Establishment of a biosafe murine model of skeletal tuberculosis using Mycobacterium smegmatis

Abstract

Background

Skeletal tuberculosis (TB) remains a persistent clinical and research challenge due to its chronic course, osteolytic destruction, and the limitations of existing animal models, which often require high‐level biosafety containment or fail to replicate human skeletal pathology.

Methods

This study developed a biosafe, accessible, and versatile murine model of skeletal TB using Mycobacterium smegmatis, a fast‐growing, nonpathogenic mycobacterial species with high genomic homology to Mycobacterium tuberculosis. Three infection routes—subperiosteal calvarial injection, intratibial injection, and intracardiac inoculation—were systematically evaluated for their ability to induce localized versus disseminated bone infection under standard biosafety level (BSL)‐1 conditions.

Results

Subperiosteal calvarial and intratibial injection of M. smegmatis induced localized bone lesions characterized by osteolysis, sequestrum formation, granulomatous inflammation, and increased osteoclast activity. Intratibial infection additionally triggered compartment‐specific immune responses, including neutrophil and macrophage expansion, transient B‐cell depletion, and activation of interferon‐γ⁺ (IFN‐γ⁺) T cells, reflecting active immune remodeling at the infection site. Systemic dissemination via intracardiac injection reproducibly generated progressive vertebral and tibial bone destruction with organized granuloma formation and immune cell infiltration but without prominent sequestrum formation. Compared to intratibial infection, intracardiac delivery exhibited lower intragroup variability and more closely recapitulated the diffuse progression of extrapulmonary skeletal tuberculosis.

Conclusions

This M. smegmatis–based murine model provides a straightforward, reliable, and immunopathologically relevant platform for exploring host–pathogen dynamics, immune‐driven bone destruction, and early‐stage therapeutic testing in skeletal TB, all within standard BSL‐1 laboratories. This model fills a critical gap by enabling BSL‐1 research into skeletal TB mechanisms and drug development.

This study developed a biosafe, accessible, and versatile murine model of bone TB using Mycobacterium smegmatis, a fast‐growing, nonpathogenic mycobacterial species with high genomic homology to Mycobacterium tuberculosis. Three infection routes—subperiosteal calvarial injection, intratibial injection, and intracardiac (left ventricular) inoculation—were systematically evaluated for their ability to induce localized versus disseminated bone infection under standard biosafety level (BSL)‐1 conditions.

1. INTRODUCTION

Tuberculosis (TB) remains a formidable global health challenge. ¹ , ² According to the WHO Global Tuberculosis Report, there were 10.8 million newly diagnosed or recurrent cases globally in 2023, with China being the second‐highest burdened country, reporting nearly 748 000 new cases each year. ³ , ⁴ TB can spread beyond the lungs through hematogenous or lymphatic routes, resulting in extrapulmonary tuberculosis (EPTB). A recent nationwide study in China estimated the prevalence of EPTB at 24.6%. ⁵ It is noteworthy that musculoskeletal tuberculosis accounts for 15.8% of EPTB cases, often involving the vertebral bodies—a condition clinically referred to as Pott's disease. ⁶ , ⁷

Skeletal tuberculosis is characterized by progressive osteolytic destruction and typically presents with persistent bone pain, pathological fractures, and joint deformities, often resulting in chronic disability. ⁸ , ⁹ The vertebral bodies and metaphyseal regions of long bones provide an ideal environment for mycobacterial colonization due to their rich vascular supply and slow blood flow. ¹⁰ Upon invasion of the bone marrow by Mycobacterium tuberculosis, an inflammatory cascade is triggered, leading to immune cell recruitment and subsequent granuloma formation. ¹¹ , ¹² As the lesion advances, altered vascularization induces tissue ischemia, which is pathologically manifested by sequestrum formation and trabecular destruction. ¹³

Despite the availability of animal models for M. tuberculosis infection, most are tailored for pulmonary TB, offering limited insights into musculoskeletal pathology. ¹⁴ These models—particularly those using virulent strains like H37Rv—also raise biosafety concerns, requiring BSL‐3 containment and incurring substantial costs. ¹⁵ Consequently, there is a critical need for biosafe, accessible models that faithfully replicate the pathological features of human skeletal TB. Existing approaches have employed attenuated strains (e.g., BCG), virulent strains (e.g., H37Rv), and nontuberculous mycobacteria (e.g., Mycobacterium marinum), each with notable limitations. BCG, though BSL‐2 compliant and widely used in immunological studies, fails to simulate the extrapulmonary dissemination and osteolytic destruction characteristic of clinical skeletal TB. ¹⁶ H37Rv models better mimic osteolytic lesions but pose significant biosafety and cost barriers. ¹⁷ , ¹⁸ Meanwhile, M. marinum–based models—mainly in zebrafish—suffer from limited replication in mammalian macrophages and significant anatomical differences from humans, reducing their translational relevance. ¹⁹ In addition, a BSL‐2 lab is necessary for M. marinum culture. ²⁰ Together, these limitations underscore the pressing need for a more suitable skeletal TB animal model.

