Shigella-Induced Emergency Granulopoiesis Protects Zebrafish Larvae from Secondary Infection

A Shigella-zebrafish infection model to study emergency granulopoiesis.As in human development, zebrafish HSCs emerge from the aorta gonad mesonephros (AGM) (18). Zebrafish HSCs are generated from 30 h post-fertilization (hpf) and migrate to the caudal hematopoietic tissue (CHT), where they begin leukocyte production. To study neutrophil production and HSC activity in response to bacterial infection, we developed a Shigella-zebrafish infection model. Larvae were injected in the hindbrain ventricle (HBV) at 2 days post-fertilization (dpf) with a low dose (0.5 × 10³ to 2.0 × 10³ CFU) of S. flexneri M90T. Survival and bacterial burden were monitored over 48 h post-infection (hpi). In this case, infection has no effect on larval mortality (~100% survival) and neutrophils successfully control Shigella infection, with the majority of bacteria cleared by 24 hpi (see Fig. S1A and B in the supplemental material). To monitor the interplay between Shigella and neutrophils, we performed HBV infections of green fluorescent protein-expressing (GFP⁺) S. flexneri in Tg(lyz::dsRed)^nz50 (herein lyz::dsRed) transgenic zebrafish larvae, in which neutrophils express dsRed fluorescent protein. Imaging of larvae by fluorescent stereomicroscopy showed neutrophil recruitment to the infection site from 1 hpi and engulfment of Shigella alongside an obvious and rapid reduction in GFP⁺ Shigella fluorescence (Fig. 1A; see Movie S1 in the supplemental material). Quantifications performed at the whole-animal level show a slight reduction in neutrophil numbers 24 hpi, and local neutrophil depletion by Shigella at infection sites can be observed in real time (Fig. S1C). These data are consistent with studies demonstrating that neutrophils are essential to control Shigella infection in vivo (13) and suggest that a successful host response to Shigella infection requires emergency granulopoiesis.

• Open in new tab
• Download powerpoint

FIG 1 

Development of a Shigella-zebrafish infection model to study emergency granulopoiesis. (A to C) At 2 dpf, lyz::dsRed zebrafish larvae with red neutrophils were injected in the HBV with PBS or a low dose (0.5 × 10³ to 2.0 × 10³ CFU) of GFP⁺ S. flexneri M90T and imaged by fluorescence stereomicroscopy. (A) Representative images of the infection site in a single larva over time (Movie S1). Scale bars, 100 µm. (B) Schematic of the zebrafish larva, highlighting the AGM region of neutrophil and HSC quantifications. Shown are representative images of the AGM of PBS-injected or Shigella-infected larvae at 48 hpi. Scale bars, 100 µm. (C) Quantifications of neutrophils from PBS-injected (open circles) or Shigella-infected (closed circles) larvae as in panel B. Circles represent counts from individual larvae. Data were pooled from 4 independent experiments using n ≥ 10 larvae per condition per experiment. Means ± SEM are shown (horizontal bars). P values between conditions at cognate time points were determined by unpaired two-tailed Student’s t test. Significance was defined as P < 0.05. ***, P < 0.001. (D) Tg(runx::mCherry)/Tg(mpx::GFP) larvae with red HSPCs and green neutrophils were injected at 2 dpf in the HBV with a low dose of S. flexneri M90T, and images of the AGM were captured by confocal microscopy (100× objective). Dashed arrows represent vasculature. DA, dorsal aorta; CV, cardinal vein. Cells expressing high levels of mCherry (*) were considered HSCs; cells expressing lower levels of mCherry were considered HPCs (i.e., HSC progeny [arrowheads]). Maximum-intensity z-projection images are shown. Scale bars, 20 µm. (E) Tg(runx::GFP) larvae with green HSCs were injected with a low dose of mCherry⁺ S. flexneri M90T as in panel D. HSCs in the AGM of PBS-injected (open circles) or Shigella-infected (closed circles) larvae were quantified at 48 hpi. Circles represent counts from individual larvae. Data were pooled from 3 independent experiments using n ≥ 6 larvae per condition per experiment. Means ± SEM are shown (horizontal bars). P values between conditions were determined by unpaired two-tailed Student’s t test. Significance was defined as P < 0.05. **, P < 0.01. (F) lyz::dsRed larvae were treated with control (Ctrl) or Gcsfr morpholino oligonucleotide (Mo). Morphants were injected in the HBV at 2 dpf with PBS (open circles) or a low dose of GFP⁺ S. flexneri M90T (closed circles). Larvae were imaged by fluorescent stereomicroscopy, and neutrophils in the AGM were quantified at 48 h following treatment. Each circle represents a count from an individual larva. Means ± SEM (horizontal bars) are shown. Data were pooled from three independent experiments using n ≥ 6 larvae per condition per experiment. P values between conditions were determined by unpaired two-tailed Student’s t test. Significance was defined as P < 0.05. ***, P < 0.001. (G) lyz::dsRed larvae were treated with control or Irf8 morpholino oligonucleotide. Morphants were injected in the HBV at 2 dpf with a low dose of GFP⁺ S. flexneri M90T. Neutrophil quantifications from the AGM of infected larvae are shown. Circles represent counts from individual larvae. Data were pooled from 3 independent experiments using n ≥ 4 larvae per condition per experiment. Means ± SEM are shown (horizontal bars). P values between conditions were determined by unpaired two-tailed Student’s t test. Significance was defined as P < 0.05.

