This study involving human participants was reviewed and approved by Istituti Fisioterapici Ospitalieri (IFO) Ethical Committee for the use of muscle samples from healthy donors undergoing surgeries (CE number: 1900/DSa IRE). Muscle biopsies were taken during orthopedical intervention, and all of the participants provided their written informed consent by the IFO in line with the Declaration of Helsinki. Istituto Superiore di Sanità Review Board approved the use of blood from healthy donors (CE number: AOO-ISS – 22/12/2023–0059852 PRE BIO CE 01.00). Written informed consent to participate in this study was provided by the personnel of the Blood Transfusion Service and Hematology Department of Umberto I Hospital (Rome, Italy).

Human foreskin fibroblasts BJ (ATCC CRL-2522) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco), 1% Glutamax (Gibco), and 1% penicillin/streptomycin (Gibco). Sub-confluent cultures (3 × 10⁶ cells/dish) were obtained in 10-cm dishes (6 × 10⁵ cells/mL), with the medium refreshed 6 hours (h) before treatment with anti-COVID-19 BNT162b2 mRNA vaccine (VAX, Pfizer-Biontech). The fibroblasts were treated with 1 µg/mL of VAX immediately after its reconstitution. Culture supernatants and cell pellets were collected at the indicated times and stored at -80 °C for later analysis as specified below.

Pericytes (myogenic progenitors, MP) were isolated from muscle biopsies obtained as res nullius from healthy donors undergoing orthopedical surgeries at IFO. Briefly, human muscle biopsies were freshly processed (within 24 h from harvesting) by enzymatic digestion with 100 U/mL of Collagenase Type II to obtain a suspension of mononucleated cells and afterward plated at low confluence (1.000 cells/cm²) for the selection of alkaline phosphatase (ALP)-positive MP and colony formation. Isolated cells were tested by cytofluorimetric (FACS) analysis to evaluate the expression of specific MP markers, namely, ALP (Novus Biologicals, NB110-3638PE), neural/glial Ag 2, melanoma cell adhesion molecule (CD146) (Novus Biologicals, NBP2-72595AF700), platelet-derived growth factor receptor b (CD140b) (BD Biosciences, 743035BV605), Thy-1 Ag (CD90; Beckman Coulter, IM1839U FITC), and neural cell adhesion molecule (CD56; Beckman Coulter, B92446 AF700) (21). To obtain a homogeneous population of MP (double-positive cells CD90⁺/CD56⁺), isolated cells were cultured into CYTO-GROW (Resnova) medium supplemented with 100 U/mL penicillin and 100 μg/ml streptomycin (Gibco) at 37°C and 5% CO₂ humid atmosphere as previously described (21, 22). To induce myogenic differentiation, 3 × 10⁵ MP were plated in six-well Petri dishes and cultured for up to 21 days in supplemented CYTO-GROW medium and placed at 37°C and 5% CO₂ humid atmosphere until reaching the complete differentiation condition (21). MP and MT (1 × 10⁶ cells at the density of 1 × 10⁶ cells/mL) were treated with VAX (1 µg/mL) immediately after its reconstitution. Culture supernatants and cell pellets were collected at the indicated times and stored at -80 °C for later analysis as specified below.

All of the experiments were performed using poly-ethylene glycol fibrinogen (PF) biomimetic matrix. In particular, for the generation of each 3D-MT, 8 mg/mL PF, supplemented with 0.1% Irgacure™ 2959 photoinitiator solution (Ciba Specialty Chemicals), was loaded with 1 × 10⁶ MP. The fork was placed in the mold, and the cell–hydrogel mixture was poured into a polytetrafluoroethylene (Teflon) caster (23). The mixture was polymerized by exposing the mold chamber to non-toxic and low-penetrating UV light for 5 min ( Supplementary Figures S1A, B ). Polymerized constructs were gently removed and transferred to normal 35-mm dishes containing 1 mL of CYTO-GROW medium supplemented as previously described. The vaccine was used fresh after reconstitution at the final concentration of 1 µg/mL. Then, 3D-MT were vaccine-injected with 100 µL of VAX dilution through an insulin syringe ( Supplementary Figure S1C ). Culture media and the corresponding 3D-MT were collected at the indicated time points and stored at -80 °C for later analysis as specified below.