In light of the limitations of existing skeletal tuberculosis models, we established a novel murine model of skeletal TB using Mycobacterium smegmatis, a nonpathogenic, fast‐growing mycobacterial species. ²¹ , ²² M. smegmatis shares over 2000 homologous genes with M. tuberculosis and has been increasingly recognized as a valuable surrogate in TB research. ²¹ , ²³ With a short generation time and minimal biosafety requirements, M. smegmatis allows for rapid, cost‐effective experimentation without the need for high‐containment facilities. Moreover, M. smegmatis has already been widely used as a surrogate organism for evaluating antituberculosis drug leads targeting M. tuberculosis, further supporting its applicability in TB‐related modeling. ²⁴ , ²⁵ Leveraging these advantages, we successfully developed a biosafe, accessible murine model that recapitulates key features of skeletal TB pathology, providing a promising platform for future mechanistic and therapeutic studies. ²⁶

Therefore, this study aims to establish a novel skeletal tuberculosis animal model using M. smegmatis, a low‐virulence strain with high genomic homology to M. tuberculosis. By employing three different inoculation routes—calvarial injection, intratibial injection, and intracardiac injection—we systematically evaluate the feasibility of using M. smegmatis to induce localized and disseminated skeletal TB lesions. This model not only offers a biosafe and accessible alternative for studying skeletal TB pathogenesis but also lays the groundwork for future investigations into underlying mechanisms and antituberculosis drug development. Our findings are expected to facilitate preclinical research by providing a versatile and reproducible platform for studying skeletal tuberculosis.

2. MATERIALS AND METHODS

2.1. Tuberculosis culture

M. smegmatis ATCC700084 and M. marinum ATCC BAA‐535 were cultured in liquid medium Middlebrook 7H9 broth (Becton Dickinson, USA, 271310) enriched with glycerol (0.4%) (Sigma Aldrich, USA, P4780) and OADC (Becton Dickinson, USA, 211886) growth supplement, 37℃, 220 rpm. To further characterize the morphology of M. smegmatis, transmission electron microscopy (TEM) was performed. The ultrastructural image (Figure S2A) revealed the characteristic rod‐shaped morphology of M. smegmatis. In addition, we cultured M. smegmatis, M. marinum, and BCG under identical conditions to compare their growth dynamics. As shown in Figure S2B, the three mycobacterial species exhibited distinct growth rates.

2.2. Establishment of animal models

2.2.1. Intracranial injection model

C57BL/6 mice (6–8 weeks old) were anesthetized with isoflurane and positioned prone. After the calvaria was disinfected with 75% ethanol, the scalp was tented using ophthalmic forceps. A 31‐G insulin syringe (8‐mm needle) was used to slowly inject 100‐μL phosphate‐buffered saline (PBS) containing 1 × 10⁷ colony‐forming units (CFU) M. smegmatis at 1 mm anterior to the coronal suture along the midline (control group received equal volume of sterile PBS). Successful injection was confirmed by wheal formation. Animals were monitored for 24 h postprocedure and euthanized at 2 weeks postinfection.

2.2.2. Intratibial injection model

After induction of standard anesthesia and disinfection, the knee was flexed. A 27‐G needle created a cortical breach at the anteromedial tibial plateau. Through this access, a 31‐G insulin syringe delivered 50 μL of bacterial suspension (1 × 10⁷ CFU in PBS; control group received 100‐μL PBS). Postoperative monitoring lasted 24 h with euthanasia at 2 and 4 weeks postinjection.

2.2.3. Left ventricular injection model

Supine‐positioned mice underwent cardiac puncture guided by anatomical landmarks. The midpoint between suprasternal notch and xiphoid process was determined using vernier calipers. Inserting a 31‐G insulin syringe (0.3 mL) at 30° angle to the right of this midpoint, left ventricular entry was confirmed by pulsatile blood return. The 100‐μL inoculum (3 × 10⁷ CFU in PBS; control: PBS only) was administered within 2 s, with position reconfirmed by final blood aspiration. Euthanasia time points were 2, 4, and 8 weeks postprocedure with 24 h acute monitoring.

2.3. Micro–computed tomography and histomorphometric analyses

Micro–computed tomography (microCT) analysis was conducted according to established protocols. ²⁷ After overnight fixation in 4% paraformaldehyde, harvested murine femurs, tibiae, and vertebral specimens underwent high‐resolution scanning using the Bruker SkyScan 1272 microCT system (Kontich, Belgium). Scanning parameters were optimized as follows: isotropic voxel resolution 7 μm, X‐ray source voltage 60 kV, current 166 μA, and exposure duration 1700 ms per projection.

For trabecular bone characterization in the distal femur, a standardized 4.0‐mm segment extending proximally from a reference point 1.0 mm above the growth plate was demarcated as the volumetric region of interest (VOI). Three‐dimensional reconstructions were generated using CTvox software (version 3.0.0), whereas quantitative morphometric analyses were performed using CTAn software (version 1.15.4.0). Evaluated parameters encompassed key morphometric indices: bone mineral density (BMD, mg HA/cm³); bone volume to total volume fraction (BV/TV, %); trabecular thickness (Tb.Th, μm); trabecular number (Tb.N, mm⁻¹); trabecular separation (Tb.Sp, μm).

