The research, led by Shujie Zeng, Leyan Cao, Wenxuan Dong, Leping Li, Zihao Zhang and Peng Jin at institutions including Shandong Provincial Hospital, Shandong Cancer Hospital and Institute, and Zhongshan Hospital of Fudan University, focuses on a protein called semaphorin 4D, abbreviated SEMA4D. Semaphorins were originally discovered as guidance cues that steer growing nerve fibers to their correct destinations, but over the past two decades they have been recognized as versatile signaling molecules in the immune system and in cancer. SEMA4D in particular has attracted attention because high levels of the protein in tumor cells are associated with an immunosuppressive tumor microenvironment, meaning a milieu in which immune cells are present but rendered ineffective, or excluded from the tumor altogether.

What has remained poorly understood is how the conversation between tumor cells and macrophages regulates SEMA4D and, crucially, how the soluble, freely circulating form of the protein is generated inside living tumors. The new work addresses this gap by combining laboratory co-culture experiments with genetically engineered mouse models, allowing the researchers to trace the entire regulatory axis from tumor-derived SEMA4D to macrophage behavior to T cell function, and finally to tumor growth itself.

The team’s in vitro experiments used co-culture systems in which tumor cells and macrophages were grown together, alongside models in which SEMA4D production was either increased or shut down. These experiments revealed that tumor-derived SEMA4D pushes tumor-associated macrophages, the macrophages that populate tumors, toward the M2 phenotype. Macrophage polarization is a central concept in tumor immunology. M1 macrophages are classically activated, inflammatory cells that present antigens, produce pro-inflammatory signals and support anti-tumor immunity. M2 macrophages, by contrast, are associated with tissue repair, wound healing and, in the context of cancer, immunosuppression. A tumor microenvironment dominated by M2 macrophages is one in which cytotoxic CD8-positive T cells, the immune system’s primary tumor-killing weapons, are inhibited in both their infiltration into the tumor and their effector function once they arrive.

The study found precisely this pattern: tumor-derived SEMA4D drove M2 polarization of tumor-associated macrophages and suppressed the infiltration and function of CD8-positive T cells. But the most consequential finding concerned the mechanism by which SEMA4D becomes soluble. SEMA4D normally sits on the cell surface as a membrane-bound protein. For it to diffuse through the tumor microenvironment and act at a distance, it must be cleaved from the cell surface, and the identity of the molecular scissors responsible in tumors has been a key open question.

The answer, according to the new research, is matrix metalloproteinase 14, or MMP14, an enzyme produced by the tumor-associated macrophages themselves. MMP14 is a membrane-anchored member of the matrix metalloproteinase family, enzymes best known for remodeling the extracellular matrix, the protein scaffold that gives tissues their structure. The study demonstrates that TAM-derived MMP14 promotes the generation of soluble SEMA4D, thereby converting a membrane-bound tumor protein into a diffusible immunosuppressive signal. In other words, the macrophages recruited into the tumor are not passive bystanders; they are equipped with the enzymatic machinery that activates the tumor’s immune-evasion program.

To establish this mechanism in living animals rather than in culture dishes, the researchers used a macrophage-specific conditional knockout mouse model, in which the gene encoding MMP14 was deleted only in macrophages, leaving the enzyme intact in other cell types. This genetic precision matters, because MMP14 is expressed in many tissues and a global knockout would produce broad developmental and physiological effects that would obscure the specific role of the macrophage-derived enzyme in tumor immunity. With MMP14 removed from macrophages, the levels of soluble SEMA4D in the tumor microenvironment dropped significantly.

The consequences of that reduction were dramatic. Genetic ablation of MMP14 in tumor-associated macrophages reversed the M2 polarization of macrophages and promoted robust infiltration of both M1 macrophages and CD8-positive T cells into the tumor core, the innermost and most immunologically hostile region of the tumor. Immune exclusion, the physical barrier that keeps cytotoxic T cells at the tumor periphery, is one of the major reasons why immunotherapies such as immune checkpoint inhibitors fail in many patients. By dissolving this barrier at its enzymatic source, the MMP14 knockout effectively opened the tumor to immune attack, and the mice showed tumor regression as a result.