PBMC were isolated from freshly collected buffy coats derived from blood donation of healthy volunteers at Blood Transfusion Service and Hematology Department of Umberto I Hospital (Rome, Italy) by density gradient centrifugation using lympholyte-H (Cedarlane) as previously described (24). Briefly, PBMC were cultured at 2 × 10⁶ cells/mL in RPMI 1640 supplemented with penicillin/streptomycin (100 U/mL), L-glutamine (2 mM), and 10% FBS for 24 or 72 h and then stimulated with 1 µg/mL of freshly reconstituted VAX or with conditioned media (CM) from fibroblast cultures or muscle cell models used at v/v ratio 1:5 and 1:10. For PBMC treatment, CM were collected at 24 or 72 h from unstimulated or VAX-treated fibroblasts, while CM were harvested at 24 h from not stimulated (NS) and VAX-exposed 2D MT and 3D-MT. To evaluate also a possible contribution of the residual VAX contained in CM after thawing, frozen VAX (FRZ. VAX) was also used at 1, 0.2, and 0.1 µg/mL to treat PBMC. After 24 h, culture supernatants from unstimulated and treated PBMC were collected and stored at -80 °C for later analysis, while cell pellets were employed for gene expression studies, confocal microscopy analysis, or evaluation of APC phenotype (see below).

Human BJ fibroblasts were cultured in 24-well plates attached on round glass coverslips (50.000 cells/well) and treated with VAX for 24 h as described above or NS as negative control. PBMC, NS or stimulated with VAX for 24 h as previously indicated, were collected, and 1 × 10⁵ cells were adhered on microscopy slides by cytospin. Both cell types were fixed in 4% paraformaldehyde (PFA, Sigma Aldrich) at room temperature (RT) and then permeabilized with 0.1% Triton X-100 (BioRad) and blocked with 4% bovine serum albumin (BSA, Sigma Aldrich) overnight (o.n.) at 4 °C or 1 h at RT. Immunofluorescence assay was performed by incubation with primary Ab against SARS-CoV-2 S1 protein (40150-R007, Sino Biological) diluted 1:500 followed by secondary anti-rabbit IgG conjugated with AlexaFluor 488 (Invitrogen, Thermo Fisher Scientific A11008) diluted 1:1,000. Finally, Draq5 fluorescent probe (Invitrogen, Thermo Fisher Scientific) diluted 1:2,000 was added for 10 min at RT for staining of cell nuclei. Immunofluorescence preparations were observed with Nikon Eclipse Ti2 confocal microscope, and images were acquired and analyzed with LAS V3.8 (Leica) software.

Not stimulated or VAX-exposed 2D MP and MT as well as 3D-MT were instead fixed with 2% PFA and processed for immunofluorescence analysis. Briefly, the samples were permeabilized with 0.3% Triton X-100 (Sigma Aldrich) in PBS 1X without Ca²⁺ and Mg²⁺ (Biowest) for 1 h at RT and then incubated with a blocking solution, consisting of PBS 1X, 0.1% Triton X-100, 1% glycine, and 10% natural goat serum for 1 h at RT. Afterward, 2D MP were incubated for 1 h at RT with Phalloidin Alexa Fluor 555 (Invitrogen, Thermo Fisher Scientific, A12379) for labeling F-actin of cytoskeleton, while MT and 3D-MT were incubated for 1 h and o.n. at RT, respectively, with anti-myosin heavy chain (anti-MF20) mouse primary antibody (DSHB, AB_2147781) diluted 1:200 to assess terminal myogenic differentiation. For the detection of SARS-Cov-2 S1 protein, Ab was used as previously specified by staining MP and MT for 1 h, while 3D-MT were incubated o.n. at RT.