2.4. Histopathological and immunohistochemical processing

Tissue specimens underwent 48‐h fixation in 4% paraformaldehyde (w/v). Femoral and tibial samples were subsequently decalcified in 10% ethylenediaminetetraacetic acid (EDTA, pH 7.4) with daily solution renewal for 7 days. Decalcified bones along with lung/spleen tissues were paraffin‐embedded and sectioned along sagittal or coronal planes at 5 μm thickness using a rotary microtome. Hematoxylin and eosin (H&E) staining was performed on both visceral organ (lung, spleen) and skeletal tissue (femur, tibia) sections. Tartrate‐resistant acid phosphatase (TRAP) staining was specifically conducted on bone sections for osteoclast visualization. All the quantitative analysis were performed using image J.

2.5. Enzyme‐linked immunosorbent assay

Levels of interleukin 6 (IL‐6) and IL‐1β in mouse bone marrow lavage were measured using one‐step enzyme‐linked immunosorbent assay (ELISA) kits (IL‐6: S0C3019; IL‐1β: S0C3029; Starter, China) following the manufacturer's instructions. Optical density was recorded at 450 nm using a microplate reader, and cytokine concentrations were calculated from standard curves. All samples were assayed in duplicate.

2.6. Flow cytometric analysis

Flow cytometric analysis was performed on immune cells isolated from spleen and bone marrow following experimental interventions in euthanized mice. Spleens were mechanically dissociated using syringe plungers and filtered through 40‐μm nylon mesh (Falcon, Corning, USA), whereas bone marrow cells were harvested by flushing femoral and tibial cavities with ice‐cold FACS buffer (PBS supplemented with 2% fetal bovine serum [FBS] and 1 mmol/L EDTA) until complete marrow evacuation. Cell suspensions were subjected to erythrocyte lysis using BD Pharm Lyse buffer (BD Biosciences, San Jose, CA, USA) with a 2‐min incubation at room temperature followed by two washes in PBS (300× g, 5 min, 4℃) and resuspension in FACS buffer. Prior to staining, Fc receptors were blocked using anti‐CD16/32 (Cat. no 553141, BD; 1:100 dilution, 10 min, room temperature). Cells were then incubated with fluorochrome‐conjugated antibodies (anti‐mouse Ly‐6G, FITC, Cat. no 551460, BD; anti‐mouse CD11b, Cat. no 553312, BD, 1:100 dilution; anti‐mouse CD19, Cat. no 557398, BD, 1:100 dilution; anti‐mouse B220, Cat. no 553092, BD, 1:100 dilution; anti‐mouse CD3, Cat. no 100307, Biolegend; anti‐mouse CD4, Cat. no 100515, Biolegend, 1:100 dilution; anti‐mouse CD8, Cat. no 553030, BD; anti‐mouse IFN‐γ, Cat. no 505829, Biolegend; anti‐mouse CD3, Cat. no 557235, BD, 20 min, room temperature) for 30 min at 4℃ in the dark. Samples were acquired using a BD LSRFortessa X‐20 flow cytometer (BD Biosciences) equipped with a 5‐laser configuration. Data were analyzed with FlowJo version 10.8.1 (Tree Star, Ashland, OR, USA) using doublet exclusion and live cell gating strategies.

2.7. Statistical analysis

Statistical analyses were performed using GraphPad Prism version 9. Two‐tailed Student's t‐tests compared two groups, whereas one‐way or two‐way analysis of variance (ANOVA) evaluated multiple groups. Pearson's correlation test was used for correlation analysis, which was assessed by linear regression F‐test. A significance threshold of p < 0.05 was used to indicate statistical significance.

3. RESULTS

3.1. Establishment of a localized calvarial bone infection model using M. smegmatis

To explore the feasibility of M. smegmatis as a pathogen for localized bone infection, we established a murine calvarial injection model (Figure 1A). M. smegmatis (1 × 10⁷ CFU in 100 μL PBS) was injected subperiosteally at the midline of the calvaria, anterior to the coronal suture. MicroCT performed at 2 weeks postinfection revealed multiple bone defects in the parietal bones of infected mice, which were absent in control animals (Figure 1B). Interestingly, despite the presence of focal osteolytic lesions, quantitative analysis showed a paradoxical increase in overall BMD and BV/TV compared to controls (Figure 1C). This finding agrees with the sagittal views of microCT in Figure 1B, which demonstrate evident sequestrum formation in the M. smegmatis–infected group.

FIGURE 1.

Localized calvarial bone infection induced by Mycobacterium smegmatis. (A) Schematic of subperiosteal injection of M. smegmatis into the murine calvaria. (B, C) Micro–computed tomography (CT) images at 2 weeks postinfection showing bone destruction and increased bone volume to total volume fraction (BV/TV) in infected mice; both three‐dimensional (3D) reconstruction of the calvaria and sagittal views are presented to illustrate the extent of bone lesions and sequestrum formation; quantification is shown in the right panel. (D, E) Hematoxylin and eosin (H&E) staining reveals inflammatory infiltration and granuloma‐like structures, scale bar: 0.500 mm; quantification is shown in the right panel. (F, G) Tartrate‐resistant acid phosphatase (TRAP) staining shows increased osteoclast numbers in infected calvariae, scale bar: 0.500 mm; quantification is shown in the right panel.

