Immunometabolic cues recompose and reprogram the microenvironment around implanted biomaterials

Abstract

Circulating monocytes infiltrate and coordinate immune responses in tissues surrounding implanted biomaterials and in other inflamed tissues. Here we show that immunometabolic cues in the biomaterial microenvironment govern the trafficking of immune cells, including neutrophils and monocytes, in a manner dependent on the chemokine receptor 2 (CCR2) and the C-X3-C motif chemokine receptor 1 (CX3CR1). This affects the composition and activation states of macrophage and dendritic cell populations, ultimately orchestrating the relative composition of pro-inflammatory, transitory and anti-inflammatory CCR2⁺, CX3CR1⁺ and CCR2⁺ CX3CR1⁺ immune cell populations. In amorphous polylactide implants, modifying immunometabolism by glycolytic inhibition drives a pro-regenerative microenvironment principally by myeloid cells. In crystalline polylactide implants, together with arginase-1-expressing myeloid cells, T helper 2 cells and γδ⁺ T cells producing interleukin-4 substantially contribute to shaping the metabolically reprogrammed pro-regenerative microenvironment. Our findings inform the premise that local metabolic states regulate inflammatory processes in the biomaterial microenvironment.

Control over the immune response to implanted biomaterials is required for medical devices to effectively function in therapeutic, diagnostic, tissue engineering and regenerative applications^1–4. Trafficking of monocytes from the bloodstream into tissues is mediated by chemokine receptor 2 (CCR2) signalling⁵ and is crucial to the foreign body response because extravasated monocytes develop into macrophages and dendritic cells, mediating the inflammatory response around implanted biomaterials. Recent advances in understanding metabolic adaptation in the biomaterial microenvironment reveal that changes in immune cell metabolism orchestrate immunological events in complex ways that could be leveraged to enhance regenerative medicine^6–13. However, how local immunometabolic cues in the biomaterial microenvironment regulate the trafficking of immune cells and organize the relative composition of pro-inflammatory, transitory and anti-inflammatory or pro-regenerative phenotypes has not been elucidated. Despite a developing understanding of the changing phenotype(s) of recruited immune cells expressing CCR2 and C-X3-C motif chemokine receptor 1 (CX3CR1) in response to sterile injury and in several disease models^14–19, the immunophenotypic composition of these cells in the biomaterial microenvironment has not been studied.

By fabricating amorphous polylactide (aPLA) implants, with and without embedded metabolic modulators, assessing release kinetics and locally implanting these biomaterial formulations in fluorescent reporter and wild-type mice, we demonstrate that the trafficking of immune cells to biomaterials is dependent on local metabolic cues. In addition, prevailing immunometabolic cues in the biomaterial microenvironment regulate the relative composition of polarized CCR2⁺ and CX3CR1⁺ populations, with metabolic changes able to create a pro-regenerative phenotype with the help of myeloid cells in the aPLA biomaterial microenvironment. In addition to arginase-1-expressing myeloid cells, T helper 2 cells and γδ⁺ T cells producing interleukin-4 (IL-4) contribute to shaping the pro-regenerative microenvironment surrounding crystalline polylactide (cPLA). Biomaterial-mediated delivery of metabolic modulators to direct events in the tissue microenvironment represents a substantial advancement, given the role of immunometabolism in physiologic and pathologic states^20,21. Specific to immunomodulation, our findings are clinically relevant because chronic inflammation elicited by polylactide biomaterials remains a major hurdle in soft-tissue and hard-tissue engineering applications, including laparoscopic²² and obstetric or gynaecological²³ anti-adhesive surgical barriers; cardiovascular stents²⁴; drug delivery devices^25,26; and dental²⁷, orthopaedic^28,29 and chest wall³⁰ reconstructive applications. Importantly, given the emerging role of immunometabolism in other clinically applied biomaterials (for example, polyethylene wear particles in total knee and hip replacements^6,10) known to elicit chronic inflammation, our findings will have broad translational impact.

Results

Rewiring metabolism in the aPLA biomaterial microenvironment controls CCR2- and CX3CR1-dependent trafficking

As a consequence of metabolic reprogramming and an established measure of the pro-inflammatory response³¹, we had reported a quantitative increase in F-18 fluorodeoxyglucose (FDG) uptake around polylactide implants in mice³², similar to clinical observations using positron emission tomography-computed tomography (PET-CT) in patients^22,23. Remarkably, in three-dimensional (3D) printed polylactide implants applied in rat critical-sized femoral defects, we showed that the inclusion of a metabolic inhibitor reduces FDG uptake (Supplementary Fig. 1), indicative of reduced inflammation in the musculoskeletal microenvironment, thus reproducing observations in the subcutis³². To understand the role of local metabolic cues in the trafficking of immune cells to sites of biomaterial implantation, we crossed Ccr2^RFP/RFPCx3cr1^GFP/GFP (CCR2- and CX3CR1-deficient) mice to B6 albino mice (Fig. 1a). The resulting F1 generation, Ccr2^RFP/+Cx3cr1^GFP/+ (CCR2- and CX3CR1-expressing) mice, were surgically incised (without biomaterial implantation) as sham controls; implanted with 7.5-mm-long reprocessed amorphous polylactide (hereafter, referred to as aPLA); or implanted with aPLA incorporating either 2-deoxyglucose (2DG) or aminooxyacetic acid (a.a.) at previously optimized concentrations³². In these dual-reporter mice, RFP (CCR2) is expressed in ~90% of Ly6C⁺ cells, allowing RFP to predominantly track classical monocytes, the analogue of CD14⁺ monocytes in humans^17,18. GFP (CX3CR1) is mostly expressed by Ly6C⁻ alternative or resident monocytes, the analogue of CD16⁺ monocytes in humans¹⁷, as well as a subset of natural killer and dendritic cells^17,19. Thus, intravital microscopic imaging of tissues adjacent to the implants allowed for visual monitoring of immune cells in the biomaterial microenvironment (Supplementary Fig. 2a). We used two inhibitors that act at different metabolic steps; 2DG inhibits hexokinase in the glycolytic pathway²¹, and a.a. reduces both mitochondrial uptake of glycolytic substrates³³ and glutamine metabolism³⁴. Lastly, we included sham and aPLA groups among Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (effectively knockout mice for these receptors; Extended Data Fig. 1a) to understand the contribution of CCR2 and CX3CR1 expression in the recruitment of immune cells to the biomaterial microenvironment.

Fig. 1|. Locally rewiring immune cell metabolism in the aPLA biomaterial environment affects CCR2- and CX3CR1-dependent trafficking.

a, B6 albino mice were crossed to Ccr2^RFP/RFPCx3cr1^GFP/GFP mice to generate Ccr2^RFP/+Cx3cr1^GFP/+ mice, which were surgically incised (sham group) or implanted with biomaterials. Intravital microscopy preceded flow cytometric analysis of tissues around incision sites (sham group) or biomaterials. b, Representative intravital microscopy images around incision sites (sham group), reprocessed aPLA, aPLA incorporating a.a. or 2DG; scale bars, 50 μm. c, Representative CCR2 and CX3CR1 flow cytometry plots gated on live cells. d–f, Flow cytometry quantification of CCR2⁺ (d), CX3CR1⁺ (e) and CCR2⁺CX3CR1⁺ (f) cells. g–i, Quantification of pro-inflammatory (CD86⁺CD206⁻) cells among CCR2⁺ (g), CX3CR1⁺ (h) and CCR2⁺CX3CR1⁺ (i) populations. j–l, Quantification of anti-inflammatory (CD206⁺) cells among CCR2⁺ (j), CX3CR1⁺ (k) and CCR2⁺CX3CR1⁺ (l) populations. m–o, Quantification of transition (CD86⁺CD206⁺) cells among CCR2⁺ (m), CX3CR1⁺ (n) and CCR2⁺CX3CR1⁺ (o) populations. p, Nucleated hematopoietic (CD45⁺) cells. q,r, Fold change of pro-inflammatory (H1; CD86⁺CD206⁻) CD45⁺ cells with respect to transition (T; CD86⁺CD206⁺) CD45⁺ cells (q) or anti-inflammatory (H2; CD206⁺) CD45⁺ cells (r). s,t, Fold change of T (s) or H2 (t) cells with respect to H1 CD45⁺ cells. u, Neutrophils (CD45⁺Ly6G⁺ cells). One-way ANOVA followed by Tukey’s or Newman–Keuls’ multiple comparison test, n = 3, mean (s.d.). Panel a created with BioRender.com.