The samples were then washed three times and incubated 1 h at RT with Alexa-fluor555-conjugated anti-mouse IgG Ab (Invitrogen, Thermo Fisher Scientific A21422) secondary antibody at 1:200 dilution for anti-MF20 detection and with AlexaFluor 488-conjugated anti-rabbit IgG Ab (Invitrogen, Thermo Fisher Scientific, A11008) diluted 1:1,000 for S1 protein detection. After the indicated time, the washing step was repeated as previously indicated, and labeling of nuclei was performed by incubating the samples with 0.1 μg/mL of DAPI (4′,6-diamidino-2-phenylindole dihydrochloride) for 15 min at RT. Finally, the 3D-MT samples were mounted with Aqua Poly/Mount (Polysciences, Inc). Samples were photographed using a DMI6000B fluorescent microscope (Leica), and images were acquired and analyzed with Leica Application Suite X software.

The release of soluble factors was studied in culture supernatants from 24- and 48-h vaccine-stimulated fibroblasts, MP, MT, and 3D-MT as well as in PBMC exposed for 24 h, as described above.

Secretions of IL-8 and RANTES/CCL5 were measured by ELISA using DuoSet ELISA Kit (R&D Systems), while the release of IL-6, TNF-α, MCP-1/CCL2, MIP-1β/CCL4, IP-10/CXCL10, IFNγ, and IL-15 was quantified by custom plate simple plex cartridge kit on an ELLA automated immunoassay system (Bio-Techne, San Jose, CA, USA) according to the manufacturer’s instructions and analyzed by the Simple Plex Explorer Program software (Bio-Techne) as previously described (25).

Interferon (IFN)-αs production was determined in culture supernatants by Verikine Human IFN-α multi-subtype ELISA kit (PBL Assay Science). PTX3 levels were measured, as previously described (26), using a sandwich ELISA developed in-house. The assay has a lower limit of detection of 0.1 ng/mL and inter-assay variability between 8% and 10%, and it does not cross-react with C-reactive protein.

Total RNA was isolated by Trizol Reagent (Invitrogen, Thermo Fischer Scientific) from the different cell models, quantified using a NanoDrop ONEc spectrophotometer and quality-assessed with an established cutoff of ~1.8 for 260/280 absorbance ratio. Reverse-transcription was conducted by Vilo reverse transcriptase kit (Invitrogen, Thermo Fischer Scientific).

The expression of genes encoding MX dynamin-like GTPase 1 (MX1), IFN regulatory factor 1 (IRF1), IFNB1, and IFNL1 was measured at 24 and 48 h by quantitative real-time PCR (q-PCR) using the appropriate TaqMan assay and TaqMan Universal Master Mix II (Applied Biosystems, Thermo Fisher Scientific) on a ViiA7 Instrument (Applied Biosystems, Thermo Fisher Scientific). RNA sensors were studied at 24 h when the expression of endosomal and cytoplasmic receptors can be observed in all of the analyzed cellular systems following specific stimulation. In particular, for TLR3, TLR7, TLR8, RIG-I, and MDA5 gene expression, LightCycler Fast Start DNA SYBR Green I Master Mix on a LightCycler 2.0 Instrument (Roche Diagnostics, Basel, Switzerland) was employed with the following primer pair sequences:

The housekeeping gene TATA-box-binding protein (TBP) was used as normalizer. Real-time reactions were run at least in duplicates. Sample values for each mRNA were normalized to the selected housekeeping gene using the formula 2^−ΔCt.

Migration assay was performed by plating 2 × 10⁵ PBMC on the top chamber of 5-µm-pore polycarbonate filters of a 24-well transwell chamber (Corning Inc.). In the lower chamber, 1:5 and 1:10 v/v dilutions of CM from NS and VAX-exposed fibroblasts or 3D-MT were placed. PBMC were then incubated for 4 h at 37 °C and 5% CO₂ humid atmosphere before the trans-migration assay. After the indicated time, the transwell inserts were removed and the migrated cells counted with a Countess 3 FL Automated Cell Counter (Invitrogen, ThermoFisher Scientific) following standardized protocols. The relative level of migrated cells was calculated on the number of PBMC not exposed to CM and basally migrated during the incubation time. Moreover, by flow cytometric analysis, the relative level of migrated CD14⁺ monocytes, CD11b⁺CD14^- macrophages, and CD11c⁺ myeloid dendritic cells (mDC) was assessed.