To test for emergency granulopoiesis, we injected the HBV of lyz::dsRed zebrafish larvae with phosphate-buffered saline (PBS) or a low dose of S. flexneri. To capture production of new neutrophils, we imaged the larval AGM at 24 and 48 hpi (Fig. 1B). Quantifications revealed a significant increase (2.4 ± 0.5-fold) of neutrophils in Shigella-infected larvae at 48 hpi (Fig. 1C). A similar increase in neutrophil production is observed following the clearance of systemic Shigella administered via caudal vein injection (Fig. S1D). Shigella virulence is dependent upon the bacteria’s type III secretion system (T3SS) (19). To determine the role of the T3SS in the induction of emergency granulopoiesis, we performed HBV infections of zebrafish larvae using a T3SS-deficient (T3SS⁻) strain of S. flexneri (ΔmxiD). In agreement with a role for Shigella virulence in the induction of emergency granulopoiesis, neutrophil production is only weakly induced (1.5 ± 0.2-fold) by T3SS⁻ infection (Fig. S1E). To test for a global increase in myeloid cell production following infection, Tg(mpeg1::Gal4-FF)^gl25/Tg(UAS-E1b::nfsB.mCherry)^c264 (herein mpeg1::G/U::mCherry) larvae were infected with Shigella in the HBV, and mCherry-expressing (mCherry⁺) macrophages were quantified. Macrophage counts in the AGM are not significantly affected by Shigella infection (Fig. S1F). Taken together, these results show that infection by Shigella drives a neutrophil-specific hematopoietic response and that Shigella infection of zebrafish can be used to study factors underlying emergency granulopoiesis in vivo.

Macrophage-independent signaling of Gcsf mediates stem cell-driven granulopoiesis during Shigella infection.HSCs are increasingly recognized as important reactive components to infection (20). Runx1 is a transcription factor crucial for HSC development (21). To visualize stem cell-driven granulopoiesis, we outcrossed Tg(runx::mCherry)^cz2010 zebrafish that have mCherry-expressing HSCs with Tg(mpx::GFP)¹¹⁴ zebrafish that have GFP-expressing neutrophils. Double-transgenic larvae were infected with Shigella and imaged by high-resolution confocal microscopy. Here, photostability of the mCherry protein confers a dim fluorescence onto the stem cell’s immediate progeny. During Shigella infection, we observed that multiple high-mCherry-expressing HSCs are surrounded by low-mCherry progenitor cells and proximal to progenitor cells are GFP-positive neutrophils (Fig. 1D). We therefore hypothesized that HSC proliferation and differentiation underlie neutrophil production in response to Shigella infection. To test for increased stem cell proliferation upon Shigella infection, we quantified HSCs in the AGM of Tg(runx::eGFP)^cz2009 larvae. Consistent with results obtained using Salmonella infection of zebrafish (10), we observed significantly more HSCs (5.3 ± 1.8-fold) in Shigella-infected larvae at 48 hpi compared to uninfected larvae (Fig. 1E). These results indicate that the emergency granulopoietic response to Shigella infection is stem cell driven.

