A granulin-positive macrophage subtype in mycobacterial granulomas alleviates tissue damage by limiting excessive inflammation

Significance

Granulomas play a crucial role in the pathology of tuberculosis, but the immune environment governing their formation remains largely unknown. Here, we identified an infection-induced grna.2⁺ macrophage subset within mycobacterial granulomas by single-cell RNA sequencing (scRNA-seq). We further demonstrate that depleting this macrophage subset shapes more compact granulomas with reduced T cell infiltration, ultimately resulting in necrotized granulomas with extensive macrophage lytic death. This study also reveals that a well-regulated immune response within granulomas is essential for host control of mycobacterial infection, and the controlled immune pathology is maintained by the grna.2⁺ macrophage subset through the suppression of excessive inflammation. Notably, in samples from tuberculosis patients, we also identified granulin (GRN)-positive macrophages that display similar anti-inflammatory functions.

Abstract

Granulomas play a crucial role in the pathology of tuberculosis, but the immune environment governing their formation remains largely unknown. To explore the dynamic changes in the immune microenvironment during the formation of tuberculous granulomas, we infected adult zebrafish with Mycobacterium marinum and then examined uninfected and infected kidneys, as well as large and small granulomas in the kidneys. Using single-cell RNA sequencing technology, we identified two major macrophage subpopulations in the hematopoietic tissue (kidney) of zebrafish under uninfected physiological conditions: monocyte derived and tissue-resident macrophages. Interestingly, the infection induced the emergence of epithelioid cells and a previously undescribed grna.2⁺ macrophage subpopulation. Depletion of grna.2⁺ macrophages with the nitroreductase-metronidazole ablation system resulted in shortened zebrafish survival after infection, increased bacterial load, and more granulomas, especially necrotic granulomas. Depletion of grna.2⁺ macrophages also produced a denser granuloma structure with fewer T cells. RNA-seq and flow cytometry analysis revealed that depletion of grna.2⁺ macrophages led to upregulated inflammatory signaling pathways, including tnfα and il1β, and increased macrophage lytic cell death. Similarly, in samples from tuberculosis patients, we also identified GRN-positive macrophages, which exhibit similar anti-inflammatory functions. This subset of grna.2⁺ macrophages present in developing granulomas can suppress excessive inflammatory responses to alleviate macrophage lytic death, reduce tissue damage, promote T cell infiltration and ultimately help control mycobacterial growth in vivo.

Tuberculosis remains an important threat to public health, causing 10.6 million new cases globally, with 1.3 million deaths in 2022 (1). When aerosols containing Mycobacterium tuberculosis (M.tb) enter the lung alveoli, alveolar macrophages (AM) engulf the bacteria that migrate into lung parenchyma, where the bacteria multiply intracellularly (2, 3). Infected macrophages recruit more macrophages as well as other immune cells, especially neutrophils, dendritic cells, and lymphocytes (3, 4). During this process, macrophages undergo epithelioid transformation and aggregate with other immune and nonimmune cells to form tuberculous granulomas (4–11). Macrophages are heterogenous and can be classified, based on their origins, functions, and surface markers, as tissue-resident macrophages, monocyte-derived macrophages, and the M1/M2 subtypes (12–15). Recently, single-cell RNA sequencing (scRNA-seq) technology has identified multiple macrophage subpopulations with unique gene expression profiles following M.tb infection (16–22), but the dynamics of the different macrophage subpopulations in granuloma formation and development remain unclear.

Granulins are a class of proteins found across species ranging from slime molds and green plants to mammals, and are regarded as one of the earliest extracellular regulatory proteins to have evolved (23, 24). In mammals, granulin (GRN) monomers are produced from proteolytic digestion of progranulin (PGRN) (25–27), which is secreted mainly by macrophages (28). In contrast to mammals, zebrafish possess two orthologs of Grn, namely granulin a (grna) and granulin b (grnb), as well as two paralogs, granulin 1 (grna.1) and granulin 2 (grna.2), which encode shorter proteins resembling mammalian GRN cleavage products (29, 30). GRN plays a crucial role in various physiological and pathological processes, including regulating cytokine release, recruiting immune cells, and stimulating phagocytosis (25, 31, 32). GRN has also been reported to be involved in immune-related noninfectious diseases such as rheumatoid arthritis and systemic lupus, where it reduces tissue damage by counteracting inflammatory factors such as Tumor Necrosis Factor alpha (TNFα) and Interferon gamma (IFN-γ) (33–35). However, the function of GRN in infectious diseases, including tuberculosis, is currently unclear.

To study the dynamic changes in the macrophage subpopulations during granuloma formation, we used the adult zebrafish model of tuberculosis. Zebrafish infected with M. marinum (M.m) form granuloma structures very similar to those observed in tuberculosis patients (10, 36–38). Using scRNA-seq analysis, we systematically analyzed the dynamics of the cell types and macrophage subtypes in the development of granuloma within the kidneys of M.m infected zebrafish.

Results

Immune Landscape Profiling Pre- and Postmycobacterial Granuloma Formation.

To investigate changes in the immune microenvironment during different stages of granuloma formation, we infected adult zebrafish (3 to 6 mo) with 2,000 colony-forming units (CFU) of M.m. Zebrafish were killed at 28 days postinfection (dpi) and the kidneys harvested: control, uninfected kidneys were designated as K1 and infected kidneys were designated as K2. Kidney granulomas were dissected and separated by size into those with longitudinal widths smaller (G1) or larger (G2) than 400 μm (SI Appendix, Fig. S1 A–C). scRNA-seq analysis of the four sample types (K1, K2, G1, G2) identified 15 cell types (Fig. 1A and SI Appendix, Fig. S1 D and E). While there were high proportions of macrophages and neutrophils, nonimmune clusters were also identified, including cells belonging to the vascular endothelium and the proximal and distal tubules (Fig. 1B).

Fig. 1.

Immune microenvironment analysis pre- and postmycobacterial granuloma formation. (A) UMAP plot of all cells from the four samples, with 15 generic cell types identified by different colors, as indicated on the Right side of the plot. (B) Proportions of different cell populations in the four samples, with colors of the cell groups corresponding to the annotations on the Right side. (C) Following the intraperitoneal injection of M.m into Tg(mpeg1-LRLG (mpeg1:loxP-DsRed2-loxP-GFP), macrophages labeled with red fluorescence) zebrafish, samples were obtained at 14 dpi. Flow cytometry (FACS) was used to analyze the proportion of DsRed2⁺ macrophages in the four samples. (D) Statistics of the proportions of macrophages in the four samples shown in panel (C). (D) analyzed with one-way ANOVA, *P < 0.05; **P < 0.01.