PBMC were stained with a well-established antibody cocktail to study by flow cytometry monocyte, macrophage, and mDC subsets as well as their activation status.

IgG1 or IgG2a isotype controls and monoclonal Abs (mAbs) for HLA-DR (APC-H7, #641411), CD14 (APC, #555399), CD16 (PE, #561313), CD11b (BV750, #747210), CD141 (BDCA3, BV510, #563298), CXCR-3 (PE, #557185), CXCR-4 (PE, #555974), and CD62-L (FITC, #555543) were purchased from BD Biosciences. CD11c (PeVio770, #130-113-581), CD1c (BDCA1, FITC, #130-113-301), and CCR7 (PeVio770, #130-117-396) were purchased from MACS from Miltenyi Biotech and CD86 (PE-CF594, #120-869-73) from eBioscience. To exclude dead cells from the flow cytometry analyses, Fixable Viability Dye (FvDye, eFluor450, #65-0863-14, eBioscience) was always included in the Ab cocktails. The gating strategy used is summarized in Supplementary Figure S2 . PBMC (1 × 10⁶ cells) were incubated with mAbs at 4°C for 30 min and then fixed with 2% PFA before analysis on a Cytoflex LX cytometer (Beckman Coulter). The data were analyzed by Cytexpert software v.2.1 (Beckman Coulter). The expression of analyzed cell surface molecules was evaluated using the median fluorescence intensity (MFI). Only viable and single cells were considered for further analysis.

Statistical analysis was performed using one-way repeated-measures ANOVA when three or more stimulation conditions were compared. In case of significant ANOVA, the pairwise comparisons were carried out using post-hoc approaches for multiple comparisons to test the significance of the difference between two stimulation effects. The results were shown as median values ± interquartile range (IQR) or, where indicated, as mean values ± standard error of the mean (SEM). A p-value ≤0.05 was considered statistically significant. In the figures, star scale was assigned as follows: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001. Data and statistical analyses were processed by Prism software version 9.4.1 (Graph Pad).

Assuming that tissue-resident non-immune cells—such as fibroblasts and muscle cells—are among the first cells to encounter in vivo an intramuscularly administered vaccine, like the BNT162b2 vaccine, we sought to in vitro reproduce and characterize the local response at the injection site by using different human-cell-based models. Specifically, we employed human MP, MT, and primary fibroblasts as well as 3D-MT models that more closely mimic the in vivo structure and cell composition present in the muscle tissue ( Figure 1 ). We assessed the expression of the BNT162b2-vaccine-encoded Ag, namely, the SARS-CoV-2 S protein, by immunofluorescence analysis at 24 h post-vaccine treatment. Among the analyzed cell models, human primary fibroblasts only showed detectable S protein expression ( Figure 2A ). Indeed no signal was detected in vaccine-stimulated MP and MT as well as on 3D-MT ( Figures 2B–D ).