To elucidate the underlying pathology, H&E staining was conducted. Histological examination demonstrated extensive inflammatory infiltration and formation of granuloma‐like structures at the infection site (Figure 1D); in addition, sequestrum formation was observed at the infection site. Moreover, quantitative analysis of H&E sections using ImageJ showed a significant increase in BV/TV, consistent with the microCT results (Figure 1E). The increased BV observed on microCT and H&E was attributed to the formation of sequestrum (sequestrum), a hallmark of chronic bone infection.

Furthermore, TRAP staining revealed a significant increase in osteoclast numbers in infected calvariae compared to controls (Figure 1F,G), indicating enhanced osteoclastic activity and bone resorption upon M. smegmatis infection.

Together, these findings demonstrate that subperiosteal injection of M. smegmatis into the calvaria induces localized bone infection characterized by osteolysis, sequestrum formation, granulomatous inflammation, and increased osteoclastogenesis—recapitulating key features of human bone tuberculosis. Importantly, this model operates under BSL‐1 conditions and is highly accessible, providing a practical and safe platform for mechanistic studies (see Table S1 for a comparison with classical M. tuberculosis and BCG models).

3.2. Establishment of an intratibial bone infection model using M. smegmatis

To further assess the bone‐invasive potential of M. smegmatis, we established an intratibial injection model in mice (Figure 2A). A cortical breach was created at the tibial plateau, and M. smegmatis (1 × 10⁷ CFU in 50 μL PBS) was injected into the bone marrow cavity. MicroCT analysis performed at 2 and 4 weeks postinfection revealed progressive bone destruction at the injection site (Figure 2B). Compared to healthy controls, infected tibia displayed irregular trabecular architecture, cortical thinning, and expanding osteolytic lesions; however, further in‐depth analysis of the microCT data revealed marked sequestrum formation at the site of infection (Figure 2C). Interestingly, despite the apparent bone damage, quantitative assessment showed an overall increased tendency in bone volume in the infected group (Figure 2D), which may be attributable to the accumulation of sequestrum tissue. However, due to high intragroup variability, this increase did not reach statistical significance.

FIGURE 2.

Intratibial bone infection model induced by Mycobacterium smegmatis. (A) Schematic illustration of intratibial injection of M. smegmatis into the murine tibia following cortical breach at the tibial plateau. (B–D) Micro–computed tomography (CT) images at 2 and 4 weeks postinfection showing progressive bone destruction in infected tibiae; both three‐dimensional (3D) reconstructions and cross‐sectional views are presented to highlight irregular trabecular structure, cortical thinning, and expanding osteolytic lesions. Quantifications are shown in the lower panel.

Consistent with the microCT findings, histological analysis using H&E staining also revealed similar pathological features (Figure 3A). At 2 weeks postinfection, extensive areas of sequestrum were observed, characterized by empty lacunae and loss of osteocyte nuclei. By 4 weeks, the sequestrum had undergone partial resorption, although remnants of devitalized bone were still detectable, indicating ongoing bone remodeling and inflammatory responses. Histological analysis also did not reveal statistically significant differences in BV/TV among the three groups, which may be attributed to the formation of sequestrum resulting in large intragroup variability (Figure 3C). TRAP staining further confirmed the presence of abundant osteoclasts within the infected tibiae, indicating heightened bone resorptive activity (Figure 3B,D).

FIGURE 3.

Intratibial bone infection model induced by Mycobacterium smegmatis. (A) Hematoxylin and eosin (H&E) staining demonstrates inflammatory infiltration and bone marrow remodeling in infected tibiae, scale bar: Left 0.500 mm, right 0.100 mm; (C) H&E staining quantification is shown in the right panel; (B) Tartrate‐resistant acid phosphatase (TRAP) staining reveals elevated osteoclast activity at sites of infection, scale bar: Left 0.500 mm, right 0.100 mm; (D) TRAP staining quantification is shown in the right panel. (E) ELISA analysis shows increased IL‐6 and IL‐1β levels in infected tibiae compared with controls, with further elevation from 2 to 4 weeks postinfection.

In line with these histopathological changes, analysis of bone marrow lavage fluid demonstrated a progressive increase in pro‐inflammatory cytokines. ELISA measurements revealed that both IL‐6 and IL‐1β levels were significantly elevated in infected tibiae compared to controls, and their concentrations further increased from 2 to 4 weeks postinfection (Figure 3E), highlighting sustained inflammatory activity within the infected bone microenvironment.