Previous work using zebrafish infection suggested that macrophage-derived Gcsf is required for emergency granulopoiesis (10). To test this, we used a morpholino oligonucleotide to deplete the Gcsf receptor (Gcsfr). Consistent with a role for Gcsf in Shigella-induced emergency granulopoiesis, Gcsfr morphants showed no significant increase in neutrophil production in the AGM upon infection (Fig. 1F; Fig. S1G). In support of a role for Gcsf in mediating granulopoiesis, quantitative reverse transcription-PCR (RT-qPCR) revealed significant increases in Gcsf expression in Shigella-infected larvae at 24 hpi (Fig. S1H). To test the role of macrophages in Shigella-induced emergency granulopoiesis, we used a morpholino oligonucleotide to target the myeloid lineage commitment factor Irf8 and deplete macrophages. Surprisingly, Irf8 morphants undergo robust emergency granulopoiesis during Shigella infection (Fig. 1G and Fig. S1I). Similar results are obtained using transgenic mpeg1::G/U::mCherry/Tg(mpx::eGFP)¹¹⁴ larvae, in which macrophages were ablated pharmacologically by metronidazole treatment (Fig. S1J and K). Together, these results indicate that macrophage-independent signaling of Gcsf is required for stem cell-driven granulopoiesis during Shigella infection.

Shigella-induced emergency granulopoiesis mediates long-term host defense.Emergency granulopoiesis is widely considered to be a transient homoeostatic mechanism for replacing exhausted leukocytes (1). To test if emergency granulopoiesis can also enhance innate immunity, we developed a Shigella reinfection assay (Fig. 2A). For this, larvae at 2 dpf were injected in the HBV with PBS or primed with a low dose (0.5 × 10³ to 2.0 × 10³ CFU) of GFP⁺ Shigella. At 48 hpi, control (naive) or infected larvae were reinfected with a lethal dose (>2.0 × 10⁴ CFU) of mCherry⁺ Shigella. Strikingly, priming of larvae with a low dose of Shigella rescued ~70% of animals that would have otherwise succumbed to secondary infection and significantly reduced bacterial burden compared to naive larvae (Fig. 2B and C; see Fig. S2A in the supplemental material). At the point of secondary infection (i.e., 0 h post-secondary infection [hp2i]), total neutrophil numbers are similar between naive and primed larvae, indicating that increased protection is not because of increased neutrophil numbers (Fig. S2B). To test the role of the T3SS in triggering protection against reinfection, we primed zebrafish larvae with wild-type or T3SS⁻ Shigella. Consistent with inducing only a mild granulopoietic response, T3SS⁻ Shigella provides some protection against secondary infection and rescues ~40% of lethal infections; however, the most robust protection is observed following infection with virulent Shigella (Fig. 2B and C). Collectively, these results show that emergency granulopoiesis is not solely a homoeostatic mechanism to counteract neutrophil cell death, but can also enhance innate immunity to secondary infection.