Most of the 15 cell types were present in all four samples, but their proportions varied (Fig. 1B and SI Appendix, Fig. S1E). Neutrophils were more abundant in infected compared to uninfected kidneys, with a higher proportion in large granulomas compared to small granulomas. Conversely, macrophages were significantly more prevalent in small granulomas, and the proportion of proximal tubule cells was significantly decreased in all infected kidneys (Fig. 1B and Dataset S1). By taking advantage of macrophage reporter line Tg(mpeg1-loxP-DsRed2-loxP-GFP(LRLG), macrophages labeled with DsRed2), we confirmed the proportion of macrophages by flow cytometry (Fig. 1 C and D and SI Appendix, Fig. S2A). To test whether Tg(mpeg1-LRLG) could label all macrophages during mycobacterial infection, we conducted two complementary experiments. First, we observed strong colocalization between mpeg1-LRLG and 4C4, another macrophage marker specifically expressed in microglia and some peripheral macrophages (39), in granulomas (SI Appendix, Fig. S2B). Second, we assessed the proportion of macrophages in the kidney and granulomas by flow cytometry using a different transgenic zebrafish line, Tg(mfap4-LRLG), which has been previously reported to stably label macrophages during mycobacterial infection (40–42), and obtained almost identical results (SI Appendix, Fig. S2 C and D), possibly due to the long half-life of DsRed2 (43, 44). In summary, although we cannot completely rule out the presence of some mpeg1-LRLG⁻ macrophages, we believe that mpeg1-LRLG labels the majority of macrophages in granulomas.

The Expansion of grna.2⁺ Macrophages Is Associated with Mycobacterial Granuloma Formation.

Macrophages, the major cell type in granulomas, can be divided into different subpopulations, each playing a distinct role in controlling mycobacterial infections. Although scRNA-seq has been used to describe the phenotypes of the macrophages present in granulomas formed postinfection (16, 19, 45), the dynamics of these subpopulations during the granuloma formation and progression remain unclear. Here, by using merged scRNA-seq data from the four sample types, we identified 20 macrophages subtypes (SI Appendix, Fig. S3A). The distribution was similar in the kidneys and granulomas from infected zebrafish but significantly different in uninfected kidneys (SI Appendix, Fig. S3 B and C). The expression correlation heatmap of highly variable genes in macrophages revealed four gene coexpression modules (SI Appendix, Fig. S3D), allowing classification of the 20 subpopulations into four classes that exhibited distinct, mutually exclusive clustering: Macrophage_1 (subpopulations 15, 7, 3, 9, 5, and 6), Macrophage_2 (17, 10, 13, 16, 12, 1, and 14), Macrophage_3 (4 and 8), and Macrophage_4 (11, 0, 19, 2, and 18) (Fig. 2A and SI Appendix, Fig. S3 E–I).

Fig. 2.

Analysis of macrophage subpopulations during granuloma formation. (A) UMAP plots showing Macrophage_1/2/3/4. (B) Proportions of Macrophage_1/2/3/4 in samples K1, K2, G1, and G2. The colors of the four macrophage groups correspond to those shown in panel (A). (C) Heatmap of the top 20 DEGs for Macrophage_1/2/3/4. Each row corresponds to a gene’s expression across different macrophages, with gene names annotated on the Left and expression levels on the Right. (D) Distribution of macrophages in zebrafish granulomas. Following the intraperitoneal injection of M.m into Tg(mpeg1-LRLG; grna.2-GFP-NTR, grna.2⁺ macrophages labeled with green fluorescence) zebrafish, samples were obtained at 14 dpi. Frozen sections were stained with antibodies against DsRed2 and GFP. White dashed lines indicate granulomas core. (E) Flow cytometry analysis of GFP⁺ grna.2⁺ macrophages from the kidney of M.m-infected Tg(mpeg1-LRLG; grna.2-GFP-NTR) zebrafish. [Scale bar: 50 μm (D).]

Macrophage_1 was present in all four samples, with higher proportions in the uninfected and infected kidney samples but significantly lower proportions in both small and large granulomas (Fig. 2B). These macrophages exhibited high expression of genes associated with cell proliferation processes, including DNA repair [hmg family (46–49)] and DNA replication [tubb2b (50, 51), pcna (52, 53)] (Fig. 2C, SI Appendix, Fig. S3 D and J, and Datasets S2 and S3). Given their presence under physiological conditions and high proliferation capacity, we termed these “Monocytes.” Macrophage_2 was present in the uninfected kidneys but almost absent in samples from infected kidneys and granulomas (Fig. 2B). It disappeared after infection and expressed genes encoding proteins involved in heat-shock (hsp70l) (54, 55), phagocytosis [ctss (56, 57), marco (58)], and clearing dead cells and tissue regeneration [ccl34a.4 (59), hmox1a (60–62)] (Fig. 2C, SI Appendix, Fig. S3 D and J, and Datasets S2 and S4). This expression profile closely resembles tissue-resident macrophages (TRM), leading us to label Macrophage_2 as TRM. Macrophage_3 emerges only after infection (Fig. 2B) and exhibits high expression of epithelial cell markers, including members of the krt family (63, 64). Given previous reports of a subset of macrophages in granulomas displaying epithelial cell-like characteristics (16, 41), we designated this cell population as “Epithelioid cells” (16, 41) (Fig. 2C, SI Appendix, Fig. S3 D and J, and Datasets S2 and S5). Macrophage_4 was almost exclusively present in the postinfection K2, G1, and G2 samples (Fig. 2B). These cells showed high expression of anti-inflammatory associated genes grna.2 (65, 66), lgals9l1 (67, 68), and lgmn (69, 70) (Fig. 2C, SI Appendix, Fig. S3 D and J, and Datasets S2 and S6). We found that grna.2 was the most highly expressed gene in Macrophage_4 subtypes but weakly expressed in the other subtypes (SI Appendix, Fig. S3K). Collectively, the Macrophage_4 subpopulation was thus termed “grna.2⁺ Macrophages” or “Infected-Induced Macrophages (IIM),” and possibly involved in the bacterial killing and immune homeostasis. Analysis of differentiation trajectories indicated that monocytes could differentiate into grna.2⁺macrophages, consistent with the monocyte serving as a proliferation-based subtype (SI Appendix, Fig. S3L). Further transcription factor analysis reveals significant upregulation of genes such as litaf, zgc:162730, tfa, atf4a, spi1b, and hmgb1a in grna.2⁺ macrophages compared to monocytes. Notably, spi1b (pu.1), a key regulator of myeloid development and immune responses (71–73), may drive grna.2 expression or determine its cell fate.

By integrating previously published scRNA-seq data on zebrafish granulomas with the G1 and G2 datasets (19), we found that most macrophages in the granuloma samples express grna.2, as shown in the grna.2 uniform manifold approximation and projection (UMAP) plot (SI Appendix, Fig. S4 A and B). However, even when incorporating the unsupervised clustering results, we could not identify a distinct macrophage population with high grna.2 expression. When we further combined the previously published scRNA-seq data with our four sample datasets for joint analysis, we similarly found that grna.2⁺ cells were present in the infected samples, and the composition of macrophage subpopulations in this previous dataset closely resembled that in our G1 and G2 samples (SI Appendix, Fig. S4 C–E).

Mycobacterial Infection Facilitates the Expansion of Granulin⁺ Macrophages.

