Matrix-M accumulates inside lysosomes

In order to determine the intracellular localization of Matrix-M after cellular uptake, we employed both BODIPY-labeled Matrix-M and LysoTracker™. BODIPY labeling of Matrix-A and Matrix-C particles enabled us to observe the trafficking of Matrix-M within the cells, while LysoTracker, which selectively accumulates in acidic organelles such as lysosomes upon protonation, allowed assessment of whether Matrix-M localizes to lysosomes through the endocytic pathway³¹. For this, BMDCs were stained with LysoTracker, followed by incubation with BODIPY-conjugated Matrix-M. Using time-lapse spinning-disk confocal imaging, we observed that shortly after exposure, BODIPY-labeled Matrix-M became visible inside the cells (Fig. 1A, Supplementary Video 1) with a time- and dose-dependent increase in BODIPY fluorescent signal (Fig. 1B). Image overlays displayed a substantial overlap between LysoTracker and BODIPY signals (Fig. 1A). Furthermore, by calculating Pearson’s correlation coefficient (PCC) for each frame, a marked increase in the degree of colocalization between LysoTracker and BODIPY-labeled Matrix-M was observed (Fig. 1C, Supplementary Fig. 1). Similarly, cellular uptake and lysosomal localization were observed using BODIPY-labeled Matrix-A or BODIPY-labeled Matrix-C (Fig. 1A–C, Supplementary Videos 2 and 3). Matrix-C displayed a higher fluorescence intensity compared to Matrix-A at the highest tested concentration (5 μg/mL) (Fig. 1B). By contrast, Matrix-A reached a higher level of colocalization with LysoTracker than Matrix-C (Fig. 1C). Altogether, these results indicate an accumulation of Matrix-M, Matrix-A, and Matrix-C within lysosomes upon cellular uptake.

BMDCs were loaded with LysoTracker Red DND-99, washed and imaged for 30 minutes. Following the initial pre-bleaching image acquisition, cells were treated with BODIPY-labeled Matrix-M, Matrix-A, or Matrix-C at concentrations of 1, 2.5, or 5 μg/mL, and live-cell spinning-disk confocal imaging continued for an additional 120 minutes. A Representative images showing individual channels (LysoTracker, BODIPY), overlays, and the DIC (differential interference contrast) channel captured after each treatment (scalebar = 10 μm). B The fluorescence intensity kinetics of BODIPY fluorescence for the indicated Matrix concentrations. C Pearson’s correlation coefficients (PCC) for BODIPY and LysoTracker fluorescence over time. Data shown in B and C represent smoothed averages from at least three biological replicates ±SEM.

Matrix-M induces lysosomal membrane permeabilization (LMP) leading to the release of lysosomal contents into the cytosol

To evaluate the effect of Matrix-M on lysosomal membrane integrity, which is required to maintain a low intra-lysosomal pH, we assessed lysosomal acidification by monitoring LysoTracker cellular staining using flow cytometry. In viable BMDCs (gating strategy shown in Supplementary Fig. 2), exposure to Matrix-M adjuvant resulted in a reduction of LysoTracker staining (Fig. 2A, B). Similar reductions in LysoTracker staining were observed following exposure to Matrix-A and Matrix-C, indicating that all three Matrix adjuvants disrupt the pH gradient across the lysosomal membrane, or disrupt the lysosomal membranes themselves. As expected, the positive controls, BafA1 and LLOMe, also reduced LysoTracker staining (Fig. 2A, B). BafA1 and LLOMe disrupt lysosomal acidification via distinct mechanisms. BafA1 inhibits the vacuolar ATPase (V-ATPase) proton pump responsible for maintaining low lysosomal pH³², whereas LLOMe induces LMP by forming membrane pores^33,34. Visible Matrix-M–induced reduction of LysoTracker signal could also be noted among single cells imaged by live-cell spinning-disk confocal microscopy (Fig. 2C, Supplementary Video 4).