To assess whether, irrespective of S expression, the tissue-resident non-immune cells are able to respond to the mRNA-based vaccine, the basal and vaccine-induced expression of both endosomal and cytoplasmic RNA sensors, namely, TLR3 (transmembrane receptor localized on endosomes that senses double-stranded RNA), TLR7 and TLR8 (endosomal receptors, recognizing single-stranded RNA), RIG-I (involved in cytoplasmic recognition of both single and double-stranded RNA), and MDA5 (cytoplasmic sensor detecting double-stranded RNA) as well as the secretion of immunoregulatory mediators was investigated ( Figure 3 ). With the exception of TLR7 poorly detectable under basal and vaccine-stimulated conditions, all the other analyzed molecules resulted to be expressed, although at a different extent, at the steady-state level in fibroblasts, MP, and MT ( Figure 3A ). While no variation in RNA sensors’ expression occurred in fibroblasts, both RIG-I and MDA5 expression increased in MP, MT, and 3D-MT in a vaccine-dependent manner. In the latter cell model, BNT162b2 vaccine injection also induced the endosomal receptor TLR3 ( Figure 3A ). Being the intracellular sensors involved in nucleic acid recognition mainly expressed in immune cells, PBMC were used as control. PBMC showed at least a 10-fold induction of all the analyzed sensors at 24 h post-vaccine treatment ( Supplementary Figure S3A ). However, these inductions occurred in the absence of S Ag expression as analyzed by confocal microscopy ( Supplementary Figure S3B ). We then investigated the responsiveness of fibroblasts and muscle cell models in terms of cytokine and chemokine production. As expected, due to nucleoside modifications in the mRNA molecules, IFN-αs production was faint and observed in fibroblasts only ( Supplementary Figure S4A ). Nevertheless, the IFN-inducible gene MX1 was induced in muscle cells ( Figure 3B ), suggesting the involvement of other IFNs, such as IFN-β and IFN-λ. Indeed the vaccine-dependent expression of IFNB1 accounts for MX1 upregulation at 24 and 48 h in 3D-MT, while in MP and MT, the MX1 levels appeared to be likely regulated by the combined action of IFNB1 and IFNL1 ( Figure 3B , Supplementary Figures S4B, C ).

Then, to study innate-like responses, a panel of secreted pro-inflammatory and chemotactic mediators was analyzed post-vaccine stimulation. In all 2D cultures, most changes occurred at 48 h, whereas in 3D-MT, modulations were detectable as early as 24 h ( Figures 3C–H ). In particular, IL-6, an inflammatory and pyrogenic cytokine involved in the acute-phase protein response, was slightly upregulated in fibroblasts, MP and MT, but significantly induced in 3D-MT at both time points ( Figure 3C ).

PTX3, a soluble receptor participating in peripheral immunity and inflammation, was significantly released by fibroblasts and MP 48 h post-BNT162b2 vaccine treatment, while it resulted poorly or was not stimulated in MT and 3D-MT ( Figure 3D ).

The profiling of IL-8/CXCL8—a neutrophil chemotactic factor and one of the major mediators of inflammatory response, IP-10/CXCL10—an IFN-inducible chemokine involved in the inflammation and recruitment of APC, NK, and T cells, RANTES/CCL5—a potent chemoattractant for leukocytes including monocytes, and MCP-1/CCL2—a monocyte/macrophage chemotactic factor was also performed ( Figures 3E–H ). The data showed a cell-type-peculiar chemokine pattern whose level very often reached significant production following 48 h of vaccine stimulation. Specifically, fibroblasts constitutively secreted MCP-1/CCL2 and released IL-8/CXCL8 in a vaccine-dependent manner ( Figures 3E, H ); MP were a major producer of IP-10/CXCL10 and also contributed to the overall vaccine-induced MCP-1/CCL2 ( Figures 3F, H ); MT secreted high basal levels of MCP-1/CCL2 and released IL-8/CXCL8, IP-10/CXCL10, and RANTES/CCL5 upon vaccine stimulation ( Figures 3E–G ); the 3D-MT model showed a robust induction of all analyzed chemokines at 24 and 48 h post-vaccine injection, with MCP-1/CCL2 reaching significance at 48 h only ( Figures 3E–H ). It is noteworthy that the early upregulation of IFN-inducible chemokines IP-10 and RANTES mirrored the increase in MX1 expression observed in vaccine-injected 3D-MT at 24 h ( Figures 3E, F, H ).

Overall, these data indicate that the immune factors produced by 3D-MT upon vaccine exposure are likely to play a major role in shaping the local immune milieu post-vaccination. Nevertheless, collaborative interactions among the other tissue-resident non-immune cells contribute to sense the mRNA-based vaccine and to initiate a rapid recruitment and activation of immune cells at the injection site.