We observed an initial increase in CCR2 and CX3CR1 expression that declined over time in all groups of Ccr2^RFP/+Cx3cr1^GFP/+ mice by intravital microscopy (Fig. 1b). CCR2⁺ and CX3CR1⁺ cells were observed in tissues of Ccr2^RFP/+Cx3cr1^GFP/+ mice, including sham (Supplementary Video 1), aPLA (Supplementary Video 2), aPLA + a.a. (Supplementary Video 3) and aPLA + 2DG (Supplementary Video 4) groups 10 weeks after surgeries. Notably, CCR2 and CX3CR1 expression appeared elevated in the aPLA group of Ccr2^RFP/+Cx3cr1^GFP/+ mice (Fig. 1b; Supplementary Video 2). Flow cytometric (quantitative) analyses of tissues around the implants corroborated increased CCR2⁺, CX3CR1⁺ and CCR2⁺CX3CR1⁺ cell populations in the aPLA group compared with sham controls; elevated levels were decreased by the incorporation of 2DG but not by a.a. in Ccr2^RFP/+Cx3cr1^GFP/+ mice (Fig. 1c–f and Supplementary Fig. 2b). Among Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Extended Data Fig. 1b), there were no notable differences in CCR2 or CX3CR1 expression between sham (Supplementary Video 5) and aPLA groups (Supplementary Video 6), and flow cytometry data corroborated our visual observations (Extended Data Fig. 1c,d). Interestingly, CCR2⁺CX3CR1⁺ cells were decreased with aPLA implantation relative to sham controls in Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Extended Data Fig. 1e).

Next, we sought to elucidate the composition of the different CCR2⁺, CX3CR1⁺ and CCR2⁺CX3CR1⁺ populations in the implant microenvironment based on CD86 and CD206 expression. We classified pro-inflammatory populations³⁵ as CD86⁺CD206⁻, anti-inflammatory populations³⁶ as CD206⁺, and transition populations moving from pro-inflammatory to anti-inflammatory states^37,38 as CD86⁺CD206⁺. Among CCR2⁺, CX3CR1⁺ and CCR2⁺CX3CR1⁺ populations, the proportion of pro-inflammatory cells was higher in the aPLA group compared with sham controls in Ccr2^RFP/+Cx3cr1^GFP/+ mice (Fig. 1g–i), whereas there were no differences with Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Extended Data Fig. 1f–h). Incorporating a.a. decreased pro-inflammatory levels among CX3CR1⁺ and CCR2⁺CX3CR1⁺ populations in Ccr2^RFP/+Cx3cr1^GFP/+ mice (Fig. 1h,i). The proportion of anti-inflammatory cells among CCR2⁺ cells was decreased with aPLA implantation compared with sham controls in Ccr2^RFP/+Cx3cr1^GFP/+ mice, while incorporating a.a. restored the proportion of these cells to a similar level as the sham controls (Fig. 1j). However, no difference was observed in the proportion of anti-inflammatory CCR2⁺ cells in Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Extended Data Fig. 1i). Proportions of anti-inflammatory cells among CX3CR1⁺ and CCR2⁺CX3CR1⁺ populations were similar across all groups in Ccr2^RFP/+Cx3cr1^GFP/+ mice (Fig. 1k,l) and Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Extended Data Fig. 1j,k). Among CCR2⁺ populations, the proportion of transition cells was reduced by the implantation of aPLA compared with sham controls in Ccr2^RFP/+Cx3cr1^GFP/+ mice (Fig. 1m), but not Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Extended Data Fig. 1l). The incorporation of either a.a. or 2DG increased proportions of transition cells (Fig. 1m). Similar to Ccr2^RFP/+Cx3cr1^GFP/+ mice (Fig. 1n,o), Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Extended Data Fig. 1m,n) showed no notable differences in proportions of transition cells among CX3CR1⁺ and CCR2⁺CX3CR1⁺ populations.

Compared with sham controls, the proportion of CD45⁺ cells was increased by implantation of aPLA in Ccr2^RFP/+Cx3cr1^GFP/+ mice, while incorporating either a.a. or 2DG significantly diminished this effect (Fig. 1p). By contrast, the proportion of CD45⁺ cells decreased with aPLA implantation in Ccr2^RFP/RFPCx3cr1^GFP/GFP mice compared with their sham controls (Extended Data Fig. 1o). In both Ccr2^RFP/+Cx3cr1^GFP/+ mice and Ccr2^RFP/RFPCx3cr1^GFP/GFP mice, we observed increased fold change of pro-inflammatory CD45⁺ cells with respect to transition cells or anti-inflammatory cells in the aPLA group when compared with sham controls (Fig. 1q,r and Extended Data Fig. 1p,q). Also, the fold change of transition or anti-inflammatory CD45⁺ cells with respect to pro-inflammatory cells was decreased in the aPLA group when compared with sham controls (Fig. 1s,t and Extended Data Fig. 1r,s). Interestingly, incorporating a.a. had the dual effect of decreasing pro-inflammatory and increasing anti-inflammatory proportions of cells (Fig. 1q–t). Compared with sham controls, Ly6G⁺ neutrophils¹⁵ were increased following aPLA implantation in Ccr2^RFP/+Cx3cr1^GFP/+ mice (Fig. 1u), whereas there were no changes in Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Extended Data Fig. 1t). Incorporating either a.a. or 2DG decreased elevated neutrophil levels (Fig. 1u).

The composition of myeloid cells is reorganized towards pro-regenerative states by immunometabolic cues in the aPLA microenvironment

Having observed that CD45⁺ populations were strikingly different among groups in our study (Fig. 1p), we sought to uncover the relative contribution of myeloid cells, including CD11b⁺ monocytes^39–41 (CD11b could also be expressed on subsets of B cells, neutrophils and macrophages⁴²), F4/80⁺ macrophages⁴² and CD11c⁺ dendritic cells^43,44, to the constitution and polarization of the biomaterial microenvironment. Compared with sham controls, we observed elevated monocyte populations following aPLA implantation in Ccr2^RFP/+Cx3cr1^GFP/+ mice, but not in Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Fig. 2a,b and Extended Data Fig. 1u). Incorporating either a.a. or 2DG reduced the frequency of monocytes compared with aPLA alone (Fig. 2a,b). Interestingly, there was increased fold change of pro-inflammatory cells with respect to transition cells or anti-inflammatory monocytes in the aPLA group when compared with sham controls, with either a.a. or 2DG incorporation modulating pro-inflammatory monocyte levels (Fig. 2c,d). Furthermore, the fold change of transition or anti-inflammatory monocytes with respect to pro-inflammatory cells was decreased in the aPLA group when compared with sham controls, with either a.a. or 2DG increasing anti-inflammatory monocyte levels (Fig. 2e,f).

Fig. 2|. Polarization states of myeloid cells are regulated by targeting immunometabolism in the aPLA biomaterial microenvironment.

a, Representative flow cytometry plots gated on CD45. b, Monocytes (CD45⁺CD11b⁺ cells). c,d, Fold change of pro-inflammatory (M1; CD86⁺CD206⁻) monocytes with respect to transition (T; CD86⁺CD206⁺) monocytes (c) or anti-inflammatory (M2; CD206⁺) monocytes (d). e,f, Fold change of T (e) or M2 (f) monocytes with respect to M1 monocytes. g, Macrophages (CD45⁺F4/80⁺ cells). h,i, Fold change of pro-inflammatory (M1; CD86⁺CD206⁻) macrophages with respect to transition (T; CD86⁺CD206⁺) macrophages (h) or anti-inflammatory (M2; CD206⁺) macrophages (i). j,k, Fold change of T (j) or M2 (k) macrophages with respect to M1 macrophages. l, Dendritic (CD45⁺CD11c⁺) cells. m,n, Fold change of pro-inflammatory (D1; CD86⁺CD206⁻) dendritic cells with respect to transition (T; CD86⁺CD206⁺) dendritic cells (m) or anti-inflammatory (D2; CD206⁺) dendritic cells (n). o,p, Fold change of T (o) or D2 (p) dendritic cells with respect to D1 dendritic cells. q, Dendritic cells expressing MHCII molecules (CD45⁺CD11c⁺MHCII⁺ cells). r,s, Fold change of pro-inflammatory (D1; CD86⁺CD206⁻) MHCII⁺ dendritic cells with respect to transition (T; CD86⁺CD206⁺) MHCII⁺ dendritic cells (r) or anti-inflammatory (D2; CD206⁺) MHCII⁺ dendritic cells (s). t,u, Fold change of T (t) or D2 (u) MHCII⁺ dendritic cells with respect to D1 MHCII⁺ dendritic cells. One-way ANOVA followed by Tukey’s or Newman–Keuls’ multiple comparison test, n = 3, mean (s.d.); reprocessed aPLA; a.a.; 2DG.