Immunofluorescent staining of transgenic zebrafish Tg(mpeg1-LRLG; grna.2-GFP-NTR) confirmed that most of the mpeg1-LRLG⁺ macrophages in the granulomas were Grna.2⁺ (Fig. 2D). In addition, flow cytometry showed a significant increase in the proportion of Grna.2⁺ macrophages in kidneys following M.m infection (Fig. 2E and SI Appendix, Figs. S5 and S6A), along with elevated expression of grna.2 mRNA (SI Appendix, Fig. S6B). Similarly, M.m-infected zebrafish larvae showed >600-fold higher expression of grna.2 mRNA compared to uninfected larvae at 5 dpi (SI Appendix, Fig. S6C).

Depletion of grna.2⁺ Macrophages Leads to Exacerbated Infection, Induced Tissue Damage, and Accelerated Mycobacterial Growth In Vivo.

To investigate the role of grna.2⁺ macrophages in mycobacterial infection, we utilized CRISPR-Cas9 to knock out the grna.2 gene and assessed its impact on the susceptibility of zebrafish embryos’ to M.m. The knockout of grna.2 alone did not affect susceptibility to M.m (SI Appendix, Fig. S7 A and B). Given the presence of two orthologs (grna and grnb) and two paralogs (grna.1 and grna.2) in zebrafish, we also performed individual knockouts of the other three genes and observed no significant differences in survival curves or bacterial load between crispants and control embryos (SI Appendix, Fig. S7 A and B). Similarly, in adult zebrafish there were no significant differences in the survival curves, bacterial loads, or the number of granulomas between grna.2 knockouts and control (SI Appendix, Fig. S7 C–E).

To address potential redundancy between these four granulin genes, we generated double knockouts of grna.1 and grna.2, which are both highly expressed in granulomas (SI Appendix, Fig. S7F). However, the survival curves and bacterial loads of these double knock-out crispants showed no significant differences compared to control embryos (SI Appendix, Fig. S7 B and G). Attempts to simultaneously knock out all four homologous genes in zebrafish embryos using 8 gRNAs resulted in a high proportion of severe deformities or mortality (SI Appendix, Fig. S7H). Thus, it remains inconclusive from knockout experiments alone whether these genes exhibit functional redundancy or if grna.2 serves as a nonfunctional marker gene.

An alternative approach to studying the role of grna.2⁺ macrophages during mycobacterial infection is to deplete these cells. To eliminate cells expressing Grna.2, we utilized the nitroreductase (NTR)-metronidazole (MTZ) cell depletion system. Prior to performing cell depletion using the previously described Tg(grna.2-GFP-NTR) line, we first confirmed that Grna.2 expression is predominantly restricted to macrophages in kidney and granuloma samples. scRNA-seq data revealed marked differences in grna.2 expression between uninfected (K1) and infected (K2, G1, G2) samples, with the expression being highly concentrated in macrophages (SI Appendix, Fig. S8A), with other cell types exhibiting minimal expression (Fig. 1B and SI Appendix, Figs. S7F and S8B). Flow cytometry analysis also demonstrated a negligible percentage of GFP⁺ cells in nonmyeloid populations (SI Appendix, Fig. S8C). Additionally, scRNA-seq data also confirmed that the grna.2⁺ cells in our four samples are not metaphocytes (59, 74–76), as the expression of marker genes typical for metaphocytes were almost absent in grna.2⁺ cells (SI Appendix, Fig. S8 D–I) (59). A comparison of signaling pathways and differentially expressed genes (DEGs) further distinguished grna.2⁺ cells from metaphocytes (SI Appendix, Fig. S8 J and K) (59, 74).

Next, the transgenic line Tg(mpeg1-LRLG; grna.2-GFP-NTR) (referred to as NTR) was exposed to MTZ, as described previously (SI Appendix, Figs. S9 and S10) (74), to assess phenotypic changes following M.m infection. Without MTZ treatment, there was no significant difference in the survival times after infection between the NTR strain and the Tg(mpeg1-LRLG), control zebrafish (Fig. 3A and SI Appendix, Fig. S9). With MTZ treatment, however, the survival time after M.m infection was significantly shorter in the NTR strain (Fig. 3B). In addition, the bacterial loads at 14 dpi increased significantly in the NTR group (Fig. 3C), along with a wider distribution of granulomas, especially necrotic granulomas, and more intense immune cell infiltration (Fig. 3 D–L and SI Appendix, Fig. S11), suggesting more severe tissue damage. These results suggest that depletion of grna.2⁺ cells leads to increased tissue damage. HE staining and E-cadherin antibody staining confirmed that MTZ treatment did not alter the structure of the epithelioid cell layer within the granulomas (Fig. 3E and SI Appendix, Fig. S12A), consistent with bioinformatics analysis showing that grna.2 is not expressed in epithelioid cells (SI Appendix, Figs. S3K and S12B). Using the larval infection model, we found that MTZ treatment slightly, but significantly, extended zebrafish survival after M.m infection, despite similar bacterial loads in the two groups (SI Appendix, Fig. S13 A and B). Collectively, these results suggest that grna.2⁺ macrophages play an important role in host resistance to mycobacterial infection.

Fig. 3.

Functional assay suggesting grna.2⁺ macrophages are protective during mycobacterial infection. (A) Survival curves of zebrafish in the Ctr, NTR, Ctr + M.m, and NTR + M.m groups. Each group contains 10 fish. (B) Survival curves of zebrafish in the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. The data shown combine three independent biological replicates. (C) Differences in bacterial load in zebrafish from the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. Each point represents the bacterial load in a single zebrafish from one biological replicate. Different colors indicate independent biological replicates. (D–H) HE staining of paraffin sections of zebrafish from the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. Blue arrows indicate granulomas. The Top Right panel shows an enlarged view of the region within the black box. Blue dashed lines indicate granulomas. Panel (G–I) show representative images of early immune cell infiltration, nonnecrotic granulomas, and necrotic granulomas, respectively. Blue dashed lines indicate granulomas, and yellow dashed lines indicate necrotic cores. (I) Quantification of granulomas in zebrafish from the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. Each point represents the total number of granulomas in one zebrafish. (J) Quantification of granulomas at different developmental stages from panel (J). (K) Number of immune cells around granulomas of zebrafish from the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. (L) Area of immune infiltration of the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. (A and B) analyzed with simple survival analysis (Kaplan–Meier), and (C, I, J, K, and L) analyzed with the unpaired two-tailed t test, **P < 0.01, ***P < 0.001, and ns = nonsignificant. [Scale bar: 500 μm (D and E), 50 μm (D and E (Top Right panel)), 25 μm (F, G, and H).]

grna.2⁺ Macrophages Can Re-Expand from grna.2⁻ Cells within Granulomas.