Perhaps the most convincing element of the study is the rescue experiment. To demonstrate that soluble SEMA4D is not merely correlated with the observed immune changes but is causally responsible for them, the researchers supplemented the MMP14 knockout mice with recombinant soluble SEMA4D, the purified protein produced in the laboratory. The result was unambiguous: exogenous soluble SEMA4D fully abrogated the anti-tumor benefits of MMP14 deletion, re-establishing the immunosuppressive tumor microenvironment despite the absence of the macrophage enzyme. This kind of loss-of-function and rescue design is the gold standard for establishing causal mechanisms in vivo, and it strongly supports the conclusion that the macrophage-MMP14-SEMA4D axis is a genuine driver of immune exclusion rather than a secondary correlate of tumor biology.

The therapeutic implications are considerable. Current immunotherapies, including checkpoint inhibitors targeting PD-1 and PD-L1, depend on the presence of functional T cells within the tumor. Patients whose tumors are immunologically cold, lacking T cell infiltration, rarely respond. The new findings suggest that targeting the macrophage-MMP14-SEMA4D regulatory axis could convert cold tumors into hot ones by removing the soluble signal that keeps T cells out, potentially sensitizing tumors to existing immunotherapies. SEMA4D-targeting agents are already being explored in clinical development for other indications, and the study provides a mechanistic rationale for combining such approaches with strategies that modulate macrophage behavior. The authors note that targeting this axis represents a promising therapeutic strategy for enhancing antitumor immunity. As with all preclinical mouse studies, translation to human cancers will require further validation, but the identification of a specific, druggable enzyme acting as the gatekeeper of an immunosuppressive signal offers a concrete molecular target in the ongoing effort to make immunotherapy work for more patients.

Subject of Research: The role of macrophage-derived MMP14 in generating soluble semaphorin 4D and shaping immunosuppression in the tumor microenvironment

Article Title: Macrophage-derived MMP14 fosters immunosuppressive tumor microenvironment by regulating soluble semaphorin 4D

Article References: Zeng, S., Cao, L., Dong, W., Li, L., Zhang, Z., & Jin, P. (2026). Macrophage-derived MMP14 fosters immunosuppressive tumor microenvironment by regulating soluble semaphorin 4D. Cell Death Discovery. https://doi.org/10.1038/s41420-026-03393-4

Image Credits: AI Generated

DOI: 10.1038/s41420-026-03393-4

Keywords: macrophages, MMP14, semaphorin 4D, tumor microenvironment, immunosuppression, CD8-positive T cells, M2 polarization, immune exclusion, cancer immunotherapy, tumor-associated macrophages, Cell Death Discovery, conditional knockout

Acute pancreatitis is a sudden inflammation of the pancreas that ranges from a brief, self-limiting illness to a catastrophic, multi-organ failure. Severe acute pancreatitis, or SAP, carries a substantial risk of death, and clinicians have long observed that patients with obesity fare markedly worse. Excess white adipose tissue is not a passive energy store; it is metabolically active and densely populated with macrophages, the scavenger immune cells that can either fuel inflammation or resolve it. In obesity, the fat depots surrounding and infiltrating the pancreas become a reservoir of inflammatory signaling, and leaked fatty acids can poison pancreatic tissue directly. Understanding why obesity amplifies pancreatic injury, and how to interrupt that process, has become a pressing question in gastroenterology and intensive care.

The Xiamen team approached the problem from an unusual angle: the microbial communities living in the mouth and the gut. The researchers first mined public microbiome datasets from patients with acute pancreatitis, comparing those with obesity to those without. The comparison revealed that the oral microbiota of obese patients with pancreatitis was distinctly enriched, with computational functional prediction pointing to heightened activity in ubiquinol biosynthesis and lipid-related metabolic pathways. In other words, the microbial ecosystems of obese and non-obese patients were not just different in composition but potentially different in what they were doing biochemically, setting the stage for the team’s animal experiments.