We observed decreased macrophage expression in the aPLA group compared with sham controls of Ccr2^RFP/+Cx3cr1^GFP/+ or Ccr2^RFP/RFP Cx3cr1^GFP/GFP mice, whereby incorporating either a.a. or 2DG increased macrophage expression (Fig. 2a,g and Extended Data Fig. 1v). Yet, exploring polarization states revealed increased fold change of pro-inflammatory macrophages with respect to transition or anti-inflammatory macrophages in the aPLA group when compared with sham controls, with either a.a. or 2DG incorporation reducing pro-inflammatory macrophage expression (Fig. 2h,i). Moreover, the fold change of transition or anti-inflammatory macrophages with respect to pro-inflammatory macrophages was decreased in the aPLA group when compared with sham controls, with a.a. increasing anti-inflammatory levels (Fig. 2j,k).

Dendritic cell populations were decreased in the aPLA group compared with sham controls of Ccr2^RFP/+Cx3cr1^GFP/+ mice, but increased in Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (Fig. 2l and Extended Data Fig. 1w). Incorporating either a.a. or 2DG increased dendritic cell expression (Fig. 2l). Furthermore, we observed increased fold change of pro-inflammatory dendritic cells with respect to transition or anti-inflammatory dendritic cells in the aPLA group when compared with sham controls, with a.a. reducing pro-inflammatory marker expression (Fig. 2m,n). There were no differences in the fold change of transition or anti-inflammatory dendritic cells with respect to pro-inflammatory dendritic cells between groups (Fig. 2o,p). Observed trends in dendritic cells were similar to results of dendritic cells expressing class II major histocompatibility complex (MHCII) molecules (Fig. 2q–u and Extended Data Fig. 1x).

Leveraging immunometabolism with aPLA favourably compares with currently used techniques

To determine how inhibiting the metabolism of immune cells using 2DG or a.a. compares to clinically used neutralization strategies⁴⁵, we fabricated aPLA incorporating hydroxyapatite (HA)³². Wild-type B6 mice were surgically incised as sham controls or implanted with 1-mm-long aPLA formulations (Fig. 3a). Flow cytometric analyses demonstrated elevated neutrophils with aPLA compared with sham controls among CD45⁺ cells; elevated levels were decreased by either a.a. or 2DG, but not by HA (Fig. 3b). There was increased fold change of pro-inflammatory CD45⁺ cells with respect to transition cells or anti-inflammatory cells in the aPLA group when compared with sham controls (Fig. 3c,d). Incorporating a.a., 2DG or HA each decreased the fold change of pro-inflammatory CD45⁺ cells with respect to transition cells (Fig. 3c), but only a.a. or 2DG decreased the fold change of pro-inflammatory cells with respect to anti-inflammatory cells (Fig. 3d). The fold change of transition or anti-inflammatory CD45⁺ cells with respect to pro-inflammatory cells was decreased in the aPLA group when compared with sham controls (Fig. 3e,f). Whereas incorporating a.a. or 2DG increased transition and anti-inflammatory marker expression, HA did not (Fig. 3e,f). Importantly, a.a. had greater effects than HA at decreasing pro-inflammatory and increasing transition or anti-inflammatory CD45⁺ proportions (Fig. 3c–f).

Fig. 3|. Using an acid more favourably modulates activation states of immune cells around aPLA biomaterials compared with traditional neutralization techniques.

a, Wild-type B6 mice were surgically incised (sham group) or implanted with reprocessed aPLA, aPLA incorporating a.a., 2DG or HA. Afterwards, flow cytometric analysis of tissues around incision sites (sham group) or biomaterials was undertaken. b, Neutrophils (CD45⁺Ly6G⁺ cells). c,d, Fold change of pro-inflammatory (H1; CD86⁺CD206⁻) cells with respect to transition (T; CD86⁺CD206⁺) cells (c) or anti-inflammatory (H2; CD206⁺) cells (d), gated for nucleated hematopoietic (CD45⁺) populations. e,f, Fold change of T (e) or H2 (f) cells with respect to H1 cells. g,h, Fold change of pro-inflammatory (M1; CD86⁺CD206⁻) monocytes (CD45⁺CD11b⁺) with respect to transition (T; CD86⁺CD206⁺) monocytes (g) or anti-inflammatory (M2; CD206⁺) monocytes (h). i,j, Fold change of T (i) or M2 (j) monocytes with respect to M1 monocytes. k,l, Fold change of pro-inflammatory (M1; CD86⁺CD206⁻) macrophages (CD45⁺F4/80⁺) with respect to transition (T; CD86⁺CD206⁺) macrophages (k) or anti-inflammatory (M2; CD206⁺) macrophages (l). m,n, Fold change of T (m) or M2 (n) macrophages with respect to M1 macrophages. o, Dendritic (CD45⁺CD11c⁺) cells. p,q, Fold change of pro-inflammatory (D1; CD86⁺CD206⁻) dendritic cells with respect to transition (T; CD86⁺CD206⁺) dendritic cells (p) or anti-inflammatory (D2; CD206⁺) dendritic cells (q). r,s, Fold change of T (r) or D2 (s) dendritic cells with respect to D1 dendritic cells. t, Dendritic cells expressing MHCII molecules (CD45⁺CD11c⁺MHCII⁺ cells). u,v, Fold change of pro-inflammatory (D1; CD86⁺CD206⁻) MHCII⁺ dendritic cells with respect to transition (T; CD86⁺CD206⁺) MHCII⁺ dendritic cells (u) or anti-inflammatory (D2; CD206⁺) MHCII⁺ dendritic cells (v). w,x, Fold change of T (w) or D2 (x) MHCII⁺ dendritic cells with respect to D1 MHCII⁺ dendritic cells. One-way ANOVA followed by Tukey’s multiple comparison test, n = 3, mean (s.d.). Panel a created with BioRender.com.

The fold change of pro-inflammatory monocytes with respect to transition or anti-inflammatory monocytes was increased in the aPLA group when compared with sham controls (Fig. 3g,h). Elevated pro-inflammatory monocyte proportions were consistently decreased by incorporating a.a. or 2DG, with HA only decreasing elevated proportions with respect to transition cells (Fig. 3g,h). However, the fold change of transition or anti-inflammatory monocytes with respect to pro-inflammatory monocytes was decreased in the aPLA group when compared with sham controls (Fig. 3i,j). Whereas incorporating a.a. or 2DG increased the proportion of transition monocytes, only a.a. increased the proportion of anti-inflammatory monocytes (Fig. 3i,j). Notably, a.a. was more effective than HA at consistently reducing pro-inflammatory and increasing transition or anti-inflammatory monocyte proportions (Fig. 3g–i).

With macrophages, the fold change of pro-inflammatory cells with respect to transition or anti-inflammatory cells was increased in the aPLA group when compared with sham controls (Fig. 3k,l). Increased pro-inflammatory macrophage proportions were reduced by incorporating a.a. or 2DG, with HA only decreasing elevated proportions with respect to transition cells (Fig. 3k,l). We observed that a.a. was more effective than HA at reducing pro-inflammatory macrophage proportions (Fig. 3k,l). In addition, the fold change of transition or anti-inflammatory macrophages with respect to pro-inflammatory macrophages was decreased in the aPLA group when compared with sham controls (Fig. 3m,n). As with monocytes, incorporating a.a. or 2DG increased the proportion of transition macrophages, with only a.a. increasing the proportion of anti-inflammatory macrophages (Fig. 3m,n).

While aPLA implantation did not reduce dendritic cell populations when compared with sham controls, incorporating a.a., 2DG or HA increased dendritic cell populations (Fig. 3o). Interestingly, the fold change of pro-inflammatory dendritic cells with respect to transition or anti-inflammatory dendritic cells was increased in the aPLA group when compared with sham controls, and incorporating a.a., 2DG or HA decreased elevated pro-inflammatory proportions (Fig. 3p,q). We also observed that the fold change of transition or anti-inflammatory dendritic cells with respect to pro-inflammatory dendritic cells was decreased in the aPLA group when compared with sham controls (Fig. 3r,s). Incorporating a.a. or HA increased the proportion of transition dendritic cells, with a.a., 2DG or HA increasing the proportion of anti-inflammatory dendritic cells (Fig. 3r,s). When compared with HA, a.a. decreased pro-inflammatory and increased transition or anti-inflammatory proportions (Fig. 3q–s).