We next investigated the origin of grna.2⁺ macrophages in vivo. Under normal physiological conditions, grna.2⁺ macrophages reside in mucosal tissues such as the gill, intestine, and epidermis (74). These grna.2⁺ macrophages can regenerate after MTZ depletion and maintain relatively stable populations (SI Appendix, Fig. S10) (59, 74–76). To determine whether grna.2⁺ macrophages in granulomas similarly regenerate postdepletion, we used an ex vivo granuloma culture system (77). Granulomas were dissected from Tg(grna.2-GFP-NTR) zebrafish kidneys infected with the M.m-tdTomato strain and used to determine that 2 mM MTZ was the minimum effective dose for depleting grna.2⁺ macrophages within granulomas (SI Appendix, Fig. S14A). Interestingly, bacterial fluorescence intensity increased rapidly in freshly dissected granulomas, indicating rapid folding and maturation of the fluorescent proteins upon transitioning from the hypoxic microenvironment within the host to the oxygen-rich environment outside (SI Appendix, Fig. S14B). Further investigation revealed that the maturation time for these fluorescent proteins is approximately 6 h (SI Appendix, Fig. S14C). Therefore, to ensure accurate measurement of bacterial fluorescence intensity, granulomas were cultured ex vivo for 6 h before measurements. Live imaging revealed that prolonged MTZ treatment led to a gradual disappearance of GFP-positive macrophages, whereas GFP intensity remained stable in the untreated group (Fig. 4A, SI Appendix, Fig. S15A, and Movie S1). After replacing the MTZ-containing L15 medium with fresh L15 medium and conducting continuous live imaging, GFP⁺ macrophages gradually reappeared in granulomas about 10 h post-MTZ removal, whereas fluorescence intensity in the control group remained stable (Fig. 4B, SI Appendix, Fig. S15A, and Movie S2). The change in the intensity of bacterial fluorescence in the MTZ-treated group was significantly higher than in the control group starting around 15 hpt (SI Appendix, Fig. S15B). These ex vivo results demonstrate that grna.2⁺ macrophages can develop from grna.2⁻ cells within granulomas and they help to control bacterial growth.

Fig. 4.

Investigation of the origin and the role of grna.2⁺ macrophages in host defense ex vivo. (A) Fluorescence changes of grna.2⁺ macrophages (green) and M.m (red) in granulomas at different time points in the control and 2 mM MTZ treatment groups. Time-lapse intervals were 10 min. (B) Fluorescence changes of grna.2⁺ macrophages (green) and M.m (red) in granulomas at different time points after removal of MTZ. Time-lapse intervals were 10 min. The Bottom Left panel shows an enlarged view of the region within the blue box. Red arrows indicate reappear grna.2⁺ macrophages. [Scale bar: 50 μm (A and B).]

Depletion of grna.2⁺ Macrophages Leads to a More Compact Structure of Granulomas and Impedes T Cell Infiltration.

The spatial distribution of immune cells within granulomas has been associated with the host’s ability to control mycobacterial infection, with different macrophage subtypes exhibiting distinct morphologies and distributions (38, 41, 78, 79). Epithelioid macrophages, for example, are found mainly around the necrotic core, perhaps obstructing the entrance of neutrophils into the core area and thereby reducing the elimination of mycobacteria (41). By employing frozen sections and immunofluorescence staining after the depletion of grna.2⁺ macrophages, we observed changes in the distribution of immune cells and their interactions with mycobacteria within the granulomas. After depleting grna.2⁺ macrophages (NTR + M.m + MTZ), the grna.2^- macrophages aggregated more tightly, especially around the granuloma cores (Fig. 5A), and this denser distribution of mpeg1⁺ macrophages was confirmed by statistical analysis (Fig. 5 B and C). To explore the effects of the altered macrophage distribution on T lymphocytes, we performed in situ hybridization to lck mRNA, a marker of mature T lymphocytes. Compared to the controls, the presence of lck⁺ T lymphocytes in the granuloma core areas decreased after the depletion of grna.2⁺ macrophages (Fig. 5 D and E). Real-time quantitative PCR analysis of T cell receptor (TCR) mRNA, showed a significant reduction in the expression of TCRα constant region (trac) and CD3 chain (cd3) after grna.2⁺ macrophage depletion, consistent with the lck in situ results (Fig. 5F). The decreased T cell infiltration was associated with an approximate doubling of the number of mycobacteria in the granuloma (Fig. 3I), consistent with reports that T lymphocytes play a role in controlling Mycobacterium proliferation (4, 10, 20, 80, 81).

Fig. 5.

Depletion of grna.2⁺ macrophages altered the distribution of immune cells in granulomas. (A) Distribution of macrophages in granulomas of zebrafish from the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. Following intraperitoneal injection of M.m into Tg(mpeg1-LRLG; grna.2-GFP-NTR) zebrafish, samples were obtained at 14 dpi. Frozen sections were stained with DAPI and antibodies against DsRed2. (B) Density value distribution of macrophages in panel (A). Each point represents the relative coordinates of a single macrophage. The R script was used to calculate the density value of each point relative to all other points. (C) Proportion of HDV macrophages in panel (B). Each point represents the statistical value for a single granuloma, with different colors indicating three independent biological replicates, with data from three zebrafish per experiment (each zebrafish having 2 to 3 granulomas). (D) Representative images of lck⁺ in situ hybridization signals in granulomas of zebrafish from the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. Red dashed lines outline the granulomas, yellow dashed lines indicate necrotic cores, and red arrows point to lck⁺ in situ hybridization signals. (E) Quantification of the number of lck⁺ in situ hybridization signals in different types of granulomas from the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. Each point represents the number of lck⁺ in situ hybridization signals in a single granuloma. (F) Relative mRNA expression levels of TCR-related genes trac and cd3 in granulomas from the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. Each group consists of three granuloma samples, with each sample derived from three zebrafish. (C) analyzed with the unpaired two-tailed t test, and (E and F) analyzed with one-way ANOVA, *P < 0.05, **P < 0.01, ***P < 0.001, and ns = nonsignificant. [Scale bar: 50 μm (A and D).]

Depletion of grna.2⁺ Macrophages Causes Early Cytokine Surge and Severe Tissue Damage.

The twofold increase of the bacterial load in zebrafish depleted of grna.2⁺ macrophages was accompanied by more severe tissue damage (Fig. 3 C–E). To further explore the effects of grna.2⁺ macrophage depletion, we examined the bacterial load and pathology at early time points postinfection. At 7 dpi, the bacterial load in grna.2⁺ macrophage depleted zebrafish was not significantly different from untreated zebrafish (Fig. 6A), but there were more granulomas that developed more rapidly, and some showed early necrotic cores (Fig. 6 B and C). The mRNA expression levels of proinflammatory cytokines such as il1β, tnfα, and il6 were also significantly higher in zebrafish depleted of grna.2⁺ macrophages (Fig. 6D), suggesting that grna.2⁺ macrophages mitigate the pathological damage and the increase in bacterial load by suppressing excessive, infection-induced inflammatory responses.

Fig. 6.

