A peptide from the COVID-19 spike protein boosts immune responses and improves anti-PD-L1 cancer therapy against melanoma

The amino acid sequence and domain information of mouse PD-L1 were obtained from the UniProt database (www.uniprot.org↗). The extracellular domain (a.a. 19–239) was loaded in the IEDB database (www.iedb.org↗) and assessed to determine the most suitable epitopes of C57BL/6J mouse MHC-I (H-2Kb) and -II (I-Ab). A fragment of the SARS spike glycoprotein had been verified to efficiently induce IFN-g secretion in a mouse model [11]. Similarly, the epitope was obtained as described above. Table S1 shows that the PD-L1 sequence was ELIIPELPATHPPQNRTSNFRGRASL, and the PD-L1-SARS sequence was ELIIPELPATHPPQNRTAAPHGVVFLHVTYVSNFRGRASL. The PD-L1 and PD-L1-SARS peptides were each coupled with KLH carrier proteins for mouse immunization.

Female C57BL/6J mice (4–5 weeks old) were divided into four groups, and each group was immunized via subcutaneous injection with 100 µL of PBS, 100 µg of KLH, 100 µg of KLH-PD-L1, or 100 µg of KLH-PD-L1-SARS. Peptide vaccines were formulated in PBS; when DMSO was required for solubilization, the final DMSO concentration was ≤ 2% (v/v), with a 100 µL injection volume. The mice were boosted every two weeks after immunization for a total of 12 weeks. In addition, blood was collected by cheek bleeding, and the titer of PD-L1-specific antibodies in the serum samples was determined. The sera (1:8000 dilution) was evaluated. Similarly, the antibody isotypes in the sera of the mice were measured.

Serum samples were collected in the 12th week of the vaccination process, and ten microliters of serum was incubated with 2 × 10⁵ B16F10 cells at 4 °C for 1 h. The binding activity of the anti-PD-L1 antibody was determined by BD Accuri™ C6 flow cytometry (BD Biosciences, San Jose, CA, USA).

The mice were vaccinated as described above. Serum samples were collected and analyzed at a tenth-week time point during the vaccination process. Besides, these mice splenocytes were isolated and seeded at 1 × 10⁶ cells/well in a 24-well culture plate and then treated with 0.1 µg/mL different peptides for 72 h. The sera or supernatants in each group were collected for analysis of cytokine levels via a cytokine multiplex assay using the Bio-Plex Pro Mouse Cytokine Standard Group I 23-Plex following the manufacturer’s protocol (Bio-Rad, Hercules, CA, USA).

B16F10 cells were seeded in a 96-well plate (1 × 10³ cells/well) and incubated overnight. Mouse splenocytes isolated from different treatment groups were added to the wells as effector cells at E: T ratios of 5:1, 10:1, and 20:1. After incubation for 72 h, the supernatants were collected, and the degree of cell cytotoxicity was assessed by using the CyQUANT™ LDH Cytotoxicity Assay Kit according to the manufacturer’s protocol (Sigma-Aldrich).

Mice were immunized and boosted with KLH-PD-L1-SARS or others as described above. To evaluate the immune response, splenocytes were harvested from treated mice after immunization. Prior to analysis, the isolated splenocytes (1 × 10⁶ cells/per well) were pulsed with the SARS peptide antigen (1 µg/mL) to stimulate antigen-specific responses. Following stimulation, the number of cells expressing IL-4 or IFN-γ was quantified using a Mouse IFN-gamma/IL-4 Dual-Color ELISpot Kit (R&D Systems, MN, USA), according to the manufacturer’s protocol. Ressulting spots were visualized and counted under an inverted microscope (Olympus, Shinjuku, Tokyo, Japan). ELISpot spots were counted blinded, background-subtracted, and expressed as spot-forming cells (SFCs)/10⁶ cells; analyses included the IFN-γ/IL-4 ratio to summarize Th1/Th2 balance.

As described above, when the mice were immunized with KLH-PD-L1-SARS for 12 weeks, female C57BL/6J mice (4–5 weeks old; n = 7–9 per group) were subcutaneously implanted with 1 × 10⁴ B16F10 cells in 100 µL of PBS. To facilitate reproducible caliper measurements, the fur over the right dorsal flank was clipped immediately before tumor inoculation (the sparse hair visible in images reflects this planned clipping, not spontaneous alopecia). Tumor size was measured with a caliper, and the tumor volume was calculated as L × H × W × 0.5236. The mice were sacrificed when the tumor size exceeded 1,000 mm³. Additionally, body weight was recorded at each measurement. Local reactogenicity (erythema/induration) was also monitored.