BMDCs were treated with either Matrix-M, Matrix-A, or Matrix-C (for 1 h), bafilomycin (BafA1, for 1 h), or L-Leucyl-L-Leucine methyl ester (LLOMe, for 15 min) at the indicated concentrations followed by staining with LysoTracker Deep Red and FVS520 for 15 and 7 min, respectively. Stained cells were analyzed by flow cytometry. A Flow cytometry histograms showing LysoTracker fluorescence intensity in viable BMDCs from representative samples (gating shown in Supplementary Fig. 2). B Normalized geometric mean fluorescence intensity (gMFI) of LysoTracker Deep Red in BMDCs. Data were normalized to untreated control samples and are shown as the mean ± SD; pooled from three independent experiments. Statistical differences between untreated control and treated samples were determined by one-way ANOVA followed by Dunnett’s multiple comparisons test (****p < 0.0001). C Images of LysoTracker-stained BMDCs before and after treatment with BODIPY-labeled Matrix-M, as described in Fig. 1. Scalebar = 10 μm.

We next sought to investigate whether the Matrix-induced loss of lysosomal acidification could be attributed to pore-forming activity of Matrix-M, leading to LMP. To this end, we monitored induction of LMP by tracking the translocation of fluorescent dextran particles (3- and 40-kDa) from lysosomes to the cytosol using time-lapse spinning-disk confocal imaging³⁵. Before the addition of Matrix-M, cells had intact lysosomes as indicated by the punctate morphology of the fluorescent dextran particles (Fig. 3A). However, a diffuse cytosolic staining pattern for both 3- and 40-kDa dextran particles appeared in the cells upon Matrix-M treatment, demonstrating the induction of pores permeable for particles up to 40-kDa size (Fig. 3A, Supplementary Video 5). Similarly, the treatment of dextran-loaded BMDCs with Matrix-A or Matrix-C resulted in the release of lysosomal content into the cytosol (Fig. 3A, Supplementary Videos 6 and 7). Of note, Matrix-M and Matrix-C were found to be more potent than Matrix-A in promoting LMP and dextran release into the cytosol (Fig. 3B, note different concentrations of Matrix adjuvants). Given that lysosomal hydrolytic enzymes depend on acidic pH for optimal activity, we next assessed whether lysosomal acidification is required for the LMP-inducing effect of Matrix adjuvants. For this, BMDCs preloaded with dextran particles were incubated with the lysosomal acidification inhibitor BafA1 for 1 h before exposure to Matrix-M, Matrix-A, or Matrix-C. Notably, inhibition of lysosomal acidification completely blocked the release of both 3- and 40-kDa dextran particles from the lysosome to the cytosol in response to Matrix-M and Matrix-A (Fig. 3B). In response to Matrix-C, blockade of lysosomal acidification resulted in complete inhibition of 40-kDa dextran release, but only partial inhibition of 3-kDa dextran release into the cytosol (Fig. 3B). These results demonstrate that all three Matrix adjuvants induce LMP, albeit with distinct potencies, as Matrix-C and Matrix-M exhibited stronger LMP-inducing effects compared to Matrix-A.

BMDCs were preloaded overnight with dextran particles of different sizes and fluorochromes (Alexa Fluor™ 488–conjugated 3-kDa dextran and tetramethylrhodamine-conjugated 40-kDa dextran). The following day, cells were pretreated with bafilomycin A1 (BafA1, 30 nM) or left untreated for 1 hour. Live-cell spinning-disk confocal imaging was performed at 3-min intervals for up to 5 h, before and after adding Matrix-M, Matrix-A (both at 10 μg/mL), or Matrix-C (1 μg/mL), to monitor lysosomal release of dextran particles into the cytosol. A Representative images showing individual channels (3- and 40-kDa dextrans) and overlays at the indicated time points and treatment conditions (scalebar = 20 μm). B The percentage of cells with intact lysosomes over time for each treatment condition. Data are pooled from three independent experiments and presented as percent cells with intact lysosomes ±95% confidence interval.