The expression of MHCII⁺ dendritic cells was decreased with aPLA implantation when compared with sham controls, and incorporating a.a., 2DG or HA increased MHCII⁺ dendritic cell expression (Fig. 3t). The fold change of pro-inflammatory MHCII⁺ dendritic cells with respect to transition or anti-inflammatory MHCII⁺ dendritic cells was increased in the aPLA group when compared with sham controls, and incorporating a.a., 2DG or HA decreased elevated pro-inflammatory proportions (Fig. 3u,v). Also, the fold change of transition or anti-inflammatory MHCII⁺ dendritic cells with respect to pro-inflammatory MHCII⁺ dendritic cells was decreased in the aPLA group when compared with sham controls (Fig. 3w,x). Incorporating a.a. or HA increased the proportion of transition MHCII⁺ dendritic cells, with a.a., 2DG or HA increasing the proportion of anti-inflammatory MHCII⁺ dendritic cells (Fig. 3w,x). Compared with HA, a.a. decreased pro-inflammatory and increased transition or anti-inflammatory proportions (Fig. 3v–x).

Immunometabolic cues from cPLA create a pro-regenerative microenvironment with the help of T cells

Compared with aPLA, we have observed by electrospray ionization-mass spectrometry that (semi-) cPLA formulations degrade more slowly³². Furthermore, the different stereochemical compositions of aPLA and cPLA could elicit differential immune cellular responses by triggering varied bioenergetic signatures⁷. As such, we sought to uncover the composition and phenotypes of immune cells in the cPLA microenvironment. Wild-type B6 mice were surgically incised as sham controls or implanted with 1-mm-long cPLA formulations with and without incorporating a.a. and 3-(3-pyridinyl)-1-(4-pyridinyl)-2-propen-1-one (3PO; Fig. 4a), with 3PO being a small-molecule inhibitor of 6-phosphofructo-2-kinase⁴⁶, the rate-limiting enzyme in glycolysis. Flow cytometric analyses demonstrated that there was increased CD45 expression with cPLA implantation compared with sham controls, and elevated levels were decreased by incorporating a.a. (Fig. 4b). There was increased fold change of pro-inflammatory CD45⁺ cells with respect to transition cells or anti-inflammatory cells in the cPLA group when compared with sham controls (Extended Data Fig. 2a,b). Also, the fold change of transition or anti-inflammatory CD45⁺ cells with respect to pro-inflammatory cells was decreased in the cPLA group when compared with sham controls (Extended Data Fig. 2c,d). Interestingly, the incorporation of a.a. or 3PO did not decrease pro-inflammatory nor increase transition or anti-inflammatory CD45⁺ proportions (Extended Data Fig. 2a–d).

Fig. 4|. Locally targeting immunometabolism in the cPLA environment elevates IL-4-expressing T-cell subsets with differential effects on myeloid populations.

a, Wild-type B6 mice were surgically incised (sham group) or implanted with reprocessed cPLA, cPLA incorporating a.a. or 3PO. Afterwards, flow cytometric analysis of tissues around incision sites (sham group) or biomaterials was undertaken. b, Nucleated hematopoietic (CD45⁺) cells. c, Neutrophils (CD45⁺Ly6G⁺ cells). d, Monocytes (CD45⁺CD11b⁺ cells). e, Macrophages (CD45⁺F4/80⁺ cells). f, Dendritic (CD45⁺CD11c⁺) cells. g, Dendritic cells expressing MHCII molecules (CD45⁺CD11c⁺MHCII⁺ cells). h–l, Nucleated hematopoietic cells (h), monocytes (i), macrophages (j), dendritic cells (k) and MHCII⁺ dendritic cells (l) expressing arginase 1 (Arg1⁺). m, Cytotoxic T lymphocytes (CD45⁺CD3⁺CD8⁺ cells). n, T helper lymphocytes (CD45⁺CD3⁺CD4⁺ cells). o, T helper 1 cells expressing IFNγ. p, T helper 2 cells expressing IL-4. q, T helper 17 cells expressing interleukin-17A (IL-17A). r, Gamma delta (γδ) T (CD45⁺CD3⁺γδ⁺) cells. s–u, γδ⁺ T cells producing IFNγ (s), IL-4 (t) and IL-17A (u). v, Innate lymphoid cells (CD45⁺CD3⁺Thy1.2⁺) producing IL-17A. w, B cells (CD45⁺CD11b⁻CD19⁺). One-way ANOVA followed by Tukey’s multiple comparison test, n = 3, mean (s.d.). Panel a created with BioRender.com.

There was no difference in neutrophil expression in the cPLA group when compared with sham controls (Fig. 4c). We observed increased monocyte expression in the cPLA group when compared with sham controls; elevated monocyte expression was decreased by incorporating a.a. (Fig. 4d). While there was increased fold change of pro-inflammatory monocytes with respect to transition or anti-inflammatory monocytes, the incorporation of a.a. or 3PO did not reduce elevated pro-inflammatory levels (Extended Data Fig. 2e,f). The fold change of transition or anti-inflammatory monocytes with respect to pro-inflammatory monocytes was decreased in the cPLA group when compared with sham controls, and the incorporation of a.a. or 3PO did not have significant effects (Extended Data Fig. 2g,h).

Macrophage expression was increased by cPLA implantation compared with sham controls, with the incorporation of a.a. or 3PO reducing the elevated expression (Fig. 3e). Furthermore, there was increased fold change of pro-inflammatory macrophages with respect to transition or anti-inflammatory macrophages in the cPLA group when compared with sham controls (Extended Data Fig. 2i,j). The fold change of transition or anti-inflammatory macrophages with respect to pro-inflammatory macrophages was decreased in the cPLA group when compared with sham controls (Extended Data Fig. 2k,l). The incorporation of a.a. or 3PO neither decreased pro-inflammatory nor increased anti-inflammatory proportions (Extended Data Fig. 2i–l).

Dendritic cell levels were increased in the cPLA group compared with sham controls, with a.a. or 3PO further increasing dendritic cell expression (Fig. 4f). Whereas there were no changes in the fold change of pro-inflammatory dendritic cells with respect to transition cells, the fold change of pro-inflammatory dendritic cells with respect to anti-inflammatory cells was increased in the cPLA group when compared with sham controls, with the incorporation of 3PO reducing elevated levels (Extended Data Fig. 2m,n). The fold change of transition or anti-inflammatory dendritic cells with respect to pro-inflammatory cells was decreased in the cPLA group when compared with sham controls (Extended Data Fig. 2o,p).

The expression of MHCII⁺ dendritic cells with cPLA implantation was similar to sham controls, and MHCII⁺ expression was increased by the incorporation of a.a. (Fig. 4g). In addition, the fold change of pro-inflammatory MHCII⁺ dendritic cells with respect to transition or anti-inflammatory MHCII⁺ dendritic cells was increased in the cPLA group when compared with sham controls (Extended Data Fig. 2q,r). Also, the fold change of transition or anti-inflammatory MHCII⁺ dendritic cells with respect to pro-inflammatory MHCII⁺ dendritic cells was decreased in the cPLA group when compared with sham controls (Extended Data Fig. 2s,t). The incorporation of a.a. or 3PO neither decreased pro-inflammatory nor increased anti-inflammatory levels (Extended Data Fig. 2q–t). While arginase 1 (Arg1) levels were increased in the cPLA group when compared with sham controls, the incorporation of a.a. further increased Arg1 levels among CD45⁺, CD11b⁺, F4/80⁺, CD11c⁺ and CD11c⁺MHCII⁺ populations (Fig. 4h–l).