H&E and immunohistochemistry (IHC) staining analyzed the tumors and organs according to the previous study [28]. The tissue slides were photographed to observe the pathology and metastatic colonies under a microscope (Olympus, PA).

Antigen Retrieval Buffer (Abcam, Cambridge, UK) and a Leica Novolink max polymer detection system (RE7140-K, Wetzlar, Germany) for IHC staining were used following the manufacturer’s protocols. Finally, the sections were probed with specific commercial antibodies (Table S2), and the samples were observed under a light microscope and photographed.

Female C57BL/6J mice (4–5 weeks old; n = 8–10 per group) were subcutaneously inoculated with 1 × 10⁴ B16F10 cells in the right dorsal flank. To facilitate reproducible caliper measurements, the fur over the right flank was clipped immediately before tumor inoculation. Tumor growth was monitored, and when an individual tumor reached ~ 30 mm³ (pre-specified 20–40 mm³ window), animals were block-randomized stratified by baseline tumor volume; the day of randomization was defined as Day 0. Beginning on Day 0, mice received weekly intratumoral (in situ) injections of the peptide vaccine (100 µg per dose) according to group assignment. Tumor volume was then measured every two days. Where applicable, tumor measurements were performed by an investigator blinded to allocation, and mice were euthanized when tumor volume exceeded 1,000 mm³ or if other humane endpoints were met.

All procedures for tumor inoculation, fur clipping, randomization, Day-0 definition, tumor measurement schedule, volume calculation, blinding, and humane endpoints were as described above. Briefly, when individual tumors reached ~ 30 mm³ (20–40 mm³ window), mice (n = 10 per group) were block-randomized and treatment was initiated (Day 0). Beginning on Day 0, animals received weekly intratumoral peptide vaccine (100 µg) and anti-mouse PD-L1 antibody (clone 10 F.9G2, Bio X Cell) intraperitoneally at 10 mg/kg every two days according to group assignment. Tumor volumes were measured every two days.

We implemented a peptidomimetic-guided, reverse-screening workflow to contextualize potential receptor engagement for the SARS-derived peptide. The 14-aa sequence was partitioned into overlapping 6-aa subfragments (sliding window). For each subfragment, side-chain topology/pharmacophore features were mapped to small-molecule–like surrogates (SMILES) and queried against BindingDB (https://www.bindingdb.org↗) via structure/similarity search to retrieve published experimental binding affinities of literature small-molecule ligands for the corresponding protein targets.

Importantly, these BindingDB affinities pertain to small molecules, not our peptide. In our dataset, the provider reported affinities as inverse values (e.g., 1/Kd in nM⁻¹); when values were available as Kd/Ki/IC₅₀ in nM, we inverted them so that larger numbers correspond to stronger affinity. To enable unitless visualization for each fragment–target pair, these affinity-derived numbers were converted to a Predicted binding index (unitless). This index is used descriptively to indicate predicted interaction propensity for each subfragment with each candidate protein; no statistical ranking/thresholding is applied, and the index does not represent experimental affinity of the peptide nor imply causality.

Independent protein–peptide docking was performed with HDOCK (http://hdock.phys.hust.edu.cn↗) [29] using default parameters to visualize poses and interface residues (peptide–protein residue pairs within 3.5 Å). Docking outputs were used qualitatively to support candidate contextualization and were not converted to experimental affinities. Binding confidence was categorized as follows: confidence scores above 0.7 indicated a high likelihood of binding; scores between 0.5 and 0.7 suggested possible binding; and scores below 0.5 indicated that binding was unlikely.