3.3. Immune responses induced by intratibial infection

To further characterize the immune alterations associated with intratibial M. smegmatis infection, we next examined immune cell dynamics in both bone marrow and spleen compartments (Figure 4). Flow cytometry analysis revealed a pronounced expansion of neutrophils (CD11b⁺Ly6G⁺) in the bone marrow throughout the infection course, with levels significantly elevated compared to controls at both 2 and 4 weeks, although no further increase was observed between these two time points. In the spleen, neutrophil proportions peaked at 2 weeks and subsequently declined by week 4 while remaining above baseline (Figure 4A). Macrophages (CD11b⁺F4/80⁺) displayed a different trend: in the bone marrow, their numbers rose significantly during infection but partially decreased at 4 weeks, though still above control levels; conversely, splenic macrophages exhibited a sustained reduction over time (Figure 4B). B cells (CD19⁺B220⁺) were markedly depleted in the bone marrow at 2 weeks, suggesting acute B‐cell exhaustion but recovered to baseline levels by week 4. A similar trend was observed in the spleen, although B‐cell levels at 4 weeks remained slightly below control values (Figure 4C).

FIGURE 4.

Immune cell alterations in the bone marrow and the spleen following intratibial infection with Mycobacterium smegmatis. (A) Representative flow cytometry plots and quantification of neutrophils (CD11b⁺Ly6G⁺) in the bone marrow and the spleen at 2 and 4 weeks postinfection. (B) Representative flow cytometry plots and quantification of macrophages (CD11b⁺F4/80⁺) in the bone marrow and the spleen. (C) Representative flow cytometry plots and quantification of B cells (CD19⁺B220⁺) in the bone marrow and the spleen. Data are shown as mean ± standard deviation (SD).

We next analyzed T‐cell responses with a particular focus on IFN‐γ production (Figure 5). In the bone marrow, both CD4⁺ and CD8⁺ T‐cell frequencies progressively declined with infection, whereas the proportion of IFN‐γ⁺ CD4⁺ and CD8⁺ T cells significantly increased in a time‐dependent manner, reflecting activation despite numerical contraction (Figure 5A,B). In the spleen, the response was less consistent: CD4⁺ T cells decreased at 2 weeks but partially recovered by week 4, whereas CD8⁺ T cells gradually increased over time. Notably, IFN‐γ⁺ CD4⁺ T cells in the spleen showed signs of exhaustion beginning at 2 weeks, whereas IFN‐γ⁺ CD8⁺ T cells rose at 2 weeks before declining at 4 weeks (Figure 5C,D). Given that this model represents a locally restricted bone infection, we considered the bone marrow data to more accurately reflect the immunological microenvironment at the site of infection.

FIGURE 5.

T‐cell dynamics and interferon γ (IFN‐γ) production during intratibial infection with Mycobacterium smegmatis. (A) Frequencies of CD4⁺ T cells and IFN‐γ⁺ CD4⁺ T cells in the bone marrow at 2 and 4 weeks postinfection. (B) Frequencies of CD8⁺ T cells and IFN‐γ⁺ CD8⁺ T cells in the bone marrow. (C) Quantifications of CD4⁺ T cells and IFN‐γ⁺ CD4⁺ T cells in the spleen. (D) Quantifications of CD8⁺ T cells and IFN‐γ⁺ CD8⁺ T cells in the spleen. Data are shown as mean ± standard deviation (SD).

Collectively, these findings indicate that intratibial M. smegmatis infection triggers a robust yet compartment‐specific immune response characterized by neutrophil and macrophage expansion, transient B‐cell depletion, and dynamic T‐cell remodeling with enhanced IFN‐γ production. These immune alterations contribute to the inflammatory and osteolytic pathology observed in this model.

However, we noted considerable intragroup variability in both radiological and histological outcomes within the intratibial model. MicroCT and H&E analyses revealed inconsistent degrees of bone destruction among mice. This variability may be attributed to the technical complexity of tibial injections, including challenges in achieving uniform bone marrow delivery and the potential for bacterial leakage into surrounding tissues during or after drilling. These limitations reduce the reproducibility and reliability of this model for studying skeletal TB. Moreover, the intratibial injection primarily induces a localized bone infection, which closely resembles acute osteomyelitis caused by pathogens such as Staphylococcus aureus, ²⁸ rather than the chronic, disseminated pathology typically observed in extrapulmonary tuberculosis. The resulting lesions are often abrupt and focal, with rapid bone necrosis and inflammation, which do not fully recapitulate the gradual and systemic nature of skeletal TB in humans. Therefore, to overcome these technical issues and better mimic the clinical course of extrapulmonary TB, we proceeded to establish a third model via intracardiac injection.

3.4. Systemic skeletal tuberculosis model via intracardiac injection of M. smegmatis

To establish a reproducible systemic model of bone tuberculosis, we performed left ventricular injections of M. smegmatis (3 × 10⁷ CFU in 100 μL PBS), allowing hematogenous dissemination of the pathogen (Figure 6A). Histological analysis of lung tissue revealed that M. smegmatis had successfully colonized the pulmonary parenchyma as early as 2 weeks postinfection, indicative of systemic dissemination (Figure S1A).

FIGURE 6.