Matrix-M induces inflammasome activation and IL-1β secretion

Release of lysosomal contents into the cytosol following LMP can initiate inflammasome activation, a process essential for converting pro–IL-1β into its active form and the subsequent release of biologically active IL-1β. To assess the ability of Matrix-M to induce inflammasome activation, non-primed and LPS-primed BMDCs were treated with increasing doses of Matrix-M for 5 h, and released IL-1β levels were measured in supernatants by ELISA. Matrix-M induced the release of IL-1β from LPS-primed BMDCs, with higher doses of Matrix-M resulting in greater IL-1β secretion (Fig. 4A). Similarly, a dose-dependent secretion of IL-1β was observed following exposure of the cells to Matrix-A or Matrix-C. Notably, Matrix-M induced higher levels of IL-1β than either Matrix-A or Matrix-C. By contrast, non–LPS-primed BMDCs did not release IL-1β upon treatment with any of the Matrix adjuvants. To investigate whether lysosomal acidification and thereby LMP is required for Matrix-induced IL-1β secretion, LPS-primed BMDCs were pretreated with increasing doses of lysosomal acidification inhibitor BafA1 before Matrix adjuvant treatment. BafA1 blocked the release of IL-1β induced by Matrix-M, Matrix-A, and Matrix-C in a dose-dependent manner (Fig. 4B), indicating that lysosomal acidification is crucial for Matrix-induced IL-1β secretion and suggesting a link to LMP. Of note, blockade of lysosomal acidification also prevented the reduction in cell viability observed in response to Matrix adjuvants (Supplementary Fig. 3A).

BMDCs were either non-primed (A) or LPS-primed (A–G) and treated with increasing doses of Matrix-M, Matrix-A, or Matrix-C for 5 hours. Released IL-1β levels were measured in the supernatants by ELISA. Before treatment with Matrix adjuvants, cells were pretreated for 1 h with no inhibitor (A–G), V-ATPase inhibitor bafilomycin A1 (BafA1) (B), cathepsin B inhibitor CA074-Me (C), NLRP3-related inhibitors MCC950 (D) and Ac-YVAD-cmk (F), or gasdermin D pore formation inhibitor disulfiram (G) at the indicated concentrations. Additionally, IL-1β levels were measured in NLRP3-deficient and wild-type (WT, C57BL/6) BMDCs after Matrix adjuvant treatment (E). Results are shown as the mean ± SD from at least three independent experiments.

One key lysosomal mediator known to have inflammasome activating capability after its release into the cytosol is cathepsin B (CatB)³⁶. To test whether CatB contributes to Matrix-induced IL-1β release, LPS-primed BMDCs were pretreated with the CatB inhibitor CA074-Me before Matrix-M, Matrix-A, or Matrix-C treatment. Inhibition of CatB substantially lowered IL-1β levels induced by Matrix-M and Matrix-A, whereas Matrix-C–induced IL-1β was only marginally affected (Fig. 4C).

Among the various inflammasomes, the NLRP3 inflammasome is of particular interest, as lysosomal damage is one of its known activating triggers³⁷. To further investigate whether Matrix-induced IL-1β release occurs through NLRP3 inflammasome activation, LPS-primed BMDCs were pretreated with MCC950, an inhibitor of NLRP3 inflammasome assembly. This effectively blocked Matrix-induced IL-1β release (Fig. 4D). In line with this, NLRP3-deficient BMDCs failed to release IL-1β upon exposure to Matrix adjuvants (Fig. 4E). Given that NLRP3 inflammasome activation results in the activation of caspase-1, which processes pro–IL-1β to mature IL-1β, the dependency of IL-1β release on caspase-1 activation was tested. Pretreatment with the caspase-1 inhibitor Ac-YVAD-cmk resulted in a marked reduction of IL-1β secretion in response to Matrix-M, Matrix-A, and Matrix-C without affecting cell viability (Fig. 4F, Supplementary Fig. 3B). This demonstrates that NLRP3 inflammasome activation and caspase-1 are crucial for Matrix-induced IL-1β secretion.

IL-1β lacks the signal sequence required for ER/Golgi-mediated secretion and is consequently released from cells through an unconventional secretion pathway involving gasdermin D (GSDMD) pores³⁸. To investigate the potential role of GSDMD in Matrix-induced IL-1β release, LPS-primed BMDCs were pretreated with disulfiram and necrosulfonamide (NSA), both inhibitors of GSDMD that block pore formation by preventing GSDMD oligomerization^39,40. Both inhibitors dose-dependently blocked IL-1β release induced by Matrix-M, Matrix-A, and Matrix-C (Fig. 4G, Supplementary Fig. 3C), showing that Matrix-induced IL-1β is released through GSDMD pores. Inflammasome-mediated GSDMD pore formation can result in either sublytic IL-1β release without causing cell death or lytic pyroptosis⁴¹. Blocking GDSMD pore formation did only marginally or not at all change cell viability observed after exposure to Matrix adjuvants (Supplementary Fig. 3D, E).