Naïve mouse splenocytes were treated with PD-L1 or PD-L1-SARS peptides for five days, which was according to the RT-PCR results (Fig. S1A), and splenocyte RNA was extracted using a PureLinkTM RNA Mini kit (Thermo Fisher Scientific Inc., Waltham, MA). NGS was subsequently performed with a NovaSeq 6000 system (Illumina, CA). The raw counts were estimated and normalized as transcripts per million (TPM) values. Gene expression was further analyzed via t-test and ANOVA. The genes with significant differences (P < 0.01) through t-test, and then the 851 genes occupying the top 70% of the expression amount were selected for subsequent CIBERSORTx, CD marker and immune cytokine expression analyses (Fig. S1B). All supporting NGS data were available to researchers in the Gene Expression Omnibus (GEO) database, and the accession number was PRJNA1007364.

Female C57BL/6J mice (n = 4) were inoculated subcutaneously with 1 × 10⁴ B16F10 cells. When the average tumor volume in each group reached 50 mm³, the mice were injected with 100 µg of peptide vaccine in situ. After seven days, the tumor tissues were collected for LC‒MS/MS analysis. The raw files were searched against a UniProt mouse database via the Sequest and Mascot search engines in Proteome Discoverer (version 2.5).

Numeric data were expressed as the mean ± SD. Data were analyzed using the SAS statistical package (SAS Institute, Inc., Cary, NC, USA). A Student’s t-test was used when comparing two independent samples, and ANOVA was used for comparing multiple samples. P < 0.05 was considered to indicate statistical significance.

Melanoma cells overexpress PD-L1 on their surface; therefore, PD-L1 is considered a tumor-associated antigen. Furthermore, T-cell epitopes in PD-L1 were determined through computational analysis using the IEDB database, as antigens. Similarly, the peptide sequences were derived from the SARS-CoV-2 spike protein as an adjuvant, and these peptide sequences were shown in Table S1.

After immunization with the different peptide vaccines, antibody titers and recognized activity were determined against peptide antigens or the cell surface antigen. The data indicated that KLH-PD-L1-SARS could dramatically induce the production of anti-PD-L1 antibodies against peptides, while KLH-PD-L1 had a minor effect (Fig. 1A). However, the result showed that the sera antibodies of all vaccinated mice did not recognize the PD-L1 antigen on the cell surface (Fig. 1B).

The cytotoxicity assay results indicated that splenocytes from mice immunized with KLH-PD-L1-SARS exhibited greater dose-dependent cytotoxicity than did those from the control groups (Fig. 1C). Although immunization with KLH-PD-L1 peptides resulted in a slight increase in cytotoxicity, this increase was not statistically significant. ELISpot showed that KLH-PD-L1-SARS increased PD-L1–specific IFN-γ and IL-4 SFCs (per 10⁶ cells, background-subtracted) versus controls (Fig. 1D). To summarize polarization, we computed a Th1/Th2 balance index (IFN-γ/IL-4; mean of individual ratios), which was 1.39 ± 0.08 (n = 6), indicating a Th1-leaning cellular response.

Changes in cytokines and chemokines in the blood and lymphoid organs (spleen) were monitored. Compared with other treatments, immunization with KLH-PD-L1-SARS efficiently led to higher expressions of proinflammatory cytokines in the blood (Fig. 1E). Unlike the systemic effect, immunization with KLH-PD-L1-SARS increased the IL-17 and IFN-γ levels in cultured immunized splenocytes. Additionally, it promoted the production of G-CSF. Furthermore, these immune cells secreted inflammatory cytokines (Fig. 1F). These results indicated that this SARS-CoV-2 peptide exhibited a potent adjuvant effect, which could promote immune cells to release proinflammatory cytokines, IL-17, and IFN-γ.

To validate the efficacy of the peptide vaccine for tumor prevention, Fig. 2A showed that the KLH-PD-L1-SARS vaccination significantly inhibited tumor growth compared with the other groups. Interestingly, two mice in the KLH-PD-L1-SARS vaccination group were tumor-free on the 30th day after tumor implantation (Fig. 2B). On the 27th day, the mice in the PBS group were all humanely sacrificed. The survival rate was 60% for the KLH-PD-L1 group and only 30% for the KLH group; however, all the KLH-PD-L1-SARS-vaccinated mice survived (Fig. 2C). For transparency, a KLH-SARS-only arm was not carried forward into this confirmatory experiment because pilot studies showed no measurable benefit under identical conditions (Fig. S5); in line with 3R (Reduction/Refinement), animals were concentrated on the antigen-matched comparison.