To test causality, the researchers turned to a mouse model of diet-induced obesity. One group of mice was fed a high-fat diet until they became obese, while controls ate a normal diet; both were then induced to develop severe acute pancreatitis. The results were stark. Obese mice suffered significantly worse pancreatic and lung injury than their lean counterparts, confirming that obesity actively aggravates the disease rather than merely correlating with it. Deep sequencing of fecal samples showed that obesity also reshaped the gut microbiota, reducing its alpha diversity, a standard measure of ecological richness that is frequently lower in inflammatory and metabolic disease. Meanwhile, metabolomic profiling of epididymal white adipose tissue revealed that the metabolite landscape inside the fat itself had shifted dramatically in obese animals with pancreatitis.

The pivotal step came when the team integrated the two datasets. Correlation analyses linking the gut microbiome to the adipose tissue metabolome uncovered meaningful associations between specific microbial patterns and specific fat-derived metabolites, suggesting that gut microbes and fat chemistry are in conversation during severe pancreatitis. This gut-fat axis, the researchers reasoned, might be a lever that could be pulled therapeutically. Enter metformin. When obese mice with severe pancreatitis were treated with the drug, the consequences were unmistakable: pancreatic and lung injury were reduced, and the animals’ inflammatory burden eased compared with untreated obese controls.

How was a diabetes drug calming a pancreatic storm? The answer, at least in part, lay in the immune cells embedded in fat. Using flow cytometry, the team showed that metformin promoted M2 polarization of adipose tissue macrophages, shifting these cells from a pro-inflammatory, tissue-damaging phenotype toward an anti-inflammatory, repair-oriented one. Immunohistochemistry backed this up at the tissue level: levels of interleukin-10, a signature anti-inflammatory cytokine produced by M2 macrophages, rose in both the pancreas and the white adipose tissue of treated animals. In effect, metformin appeared to recruit the fat’s own immune residents to the side of resolution rather than destruction.

The metabolomic data offered a concrete candidate for the messenger. Metformin treatment significantly altered the composition of the gut microbiota, though notably it did not restore alpha diversity, and it modified obesity-associated changes in fat metabolites that were tied to the microbiome. Two metabolites stood out as upregulated: hexanoylcarnitine, an acylcarnitine with microbial associations, and isomaltose, a disaccharide sugar. To determine whether hexanoylcarnitine was merely a bystander or an active player, the researchers moved to the culture dish. They modeled pancreatitis-like conditions in RAW264.7 macrophage cells by treating them with lipopolysaccharide, a bacterial toxin, together with palmitic acid, a saturated fatty acid abundant in obese adipose tissue. When they then added L-hexanoylcarnitine to these stressed cells, the macrophages shifted toward the M2 phenotype, showing increased expression of Pparg mRNA, a master transcriptional regulator of M2 identity, and elevated levels of CD206, a canonical M2 surface protein.

That single experiment ties the whole story together into a coherent, if still preliminary, mechanistic chain. Obesity disrupts the gut microbiota and rewrites the metabolite profile of white adipose tissue, worsening pancreatic and lung injury in severe pancreatitis. Metformin intervenes by further remodeling the microbial community, which raises levels of microbe-associated hexanoylcarnitine, and that metabolite in turn pushes adipose tissue macrophages into their anti-inflammatory M2 state, boosting interleukin-10 and damping tissue damage. It is a striking illustration of how a drug’s benefits can travel through an ecosystem rather than a single molecular target, and it reframes the fat depot not just as a victim of inflammation but as a modifiable immune organ.