Matrix-M induces IL-18 release independently of NLRP3 inflammasome activation

In addition to IL-1β, IL-18 can be activated through NLRP3 and other inflammasomes. To evaluate the role of Matrix-M, Matrix-A, and Matrix-C in IL-18 release, we measured the levels of IL-18 secreted by non-primed and LPS-primed BMDCs using ELISA. IL-18 release was observed in response to all three adjuvants, with non-primed cells secreting higher levels than LPS-primed cells (Fig. 5A). Inhibition of lysosomal acidification again completely blocked Matrix-induced IL-18 release, linking this process to Matrix-induced LMP (Fig. 5B). However, and in contrast to observations for IL-1β, Matrix-induced IL-18 secretion was not dependent on NLRP3 inflammasome activation. In fact, inhibition of NLRP3 with the assembly inhibitor MCC950, or the use of NLRP3-deficient BMDCs, resulted in elevated IL-18 secretion responses compared to controls (Fig. 5C, D).

BMDCs were either non-primed (A–D) or LPS-primed (A). BMDCs were treated with increasing doses of Matrix-M, Matrix-A, or Matrix-C for 24 hours. Released IL-18 levels were measured by ELISA. Before treatment with Matrix adjuvants, cells were pretreated for 1 h with V-ATPase inhibitor bafilomycin A1 (BafA1) (B) or NLRP3-assembly inhibitor MCC950 (C) at the indicated concentration. Similarly, IL-18 levels were measured in NLRP3-deficient and wild-type (WT, C57BL/6) BMDCs after Matrix adjuvant treatment (D). Results are shown as mean ± SD of three independent experiments.

NLRP3 inflammasome activation is not required for Matrix-M adjuvanticity in vivo

Given the observed Matrix-induced NLRP3 inflammasome activation in vitro, we investigated whether this activation was essential for the adjuvanticity of Matrix-M in vivo. Wild-type (WT, C57BL/6 mice) and NLRP3-deficient mice (Nlrp3 KO mice) were immunized in a two-dose immunization regimen with the SARS-CoV-2 rS antigen. At days 13 and 20 post-primary immunization as well as at day 28 (7 days post-secondary immunization), similar titers of rS-specific IgG1, IgG2c, and functional human angiotensin converting enzyme (hACE)-2 receptor-binding blocking antibodies were found in sera of WT and Nlrp3 KO mice, with both adjuvanted groups showing superior responses compared to their respective unadjuvanted antigen group (Fig. 6A–C). Comparable results were observed using Matrix-A and Matrix-C as adjuvants, although Nlrp3 KO mice displayed higher hACE2 binding inhibition titers than WT mice in the Matrix-A–adjuvanted group at day 13 (Fig. 6C).

C57BL/6 (WT) and Nlrp3 KO mice (n = 10/group) were immunized with 0.1 µg SARS-CoV-2 rS (rS) alone or adjuvanted with 5 μg Matrix-M, Matrix-A, or Matrix-C. Serum samples obtained 13, 20, and 28 days after the primary immunization were evaluated by ELISA for IgG1 (A) and IgG2c (B) antibody titers against rS protein and for hACE2 receptor-inhibiting antibody titers (C). At day 28, splenocytes from individual mice were isolated and the number of cells producing IFN-γ, IL-2, or IL-4 in response to rS protein restimulation was determined by FluoroSpot assay (D). Individual titers (ELISA) or spot forming units (SPU)/10⁶ splenocytes are shown as individual symbols, with horizontal bars representing geometric mean values, and error bars represent their 95% confidence intervals. Additionally, splenocytes were analyzed by flow cytometry (intracellular staining) to access frequencies of rS-specific CD4⁺ and CD8⁺ T cells producing IFN-γ and IL-4 (E). Data for each individual mouse response are shown with symbols and group means are represented by horizontal bars. The data (A–D) were log-transformed before being analyzed. The data (A–E) were analyzed for differences between WT and and Nlrp3 KO mice within each adjuvanted group by one-way ANOVA with Šidák’s multiple comparisons test (*for p < 0.05, ** for p < 0.01).