In the KLH-PD-L1-SARS group, the H&E stain images revealed that the tumor tissues were significantly more damaged than those in the other treatments were (Fig. 2D). The IHC results showed that the expression of the PD-L1 antigen was lower in the KLH-PD-L1-SARS group than in the other groups. Additionally, the IHC results indicated an increase in CD4, CD8, and IL-17 expression in the tumor area of the mice that received the KLH-PD-L1-SARS vaccine (Fig. 2E). In contrast, the presence of FOXP3 molecules was lower in the KLH-PD-L1-SARS group than in the PBS, KLH, and KLH-PD-L1 groups (Fig. 2E). These findings suggested that KLH-PD-L1-SARS treatment promoted the infiltration of CD4⁺ T, CD8⁺ T, and Th17 cells while suppressing the infiltration of Treg cells into the tumor area.

In addition, the impact of KLH-PD-L1-SARS vaccination on distant metastasis was assessed, and the result indicated a significant reduction in the sizes and numbers of metastatic colonies in the liver (Fig. 2F and G). Figure 2H demonstrated that there were no visible pathological differences in the vital organs among all the treatment groups, and the body weights of the mice receiving the KLH-PD-L1-SARS vaccine were not significantly different from those of mice in the other groups (Fig. 2I). These results suggested that KLH-PD-L1-SARS vaccination did not induce severe or significant side effects.

Furthermore, healthy and non-tumor-implanted C57BL/6J mice were vaccinated once every two weeks until the 30th week. Figure S2A showed that the body weights of the mice still steadily increased under high anti-PD-L1 antibody titers during a long period of vaccination. Finally, the pathological images of the organs were examined, and they were not visibly damaged in any of the groups (Fig. S2B). Therefore, the abovementioned results indicated that the KLH-PD-L1-SARS vaccine exhibited effective prevention activity and safety in mice.

The therapeutic effects of KLH-PD-L1-SARS peptide vaccination on tumor growth and metastasis were subsequently evaluated via in situ injection. When the average volume of B16F10 melanoma in mice reached 30 mm³, different treatments were injected into the tumors to validate their activities in suppressing tumor growth. Figure 3A showed that B16F10 tumor growth did not significantly differ among the groups treated with PBS, KLH, and KLH-PD-L1 (Fig. 3A and B). However, compared with other treatments, KLH-PD-L1-SARS peptide treatment significantly suppressed tumor growth. Interestingly, it also efficiently prolonged survival (Fig. 3C).

H&E stain images showed that the in situ KLH-PD-L1-SARS caused more tumor cell death and bleeding in the tumor tissue compared with the KLH and KLH-PD-L1 groups (Fig. 3D). Additionally, KLH-PD-L1-SARS reduced the expression of PCNA in tumor cells, indicating a decrease in tumor cell proliferation. Compared with KLH and KLH-PD-L1, KLH-PD-L1 also had greater ability to activate caspase 3, leading to cell apoptosis (Fig. 3D).

Compared with other treatments, KLH-PD-L1-SARS treatment increased the levels of CD68 and CD86. However, it reduced the expression of CD163 antigens in tumor areas, indicating increased macrophage infiltration into the tumor area and differentiation into the M1 subtype of macrophages (Fig. 3D). Interestingly, the in situ therapeutic effect of KLH-PD-L1-SARS on distant metastasis to the liver revealed that the sizes and numbers of metastatic colonies were markedly lower than those in the other groups (Fig. 3E and F, and S4).

The anti-PD-L1 therapeutic antibodies as immune checkpoint inhibitors (ICIs) have been applied in human melanoma clinical therapy. Notably, the PD-L1-SARS peptide vaccine couldn’t produce antibodies recognized against PD-L1 antigens of the tumor surface, determined its adjuvant effect when combined with such ICIs therapy. The result indicated that treatment with the anti-PD-L1 antibody alone indeed suppressed tumor growth; however, tumor growth inhibition was further enhanced by the combination of this antibody with in situ injection of the PD-L1-SARS peptide vaccine (Fig. 4A and B). The survival rate of the combination therapy group was greater than that of the single therapy group (Fig. 4C).