The caveats deserve equal emphasis. The mouse findings come from a diet-induced model, and the human data were limited to reanalyzed public microbiome datasets rather than a prospective clinical trial of metformin in obese patients with pancreatitis. Macrophage behavior in cultured RAW264.7 cells, while informative, does not capture the full complexity of living adipose tissue. The authors are transparent that their work identifies a potential mechanism, not a validated therapy, and metformin is not currently an established treatment for acute pancreatitis in any patient group. Clinical translation would require carefully designed trials in obese patients at high risk of severe disease, with attention to dosing, timing, and the gastrointestinal tolerability issues that can complicate metformin use in acutely ill people.

Even so, the study’s implications are hard to ignore. Metformin is inexpensive, generically available, and already taken by hundreds of millions of people worldwide, with a long safety record and a growing portfolio of proposed anti-inflammatory effects. If the gut-microbiota-to-adipose-macrophage pathway described here holds up in humans, it could open a rational route to protecting a vulnerable patient population from one of pancreatology’s worst outcomes, using a drug that costs pennies. It also adds to a broader shift in biomedical thinking: the recognition that metabolites produced in dialogue with our microbial residents can act as chemical instructions for the immune system. In this study, one such instruction, hexanoylcarnitine, told fat-resident macrophages to stand down. For patients with obesity facing severe acute pancreatitis, learning to send that message deliberately could one day make the difference between recovery and catastrophe.

Subject of Research: Metformin, gut microbiota-derived metabolites, and adipose tissue macrophage polarization in obesity-related severe acute pancreatitis

Article Title: Metformin ameliorates injury in obesity-related severe acute pancreatitis and modulates a gut microbiota-associated metabolite that promotes M2 polarization of adipose tissue macrophages

Article References: Guo, X.-W., Sui, Y.-C., Zhang, J.-Y., Hao, Y., Li, J., & Sheng, L.-P. (2026). Metformin ameliorates injury in obesity-related severe acute pancreatitis and modulates a gut microbiota-associated metabolite that promotes M2 polarization of adipose tissue macrophages. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-08993-3

Image Credits: AI Generated

DOI: 10.1186/s12967-026-08993-3

Keywords: metformin, severe acute pancreatitis, obesity, gut microbiota, adipose tissue macrophages, M2 polarization, hexanoylcarnitine, metabolomics, interleukin-10, white adipose tissue, inflammation, Journal of Translational Medicine

The research, led by Hao Pan, Jiefeng Huang, Haoze Zhu, Xiaokun Li, and Cailong Liu, builds on a conceptual shift that has been gathering momentum in regenerative medicine. Bone repair is increasingly understood not as a purely osteogenic event but as an immunoregenerative one. Successful healing requires the temporally orchestrated resolution of inflammation, recruitment and differentiation of osteoprogenitor cells, formation of vascular networks, deposition of extracellular matrix, and eventual mineral maturation. At the center of this cascade sit macrophages—the immune cells that flood any injury site within hours. Early on, pro-inflammatory macrophages clear debris and defend against pathogens, but if that inflammatory state persists, it amplifies cytokine release, kills off osteoprogenitors, suppresses angiogenesis, and stalls bone bridging. When macrophages transition in time toward a reparative, so-called M2 phenotype, they instead secrete signals that resolve inflammation, remodel matrix, stabilize vessels, and drive osteogenic differentiation.

Previous attempts to exploit this plasticity have relied on small molecules, inorganic ions, or cytokines embedded in scaffolds. Such agents diffuse rapidly, act through narrow signaling channels, and cannot reproduce the rich paracrine language that reparative immune cells actually speak. The new work turns instead to small extracellular vesicles, or sEVs—nanoscale, membrane-bound packets that cells use to ship proteins, lipids, mRNAs, and microRNAs to one another. Because their cargo reflects the physiological state of the parent cell, sEVs can regulate recipient cells through multiple targets and pathways simultaneously. Macrophage-derived vesicles are especially attractive: their composition and function track the macrophage’s polarization status, and M2 macrophage exosomes have already been shown to restore osteoimmune balance and accelerate fracture healing, even under diabetic conditions where inflammatory dysregulation cripples bone repair.