• Open in new tab
• Download powerpoint

FIG 2 

Emergency granulopoiesis mediates long-term host defense. (A) Schematic of reinfection assays. At 2 dpf, wild-type (WT) AB zebrafish larvae were injected in the HBV with PBS or a low dose (0.5 × 10³ to 2.0 × 10³ CFU) of GFP⁺ S. flexneri M90T. At 4 dpf, i.e., 48 h post-primary injection (hp1i), all larvae were injected with a high dose (>2.0 × 10⁴ CFU) of mCherry-expressing (mCherry⁺) S. flexneri M90T. Analyses were performed on larvae up to 72 h post-secondary infection (hp2i). (B and C) WT AB larvae were injected with PBS (open circles) or “primed” with wild-type or T3SS⁻ GFP⁺ S. flexneri M90T (closed circles), prior to a high dose of mCherry⁺ S. flexneri M90T at 48 hpi, as described above. (B) Survival curves pooled from 4 independent experiments using n ≥ 9 larvae per condition per experiment. Up to three larvae per condition were taken for CFU at the 24 and 48 h time points. The top graph represents collated data. The bottom graph represents only Shigella-primed larvae, a subset of the above data. The P value between conditions was determined by log-rank Mantel-Cox test. Significance was defined as P < 0.05 (*). (C) Fluorescent mCherry⁺ S. flexneri M90T burden of larvae was imaged by stereomicroscopy over time, and images were analyzed to produce fluorescence intensity measurements (as in Fig. S2A). Data were pooled from 4 independent experiments with n ≥ 4 per time point per condition per experiment. The top graph represents collated data. The bottom graph represents only Shigella-primed larvae, a subset of the above data. P values between conditions at cognate time points were determined by unpaired two-tailed Student’s t test. Significance was defined as P < 0.05 (*).

Ethics statement.Animal experiments were approved by the Home Office (project licenses PPL 70/7446 and PPL P84A89400) and performed in accordance with the Animals (Scientific Procedures) Act 1986.

Zebrafish husbandry.Wild-type AB zebrafish were purchased from the Zebrafish International Resource Center (Eugene, OR). The transgenic lines Tg(lyz::dsRed)^nz50, Tg(mpx::GFP)ⁱ¹¹⁴, Tg(runx::mCherry)^cz2010, Tg(runx::eGFP)^cz2009, and Tg(mpeg1::Gal4-FF)^gl25/Tg(UAS-E1b::nfsB.mCherry)^c264 were previously published (27–30). Embryos were obtained from naturally spawning zebrafish, and both control and infected larvae were maintained at 28.5°C in embryo medium (0.5× E2 medium supplemented with 0.3 µg/ml methylene blue) (31). Larvae were staged according to Kimmel et al. (32). For injections and live microscopy, larvae were anesthetized with 200 µg/ml tricaine (Sigma-Aldrich) in embryo medium.

Zebrafish infection.Bacterial strains used include wild-type invasive S. flexneri serotype 5a M90T expressing GFP or mCherry protein and a T3SS⁻ noninvasive ΔmxiD variant expressing mCherry (33). Shigella cells were cultured overnight in Trypticase soy broth (TSB), diluted 50× in fresh TSB, and grown at 37°C until they reached an A₆₀₀ of 0.6. Bacteria were harvested by centrifugation, washed, and resuspended in PBS to achieve the desired concentration. For infection, larvae were microinjected in the hindbrain ventricle (HBV) or caudal vein at 2 dpf with up to 1 nl of PBS or a low dose (0.5 × 10³ to 2.0 × 10³ CFU) of S. flexneri. For reinfection assays, larvae at 2 dpf were injected with either PBS or a low dose of GFP⁺ S. flexneri in the HBV; larvae were confirmed to have cleared GFP⁺ Shigella infection (as determined by fluorescence stereomicroscopy) and were infected at 4 dpf (i.e., 48 hpi) with a high dose (>2.0 × 10⁴ CFU) of mCherry⁺ S. flexneri. Injection protocols are as described previously (34). Larvae were maintained in individual wells of a 12-well culture dish for assessment.

Survival assays.Larvae were imaged using a light stereomicroscope at time points following infection. Larvae failing to produce a heartbeat or larvae in which bacteria had compromised the hindbrain were considered nonviable.