Given the distinct observation made with cPLA implants, we sought to elucidate the role of the adaptive immune response, including CD19 B-cell activity⁴⁷, CD4 T helper and CD8 cytotoxic T cells^36,48. We suspected a scenario where T helper 2 pathway (IL-4) signalling preceded macrophage polarization in the biomaterial microenvironment³⁶. With cPLA implantation, CD8 expression was increased when compared with sham controls, and the incorporation of a.a. decreased elevated levels (Fig. 4m). Similarly, CD4 expression was higher in the cPLA group compared with sham controls, with 3PO reducing elevated levels (Fig. 4n). CD4⁺ cells expressing interferon-γ (IFNγ) were reduced in the cPLA group when compared with sham controls; the incorporation of a.a. or 3PO tended to increase IFNγ expression only to levels similar to the sham group (Fig. 4o). By contrast, CD4⁺ cells expressing IL-4 were similar between the cPLA group and sham controls, with incorporation of a.a. or 3PO increasing IL-4 levels (Fig. 4p). Assessing CD4⁺ cells expressing IL-17 revealed reduced expression in the cPLA group when compared with sham controls, and the incorporation of a.a. or 3PO tended to increase IL-17 expression only to levels similar to the sham group (Fig. 4q). While the γδ repertoires were decreased in the cPLA group when compared with sham controls, the incorporation of 3PO but not a.a. tended to increase the T-cell receptor repertoires (Fig. 4r). In addition, while there were no changes in IFNγ and IL-17 expression from γδ⁺ T cells, the incorporation of a.a. or 3PO increased IL-4 expression from γδ⁺ T cells (Fig. 4s–u). There were no changes in IL-17 expression from innate lymphoid cells (Fig. 4v), and we did not observe changes in CD19 expression between the cPLA group and sham controls; however, the incorporation of 3PO tended to increase CD19 expression (Fig. 4w).

Incorporated metabolic inhibitors are released faster from aPLA than cPLA formulations

Consistent with the slower hydrolytic degradation rate of cPLA, drug release kinetics were slower from cPLA formulations compared with aPLA formulations (Supplementary Fig. 3a–f). By 12 weeks, approximating the in vivo study timeline, 25.96% of a.a. was released from aPLA compared with 0% from cPLA (Supplementary Fig. 3a,b). While aPLA + 2DG and aPLA + 3PO were not assessed in vivo, we included their release kinetics to enable direct comparison to their cPLA counterparts, especially because the mass spectrometry technique applied is unable to measure concentrations <0.078 μM of a.a. or <0.0002 μM 3PO. Our data showed that, although 10.99% of 2DG was released from aPLA by 12 weeks, only 0.03% of 2DG was released from cPLA (Supplementary Fig. 3c,d). Following the same trend, 0.17% of 3PO was released from aPLA by the 12-week time point, when only 0.02% had been released from cPLA formulations (Supplementary Fig. 3e,f). Total content of metabolic inhibitors (mean ± s.d.) determined by complete dissolution of 200 mg biomaterial formulations (refer to Methods) were cPLA + a.a. (211.05 ± 15.91 ng), aPLA + a.a. (1,565.41 ± 142.98 ng), cPLA + 2DG (160,171.98 ± 46,881.94 ng), aPLA + 2DG (205,080.32 ± 48,295.19 ng), cPLA + 3PO (1,633.57 ± 221.77 ng) and aPLA + 3PO (2,732.97 ± 1,181.56 ng).

Discussion

Although polylactide-based medical devices are approved by the US Food and Drug Administration for applications in adult reconstructive surgery, drug delivery and nanotechnology, their clinical utility is considerably limited by long-term, sterile inflammation, which is poorly understood^3,49. Our findings suggest that the activation states and trafficking of immune cells to the aPLA biomaterial microenvironment are dependent on CCR2 and CX3CR1 expression, with CCR2⁵ likely to play a greater role over CX3CR1^18,19 expression. We reveal that both CCR2 and CX3CR1 signalling could be regulated by locally controlling glycolytic flux through targeting hexokinase in the biomaterial microenvironment. In liver models of sterile injury as well as in the heart following myocardial infarction, CCR2⁺ monocytes dominate the pro-inflammatory phase, with CCR2⁺CX3CR1⁺ and CX3CR1⁺ monocytes and macrophages playing a greater role during the anti-inflammatory or reparative phase of healing^14,50. In our study, however, sterile aPLA implants increase pro-inflammatory (CD86⁺CD206⁻) proportions of CCR2⁺, CX3CR1⁺ and CCR2⁺CX3CR1⁺ cell populations, while decreasing transition (CD86⁺CD206⁺) and anti-inflammatory proportions (CD206⁺) of CCR2⁺ populations, an effect that required CCR2 and CX3CR1 competency. Against our hypothesis, the incorporation of a.a. did not reduce monocyte recruitment around biomaterials; however, it reshaped the composition of recruited CCR2⁺, CX3CR1⁺ and CCR2⁺CX3CR1⁺ cell populations to comprise reduced pro-inflammatory and elevated transition and anti-inflammatory proportions. These transition immune cell populations suggest a reversal of pro-inflammatory biomaterial responses, playing a crucial role in angiogenesis and tissue regeneration^37,38. Consistent with our findings, an increase in CX3CR1⁺ populations is observed around polylactide copolymers⁵¹.

Myeloid cell recruitment to aPLA-based implants appears to be regulated by immunometabolism in the biomaterial microenvironment. We observed a two-pronged effect of aPLA in constituting a pro-inflammatory biomaterial microenvironment—aPLA elevated the relative levels of pro-inflammatory monocytes while concurrently decreasing the relative proportions of transition and anti-inflammatory monocytes. This trend is observable in vitro with polyethylene wear particles^6,10 and other types of polylactide materials⁷, whose implantation results in chronic inflammatory responses. Consequently, it is likely that this two-pronged effect occurs with other classes of biomaterials, providing new insight into complex ways that biomaterials modify their immune microenvironment.

During the inflammatory response to implants, the biomaterial microenvironment is glycolytically reprogrammed, showing enhanced radiolabelled glucose uptake in mice³², an observation made by PET imaging in implanted human patients^22,23. Our findings reveal that disrupting elevated glycolytic flux using 2DG or a.a. decreased pro-inflammatory monocytes, increased transition and anti-inflammatory monocytes, and decreased neutrophil recruitment to the biomaterial microenvironment, likely accounting for the reduction in FDG uptake in rat critical-sized femoral defects. Paradoxically, aPLA implantation decreased macrophage and dendritic cell recruitment while lowering MHCII expression on dendritic cells, an observation that is thought to imply an immunomodulatory role of polylactide implants⁴³. We reproduce these findings but also uncover that the relative proportion of pro-inflammatory macrophages, dendritic cells or dendritic cells expressing MHCII is increased by aPLA implantation. With macrophages and, to a lesser extent, dendritic cells and dendritic cells expressing MHCII, inhibiting glycolysis reduced pro-inflammatory and increased transition and anti-inflammatory proportions, providing new insight on how biomaterials could result in chronic inflammation all the while orchestrating seemingly conflicting immunological events. Of note, CCR2 and CX3CR1 deficiency prevents the recruitment of neutrophils, monocytes, macrophages, dendritic cells and MHCII expression on dendritic cells in the aPLA microenvironment, which may be due to the role of CCR2 in promoting the local activation and maturation of immune cells^52,53.

For decades, chronic inflammation by polylactide biomaterials was thought to be due to acidic degradation products lowering surrounding pH⁴⁵. The correlation of reduced bioluminescence, a measure of toxicity in Photobacterium phosphoreum, to low pH informed this dogma in biomedical engineering⁵⁴. This correlational observation led to current efforts aimed at neutralizing pH in the polylactide microenvironment by incorporating alkalinizing agents such as magnesium hydroxides, calcium hydroxyapatite, calcium carbonate, sodium bicarbonate, bioglass and polyphosphazene^45,55–58. While these strategies remain relevant in bone and dental tissue engineering, where the bioactivity of ceramic biomaterials and alkaline microenvironments enhance osteoblastogenesis⁵⁹, composites of polylactide and clinically applied HA could still result in adverse responses^9,60, and their application in soft-tissue applications is limited by the mechanical properties of polymer–ceramic composites. Contrary to HA’s effects, a.a. further makes the biomaterial microenvironment acidic³²; therefore, in principle, a.a. would worsen aPLA-induced inflammation. We show that while incorporating a.a. reduced neutrophil levels, HA increased neutrophil recruitment, accentuating inflammation. To extents greater than HA’s, incorporating a.a. in aPLA implants reduced the relative proportion of pro-inflammatory immune cells, including overall nucleated hematopoietic cell populations, monocytes, macrophages, dendritic cells and dendritic cells expressing MHCII. Furthermore, a.a. largely increased transition and anti-inflammatory cellular proportions more than HA could. This suggests that while pH may exert some role in the process, it is not the sole regulator of immunological events in the polylactide microenvironment. Unlike current, pH-based strategies, our findings support a novel mechanism based on metabolic reprogramming as a primary driver of the adverse immune responses to polylactide implants.