Native sEVs, however, suffer from well-known weaknesses: modest yields, heterogeneous bioactivity, inefficient uptake by target cells, and rapid clearance after injection. The team’s solution was a two-step engineering scheme. First, they polarized RAW264.7 murine macrophages with interleukin-4 to generate reparative M2-like cells, confirmed by upregulation of the M2 markers Mrc1 and Arg1, elevated CD206 expression, and increased secretion of the anti-inflammatory cytokines IL-10 and TGF-β. Then they subjected these cells to 24 hours of hypoxic preconditioning in a sealed low-oxygen system, mimicking the low oxygen tension characteristic of injury niches. Western blotting confirmed accumulation of HIF-1α, the master hypoxia-responsive transcription factor, verifying that the donor cells had mounted a hypoxia program before releasing their vesicles.

The hypoxic treatment produced a striking and therapeutically useful side effect: apoptotic mimicry. Annexin V staining revealed a time-dependent externalization of phosphatidylserine on the donor macrophages, reaching 50.6 percent Annexin V-positive, membrane-intact cells at 24 hours. Phosphatidylserine is the classic eat-me lipid that dying cells display to trigger efferocytosis—the engulfment of apoptotic material—and, crucially, this recognition is not merely a cleanup mechanism. PS-mediated uptake actively drives anti-inflammatory, pro-resolving responses in phagocytes. Confocal imaging confirmed that the purified vesicles, dubbed M2-HP-Apo-sEVs, retained this PS-enriched membrane. Transmission electron microscopy showed typical rounded, cup-shaped nanovesicles of roughly 148 to 151 nanometers, and marker analysis confirmed enrichment of the EV proteins CD81 and ALIX with no detectable calnexin contamination. Notably, hypoxia also boosted vesicle yield per donor cell—without degrading purity, as shown by a stable particle-to-protein ratio—likely because hypoxic signaling accelerates endosomal trafficking and multivesicular body turnover.

Functionally, the engineered vesicles outperformed their normoxic counterparts at every checkpoint. In flow-cytometric uptake assays, macrophages internalized M2-HP-Apo-sEVs far more readily than standard M2-sEVs, an advantage the team traced in part to the MerTK receptor, a phosphatidylserine-binding efferocytic receptor. When the researchers partially silenced Mertk with siRNA, the uptake advantage vanished and the two vesicle types became indistinguishable in their immunomodulatory effects—strong evidence that PS-MerTK recognition underlies the enhanced activity, though the authors caution that incomplete knockdown and the absence of direct PS-blocking experiments mean other pathways likely contribute. In lipopolysaccharide-stimulated inflammatory macrophages, M2-HP-Apo-sEVs suppressed the pro-inflammatory genes Il1b and Nos2 more strongly than ordinary M2-sEVs while boosting Arg1 and Mrc1, shifting secreted cytokines away from IL-6 and IL-1β and toward IL-10 and TGF-β, and flipping surface marker expression from CD86 toward CD206. In short, the vesicles did not merely dampen inflammation—they actively converted inflammatory macrophages into a pro-resolving, tissue-repairing state.

The benefits rippled outward to the two other cell lineages that bone regeneration depends on. In bone marrow mesenchymal stem cells, M2-HP-Apo-sEVs produced the strongest alkaline phosphatase activity at day seven of osteogenic induction and the most abundant mineralized nodule formation by day fourteen, alongside coordinated upregulation of Alpl, Runx2, Opn, Ocn, and Bmp2 and intensified BMP2 protein staining. In human umbilical vein endothelial cells, the vesicles drove proliferation in EdU assays, accelerated migration in Transwell chambers, and produced denser, more interconnected capillary-like networks with significantly more nodes, junctions, and branches. This triple action—immune resolution, osteogenesis, and angiogenesis—reflects the emerging principle of osteo–immuno–angiogenic coupling, in which macrophages, stem cells, and vessels reinforce one another rather than operating as isolated modules.