Measurement of inocula and bacterial burden.For enumeration of live bacteria by CFU plating, larvae were mechanically homogenized in lysis buffer (0.4% Triton X-100, PBS). Homogenates were serially diluted and plated on lysogeny broth (LB) agar supplemented with 50 µg/ml carbenicillin. Fluorescent colonies were scored after incubation of plates for 24 h at 37°C. For reinfection assays, S. flexneri burden was determined by fluorescence stereomicroscopy. Here, the larval hindbrain was defined as a “region of interest” and was subjected to thresholding to give the percentage of fluorescence of the hindbrain. Only viable larvae were used for CFU and image fluorescence analysis.

Morpholino oligonucleotide injection.Antisense morpholino oligonucleotides were purchased from GeneTools (http://www.gene-tools.com). gcsfr (ENSDARG00000045959) was targeted using published morpholino sequence 5′ ATTCAAGCACATACTCACTTCCATT 3′ to block mRNA splicing (10). Macrophages were depleted by targeting irf8 (ENSDARG00000056407) using published morpholino sequence 5′ AATGTTTCGCTTACTTTGAAAATGG 3′ to block mRNA splicing (35). To control for nonspecific effects, a standard morpholino oligonucleotide with no known target in the zebrafish genome was used (13). Morpholino oligonucleotide solutions were diluted to the desired concentration (1 mM) in 0.1% phenol red solution (Sigma-Aldrich), and 0.8 nl was microinjected into the yolk sack at the 1- to 2-cell stage.

Microscopy and image analysis.For in vivo time-lapse imaging, larvae were immobilized in 1.5% low-melting-point agarose as previously described (34). Stereomicroscopy was performed using a Leica M205FA microscope and 10× (NA 0.5) dry objective. For Movie S1, z-stacks were acquired every 15 min. For high-resolution confocal microscopy, larvae were positioned in 35-mm-diameter glass-bottom MatTek dishes and imaging was performed using a Zeiss LSM 710 and 10×, 20×, and 40× oil or 63 × oil immersion objectives. Image files were processed using ImageJ/FIJI software (36). Leukocyte and HSC quantifications were performed manually from images taken by stereomicroscopy.

RT-qPCR.RNA was extracted from 10 snap-frozen larvae with an RNeasy minikit (Qiagen) and reverse-transcribed using a QuantiTect reverse transcription kit (Qiagen) as per the manufacturer’s instructions. Template cDNA was subjected to PCR using primers for gcsf (ENSDARG00000102211), as previously described (10). Quantitative PCR (qPCR) was performed on a Rotor-GeneQ thermocycler (Qiagen) and samples run in technical duplicate with SYBR green master mix (Applied Biosystems). Primers against housekeeping gene ef1a1l1 (13) and the threshold cycle (2^−ΔΔCT) method (37) were used to normalize cDNA.

Validation of morpholino oligonucleotide depletion.For validation of alternative Gcsfr splicing, RNA was isolated from a pool of control and Gsfr morphants at 2 dpf. cDNA was prepared as described above in “RT-qPCR” and used as a template for RT-PCR using OneTaq DNA polymerase (New England Biolabs) and primers Gcsfr-FW (5′ CATCCGTCTCGCTTGTGCTT 3′) and Gcsfr-Rv (5′ GGTGGGACCGCATAAACCTT 3′).

Drug treatments.To deplete macrophages, mpeg1::G/U::mCherry larvae were treated with 10 mM metronidazole (Sigma-Aldrich) in embryo medium supplemented with 1% dimethyl sulfoxide (DMSO) (Sigma-Aldrich) from 24 hpf, as previously described (28).

Statistical analyses.Statistical tests were performed using Prism software (GraphPad Software, Inc.). Statistical significance of survival curves was determined using the log rank Mantel-Cox test. In all other cases, statistical significance was determined using unpaired two-tailed Student’s t test; analyses were performed on raw values for cell counts, log₁₀ values for CFU counts, and log₂ values for gene expression data. Bonferroni’s posttest was applied in cases of multiple testing, as specified in the figure legends. Data are represented as mean ± standard errors of the mean (SEM).