To assess whether and how local tissue response shapes the composition of the immune infiltrate, CM from fibroblasts and 3D-MT, either NS or stimulated with the BNT162b2 vaccine, were used in the migration assays ( Figure 4 ).

Among the different in vitro muscle models, 3D-MT was chosen for collecting CM since its 3D structure more closely resembles the predominant tissue and physiological environment encountered by an intramuscular administered vaccine. Moreover, with this model, several innate immune mediators were found to be rapidly and significantly regulated in response to BNT162b2 vaccine (see Figure 3 ). Thus, PBMC, containing the main immune cells that a vaccine formulation aims to stimulate, were distributed in the upper side of a 5-µm cell culture insert, while in the lower part of the transwell chamber the 1:5 and 1:10 dilutions of CM from fibroblasts and 3D-MT were placed. After 4 h, the total number as well as the type of migrated cells were assessed by hemocytometer counting and cytofluorimetric analysis, respectively. PBMC migration was significantly enhanced by CM from vaccine-injected 3D-MT only at a 1:10 ratio compared to mock-injected 3D-MT (NS) CM, while no increase in the number of migrated PBMC was found in response to fibroblast CM ( Figure 4A ). The cytofluorimetric analysis of migrated cells revealed that CM from both fibroblasts and 3D-MT promoted the migration of APC involved in initiating an Ag-specific adaptive immune response, although at a different extent ( Figures 4B–D ). Irrespective of vaccine-stimulation, soluble factors released under steady-state conditions by fibroblasts and 3D-MT promoted mDC recruitment ( Figure 4B ). On the contrary, CD14⁺ monocytes ( Figure 4C ) and, particularly, CD11b⁺CD14^- macrophages ( Figure 4D ) were significantly recruited by CM from both vaccine-exposed fibroblasts (at 1:5 ratio only) and 3D-MT (at both ratio) compared to CM from NS cells. These findings suggest that the combined action of soluble factors released by stroma and tissue compartments impacts on the composition of the immune infiltrate at the injection site, which is characterized by constitutive mDC migration and vaccine-driven recruitment of monocytes and macrophages.

Once recruited at the vaccine injection site, APC can get in contact either with the mediators released by vaccine-stimulated muscle cells or fibroblasts as well as with the vaccine itself present in the extracellular environment.

Firstly, we conducted a comparative immunophenotypic analysis of APC populations—namely, mDC ( Figure 5 ), monocytes ( Figure 6 ), and macrophages ( Figure 7 )—in PBMC stimulated for 24 h with BNT162b2 vaccine (1 µg/mL) or with CM from fibroblasts and 3D-MT (1:5 and 1:10 dilutions).

mDC were considered as CD11c⁺ cells in the CD14^- gate and further classified in the two main BDCA1⁺ and BDCA3⁺ subsets, known respectively for initiating T cell responses (both CD4⁺ and CD8⁺) and non-self Ag cross-presentation to CD8⁺ T cells (27) ( Figures 5A, B ). In vaccine-treated PBMC, a modest increase in the percentage of total CD11c⁺ mDC was observed ( Figure 5A ), mainly reflecting the variation in BDCA1⁺ subpopulation, the most representative mDC ( Figure 5B , in light blue). A different picture emerged in PBMC stimulated with CM collected from fibroblast cultures treated for 24 and 72 h with the vaccine. In particular, CM from fibroblasts treated for both 24 and 72 h increased the total CD11c⁺ mDC percentage ( Figures 5A, C ). Although an enhancement in BDCA1⁺ cell frequency appeared upon 24 h of stimulation with CM from 24-h vaccine-treated fibroblasts, the 72-h incubation of PBMC with CM collected from 72 h BNT162b2-vaccine-exposed fibroblasts fostered the enrichment of the BDCA3⁺ mDC subpopulation ( Figure 5B in light green and Figure 5D in green) and reduction of the BDCA1⁺ mDC subset ( Figure 5E , in green). However, CM from 24-h-treated fibroblasts only (1:10 ratio) drive a vaccine-specific enhancement of the co-stimulatory marker CD86 MFI in total CD11c⁺ mDC, indicating APC activation ( Figures 5F, H ). Although CM from BNT162b2-treated 3D-MT did not modify mDC frequency ( Figures 5A, B , in purple), a significant enhancement of the CD86 expression was observed in total CD11c⁺ mDC at both 1:5 and 1:10 dilutions ( Figures 5G, H ). It is worth noting that CM from vaccine-stimulated 2D MT did not affect either the mDC frequency ( Supplementary Figures S5A, B ) or the enhancement in CD86 expression on treated PBMC ( Supplementary Figure S5C ).