Depletion of grna.2⁺ macrophages led to a hyperinflammatory granuloma microenvironment and promoted macrophage lytic cell death and tissue damage. (A) The bacterial load in the early stages of infection. Each point represents the bacterial load in an individual fish. (B) The number of granulomas in the early stages of infection. Each group consists of 5 zebrafish. (C) Representative morphology of granulomas shown in panel (B). Red dashed lines indicate granulomas, yellow dashed lines indicate necrotic cores. (D) Relative mRNA levels of proinflammatory cytokines il1β, tnfα, and il6 mRNA from the Ctr + M.m + MTZ and NTR + M.m + MTZ groups in the early stages of infection. Each group consists of three kidney samples, each sample derived from three zebrafish. (E) Schematic diagram of experimental design for RNA-seq analysis of kidney samples from Ctr + M.m + MTZ and NTR + M.m + MTZ groups of zebrafish. (F) GO enrichment analysis of the top 10 upregulated pathways in the NTR + M.m + MTZ group. The chart displays the top 10 most significant BP terms. (G) Proportion of dead macrophages (7-AAD⁺) in the Ctr + M.m + MTZ and NTR + M.m + MTZ groups. Following the intraperitoneal injection of M.m into Tg(mpeg1-LRLG; grna.2-GFP-NTR) zebrafish, samples were obtained at 14 dpi. (H) Statistical analysis of the proportion of dead macrophages in panel (G). Each point represents the proportion of dead cells in an individual zebrafish. (I) Schematic diagram of experimental design for sorting of kidney samples from NTR + M.m + MTZ and NTR + M.m groups of zebrafish. (J–L) Following the intraperitoneal injection of Katushka-labeled M.m into Tg(mpeg1-LRLG; grna.2-GFP-NTR) zebrafish, samples were obtained at 7 dpi. FACS was used to analyze the proportion of Katushka⁺ macrophages in grna.2⁻ macrophages from NTR + M.m + MTZ group, grna.2⁺ and grna.2⁻ macrophages from NTR + M.m group. (M) Statistical analysis of the MFI of Katushka⁺ macrophages in three samples shown in panel (J–L). (A, B, D, and H) analyzed with the unpaired two-tailed t test, and (M) analyzed with one-way ANOVA, *P < 0.05, **P < 0.01, ***P < 0.001, and ns = nonsignificant. [Scale bar: 25 μm (C).]

Depletion of grna.2⁺ Macrophages Leads to a Hyperinflammatory Microenvironment in the Granuloma and Promotes Lytic Cell Death of Macrophages.

To further investigate the mechanisms underlying the exacerbated tissue damage, we examined the differential expression signaling pathways in the kidneys of zebrafish after M.m infection (Fig. 6E). Compared to the control (Ctr + M.m + MTZ) zebrafish, those deleted for grna.2⁺ macrophages (NTR + M.m + MTZ) had 5,382 DEGs (q < 0.05; |log2FC| > 1), with 3,596 genes upregulated and 1,786 genes downregulated. The sample correlation tests, principal component and clustering analyses, showed very low correlation between the two groups, indicating significant differences in the signaling pathways (SI Appendix, Fig. S16 A–C). Gene Ontology (GO) enrichment analysis revealed that most of the pathways upregulated in samples deleted for grna.2⁺ macrophages are involved with cellular immune and inflammatory responses, suggesting extensive activation of the host’s inflammatory responses (Fig. 6F). Similarly, KEGG analysis indicated that grna.2⁺ macrophage depletion led to enrichment of immune cell activation-related pathways such as NOD-like receptors, C-type lectin receptors, phagosome, and MAPK pathways (SI Appendix, Fig. S16D). Comparison of DEGs in the top 10 GO pathways revealed that tnfsf1a appeared in 5 enriched pathways, while il1β, tnfα, pycard, and jak2a appeared in 4 pathways (SI Appendix, Fig. S17A), suggesting that the function of grna.2⁺ macrophages may be linked to the tnfα-tnfsf1a and il1β-pycard signaling pathways. The mRNA expression levels of genes in these pathways were significantly higher after grna.2⁺ macrophage depletion (SI Appendix, Fig. S17B). STRING analysis of the top 300 DEGs also revealed TNFα as a key protein in samples deleted for grna.2⁺ macrophages (SI Appendix, Fig. S17C). Finally, the deleterious effects of decreased GRN were demonstrated by the significantly higher proportion of macrophage lytic cell deaths observed in infected kidneys depleted of grna.2⁺ macrophages compared to the untreated, infected controls (Fig. 6 G and H and SI Appendix, Fig. S18). Additionally, by using flow cytometry, we observed a significantly higher mean fluorescence intensity (MFI) of bacteria in the grna.2⁺ macrophages, suggesting weaker bactericidal/bacteriostatic capacity compared to that of grna.2^- macrophages (Fig. 6 I–M and SI Appendix, Fig. S19).

GRN Exerted an Anti-Inflammation Function in Human Mycobacterial Granulomas.

To further investigate whether granulin plays a similar function in human tuberculosis patients, we first examined GRN protein expression in their granulomas and found it to be abundantly expressed (Fig. 7A). Additionally, GRN mRNA expression levels were analyzed using publicly available scRNA-seq data from human tuberculosis granulomas (Fig. 7B) (82). In that study, PET-CT scans were employed to locate active lesions with inflammation by tracking regions with increased glucose uptake, using ¹⁸F-labeled fluorodeoxyglucose (¹⁸F-FDG) as a tracer. GRN mRNA expression was significantly higher in patients with low pulmonary inflammation (PET-CT_low) compared to those with high pulmonary inflammation (PET-CT_high) (Fig. 7C) (82). We further examined GRN mRNA expression across different macrophage subpopulations and found that GRN mRNA is highly expressed in subpopulations 0, 4, and 5 (Fig. 7 D and E). Notably, the subpopulations with high GRN expression were significantly more prevalent in patients with low pulmonary inflammation compared to those with high pulmonary inflammation (Fig. 7 F and G), which is consistent with our findings in zebrafish.

Fig. 7.

GRN exerted an anti-inflammation function in human mycobacterial granulomas. (A) Representative images of GRN expression in lung granulomas from a tuberculosis patient. The Left panel shows the overall granuloma structure, while the Right panel provides an enlarged view of the red boxed region. Brown spots indicate GRN positive signals, and blue spots indicate the nucleus. GRN-positive signals are predominantly located at the periphery of the granuloma. (B) Schematic diagram of TB patients with PET-CT and scRNA-seq of granulomas (82). (C) Expression level of GRN mRNA in the lung of tuberculosis patients with high ¹⁸F-FDG-avidity and low ¹⁸F-FDG-avidity. Data are sourced from ref. 82. PKM, Per Kilobase of transcript per Million mapped reads. (D) UMAP plot of scRNA-seq data of all macrophages from publicly available scRNA-seq data about human tuberculosis granulomas. (E) Expression map of GRN in macrophages. The red dashed line indicates the area of macrophages with high expression of GRN. (F) UMAP plot of scRNA-seq data of all macrophages in PET-CT_high and PET-CT_low groups. The black dashed line indicates the area of macrophages with high expression of GRN. (G) The GRN high macrophage ratios in PET-CT_high (n = 5) and PET-CT_low groups (n = 6). (C) analyzed with the unpaired two-tailed t test, ***P < 0.001. [Scale bar: 50 μm (A).]