Systemic bone infection model established via intracardiac injection of Mycobacterium smegmatis. (A) Schematic illustration of left ventricular injection procedure. (B–D) Representative micro–computed tomography (CT) images of vertebrae at 2, 4, and 8 weeks postinfection; three‐dimensional (3D) reconstructions and sagittal views are shown. Quantifications are provided in the lower panel.

MicroCT analysis revealed that no significant skeletal changes were observed at 2 weeks postinfection. However, for 4 and 8 weeks, mice developed severe vertebral bone destruction characterized by cortical erosion and trabecular rarefaction, with severity increasing over time (Figure 6B–D). Consistently, microCT assessment of the tibiae also demonstrated progressive bone loss, with statistically significant reduction in bone volume at 8 weeks postinfection compared to controls (Figure 7A–C). Interestingly, we did not observe prominent features of sequestrum formation in the tibiae of mice receiving intracardiac injection (Figure 7B). This contrasts with the intratibial injection model, where extensive areas of dead bone were evident. The absence of localized necrosis in the systemic dissemination model suggests that intracardiac injection may better recapitulate the gradual and diffuse progression of extrapulmonary tuberculosis, without inducing the acute, focal damage seen in direct bone inoculation.

FIGURE 7.

Micro–computed tomography (CT) evaluation of tibial bone changes following systemic Mycobacterium smegmatis infection. (A–C) Representative tibial microCT images at 2, 4, and 8 weeks postinfection; three‐dimensional (3D) and cross‐sectional views are shown. Quantifications are provided in the lower panel.

Histological examination of tibial sections revealed that although the marrow cavity appeared intact for 2 weeks, prominent granuloma formation was observed by weeks 4 and 8. These granulomas were composed of dense infiltrates of immune cells and visible clusters of M. smegmatis, indicating delayed but progressive invasion of the bone marrow (Figure 8A). Furthermore, ImageJ‐based quantification of bone mass demonstrated a progressive decline in BV/TV ratio with tuberculosis progression (Figure 8C). TRAP staining further revealed a significant increase in osteoclast activity at infected sites, supporting enhanced bone resorption associated with the infection (Figure 8B,D).

FIGURE 8.

Histological and osteoclast assessment of tibiae from systemically infected mice. (A) Hematoxylin and eosin (H&E) staining of tibial sections at 2, 4, and 8 weeks postinfection, scale bar: Left 0.500 mm, right 0.100 mm. (B) Tartrate‐resistant acid phosphatase (TRAP) staining of tibial sections from the same point, scale bar: Left 0.500 mm, right 0.100 mm. (C, D) Quantifications are provided on the right panel.

In addition to skeletal lesions, we next assessed systemic immune alterations induced by intracardiac M. smegmatis infection at 8 weeks. Flow cytometric analysis revealed profound remodeling of both myeloid and lymphoid compartments (Figures S3 and S4). Specifically, the proportion of CD11b⁺Ly6G⁺ neutrophils was significantly increased in the bone marrow but markedly decreased in the spleen, whereas CD11b⁺F4/80⁺ macrophages were consistently reduced in both bone marrow and spleen (Figure S3A–D). Similarly, CD19⁺B220⁺ B cells were reduced in both compartments at 8 weeks (Figure S3E,F). These alterations resemble the trends observed in the intratibial infection model, where neutrophil expansion in the bone marrow and macrophage/B‐cell depletion were also prominent, indicating that both localized and systemic infection converge on a shared pattern of myeloid dominance and adaptive lymphoid suppression.

We further examined T‐cell responses. In the bone marrow, CD4⁺ T‐cell frequency was elevated, whereas CD8⁺ T cells remained largely unchanged; nevertheless, both CD4⁺IFN‐γ⁺ and CD8⁺IFN‐γ⁺ subsets significantly increased at 8 weeks postinfection (Figure S4A,B). Comparable trends were observed in the spleen, where CD4⁺ T cells increased, CD8⁺ T cells slightly declined, and both IFN‐γ–producing CD4⁺ and CD8⁺ T cells markedly expanded (Figure S4C,D). This contrasts with the intratibial model, where T‐cell frequencies in the bone marrow declined despite enhanced IFN‐γ production, suggesting that focal infection drives local T‐cell contraction with functional activation, whereas systemic dissemination maintains T‐cell numbers while promoting sustained IFN‐γ responses.

Together, these findings demonstrate that intracardiac infection not only reproduces the skeletal pathology of tuberculosis, including vertebral and tibial bone destruction and granuloma formation in bone and lung tissues, but also induces systemic immune dysregulation characterized by bone marrow neutrophil expansion, macrophage/B‐cell depletion, and sustained IFN‐γ–producing T‐cell responses. Compared to the intratibial model, the intracardiac approach better mimics the hematogenous dissemination of extrapulmonary TB, exhibits lower intragroup variability, and provides a physiologically relevant platform for investigating host–pathogen interactions and evaluating therapeutic interventions in disseminated mycobacterial bone infections.