To keep the vesicles where they are needed, the team embedded them in gelatin methacryloyl (GelMA), a photocrosslinkable hydrogel that combines cell-adhesive motifs with tunable degradation. Scanning electron microscopy showed that vesicle loading did not disrupt the hydrogel’s interconnected porous architecture, and release experiments in collagenase-containing medium demonstrated progressive, sustained vesicle liberation over seven days, governed by enzymatic degradation of the matrix. The construct was then tested in a demanding model: a five-millimeter critical-sized calvarial defect in mice, filled in situ with photocrosslinked hydrogel. Twelve weeks later, micro-computed tomography revealed minimal spontaneous healing in the GelMA-only group, substantial improvement with ordinary M2-sEVs, and the most extensive defect filling and radiographic bridging with the hypoxia-programmed vesicles. Quantitative analysis confirmed significantly better bone volume fraction, mineral density, trabecular number, and trabecular thickness, with reduced trabecular separation.

Tissue-level analyses told a coherent mechanistic story. Immunofluorescence of regenerated calvarial tissue showed reduced CD86 and increased CD206, indicating that the macrophage phenotype shift survived implantation; BMP2 and the endothelial marker CD31 were both elevated, evidencing coupled osteogenesis and vascularization; and H&E plus Masson’s trichrome staining revealed more continuous tissue bridging and richer collagen matrix in the M2-HP-Apo-sEV group. Gene expression in the defect tissue mirrored the in vitro findings, with inflammatory transcripts down and reparative, angiogenic, and osteogenic transcripts up. Examination of heart, liver, spleen, lung, and kidney showed no overt histopathological abnormalities at the twelve-week endpoint, offering preliminary reassurance on biocompatibility.

The authors are candid about limitations: the in vivo evidence rests on a single murine model at a single time point; the vesicles came from a cell line rather than primary human macrophages; the oxygen level during preconditioning was not continuously controlled; ultracentrifugation cannot fully resolve apoptosis-associated vesicle subpopulations; and the specific cargo—protein, lipid, or RNA—responsible for the effects remains unidentified. Clinical translation will also demand GMP-compatible manufacturing, closed processing systems, and rigorous lot-release criteria. Even so, the study delivers a compelling proof of concept: by programming donor macrophages with hypoxia, borrowing the apoptotic eat-me signal to hijack the immune system’s own resolution machinery, and parking the resulting nanovesicles in a degradable hydrogel, the researchers have assembled a cell-free platform that teaches the wound to heal itself. If the approach generalizes to load-bearing defects and larger animals, it could reshape how surgeons think about the impossible hole in the bone—not as a gap to be filled, but as an immune conversation waiting to be redirected.

Subject of Research: Hypoxia-programmed apoptotic-mimetic M2 macrophage-derived small extracellular vesicles delivered in GelMA hydrogels for immunoregenerative repair of critical-sized bone defects

Article Title: Hypoxia-programmed apoptotic-mimetic M2 macrophage-derived small extracellular vesicles for hydrogel-mediated immunoregenerative bone repair

Article References: Pan, H., Huang, J., Zhu, H., Chen, L., Huang, S., Huang, Y., Zhu, X., Li, X., & Liu, C. (2026). Hypoxia-programmed apoptotic-mimetic M2 macrophage-derived small extracellular vesicles for hydrogel-mediated immunoregenerative bone repair. Materials Today Bio, 41, Article 103696. https://doi.org/10.1016/j.mtbio.2026.103696

Image Credits: AI Generated

DOI: 10.1016/j.mtbio.2026.103696

Keywords: extracellular vesicles, macrophages, M2 polarization, hypoxia preconditioning, phosphatidylserine, efferocytosis, MerTK, GelMA hydrogel, bone regeneration, osteogenesis, angiogenesis, calvarial defect