Discussion

In this study, we used scRNA-seq technology to compare the immune microenvironment of granulomas before and after mycobacterial infection. A subset of grna.2⁺ macrophages was identified in M.m-infected zebrafish, and their depletion via NTR-MTZ system resulted in shorter survival times, higher bacterial loads, and more granulomas, especially necrotic granulomas, demonstrating that grna.2⁺ macrophages play a role in controlling mycobacterial infections and their associated pathology. Furthermore, the presence of grna.2⁺ macrophages is associated with a looser granuloma structure containing more T cells than granulomas depleted of grna.2⁺ macrophages (SI Appendix, Fig. S20). Additionally, using ex vivo granuloma culture techniques, we showed that grna.2⁺ macrophages can develop from grna.2⁻ cells. Although the depletion of grna.2⁺ macrophages increased proinflammatory cytokines and promoted granuloma development, it did not significantly affect early mycobacterial proliferation. RNA-seq and flow cytometry analyses confirmed that depletion of grna.2⁺ macrophages created a hyperinflammatory microenvironment in the kidney, increasing macrophage lytic cell death and potentially enhancing mycobacterial spread (SI Appendix, Fig. S20). The anti-inflammatory role of granulin was also confirmed in tuberculosis patients.

The immune microenvironment of granulomas is crucial for the host’s defense against M.tb infection (16, 17, 20, 83, 84). However, most studies that have used scRNA-seq focused on granuloma heterogeneity postformation, with limited exploration of dynamic cell-type changes during granuloma development. The current study found grna.2⁺ macrophages emerging mainly after infection. However, because grna.2 is expressed in the majority of macrophages postinfection, it is not suitable for distinguishing macrophage subpopulations, possibly explaining why this specific subset was previously overlooked.

Macrophages play an important but complex role in shaping the immune microenvironment within tuberculous granulomas. They can kill mycobacteria (4, 10, 85, 86) but also provide a niche for mycobacteria to survive, potentially facilitating the spread of infection to other tissues (2, 4, 10, 86). Advances with scRNA-seq have revealed deeper insights into the heterogeneity and functions of macrophages during mycobacterial infections. In zebrafish, for example, a subpopulation of epithelioid macrophages producing a type 2 inflammatory immune response was associated with granuloma necrosis (16). An IFN-responsive macrophage subset (CD163⁺MRC1^low) was identified in macaques with pulmonary TB (18). Another macrophage subtype, also identified in M.tb infected macaques, was associated with high bacterial loads and upregulated interferon related genes (e.g., NFKBIA), proinflammatory related genes (e.g., IL-1β), and complement activation genes (e.g., C1QA) (20). In bronchoalveolar lavage fluid from latent TB patients, a macrophage subset was identified that secretes chemokines to recruit CD8 T cells and enhance innate and adaptive immunity crosstalk (21). By using the Tg(mpeg1-LRLG) transgenic line to label macrophages, we identified a subset of grna.2⁺ macrophages and investigated their behavior and function in vivo. Although mpeg1 RNA was reported as significantly downregulated following M.m infection (87), we observed that it still effectively labeled macrophages in our experiment, likely due to the long half-life of the DsRed2 protein that retains fluorescence for 1 to 2 wk (43, 44). However, future studies using short half-life proteins, such as eGFP linked to the mpeg1 promoter, will require careful interpretation. This subset of grna.2⁺ macrophages appeared capable of suppressing excessive inflammatory responses and decreasing macrophage lytic cell death and tissue damage, while simultaneously promoting T cell infiltration and controlling Mycobacterium proliferation in vivo. Similarly, a previous study on pulmonary granulomas in TB patients found high GRN mRNA expression in low-inflammation regions showing low PET-CT activity, suggesting that GRN also reduces inflammation in tuberculosis patients (82).

As observed in previous studies, we found that the spatial distribution of immune cells in granulomas is related to the host’s resistance to mycobacterial infection (78, 79). Depletion of grna.2⁺ macrophages resulted in granulomas with a denser distribution of the remaining macrophages and significantly fewer T lymphocytes, suggesting that grna.2⁺ macrophages help maintain a granuloma cellular structure conducive to the entry of lymphocytes into the core. The arrangement of other immune cells also influences granuloma’s function. Notably, grna.2⁺ macrophages depletion did not affect the epithelial structure of granulomas, which is consistent with bioinformatics analysis and antibody staining results showing that epithelioid cells do not express grna.2, thus ruling out epithelial layer disruption as a cause of M.m proliferation and dissemination. In tumor immunology, tumors with T cell infiltration are commonly referred to as “hot tumors,” while those lacking such infiltration are termed “cold tumors” (88–90). Similarly, we define granulomas with T cell infiltration as “hot granulomas,” while granulomas lacking T cell infiltration are referred to as “cold granulomas,” with granulins potentially regulating this transition. B cell distribution in granulomas was associated with necrosis (78), and appears to inhibit T cell activation (79). Thus, host genes or mycobacterial virulence factors may regulate the distribution of immune cells within granulomas, providing new insights into the host’s responses to mycobacterial infections.

To further investigate how grna.2⁺ macrophages mediate host protective immunity and their origin, we used an ex vivo granuloma culture system. Using fluorescently labeled bacteria and host cells, we dynamically observed bacteria–host interactions. During the ex vivo experiments, DsRed2-labeled M.m fluorescence intensity increased twofold to eightfold within 6 h postdissection, exceeding the typical M.m proliferation time of 4 to 6 h (91, 92). This is likely because granulomas in vivo exist in a hypoxic environment, slowing the protein’s proper folding and maturation required for fluorescence (93, 94). After dissection, exposure to an oxygen-rich environment significantly accelerates this process. Therefore, fluorescence-based quantification for ex vivo experiments requires a preculture period in normal oxygen conditions to ensure accurate measurements.

Macrophage cell death plays a crucial role in controlling mycobacterial infections (3, 4, 8, 10, 95–97). In zebrafish depleted for grna.2⁺ macrophages, early bacterial loads were unchanged despite increased proinflammatory cytokines and granuloma formation. While proinflammatory cytokines generally enhance macrophage-mediated bacterial killing, flow cytometry analysis showed that grna.2⁻ macrophages harbored significantly fewer bacteria than their grna.2⁺ counterparts during early infection (Fig. 6 I–M). However, the elevated cytokines levels secreted by lacking grna.2⁺ macrophages likely caused cell death and tissue damage, balancing their increased bactericidal activity. At later infection stages, depletion of grna.2⁺ macrophages induces a hyperinflammatory microenvironment, accelerating tissue damage and disease progression. Although we did not observe any notable upregulation of phagocytosis-related genes in these macrophages, the potential contribution of enhanced phagocytosis in grna.2⁺ macrophages cannot be entirely ruled out and warrants further investigation to clarify its role. TNFα is among the proinflammatory cytokines whose expression is significantly elevated following depletion of grna.2⁺ macrophages. TNFα is known to play a versatile role in the antituberculosis immune response; insufficient TNFα expression fails to control mycobacteria, while excessive TNFα leads to macrophage necrosis and mycobacterial dissemination (98–103). Although TNF is not essential for granuloma formation, it can influence granuloma structure (104–106), as shown by the altered granuloma architecture and cell composition after anti-TNF treatment with Infliximab or Adalimumab (107, 108). After TNF neutralization, about 20% granulomas in mice lack epithelioid cells (104). Cellular infiltrates in TNF^−/− mice were poorly formed, with extensive regions of necrosis and neutrophilic infiltration of the alveoli (105). Our findings highlight the role of grna.2⁺ macrophages in balancing immunity and minimizing tissue damage during infection. Further studies are needed to explore whether GRN expression could serve as a clinical marker or a target for host-directed therapy (HDT) strategies in tuberculosis.