4. DISCUSSION

In this study, we establish M. smegmatis as a biosafe and effective surrogate pathogen for modeling human skeletal tuberculosis in mice. By employing three distinct inoculation routes—calvarial, intratibial, and intracardiac—we demonstrate that M. smegmatis repeat key pathological hallmarks of skeletal TB, including osteolytic destruction, granulomatous inflammation, sequestrum formation, and dynamic immune responses. Importantly, the intracardiac injection model emerged as a robust platform for studying systemic skeletal TB pathogenesis under standard biosafety conditions (BSL‐1), addressing critical limitations of existing models reliant on virulent M. tuberculosis strains or nonmammalian hosts.

A major advantage of the M. smegmatis model lies in its favorable biosafety profile. Unlike virulent M. tuberculosis, which requires BSL‐3 containment, M. smegmatis allows for large‐scale, cost‐effective preclinical studies within standard laboratory settings. ²⁰ , ²¹ , ²⁹ Although Mycobacterium bovis BCG, a BSL‐2‐compliant strain, has been previously used for TB research, it is primarily utilized in vaccine development and fails to mimic the chronic and granulomatous nature of skeletal TB. ³⁰ Indeed, BCG‐induced osteolytic lesions more closely resemble acute osteomyelitis caused by S. aureus. ³¹ Moreover, M. smegmatis shares over 2000 homologous genes ²¹ , ²³ with M. tuberculosis, including conserved virulence pathways involved in macrophage invasion and intracellular persistence. ¹⁹ This enhances its translational relevance when compared to more phylogenetically distant surrogates like M. marinum. ²¹ Taken together, these features position M. smegmatis as a promising and practical alternative for modeling skeletal TB.

Each inoculation route in this study offers distinct advantages and limitations, providing complementary perspectives on skeletal TB pathology. The calvarial injection model presents a technically simple and minimally invasive approach for inducing localized bone lesions. ³² Its high reproducibility and ease of operation make it suitable for short‐term studies of osteolytic mechanisms and granuloma formation. Unlike the intratibial model, it avoids cortical breaching, thus reducing the risk of bacterial leakage and interanimal variability. However, the calvaria lacks a true bone marrow niche, limiting its utility in studying immune dynamics within marrow compartments—a crucial aspect of TB pathogenesis. Additionally, this model does not mimic systemic dissemination or multisite skeletal involvement, which are characteristic of advanced TB cases.

In contrast, the intratibial injection method directly delivers bacteria into the marrow cavity, effectively simulating localized bone marrow infection. This approach led to rapid osteolysis and prominent sequestrum formation, with histopathological features closely resembling those observed in human TB. Nevertheless, the procedure requires cortical drilling, introducing technical variability and potential bacterial leakage into periosteal tissues, which may contribute to inconsistent lesion severity across animals. Furthermore, although intratibial infection triggers strong local immune responses—including neutrophil and macrophage expansion, transient B‐cell depletion, and functional T‐cell activation—it fails to capture hematogenous spread to distant skeletal sites, limiting its representation of systemic skeletal TB.

To address these limitations, we developed a systemic infection model via intracardiac injection, ³³ which achieved hematogenous dissemination of M. smegmatis. This approach bypassed the technical pitfalls of localized models, demonstrating high reproducibility and consistent progression of osteolytic lesions in both tibiae and vertebrae over time. The delayed onset of vertebral destruction (4–8 weeks postinfection) mirrors the insidious progression of human skeletal TB, where hematogenous seeding leads to preferential colonization of vascular‐rich metaphyseal and vertebral bone marrow. Importantly, this model circumvented the biosafety and logistical barriers of M. tuberculosis while retaining pathological fidelity, making it a reliable tool for preclinical research.

Our study also provides insight into the immune landscape associated with localized versus systemic skeletal TB. Intratibial infection induced local T‐cell contraction with increased IFN‐γ⁺ subsets, reflecting functional activation amid numerical decline, whereas intracardiac infection maintained or increased T‐cell frequencies with sustained IFN‐γ production, indicative of persistent systemic immune engagement. Both models exhibited bone marrow neutrophil expansion and macrophage/B‐cell depletion, paralleling observations in human bone TB where myeloid dominance and adaptive lymphoid suppression contribute to immune dysregulation. ³⁴ , ³⁵ These findings underscore the utility of the M. smegmatis model for dissecting compartment‐specific immune responses and their contribution to osteolytic pathology.

The M. smegmatis model is safe, reproducible, and effectively recapitulates key features of acute and subacute human skeletal TB. Nevertheless, it has important limitations. M. smegmatis grows faster and differs metabolically from M. tuberculosis, and it cannot establish latent infection or persist long‐term intracellularly, limiting its utility for studying TB reactivation, chronic progression, or drug tolerance. In addition, intracardiac injection requires technical precision, and extending observations beyond 8 weeks (e.g., to 10 weeks) led to high mortality due to pulmonary fibrosis and cachexia, raising ethical and reproducibility concerns. As for functional outcomes, such as biomechanical testing for fracture risk and long‐term bone remodeling, these were not assessed. Future studies could extend time points, include functional assays, and integrate complementary models to investigate chronic sequelae, evaluate drug efficacy, and explore immunomodulatory interventions, providing a more complete representation of human skeletal TB across disease stages.