Analysis of splenocytes producing IFN-γ, IL-2, or IL-4 following rS restimulation at day 28 showed clear adjuvant effects across Matrix-M–, Matrix-A–, and Matrix-C–adjuvanted groups, regardless of the presence or absence of the NLRP3 inflammasome (Fig. 6D). ICS analysis revealed similar frequencies of antigen-specific IFN-γ–producing CD4⁺ and CD8⁺ T cells, as well as IL-4–producing CD4⁺ T cells in WT and Nlrp3 KO mice in the non-adjuvanted and Matrix-M– and Matrix-A–adjuvanted groups (Fig. 6E). In the Matrix-C–adjuvanted groups, a higher frequency of antigen-specific IFN-γ–producing CD4⁺ T cells was observed in Nlrp3 KO mice, while similar frequencies of IFN-γ–producing CD8⁺ and IL-4–producing CD4⁺ T cells were found in Matrix-C–adjuvanted WT and Nlrp3 KO mice. Collectively, the data demonstrate that Matrix-M, Matrix-A, and Matrix-C show potent adjuvant effects in vivo both in the presence and absence of the NLRP3 inflammasome.

Matrix-M enables antigen cross-presentation in BMDCs

The localization of Matrix-M inside lysosomes together with its observed pore-forming activity, could allow exogenous antigens to reach the cytosol. This, in turn, may enable endocytosed antigens to enter the MHCI processing pathway, thereby promoting antigen cross-presentation, a mechanism that has been exemplified by other SBAs^16,25,28. Given that Matrix-A and Matrix-C particles in Matrix-M are not physically linked to antigens and thus Matrix-M does not serve as a direct antigen delivery system, we first evaluated the colocalization of the antigen with lysosomes and Matrix-M as a prerequisite condition for cross-presentation. LysoTracker-loaded BMDCs were incubated with BODIPY-conjugated Matrix-M together with the Alexa Fluor 647–labeled SARS-CoV-2 rS antigen and monitored by time-lapse live-cell spinning-disk confocal imaging (Supplementary Fig. 4). PCC analysis revealed a marked time-dependent increase in the degree of colocalization between LysoTracker and the antigen, confirming antigen internalization into the lysosomes (Fig. 7A). In parallel, colocalization between Matrix-M and the antigen was also confirmed (Fig. 7B). These results demonstrate that both Matrix-M and the antigen accumulate within lysosomes upon cellular uptake.

BMDCs were loaded with LysoTracker Red DND-99, washed and imaged using spinning-disk confocal microscopy for 30 minutes. Following the initial pre-bleaching image acquisition, cells were treated with Alexa Fluor 647–labeled rS (0.5 or 1 μg/mL) and BODIPY-Matrix-M (1 or 2.5 μg/mL) and live-cell imaging continued for an additional 120 min. Pearson’s correlation coefficients (PCC) for the channel combination LysoTracker and Alexa Fluor 647 (A), and BODIPY and Alexa Fluor 647 (B) were determined as indicators of lysosome–antigen or Matrix‒M–antigen colocalization, respectively. Data in panels A and B are presented as smoothed averages ± SEM, pooled from at least three experiments, each performed in triplicates. C–G BMDCs were exposed to 0.8 mg/mL ovalbumin (OVA) and indicated concentrations of Matrix-M, Matrix-A, or Matrix-C for 24 hours. Cells were analyzed by flow cytometry for surface markers and for H-2kB bound to SIINFEKL complexes, with the latter indicative of antigen cross-presentation. The percentage of viable BMDCs positive for H-2kB/SIINFEKL complexes is shown as representative contour plots (C) and corresponding graphs (D) for untreated and Matrix-treated groups. E Representative flow cytometry plots and F graphs show the proportion of major subpopulations within BMDC culture, identified as non-DC-like cells (CD11c⁻ MHCII⁻, dark blue), immature DC-like cells (CD11c⁺ MHCII^−/lo, turquoise) and mature DC-like cells (CD11c⁺ MHCII^hi, gray). G The percentage of various BMDC subpopulations within cells positive for H-2Kb/SIINFEKL complexes are shown. Data in panels C–E are representative of three independent experiments, while graphs in F and G show pooled data from three independent experiments, presented as the mean ± SD.