Considering that CD86 modulation was already evident in mDC after 24 h of stimulation with CM from vaccine-exposed fibroblasts and 3D-MT, we additionally extended our analysis at this time point to include the chemokine receptors CXCR4, CCR7, and CXCR3, which control APC recruitment into peripheral lymph nodes for the former and into inflamed tissues for the latter ( Figure 5I ). While the expression of these chemokine receptors was not markedly modulated in mDC by either direct vaccine exposure or fibroblast-derived CM treatment ( Figure 5I ), a significant enhancement was instead observed after the treatment with CM from vaccine-injected 3D-MT ( Figure 5I ).

In monocytes, the direct BNT162b2 vaccine stimulation leads to a modest reduction of total CD14⁺ cells and of the inflammatory monocyte CD14⁺CD16⁺ subset ( Figures 6A, B ). A significant decrease in CD14⁺CD16⁺ cells coupled with an increase in classical, not inflammatory, CD14⁺CD16^- monocytes was observed after the exposure of PBMC to CM from BNT162b2-vaccine-treated fibroblasts or 3D-MT as compared to the control CM ( Figures 6A, B ). No significant variations in the frequency of total CD14⁺ monocytes or in CD14⁺ subsets were triggered by CM from 2D MT ( Supplementary Figures S5D, E in pink). By evaluating CD86 expression, no modulation was detected in total CD14⁺ ( Figure 6C ), while a significant enhancement was observed in the inflammatory CD14⁺CD16⁺ subset only in response to the BNT162b2 vaccine ( Figure 6D , Supplementary Figure S5F ). The expression of the chemokine receptors CXCR3, CXCR4, and CCR7 was instead mainly influenced in CD14⁺ monocytes by CM rather than the direct vaccine treatment ( Figure 6E ). CM from vaccine-treated fibroblasts robustly upregulated CXCR3, while CM from vaccine-injected 3D-MT significantly increased CXCR4 but downregulated the CCR7 expression ( Figure 6E ).

To further dissect in vitro the local immune response and assess whether the observed reduction in CD14⁺CD16⁺ monocytes could reflect differentiation into macrophages (28, 29), a cytofluorimetric analysis was conducted and revealed that the percentage of CD14^-CD11b⁺ macrophages doubled in BNT162b2-stimulated PBMC compared to unstimulated cells ( Figures 7A, C ). Specifically, the frequency of activated CD16⁺ macrophages was significantly enhanced, while the immature CD16^- subset percentage was diminished ( Figures 7B, C ). A similar increase in macrophage frequency was observed upon PBMC exposure to CM from BTN162b2-stimulated fibroblasts and 3D-MT at a 1:10 ratio, with the latter inducing a significant reduction of CD16^- immature macrophages and the concomitant increase of CD16⁺ activated macrophages ( Figures 7A–C ). Although not statistically significant, fibroblast CM produced a similar trend ( Figures 7B, C ). Conversely, CM from 2D MT failed to modulate in macrophage frequency or subset balance ( Supplementary Figures S5G–I ).

Altogether these results indicate that following vaccine intramuscular administration, factors secreted at the local level by both stromal and tissue-specific cells mainly shape the APC activation profile, while the direct interaction of immune cells with BNT162b2 vaccine mostly drives the differentiation of monocytes into activated macrophages.

