4h morphants form granulomas earlier; they are significantly more likely to have at least one granuloma than WT at two dpi (P=0.042; Fisher’s exact test). (I) average number of granulomas per fish over time. P<0.001 when comparing both time post-infection and morphant status by 2-way ANOVA. Error bars, SD (J) morphometric analysis of control MO (wt) and lta4h MO granuloma size at four dpi, error bars SEM. (K) Neutral red labeling of granuloma macrophages in WT and zm5961 granulomas. Arrowheads indicate examples of neutral red positive macrophages. Fish were infected with approximately 250 bacteria then neutral red stained at four dpi, six dpf. Scale bars, 10 µm. (L) Numbers of neutral red stained cells in the tails of infected WT and zm5961 fish infected as described in (K). P=0.0003 (Student’s unpaired t-test). (M) quantitation of neutral red stained cells in the tails of uninfected WT and zm5961 fish at 6 dpf. (N) Bacterial burdens (FPC) at four dpi with 75 bacteria in control or TR1 morphants in the background of either WT or zm5961 animals. * P<0.05; ** P<0.01; ***P<0.001. (one-way ANOVA with Tukey’s post test; all other comparisons not significant). See also Figures S2 and S3.

Figure 5. lta4h Deficiency Induces an Immunoregulatory Phenotype Recapitulated by Exogenous LXA₄

(A–B) tnf mRNA expression in control and lta4h morphant fish infected with Mm, injected with LTB₄, or both. Each bar represents a mean of 3 independent pools of 20–30 animals. Error bars, SD. (A) tnf mRNA levels in control and morphant fish injected with 110 bacteria and at one dpi mock-injected or injected with 1.5×10⁻¹⁴ mol LTB₄ into the caudal vein 2.5 hours prior to RNA extraction. *, P<0.05; **, P<0.01 (one-way ANOVA. All other comparisons not significant). (B) tnf mRNA levels in uninfected three dpf control or lta4h morphant animals 2.5 hours after injection of 1.5×10⁻¹⁴ mol LTB₄ into the caudal vein. (C) diagram of LTB₄ and LXA₄ biosynthetic pathways. Reduction of LTA4H activity (red) is hypothesized to result in increased synthesis of LXA₄ through pathway marked by green arrow. (D) iNOS antibody staining in infected WT and zm5961 animals. Green represents GFP-expressing Mm and red represents iNOS staining. Animals were infected with approximately 100–150 bacteria and fixed for antibody staining and imaging at 3 dpi. Scale bars, 25 µm. (E) Numbers of iNOS positive cells in WT and zm5961 animals as described in (D) for both total number of iNOS positive cells in the tail (P=0.008) and iNOS positive cells within granulomas (P=0.01). (F) Light microscopy images of Sudan black stained-neutrophils in right ear of three dpf animals injected with vehicle (left and middle) or 3.5×10⁻¹⁴ mol of LXA₄ (right) ten hours prior and with vehicle (left) or 1.5×10⁻¹⁴ mol LTB₄ (middle and right) into the right ear four hours prior. Scale bar, 50 µm. (G) Mean number of neutrophils in ears of animals in (F). *, P<0.05; ***, p<0.001 (Kruskal-Wallis non-parametric one-way ANOVA with Dunn’s post test; other comparisons not significant). (H) Mean number of neutrophils in ears of three dpf controls, lta4h morphant and zm5961 mutant animals either infected (135 CFU) or mock-infected one day prior then injected into right ear with 1.5×10⁻¹⁴ mol LTB₄ and scored four hours later. *, P<0.05; **, P<0.01; ***, P<0.001 (One-way ANOVA with Tukey’s post test; all other pairwise comparisons not significant). (I) Mean number of neutrophils recruited to ears of three dpf uninfected animals injected into right ear with 5×10⁻¹⁴ mol LTB₄ after overnight exposure to vehicle or 1 µM PD146176. **, P<0.01; ***, P<0.001 (One-way ANOVA with Tukey’s post test). (J) mean tnf levels relative to uninfected controls in animals (25–30 per group) eight hours after single dose of 3.5×10⁻¹⁴ mol LXA₄ or vehicle injected into caudal vein three dpi with either mock-infection or with 158 bacteria. Error bars, SD. Representative of three independent experiments. (K) Mean bacterial burdens (FPC) in five dpi WT fish infected at two dpf with 212 CFU and, beginning one dpi, given injections of 3.5×10⁻¹⁴ mol of LXA₄ or vehicle every 12 hours into the caudal vein for four days (n=18 per group). P=0.0283 (Unpaired t test with Welch’s correction to account for unequal variances). (L) Examples from (K) of non-cording bacteria within granuloma in vehicle-injected fish and cording bacteria in LXA₄-injected fish at five dpi. Scale bar, 30 µm. (M) Quantitation of cording from (K,L) at five dpi. P=0.018 by Fisher’s exact test of a contingency table. See also Figure S4.

Figure 6. LTA4H and Susceptibility to Human Mycobacterial Diseases

(A) LTA4H exons are shown as purple rectangles. Below the LTA4H gene are indicated the allele at each SNP associated with higher levels of LTB₄ after ionomycin stimulation of granulocytes (Helgadottir et al., 2006) and the frequencies of these alleles among Vietnamese controls. SNPs rs1978331 and rs2660898, for which significant associations were found in this study, are in bold type. (B) Linkage disequilibrium between SNPs, based on D-prime (D’) and R-squared (R²) values, were calculated for Vietnamese controls and are shown as triangles. The minor allele frequency is shown adjacent to each corresponding SNP. (C) Mortality from meningeal TB for patients heterozygous for both LTA4H SNPs rs1978331 and rs2660898 (N=53, red curve) and for patients homozygous at one or both SNPs (N=156, blue curve). Difference between the curves is significant at P=0.025.

Figure 7. Model of lta4h Effects on Eicosanoid Balance

Variations in LTA4H levels or activity, represented by the rheostat symbol, influence the balance between pro-inflammatory LTB₄ and anti-inflammatory LXA₄. High levels of LXA₄ inhibit TNF production, resulting in exuberant intracellular bacterial growth, increased granuloma formation, increased macrophage death and the cording morphology typical of extracellular mycobacteria. Although WT levels of TNF are protective during infection, excessive levels of TNF may produce increased immunopathology, resulting in worse outcome from mycobacterial infections.