The ability of Matrix-M to induce antigen cross-presentation was subsequently investigated using the 25-D1.16 monoclonal antibody, which specifically recognizes cell surface complexes of the OVA peptide SIINFEKL bound to the MHCI molecule H-2Kb. BMDCs treated with OVA and Matrix-M for 24 h showed a Matrix-M–induced cross-presentation (Fig. 7C, D). Similarly, antigen cross-presentation was observed in BMDCs after treatment with Matrix-A and Matrix-C (Fig. 7C, D). When comparing the cross-presentation–inducing potential of the Matrix adjuvants, Matrix-C elicited the highest frequency of antigen cross-presenting cells at Matrix concentrations of 5 µg/mL or lower, followed by Matrix-M, while Matrix-A induced the lowest level of cross-presentation. Both Matrix-M and Matrix-A promoted a concentration-dependent increase in the frequency of antigen cross-presenting cells, whereas treatment with Matrix-C at concentrations above 5 µg/mL resulted in a reduced frequency of cross-presenting BMDCs. Phenotypic analysis of BMDCs, categorized into non–dendritic cell (DC)-like, immature DC-like, and mature DC-like populations (gating strategy shown in Supplementary Fig. 5), showed a decrease in the proportion of immature DC-like cells after Matrix treatment (Fig. 7E, F). Antigen cross-presentation was primarily driven by mature DC-like cells at 5 and 25 µg/mL Matrix concentrations, irrespective of the Matrix formulation (Matrix-M, Matrix-A, or Matrix-C) used (Fig. 7G). Furthermore, the effect of Matrix adjuvants on BMDC activation was investigated, with Matrix-M and Matrix-A showing the strongest effect, as indicated by increased frequencies of MHCII⁺ and CD86⁺ cells (Supplementary Fig. 6A–H). At higher concentrations (5 and 25 µg/mL), Matrix-A treatment led to higher proportions of MHCII⁺ and CD86⁺ BMDCs compared to Matrix-M (Supplementary Fig. 6C, G). The upregulation of activation markers MHCII and CD86 on BMDCs was reduced in NLRP3-deficient cells following Matrix-M, Matrix-A, or Matrix-C treatment (Supplementary Fig. 7A, B). Altogether, these results suggest that the combination of Matrix-C and Matrix-A in Matrix-M effectively merges Matrix-C’s potential for antigen cross-presentation with Matrix-A’s ability to promote APC activation.

Lysosomal acidification and proteasomal degradation contribute to Matrix-induced antigen cross-presentation

To assess the ability of APCs to cross-prime CD8⁺ T cells, BMDCs were co-cultured with B3Z cells, a co-stimulation independent reporter T-cell line⁴². The B3Z cross-presentation assay, in which the presentation of SIINFEKL via H-2Kb activates T-cell receptor signaling inducing β-galactosidase expression, showed a concentration dependent induction of cross-priming, with a higher potential of Matrix-C compared to Matrix-M and Matrix-A (Fig. 8A). Importantly, the exposure of cells to Matrix-M, Matrix-A or Matrix-C did not alter MHCI presentation or cell viability, as presentation of pulsed SIINFEKL was similar or slightly reduced within the tested concentration range (Fig. 8B). Blockade of lysosomal acidification showed that Matrix-induced cross-presentation was dependent on a low lysosomal pH, suggesting a link to LMP (Fig. 8C, D), consistent with the observed inhibition of dextran release by BafA1 (Fig. 3B). For the cross-presentation of exogenous antigens, two main intracellular pathways, the “endosomal” and “cytosolic,” have been proposed⁴³. Given that Matrix-induced dextran release indicated antigen translocation into the cytosol, the role of proteasomal degradation, a key component of the cytosolic pathway in cross-presentation, was examined. Inhibition of proteasomal activity by epoxomicin resulted in a strong reduction in cross-presentation induced by Matrix-M, Matrix-A, and Matrix-C, confirming that Matrix-induced antigen translocation into the cytosol underlies Matrix-driven cross-presentation (Fig. 8E, F).