cPLA degrades at markedly slower rates than aPLA³² and exhibits delayed adverse immune response in clinical settings³. It is likely that while the inflammatory response to aPLA is elicited by both bulk and degrading biomaterial, at least the initial (on the order of months to years) response to cPLA results from mostly bulk biomaterial. Consistent with this reasoning, release kinetics revealed that metabolic inhibitors are released at a faster rate from aPLA (up to 26% by 12 weeks) than cPLA formulations (up to 0.03%). Although aPLA formulations retained higher levels of metabolic inhibitors, the relatively minute amounts of released metabolic inhibitors measured from cPLA formulations suggest that these are likely from the surface of the bulk biomaterial. These differential degradation rates and drug release kinetics offer rare insight into how these biomaterial properties temporally affect immunometabolic events, partly explaining the unexpected results observed with cPLA biomaterials. For example, in contrast to observations made with aPLA, cPLA increased macrophage and dendritic cell recruitment, did not elevate neutrophil levels nor reduce MHCII expression on recruited dendritic cells. Also, unlike with aPLA, glycolytic inhibition did not reduce myeloid pro-inflammatory nor increase myeloid transition and anti-inflammatory proportions. However, glycolytic inhibition in the cPLA microenvironment reduced macrophage recruitment, specifically increased IL-4-expressing γδ⁺ T cells and T helper 2 cells, while keeping IFNγ at homeostatic levels, necessary for tissue physiology⁶¹; this elevated IL-4 response has been reported to precede and direct subsequent macrophage polarization in the pro-regenerative biomaterial microenvironment³⁶. Importantly, a.a. reduced monocyte recruitment and increased Arg1 expression among overall myeloid populations, including monocytes, macrophages, dendritic cells and dendritic cells expressing MHCII. This observation is consistent with a.a. inhibiting aspartate aminotransferase, a key transaminase in the aspartate-arginosuccinate shunt in pro-inflammatory macrophages during metabolic reprogramming⁶². Antigen-presenting cells are able to activate both class I and II MHC, following exposure to biomaterials⁶³. While a.a. reduced cytotoxic T lymphocytes, 3PO reduced overall T helper cell recruitment to the cPLA biomaterial microenvironment, consistent with the crucial role of metabolism on T-cell function⁶⁴.

Outlook

Our findings provide new insight into the role of immunometabolic cues on immune cellular trafficking to the biomaterial microenvironment and how this affects the composition and activation states of immune cell populations. Both a.a. and 3PO, as well as their derivatives, have been well tolerated when administered during clinical trials in cancer and Huntington’s disease^34,65, making them translatable candidates. As adverse immune response to polylactide implants is often local at anatomical sites of implantation or entrapment following injection (as in nanomedicine), it is anticipated that intervention would be local. Biodegradable materials that concurrently release metabolic inhibitors as they release immunogenic by-products constitute a promising biomaterial-based immunomodulation strategy. Towards clinical translation in human and veterinary patients, further studies are needed to (a) elucidate the spatial transcriptomic profiles in the implant microenvironment whose metabolism has been altered; (b) evaluate the role of immunometabolism in creating a pro-regenerative microenvironment in large animal models, albeit polylactide-induced adverse responses and associated interventions are reproducibly documented across mice⁵⁷, rat⁶⁶, rabbit²⁸, canine patients⁶⁷, sheep⁶⁰ and humans^22,24; and (c) more robustly evaluate how observations made subcutaneously in this study are affected by changing anatomical locations for different clinical applications. For example, intraperitoneal and intramuscular implantation of biomaterials could elicit greater inflammatory reactions than subcutaneous sites⁶⁸. Ultimately, the temporally regulated local release of these metabolic modulators from implants to program the trafficking and polarization of immune cells in the biomaterial microenvironment could offer a highly translatable opportunity that would advance regenerative engineering towards improved human and animal health.

Methods

Biomaterial formulation and metabolic modulators

aPLA (PLA 4060D) and semi-cPLA (PLA 3100HP) biomaterials from NatureWorks were used. Before using these materials, we verified their physicochemical and thermal properties using optical rotation to determine l-lactide content, gel permeation chromatography to evaluate molecular weights, differential scanning calorimetry to analyse melting temperature, glass transition temperature and percent crystallinity of the PLA grades³². As metabolic modulators and for neutralization studies, 3PO (MilliporeSigma), 2DG (MilliporeSigma), a.a. hemi-hydrochloride (Sigma-Aldrich) and HA (2.5 μm² particle sizes⁶⁹; Sigma-Aldrich) were incorporated into biomaterials by melt-blending at 190 °C for 3 min in a DSM 15 cc mini-extruder (DSM Xplore) and then made into pellets using a pelletizer (Leistritz Extrusion Technology). Afterwards, pellets were made into 1.75 mm (diameter) filaments using an extruder (Filabot EX2) at 170 °C with air set at 93. Filaments were cut into 1-mm-long or 7.5-mm-long sizes and then sterilized by ultraviolet radiation for 30 min (ref. 70). After accounting for potential thermal degradation of metabolic inhibitors during biomaterial fabrication, based on our prior study³², we estimated that 189 mg of 2DG, 4.86 mg of 3PO, 90 mg of a.a. or 200 mg of HA in 10 g of aPLA or cPLA would approximate concentrations that were effective when applied in vitro, where 1 mM concentrations of metabolic inhibitors were found to be effective with minimal cytotoxicity, following dose-dependent analyses³². To control for melt-blending as a confounder, biomaterials not incorporating metabolic modulators or HA were processed under similar conditions to make ‘reprocessed’ formulations.

Rat studies

To fill critical-sized, 5 mm, mid-diaphyseal, full thickness, left femoral defects in rats, we created and sliced a 5 mm model using Simplify3D software (version 5.0). Using fabricated 1.75 mm filaments, we 3D printed the 5 mm mid-diaphyseal model using a fused deposition modelling-type 3D printer (PRUSA i3 MK3S), with nozzle temperature set at 210 °C and bed temperature at 60 °C. Printed constructs were sterilized by ultraviolet light for 30 min before surgical implantation in rats.

Animal studies were approved by the Institutional Animal Care and Use Committee of the University of Michigan (PRO00010536). Under anaesthesia (2–3% isoflurane), we created a left mid-diaphyseal, full thickness, femoral defect in Sprague-Dawley male rats (7 months of age) weighing approximately 500 g (Charles River). In these defects, we implanted 3D printed constructs, which were fixed using Radiolucent Fixation Systems bone miniplates and screws (Translational Osteosynthesis), which facilitated artefact-free in vivo radio-wave imaging. During implantation, the screws were ultrasonically welded to the miniplates utilizing Acousteon handheld piezoelectric technology (Translational Osteosynthesis) for enhanced mechanical fixation strength needed to support load bearing throughout the study. All animals received pre- and post-operative analgesia. Following 6 weeks of implantation, rats implanted with aPLA (n = 6) or aPLA + 2DG (n = 3) were injected with 527.5 uCi of F-18 fluorodeoxyglucose (FDG; Michigan Medicine Cyclotron Facility), and animals were recovered in their home cages. After 40 min, animals were placed in the gantry of a Siemens Inveon (Siemens Medical Solutions) and a 20 min PET scan and 10 min CT in vivo images were obtained. Images were acquired with the Inveon Acquisition Workspace (IAW version 2.1.272) software; representative PET-CT images were prepared using PMOD software version 4.2 (PMOD Technologies). Following 2 days of PET-CT scan (after complete decay of FDG), rats were euthanized.

Mice

Studies involving mice were approved by the Institutional Animal Care and Use Committee at Michigan State University (PROTO202100327). Ccr2^RFP/RFPCx3cr1^GFP/GFP mice, B6(Cg)-Tyr^c−2J/J (B6 albino) mice and C57BL/6J (wild-type B6) mice were obtained from the Jackson Laboratory. To generate Ccr2^RFP/+Cx3cr1^GFP/+ mice, we crossed female Ccr2^RFP/RFPCx3cr1^GFP/GFP (8-week-old) mice to male B6(Cg)-Tyr^c−2J/J (B6 albino; 8-week-old) mice, as previously described^18,19. At 4 weeks old, generated Ccr2^RFP/+Cx3cr1^GFP/+ mice were assigned n = 3 mice (two females, one male) per group. Only female mice (n = 3 mice per group) were used for studies involving Ccr2^RFP/RFPCx3cr1^GFP/GFP mice (14 weeks old) and C57BL/6J mice (9 weeks old).