In our research, discrepancies between proportions detected by scRNA sequencing and flow cytometry are not uncommon. For example, a study on C57BL/6 mice under physiological conditions found a ratio of approximately 1:20 between AM and interstitial macrophages (IM) (109), while another flow cytometry-based study found a ratio of about 1:12 using the same mouse strain (110). Possible reasons for these discrepancies include differences in marker genes, flow cytometry antibodies, and transgenic labels; varying capture rates during single-cell library preparation, such as the lower capture rate for neutrophils in 10× sequencing platform (111); and the difference between RNA expression (scRNA-seq) and protein expression (flow cytometry).

This study had some limitations. In zebrafish, there are four grn homologous/paralogous genes. Although individual knock out of all four genes in juveniles individually and grna.2 in adults showed no phenotypes, we cannot rule out the possibility that knocking out the other three genes may show a phenotype in adults or that simultaneous knockout of multiple genes could result in a phenotype. Therefore, our current results do not provide insight into support grn as a marker gene induced by M.tb infection or as a gene involved in anti-inflammatory functions in humans. While this study focused on macrophages, further analysis could reveal subpopulations of other cell types influencing or responding to the formation of granuloma. Finally, although granulin-producing macrophages were shown to regulate immune response in mycobacterial infections, their role in other infections or in immune-related diseases remains to be explored.

Materials and Methods

Ethics Statement.

Informed written consent was obtained from all participants or their guardians before study inclusion. Participants were informed about the research purpose, procedures, risks, benefits, and their right to withdraw without consequences. The study was approved by the Research Ethics Committee of Shenzhen Third People’s Hospital, and lung biopsy sections from TB patients were also obtained with approval (2022-110). Zebrafish were maintained according to ethical guidelines (GB/T 35823-2018), with approval for experiments from the Animal Care and Use Committee of Shanghai Public Health and Clinical Center (2019-A016-01, 2022-A002-01, 2024-A20-01).

Bacterial Strains.

Single cell suspensions of M.m M strain expressing Wasabi, tdTomato, or tdKatushka2 fluorescent proteins were obtained as previously described (112). Briefly, bacteria were cultured in 7H9 medium supplemented with 10% OADC (Gene Optimal, GOMY0073), 0.5% glycerol, 0.04% Tyloxapol, and antibiotics.

Zebrafish Strain and Husbandry.

All zebrafish strains had AB wild-type backgrounds. All wild-type and Tg(mpeg1-LRLG) (59, 74–76, 113), Tg(mfap4-LRLG) (40–42) and Tg(mpeg1-LRLG; grna.2-GFP-NTR) (74–76) zebrafish were reared under standard conditions (114).

Zebrafish Infection and Bacterial Load Calculation.

Adult zebrafish were intraperitoneally injected with approximately 2,000 CFU of M.m, except for 5,000 CFU for survival experiments (38). Bacterial burden was determined by the colony counting method. Briefly, infected zebrafish were killed with tricaine, followed by treatment with 75% ethanol and PBS for 3 min each. Zebrafish were then homogenized in a tissue homogenizer (Shanghai Jingxin Experimental Technology). Homogenates were treated with 10% NaOH for 15 min, gradient diluted, and plated onto 7H10 agar plates (containing hygromycin/amphotericin B/polymyxin B/carbenicillin). Plates were incubated at 32 °C for approximately 10 d, after which M.m colonies were counted. As previously described, zebrafish embryos were infected via microinjection of ~250 CFU into the duct of Cuvier, and bacterial burden was determined either by quantified bacterial fluorescence intensity or colony counting method (114).

scRNA-seq Library Construction, Sequencing, and Data Preprocessing.

scRNA-seq libraries were prepared using Chromium Single Cell 3’ Reagent Kits v3.1 (10× Genomics, PN-1000121) following established protocols (21). Approximately 2 × 10⁶ single cells per sample were used to generate gel bead-in-emulsions, and cDNA/transcriptional libraries were processed as per 10× Genomics guidelines. Sequencing was performed on an Illumina NovaSeq6000 platform. Reads were aligned, filtered, and processed using the Cell Ranger (v5.0.1) against the Danio rerio reference genome (ensemble 99). High-quality cells with 200 to 7,500 detected genes and mitochondrial gene count below 15% were retained for downstream analysis.

Unsupervised Dimension Reduction and Clustering Analysis.

Briefly, gene expression data were normalized using the Seurat package (v4.0.3, https://satijalab.org/seurat/), including “NormalizeData,” “IntegrateData,” “FindVariableFeatures,” “RunPCA,” “RunUMAP,” “FindNeighbors,” and “FindClusters” functions.

Identification and Annotation of Cell Clusters with Gene Expression Signature Analysis.

Marker genes for each cluster were identified with the MAST algorithm in the Seurat “FindAllMarkers” with criteria of |log2FC| ≥ 0.585, p.adjust ≤ 0.05, and pct.1 > 0.25. Cell clusters were annotated using previously reported cell type–specific marker genes. Gene expression signature score was calculated using Seurat “AddModuleScore” function.

Functional Annotation Analysis.

DEGs were identified using GO and KEGG pathway enrichment analyses were performed using the clusterProfiler R package, with GO analysis covering biological processes (BP), molecular functions (MF), and cellular components (CC).

Cell Depletion.

For grna.2⁺ macrophage cell depletion, adult Tg(mpeg1-LRLG; grna.2-GFP-NTR) zebrafish were immersed in 5 mM MTZ (Sigma, M3761), which can be converted into cytotoxic agents to crosslink with DNA by NTR, causing the death of NTR⁺ cells (74–76). After 24 h, zebrafish were washed with fresh system water. The interval between the MTZ treatments was 48 h.

RNA Extraction and Real-Time PCR.

Whole zebrafish embryos, kidneys, or granulomas from adults were homogenized through a 27G needle. cDNA was synthesized from 1 µg RNA using HiScript III RT SuperMix for qPCR (Vazyme). Real-time PCR was performed with Hieff UNICON Universal Blue qPCR SYBR Green Master Mix (Yeasen) on a CFX96 Real-time system instrument (Bio-Rad).

Immunohistochemistry.

Paraffin-embedded tissue sections were dehydrated, rehydrated, and boiled in EDTA antigen retrieval solution, followed by PBS washes. Endogenous peroxidase activity was blocked with a blocking agent for 10 min at room temperature. Primary antibody (Abcam Ab169325, 1:500 for Granulin) was incubated overnight at 4 °C, followed by goat anti-mouse anti-rabbit IgG/IgM H&L (HRP polymer) (Abcam Ab2891, 100 μL) incubation at room temperature for 15 min. DAB chromogen solution was applied for 5 min, rinsed with tap water, and counterstained with hematoxylin for 30 s. Sections were then dehydrated, cleared, and mounted with neutral resin.

Construction of Gene Knockout Zebrafish.