BMDCs were exposed to 80 µg/mL OVA and indicated concentrations of Matrix-M, Matrix-A, or Matrix-C for 5 hours (A, C, E). In control experiments, 5 ng/mL SIINFEKL instead of OVA was added (B, D, F). Cells were co-cultured for 18 h with B3Z cells, which produce β-galactosidase upon TCR recognition of SIINFEKL presented on H-2kB. BMDCs were pretreated for 1 h with bafilomycin A1 (BafA1) to inhibit lysosomal acidification (C, D) or with epoxomicin (E, F) to inhibit proteasomal activity. Results shown are representative of two independent experiments, with data presented as mean values of technical triplicates ± SD.

Matrix-M induces dose-dependent CD8⁺ T-cell responses in vivo

To further evaluate the ability of Matrix-M to induce antigen cross-presentation and generate antigen-specific CD8⁺ T-cell responses, two different approaches were employed. First, antigen cross-presentation was assessed ex vivo by co-culturing sorted CD11c⁺ DCs, which were loaded in vivo with OVA either in presence or absence of Matrix-M, with naïve OT-I CD8⁺ T cells that carry an OVA peptide-specific T-cell receptor. Immunization with OVA adjuvanted with Matrix-M resulted in enhanced proliferation and activation of naïve OVA-specific CD8⁺ T cells in ex vivo co-cultures (Fig. 9A–D). Analysis of the CD11c-sorted DCs showed a shift in proportions of DC subsets, with an increase in monocyte-derived DCs (moDCs) and a decrease in conventional DC1 and DC2 (cDC1 and cDC2) subpopulations in the group receiving Matrix-M (Supplementary Fig. 8A, B). Moreover, Matrix-M treatment led to increased activation of DCs in the draining lymph nodes (dLNs) (Supplementary Fig. 8C).

C57BL/6 mice were immunized with OVA alone or adjuvanted with 5 µg Matrix-M. Pools of DCs isolated from the dLNs 24 h post-immunization were co-cultured for 3 days with naïve, CellTrace Violet (CTV)-stained CD8⁺ OT-I T cells. Representative histograms for CTV dilution are shown (A); these were used to calculate the percentage of divided cells by FlowJo proliferation analysis (B). Surface expression (median fluorescence intensity; MFI) of the T-cell activation marker CD44 within CD8⁺ T cells was analyzed, as displayed in the representative contour plot and corresponding graph (C, D). Each symbol represents one individual pool, with the solid horizontal bars representing group means ± SD. Unpaired t-test was applied for comparisons between the two groups (B, D). E, F BALB/c mice (n = 10/group) were immunized with 0.1 µg SARS-CoV-2-rS alone or adjuvanted with the indicated doses of Matrix-M (E), Matrix-A, or Matrix-C (F). Splenocytes from individual mice were isolated to determine the frequency of CD8⁺ T cells that produce IFN-γ after rS protein stimulation. Each symbol represents an individual mouse; horizontal bars represent group means. One-way ANOVA with Tukey’s multiple comparisons test was applied to compare the groups to the antigen only (p-values * for p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

In a second approach, the effect of Matrix-M adjuvant on CD8⁺ T-cell responses was investigated in vivo using a two-dose immunization scheme with SARS-CoV-2 rS protein, where different doses of Matrix-M were administered. ICS for IFN-γ demonstrated a dose dependent induction of rS-specific CD8⁺ T-cell responses (Fig. 9E). In addition, a dose response was observed for the generation of specific serum antibodies, cellular responses, and CD4⁺ T-cell responses (Supplementary Fig. 9A–D). Similarly, a dose-dependent induction of antigen-specific CD8⁺ T cells was observed when varying doses of Matrix-A and Matrix-C adjuvants were used for rS immunizations (Fig. 9F). Although non-adjuvanted rS immunization did not elicit detectable rS-specific CD8⁺ T cells, the addition of 2.5 µg Matrix-M and Matrix-C or 5 µg Matrix-A was sufficient to induce CD8⁺ T-cell responses. Altogether these data demonstrate the ability of Matrix-M, as well as that of Matrix-A and Matrix-C, to induce specific CD8⁺ T cells responses to exogenous antigens, as demonstrated with two distinct antigens, OVA and SARS-CoV-2 rS protein.