Subcutaneous surgical model

Anaesthesia was accomplished using isoflurane (2–3%). Using the aseptic technique, the skin of each mouse was shaved and disinfected using iodine and alcohol swabs. Surgical incision was made through the skin into the subcutis, with or without biomaterial implantation after a pouch had been made with forceps. Surgical glue (3M Vetbond) was used to close the skin, and each mouse received intraperitoneal or subcutaneous pre- and post-operative meloxicam (5 mg kg⁻¹) injections as well as post-operative saline. In Ccr2^RFP/+Cx3cr1^GFP/+ mice and Ccr2^RFP/RFPCx3cr1^GFP/GFP mice, the neck (just caudal to the ear; Supplementary Fig. 2a) was surgically incised (sham) and implanted using 7.5-mm-long filaments to allow for imaging of the biomaterial microenvironment. In C57BL/6J mice, the dorsum (back) of mice was incised (sham) and implanted using 1-mm-long filaments.

Intravital microscopic imaging and processing

Mice were anaesthetized using an isoflurane (2–3%), and image stacks were acquired using a Leica SP8 DIVE laser multiphoton microscope equipped with Spectra-Physics Insight X3 dual beam (630 to 1,300 nm tunable and 1,040 nm fixed) and 4Tune, tunable, super sensitive hybrid detectors (HyDs). Images on the intravital microscope were collected using Leica Application Suite X (LAS X) software (version 3.5.7.23225). To acquire serial optical sections, a laser beam (940 nm for GFP; 1,040 nm for RFP) was focused through a 25× water-immersion lens (NA 1.00 HC PL IRAPO, Leica) and scanned with a field of view of 0.59 × 0.59 mm² at 600 Hz. To visualize a larger area, 3 tiles of optical fields were imaged using a motorized stage to automatically acquire sequential fields of view. Z-stacks were acquired in 3 μm steps to image a total depth of 117 μm of tissue. To avoid fluorophore bleed-through, images were acquired using sequential scanning in between frames. Visualization of collagen was achieved via the second harmonic signal using the blue channel at 940 nm. Raw image stacks were imported into Fiji software (v1.53t; National Institutes of Health) for tile merging. The tiled images were stitched by a grid/collection stitching plugin in Fiji. The merged image stacks were then imported into Imaris software (v10.0.0; Bitplane/Oxford Instruments) for further processing. Melanin autofluorescence from mouse skin was subtracted from the green and red channels. Also, GFP-expressing cells in the epidermis were excluded from images, as these are likely dendritic epidermal T cells^71,72. The filtered red, green and blue channels were then z-projected and shown as a single 2D image, with videos created in Imaris to show the individual slices from processed z-stacks.

Tissue collection and dissociation

After 11 weeks post-operatively, mice were shaved around incision sites (sham) or biomaterials and then euthanized to obtain biopsies. In Ccr2^RFP/+Cx3cr1^GFP/+ mice and Ccr2^RFP/RFPCx3cr1^GFP/GFP mice, rectangular biopsies (9.5 mm long and 3.75 mm wide) around incision sites (sham) or biomaterials were collected. In C57BL/6J mice, circular biopsies (8 mm diameter) were collected. Tissues from different mice belonging to the same group in each study were pooled together for dissociation. Tissues were placed into 10 ml of an enzyme cocktail containing 0.5 mg ml⁻¹ Liberase (Sigma-Aldrich), 0.5 mg ml⁻¹ Collagenase Type IV (Stem Cell Technologies) and 250 U ml⁻¹ deoxyribonuclease I (Worthington Biochemical Corporation) in 25 mM HEPES buffer (Sigma-Aldrich) on a serrated Petri dish. Next, tissues were cut with surgical scissors for ~1 min and moved to an incubator at 37 °C with 5% CO₂ on top of an orbital shaker, shaking at 70 rpm for 1 h. After the incubation period, the Petri dish was removed and 5 ml of the enzyme cocktail and dissociated cells were put through a 70 μm filter into a 50 ml conical tube. In another 5 ml of enzyme cocktail, undigested (residual) tissues were mechanically dissociated by being pressed against the serrated portion of the Petri dish. Afterwards, using a 25 ml pipette, 5 ml of enzyme cocktail was filtered into a 50 ml conical tube. Any undigested tissue on top of the 70 μm filter was further mechanically digested with the thumb press of a syringe plunger for optimal extraction of cells. Using the same 25 ml pipette, 30 ml of cold Hanks’ Balanced Salt Solution without calcium, magnesium and phenol red (ThermoFisher Scientific) was used to wash the digestion Petri dish and filtered into the 50 ml conical tube. The cells in the 50 ml conical tube were centrifuged at 350 × g for 10 min and the supernatant was discarded. Sedimented cells were counted and then used for flow cytometry.

Flow cytometry

Following tissue digestion, for experiments involving Ccr2^RFP/+ Cx3cr1^GFP/+ mice and Ccr2^RFP/RFPCx3cr1^GFP/GFP mice, pooled, dissociated cells were split into n = 3, each containing 1.5 × 10⁶ cells for staining in a polypropylene 96-well round-bottom plate. All staining steps were performed in 100 μl volume in the dark at 4 °C. Samples were first incubated with LIVE/DEAD Fixable Blue Dead Cell Stain Kit (1:500, ThermoFisher, L23105) for 30 min. Cells were washed once with flow buffer (1× phosphate-buffered saline, 0.5% bovine serum albumin), followed by incubation with TruStain FcX PLUS (anti-mouse CD16/32) antibody (BioLegend, 156603; 0.25 μg per sample) for 10 min. The following antibodies were mixed and added directly to the cell suspension: BV421 CD86 (1:200, BioLegend, 105031), PacBlue Ly6G (1:150, BioLegend, 127611), BV605 CD45 (1:300, BioLegend, 103139), BV785 F4/80 (1:300, BioLegend, 123141), PerCP MHCII (1:200, BioLegend, 107623), PE-Dazzle 594 CD11c (1:500, BioLegend, 117347), APC CD206 (1:200, BioLegend, 141707) and AF700 CD11b (1:400, BioLegend, 101222). Cells and antibody mixture were incubated for 30 min. Cells were washed twice with flow staining buffer and fixed with 4% PFA for 10 min and resuspended in a final volume of 100 μl for flow cytometry analysis.

For experiments involving C57BL/6J mice, pooled, dissociated cells were split into n = 3, each containing 1 × 10⁶ cells for staining in a polypropylene 96-well round-bottom plate. All staining steps were performed in 100 μl volume in the dark at 4 °C. Samples were first incubated with LIVE/DEAD Fixable Blue Dead Cell Stain Kit (1:500, ThermoFisher, catalogue number L23105) for 20 min. Cells were washed once with flow buffer, followed by incubation with TruStain FcX (anti-mouse CD16/32) antibody (BioLegend, 101319; 1 μg per sample) in 50 μl volume for 10 min. The following antibodies were mixed together and added directly to the cell suspension: BV605 CD45 (1:500, BioLegend, 103139), AF700 CD11b (1:300, BioLegend, 101222), BV785 F4/80 (1:300, BioLegend, 123141), BV421 CD86 (1:200, BioLegend, 105031), APC CD206 (1:200, BioLegend, 141707), PerCP MHCII (1:200, BioLegend, 107623), SparkBlue 550 CD3 (1:100, BioLegend, 100259), APC-Fire 810 CD4 (1:100, BioLegend, 100479), BB700 CD8a (1:100, BioLegend, 566410), BV711 γδ TCR (1:200, BD Biosciences, 563994), BV480 Thy 1.2 (CD90.2; 1:40, BD Biosciences, 746840), BUV615 CD19 (1:80, BD Biosciences, 751213), PacBlue Ly6G (1:250, BD Biosciences, 127611) and PE-Dazzle 594 CD11c (1:500, BioLegend, 117347). Cells and antibody mixture were incubated for 30 min. Cells were washed once before fixation and permeabilization (BD Cytofix/Cytoperm kit, BDB554714) as per the manufacturer’s instructions. Cells were then resuspended in BD Perm/Wash buffer with the following antibodies: BV650 IL4 (1:50, BD Biosciences, catalogue number 564004), APC-Fire750 IFNγ (1:80, BioLegend, 505859), AF647 IL-17a (1:200, BioLegend, 506911) and PE-Cy7 Arg1 (1:100, ThermoFisher, 25–3697-80). Cells were incubated with antibody mixture for 30 min. Cells were washed twice with BD Perm/Wash buffer followed by resuspension in a final volume of 100 μl for flow cytometry analysis.