5. CONCLUSION

In summary, the M. smegmatis–based murine model offers a versatile, biosafe, and clinically relevant platform for studying skeletal TB. By providing complementary local (calvarial and intratibial) and systemic (intracardiac) infection approaches, the model enables mechanistic studies of host–pathogen interactions, immune‐mediated bone destruction, and therapeutic interventions under BSL‐1 conditions. Future work should focus on validating responses to standard TB regimens, exploring immunomodulatory interventions, and integrating complementary models to investigate latency and chronic disease features. Collectively, this platform fills a critical gap in preclinical skeletal TB research and provides a robust foundation for translational studies.

AUTHOR CONTRIBUTIONS

Yewei Jia: Conceptualization; data curation; investigation; methodology; project administration; writing – original draft. Yuhuai Guo: Data curation; formal analysis; software; validation. Yusheng Yang: Conceptualization; project administration; resources; visualization. Jie Zhang: Formal analysis; validation; visualization. Ziyang Zhang: Data curation; validation. Ying Qu: Conceptualization; data curation; formal analysis. Jiulin Tan: Validation; visualization. Jie Shen: Investigation; project administration; visualization. Nathachit Limjunyawong: Resources; writing – review and editing. Jianzhong Xu: Funding acquisition. Zehua Zhang: Investigation; supervision; visualization. Fei Luo: Funding acquisition; supervision. Ce Dou: Funding acquisition; supervision; writing – original draft; writing – review and editing.

FUNDING INFORMATION

This work was funded by the National Natural Science Foundation of China (82172489, 82172449, 82103778), NSFC Key Projects of the Regional Innovation and Development Joint Fund (U23A20413), Fundings for Young Investigators of PLA (2022‐JCJQ‐QT‐004), Southwest Hospital Boqing Innovation Fund (2024BQCXJJ‐9), and the China Postdoctoral Science Foundation (grant number 2023M744280). The Southwest Hospital Postdoctoral Starting Fund (5175ZA36BP) also supported this work.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to declare.

ETHICS STATEMENT

All animal experiments were performed according to the requirements of Animal Welfare Act and the requirements of the Ethics Committee of Army Medical University (no. AMUWEC20223775).

Supporting information

FIGURE S1. Systemic dissemination. A. Hematoxylin and eosin (H&E) staining of lung tissue at 2 weeks postinfection showing pulmonary colonization by Mycobacterium smegmatis.

FIGURE S2. Morphology and growth of Mycobacterium smegmatis. (A) Scanning electron microscopy (SEM) morphology of Mycobacterium smegmatis (scale bar: 2 μm). Branching structures are indicated by arrows. (B) Growth kinetics of M. smegmatis, Mycobacterium marinum, and BCG under standard conditions. The optical density at 600 nm (OD600) was monitored, and the culture reached the stationary phase 24 h postinoculation.

FIGURE S3. Myeloid and B‐cell alterations following intracardiac infection with Mycobacterium smegmatis. (A) Frequencies of CD11b⁺Ly6G⁺ neutrophils in the bone marrow and the spleen at 8 weeks postinfection. (B) Quantification of CD11b⁺Ly6G⁺ neutrophils in the bone marrow and the spleen. (C) Frequencies of CD11b⁺F4/80⁺ macrophages in the bone marrow and the spleen. (D) Quantification of CD11b⁺F4/80⁺ macrophages in the bone marrow and the spleen. (E) Frequencies of CD19⁺B220⁺ B cells in the bone marrow and the spleen. (F) Quantification of CD19⁺B220⁺ B cells in the bone marrow and the spleen. Data are shown as mean ± standard deviation (SD).

FIGURE S4. T‐cell dynamics and interferon γ (IFN‐γ) production following intracardiac infection with Mycobacterium smegmatis. (A) Frequencies of CD4⁺ T cells, IFN‐γ⁺ CD4⁺ T cells, CD8⁺ T cells, and IFN‐γ⁺ CD8⁺ T cells in the bone marrow at 8 weeks postinfection. (B) Quantifications of CD4⁺ T cells, IFN‐γ⁺ CD4⁺ T cells, CD8⁺ T cells, and IFN‐γ⁺ CD8⁺ T cells in the bone marrow at 8 weeks postinfection. (C) Frequencies of CD4⁺ T cells, IFN‐γ⁺ CD4⁺ T cells, CD8⁺ T cells, and IFN‐γ⁺ CD8⁺ T cells in the spleen at 8 weeks postinfection. (D) Quantifications of CD4⁺ T cells, IFN‐γ⁺ CD4⁺ T cells, CD8⁺ T cells, and IFN‐γ⁺ CD8⁺ T cells in the spleen at 8 weeks postinfection. Data are shown as mean ± standard deviation (SD).

TABLE S1.

Jia Y, Guo Y, Yang Y, et al. Establishment of a biosafe murine model of skeletal tuberculosis using Mycobacterium smegmatis . Anim Models Exp Med. 2026;9:5‐20. doi: 10.1002/ame2.70101

DATA AVAILABILITY STATEMENT

All relevant data and details of resources can be found within the article and its Supporting Information.