Guide RNAs (gRNAs) targeting grna, grnb, grna.1, and grna.2 were designed using CRISPRscan (https://www.crisprscan.org/?page=gene). gRNAs and Cas9 protein (Novoprotein, E365-01A) were microinjected into zebrafish embryos at single cell stage (16). Genomic DNA was extracted at 24 hours post fertilization (hpf), and PCR was used to amplify the CRISPR site and the mutations were confirmed with Sanger sequencing. Details of gRNA and primer sequences are provided in Dataset S7.

Survival Curve.

The death of zebrafish was recorded until all zebrafish succumbed, as previously described (38, 115).

Histology of Zebrafish Adults and HE Staining.

Infected adult zebrafish were killed with tricaine, with their scales and fins removed, and immersed in 4% paraformaldehyde for at least 72 h. Paraffin embedding, serial sectioning (4 to 5 µm), and HE staining were performed as previously described (38). The number of granulomas was counted from the largest cross-sectional slide cut from each zebrafish. After staining, slides were scanned, and granulomas were counted using ImageViewerG software. The analysis included early infiltrative lesions, nonnecrotized granulomas, and necrotized granulomas. Then, the number of immune cells surrounding each granuloma and the total area of immune infiltration in each fish were analyzed using TissueFAX200 and HistoQuest software based on HE staining (SI Appendix, Fig. S11).

Preparation of Frozen Section, Antibody Staining.

Infected zebrafish were killed with tricaine at 14 dpi, and after removing scales and fins, they were fixed in 1% paraformaldehyde for 2 h at room temperature. After washing with PBS, samples were embedded in OCT and stored at −80 °C. Serial sections (15 to 20 µm) were prepared for antibody staining. Sections were rinsed with PBS and precooled acetone, followed by blocking for 1 to 2 h. Primary antibodies [Abcam Ab6658 (goat polyclonal antibody) for GFP (1:300), Takara 632496 (rabbit polyclonal antibody) for DsRed2 (mpeg1:LRLG) (1:500), BD Pharmingen 610181 (mouse anti-E-cadherin) for E-cadherin (1:100), and Sigma ECACC 92092321 (mouse monoclonal antibody) for 4C4 (1:500) (39)] were incubated overnight at 4 °C, followed by secondary antibody incubation [Abcam Ab150129 (donkey anti-goat IgG H&L) for Alexa Fluor 488 (1:500), Yeasen 34212ES60 (donkey anti-rabbit IgG H&L) for Alexa Fluor 594 (1:500), Abcam Ab150105 (donkey anti-mouse IgG H&L) for E-cadherin (1:200) and 4C4 (1:200)] at room temperature for 2 h. After antibody staining, DAPI staining was performed at room temperature for 10 min (1:5,000). Sections were sealed with an anti-fluorescence quencher (16).

Calculation of Macrophage Density in Granuloma.

To create a spatial distribution map of immune cells, we used Fiji software (v 2.0) to convert the antibody staining signals on frozen sections into coordinate information (x and y) (116). The midpoint of all coordinates was calculated (average x and y values) and normalized by dividing them at the midpoint. R software (v 4.3.1) was used to generate plots with standardized coordinates on a shared coordinate system and colors were added to the points based on their density, as previously described (38). We defined density values greater than 90% as “high-density values (HDV)” and counted their proportions in each granuloma.

Paraffin Section In Situ Hybridization.

Zebrafish paraffin sections were hybridized overnight at 65 °C with DIG-labeled lck probes in hybridization buffer. Sections were incubated in 4× saline-sodium citrate (SSC) and 50% formamide in 2× SSC at 65 °C for 30 min, followed by TNE (Tris-HCl/NaCl/EDTA) treatment at 37 °C for 10 min. Sections were washed with 2× SSC and 0.2× SSC at 65 °C for 15 min each, and then incubated overnight at 4 °C with anti-DIG-alkaline phosphatase antibody (Roche, 11093274910). Signal detection was performed using NBT/BCIP (Roche, 11697471001) for at least 30 min, stopping when the signals were clearly visible (38).

Live Imaging of Ex Vivo Granuloma.

After M.m-tdTomato infection, Tg(grna.2-GFP-NTR) zebrafish were killed at 14 dpi, and their abdomens dissected. Granulomas were extracted under a fluorescence microscope and placed on cavity slides with L15 medium overnight (77). After adding 2 mM MTZ, fluorescence changes in granulomas were continuously observed using a Leica SP8 confocal microscope in live data mode. A control group without MTZ was treated similarly.

Flow Cytometry Sorting of Macrophages.

Zebrafish infected with Katushka-labeled M.m were intermittently immersed in 5 mM MTZ every 3 d for 24 h. At 14 dpi, zebrafish were killed, kidneys were dissected, and single-cell dissociation was performed using 1× TrypLE. After centrifugation and filtration, flow cytometry assessed differences in macrophage proportions. Suspensions from Tg(mpeg1-LRLG; grna.2-GFP-NTR) zebrafish before and after MTZ treatment were stained with 7-AAD dye for 10 min to compare the dead macrophage proportions. Cell gating was performed using FlowJo software (117, 118), with macrophages gated based on DsRed2 (mpeg1:LRLG) expression and further sorted into grna.2⁺ and grna.2⁻ populations. Katushka⁺ macrophages were analyzed for MFI at 7 dpi, comparing grna.2⁺ and grna.2⁻ macrophages from MTZ-treated and untreated Tg(mpeg1-LRLG; grna.2-GFP-NTR) zebrafish.

Processing and Analysis of RNA-seq Sequencing Data.

Briefly, RNA quality control was done with RseQC (119), and reads were aligned to the GRCz11 zebrafish reference genome (120) using HISAT2. The quantification of gene expression was obtained with HTSeq-count (121). PCA was conducted using R (v3.2.0). DEGs were identified using DESeq2 (q < 0.05, foldchange > 2, or foldchange < 0.5). DEG clustering and pathway enrichment were performed using GO (122) and KEGG (123) databases, with significant results visualized in R. Protein interaction networks were generated using the STRING database (https://string-db.org).

Statistics Analysis.

Statistical analysis was performed with GraphPad Prism software. Data are presented as the mean ± SEM. Comparisons between two groups were made using an unpaired two-tailed t test, while differences among multiple groups were assessed with a One-Way ANOVA. Statistical significance is indicated in the tables and figures as follows: ns = nonsignificant (P > 0.05); *P < 0.05; **P < 0.01; ***P < 0.001.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Live imaging observing grna.2+ macrophages depletion and M.m growth ex vivo. Granulomas were removed from Tg(grna.2-GFP-NTR) zebrafish infected with M.m-tdTomato. Time-lapse images were taken at 10-minute intervals.

Movie S2.

Live imaging observing grna.2+ macrophages regeneration and M.m growth ex vivo. Granulomas were removed from Tg(grna.2-GFP-NTR) zebrafish infected with M.m-tdTomato. Time-lapse images were taken at 10-minute intervals.

Data, Materials, and Software Availability

The raw sequence data reported in this study have been deposited in the Genome Sequence Archive (124) in the National Genomics Data Center (125), China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (BioProject Accession: PRJCA027270) and are publicly accessible at https://ngdc.cncb.ac.cn/search/specific?db=bioproject&q=PRJCA027270 (126).