All samples were analysed using the Cytek Aurora spectral flow cytometer (Cytek Biosciences), with the Cytek SpectroFlo software (version 3.0.3) for data collection. Fluorescence minus one samples were used to guide gating strategies shown in Supplementary Fig. 2b. Flow cytometry data were analysed with the software FCSExpress (De Novo Software, version 7.12.0005).

Release kinetics of metabolic modulators

To evaluate release kinetics of metabolic inhibitors, 200 mg of pelletized filament (approximately 1 mm long) was suspended in 1 ml of Milli-Q water (n = 3) at 37 °C in an orbital shaker set at 250 rpm for 12 weeks. Every 2 weeks, the releasate (supernatant) was collected and stored at −20 °C and then replaced with 1 ml of Milli-Q water. After collecting the releasate from the 12-week time point, the remaining (undissolved) pellet was suspended in 1 ml chloroform (ThermoFisher Scientific) for complete dissolution. Immediately afterwards, 0.5 ml Milli-Q water was added to partition off chloroform while dissolving the water-soluble metabolic inhibitors. Following a 2 min vortex step to vigorously shake the mixture, samples were centrifuged at 500 rpm for 3 min to separate the aqueous and organic phases, allowing the aqueous phase to be decanted and stored at −20 °C. Cumulative release was plotted (Supplementary Fig. 5) to account for remnant metabolic inhibitors present in undissolved pellets; the sum of the amount of drug released over the 12-week duration as well as remnant was used to determine the total drug content of biomaterial formulations.

Amounts of metabolic inhibitors were measured by liquid chromatography-electrospray ionization mass spectrometry (LC-ESI-MS). a.a. was assessed using a Xevo TQ-S micro Triple Quadrupole Mass Spectrometer (Waters) interfaced with a Thermo Vanquish UPLC. The sample (5 μl) was injected into the Waters Acquity HSS T3 column (2.1 × 100 mm, 1.8 micron) and a.a. was separated using the following gradient at 0.3 ml min⁻¹: initial conditions were 100% mobile phase A (10 mM perfluoroheptanoic acid in water) and 0% mobile phase B (acetonitrile), ramp to 40% B at 3 min, ramp to 90% B at 3.01 min, hold at 90% B until 5 min, return to 100% A at 5.01 min and hold until 8 min. Both 2DG and 3PO were evaluated by a Xevo TQ-XS Triple Quadrupole Mass Spectrometer (Waters) interfaced with a Thermo Vanquish UPLC. 3PO samples were injected onto a Waters Acquity BEH-C18 UPLC column (2.1 × 50 mm) and separated with the following gradient at 0.3 ml min⁻¹: initial conditions were 98% mobile phase A (water + 0.1% formic acid) and 2% mobile phase B (acetonitrile), ramp to 65% B at 3 min, ramp to 99% B at 3.5 min, hold at 99% B until 4 min, return to 98% A at 4.01 min and hold until 6 min. 2DG samples were injected onto a Waters Acquity BEH-Amide UPLC column (2.1 × 100 mm) and separated with the following gradient at 0.3 ml min⁻¹: initial conditions were 5% mobile phase A (10 mM ammonium acetate in water) and 95% mobile phase B (90:10 acetonitrile/water containing 10 mM ammonium acetate), hold at 5% A until 1 min, ramp to 60% A at 4 min and hold until 5 min, return to 10% A at 5.01 min and hold until 8 min. For MS/MS, ions were generated by electrospray ionization in positive mode (a.a., 3PO) or negative mode (2DG). Data were analysed using the TargetLynx tool in the Waters MassLynx (v4.2) software.

Statistics and reproducibility

Statistical software (GraphPad Prism version 9.5.1 (528)) was used to analyse data presented as mean with standard deviation (s.d.). Exact statistical test, P values and sample sizes are provided in figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Extended Data

Extended Data Fig. 1|. Deficiency of CCR2 and CX3CR1 differentially affects the proportion and activation states of cells in the amorphous polylactide biomaterial microenvironment.

a, Ccr2^RFP/RFPCx3cr1^GFP/GFP (CCR2- and CX3CR1-deficient) mice were surgically incised (sham group) or implanted with reprocessed amorphous polylactide (aPLA). Afterwards, intravital microscopy and flow cytometric analysis of tissues around incision sites (sham group) or implants were undertaken. b, Representative intravital microscopy images at different time points around incision (sham group) or implants (scale bars, 50 μm). c-e, Flow cytometry quantification of CCR2⁺(c), CX3CR1⁺(d) and CCR2⁺CX3CR1⁺(e) cells. f-h, Quantification of proinflammatory (CD86⁺CD206⁻) cells among CCR2⁺(f), CX3CR1⁺(g) and CCR2⁺CX3CR1⁺(h) populations. i-k, Quantification of anti-inflammatory (CD206⁺) cells among CCR2⁺(i), CX3CR1⁺(j) and CCR2⁺CX3CR1⁺(k) populations. l-n, Quantification of transition (CD86⁺CD206⁺) cells among CCR2⁺(l), CX3CR1⁺(m) and CCR2⁺CX3CR1⁺(n) populations. o, Nucleated hematopoietic (CD45⁺) cells. p-q, Fold change of proinflammatory (H1; CD86⁺CD206⁻) CD45⁺ cells with respect to transition (T; CD86⁺CD206⁺) CD45⁺ cells (p) or anti-inflammatory (H2; CD206⁺) CD45⁺ cells (q). r-s, Fold change of T (r) or H2 (s) CD45⁺ cells with respect to H1 CD45⁺ cells. t, Neutrophils (CD45⁺Ly6G⁺ cells). u, Monocytes (CD45⁺CD11b⁺ cells). v, Macrophages (CD45⁺F4/80⁺ cells). w, Dendritic cells (CD45⁺CD11c⁺ cells). x, Dendritic cells expressing MHCII (CD45⁺CD11c⁺MHCII⁺ cells). Unpaired t-test (two-tailed), n = 3, mean (SD). Extended Data Fig. 1a was created with BioRender.com.

Extended Data Fig. 2|. CD86 and CD206 proportions in myeloid populations are not modulated by locally targeting immunometabolism at 11 weeks post-implantation of crystalline polylactide biomaterials.

a-b, Fold change of proinflammatory (H1; CD86⁺CD206⁻) cells with respect to transition (T; CD86⁺CD206⁺) cells (a) or anti-inflammatory (H2; CD206⁺) cells (b), gated for nucleated hematopoietic (CD45⁺) populations. c-d, Fold change of T (c) or H2 (d) cells with respect to H1 cells. e-f, Fold change of proinflammatory (M1; CD86⁺CD206⁻) monocytes (CD45⁺CD11b⁺) with respect to transition (T; CD86⁺CD206⁺) monocytes (e) or anti-inflammatory (M2; CD206⁺) monocytes (f). g-h, Fold change of T (g) or M2 (h) monocytes with respect to M1 monocytes. i-j, Fold change of proinflammatory (M1; CD86⁺CD206⁻) macrophages (CD45⁺F4/80⁺) with respect to transition (T; CD86⁺CD206⁺) macrophages (i) or anti-inflammatory (M2; CD206⁺) macrophages (j). k-l, Fold change of T (k) or M2 (l) macrophages with respect to M1 macrophages. m-n, Fold change of proinflammatory (D1; CD86⁺CD206⁻) dendritic (CD45⁺CD11c⁺) cells with respect to transition (T; CD86⁺CD206⁺) dendritic cells (m) or anti-inflammatory (D2; CD206⁺) dendritic cells (n). o-p, Fold change of T (o) or D2 (p) dendritic cells with respect to D1 dendritic cells. q-r, Fold change of proinflammatory (D1; CD86⁺CD206⁻) MHCII⁺ dendritic cells with respect to transition (T; CD86⁺CD206⁺) MHCII⁺ dendritic cells (q) or anti-inflammatory (D2; CD206⁺) MHCII⁺ dendritic cells (r). s-t, Fold change of T (s) or D2 (t) MHCII⁺ dendritic cells with respect to D1 MHCII⁺ dendritic cells. One-way ANOVA followed by Tukey’s multiple comparison test, n = 3, mean (SD); reprocessed crystalline polylactide, cPLA; aminooxyacetic acid, a.a.; 3-(3-pyridinyl)-1-(4-pyridinyl)-2-propen-1-one, 3PO.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41551-024-01260-0.

Data availability

The main data supporting the results in this study are available within the paper and its Supplementary Information. The raw and analysed datasets generated during the study are available for research purposes from the corresponding authors on reasonable request. Source data are provided with this paper.