Having found that the BNT162b2 vaccine, either directly or through bystander effects mediated by tissue-resident cells, influences the immune cell phenotype and differentiation status, we extended our analysis to the PBMC secretome that could, in turn, act at a local or systemic level to orchestrate the protective immune response. In particular, we focused on the regulation of an early innate immune module previously shown to predict protective humoral responses to the BNT162b2 vaccine in vivo (25). To this aim, PBMC were in vitro stimulated with the anti-COVID-19 mRNA vaccine or with CM derived from vaccine-treated fibroblasts and 3D-MT ( Figure 8 ). Since thawed CM were used in our experimental setup, to evaluate the contribution of the residual BNT162b2 vaccine likely present in the supernatants from vaccine-treated fibroblasts and 3D-MT, PBMC were also stimulated with reconstituted vaccine subjected to a freeze–thaw cycle (FRZ. VAX) at 1 µg/mL (the dose used in the study) and 0.2 and 0.1 µg/mL (corresponding to the estimated vaccine content in CM at ratios of 1:5 and 1:10, respectively) ( Supplementary Figure S6 ).

The multiplex analysis of soluble factors contained in the aforementioned innate module, namely, IL-6, TNF-α, IL-15, IFN-γ, MIP-1β, MCP-1, and IP-10, indicated that stimulation of PBMC with CM from fibroblasts and 3D-MT resulted in a consistent modulation of PBMC secretome. In contrast, direct stimulation with BNT162b2 mRNA vaccine had no or limited effect except for the monocyte chemoattractant MCP-1 induction ( Figures 8A–G ). Upon treatment with CM from vaccine-injected 3D-MT, PBMC displayed a robust enhancement in the release of the pro-inflammatory cytokines IL-6 and TNF-α ( Figures 8A, B ) as well as the macrophage/monocyte chemoattractant MIP-1β ( Figure 8F ). CM from vaccine-treated fibroblasts and 3D-MT also significantly potentiated the secretion of IL-15, a pleiotropic cytokine involved in inflammation and in priming/maintenance of adaptive immunity ( Figure 8C ) and of MCP-1 ( Figure 8E ). Interestingly, CM from BNT162b2-vaccine-treated fibroblasts significantly and specifically promoted in PBMC the co-secretion of the type II IFN-γ and of the IFN-inducible chemokine IP-10 ( Figures 8D, G ). Consistent with these findings and with the observed expression of type I IFNs (IFN-αs and IFNB1) ( Supplementary Figure S4 ), we also found a significant modulation of IFN-inducible genes MX1 and IRF1. These genes are part of the innate immune panel previously identified as predictive of BNT162b2-induced protection in vivo (25) and were significantly upregulated upon PBMC exposure to CM from vaccine-stimulated fibroblasts (1:10 ratio) and vaccine-injected 3D-MT ( Figure 8H ). In contrast, FRZ. VAX had no significant effect on any of the molecules analyzed in Figure 8 , supporting the central role of local tissue responses in shaping peripheral immune cell secretome ( Supplementary Figure S6 ).

Finally, to explore the inter-relationship between direct and indirect BNT162b2-vaccine-induced effects on the early innate immune module in PBMC, we constructed a heat map using the Log2(fold change) of factors with FDR <0.01 ( Figure 8I ). Hierarchical clustering additionally highlights that (I) BNT162b2 vaccine alone induces MCP-1 chemokine, (II) tissue-resident muscle cells prime PBMC to generate an immunostimulatory soluble milieu including the innate cytokines TNF-α, IL-6, and IL-15 as well as the chemotactic factor MIP-1β, and (III) stromal compartment, here recapitulated by human primary fibroblasts, further enhances PBMC responses by promoting the expression of IFN-γ and other innate immune mediators belonging to the IFN signature, such as IP-10, MX1, and IRF1. Collectively, these findings indicate that interaction among immune, stromal, and tissue compartments can recapitulate in vitro the early serum/blood signature associated with vaccine-induced protection observed in BNT162b2 recipients in vivo.