Moreover, H&E stain images showed that the combination therapy caused more severe damage to the tumor tissue compared with the anti-PD-L1 antibody therapy (Fig. 4D). Compared with anti-PD-L1 antibody treatment, the combination therapy reduced the number of PD-L1-expressing melanoma cells more effectively. This could be attributed to the combination therapy’s more efficient reduction of cell proliferation (as indicated by PCNA expression) and promotion of cell apoptosis (as indicated by c-Casp3 expression) (Fig. 4D). Figure 4E indicated that no significant damage was observed in any of the normal organ tissues, including the heart, liver, spleen, lung, and kidney, for either of the treatments.

CIBERSORTx analysis of the NGS data revealed changes in immune cell populations, as shown in Fig. 5A. Compared with the PD-L1 peptide, the PD-L1-SARS peptide significantly increased the proportions of naïve B-cells, plasma cells, activated mast cells, follicular Th cells, activated memory CD4⁺ T cells, and memory B-cells in the assay. Conversely, compared with the PD-L1 peptide, the PD-L1-SARS peptide reduced the proportions of CD8⁺ T cells, monocytes, resting NK cells, and resting dendritic cells. These results indicated that administering PD-L1-SARS peptides stimulated B- and T-cell activity and differentiation.

As expected, the immune markers of active T cells, B cells, DCs, and macrophages, such as CD69, CD80, and CD86, were upregulated in the PD-L1-SARS group compared with the PD-L1 group. Conversely, the expression of the immunosuppressive marker FOXP3 was decreased in the PD-L1-SARS group, whereas that of CD276 was not significantly different, and that of CD274 was increased (Fig. 5B). These results suggested that the PD-L1-SARS peptide could directly activate immune cells and attenuate immunosuppression.

Furthermore, the PD-L1-SARS peptide effectively increased the levels of proinflammatory cytokines such as TNF-α, IL-1α, IL-1β, and IL-6 but not IL-13 (Fig. 5C and S3). Additionally, the expression of the Th1 activating-cytokine IL-12, namely IL-12α, was elevated in the PD-L1-SARS group, whereas the expression of the IL-12β gene was decreased (Fig. 5C). Subsequent monitoring of the IL-17 family revealed increased expression of IL-17a and IL-17f, whereas the expression of IL-17b, IL-17c, and IL-17d was not significantly different (Fig. 5C and S3). Moreover, the levels of type II IFN, IFN-γ, were also increased, while the levels of type I IFNs (IFN-α and IFN-β) and other IFN family members were not significantly different (Fig. 5C and S3). In contrast, the PD-L1-SARS peptide significantly decreased the levels of immunosuppressive cytokines, including TGF-β1 and TGF-β3, but increased the level of IL-10 (Fig. 5C). Therefore, the results indicated that the PD-L1-SARS peptide efficiently activated the antitumor immune response by increasing the levels of proinflammatory cytokines, IL-17, and IFN-γ and decreasing the levels of immunosuppressive cytokines, such as TGF-β.

LC‒MS/MS identified significant protein differences between PD-L1-SARS and PD-L1 peptide treatments in the tumor microenvironment (TME). The peptide spectrum matches (PSMs) and fold changes (FCs) of 102 significant proteins and 1254 insignificant proteins were displayed in Fig. 5D. We identified nine proteins with high FC values and significant expressions. These proteins followed the formula y = x − 1. On the basis of our calculation via the formula FC = (PSMs of the specific protein in the PD-L1-SARS group + 1)/(PSMs of the specific protein in the PD-L group + 1), the nine proteins were not expressed in the PD-L1 peptide group but were highly expressed in the PD-L1-SARS peptide group.

The names and accession numbers of the nine proteins were as follows (also listed in Table S3): haptoglobin (Q61646↗), IFN-induced guanylate-binding protein 2 (Q9Z0E6↗), IFN-γ-inducible GTPase ifggb6 protein (Q3T9E4↗), IFN-inducible GTPase 3 (Q60766↗), IFN-inducible GTPase 1 (Q9QZ85↗), E3 ubiquitin-protein ligase (E9Q555↗), H-2 class I MHC (IFN-induced protein, P01901↗), lactoylglutathione lyase (Q9CPU0↗), and IFN-induced 15 kDa protein (Q64339↗). According to the published literature, eight out of the nine proteins were involved in the IFN-γ-related pathway (Table S3). In addition, lactoylglutathione lyase catalyzed the formation of S-lactoylglutathione, which was involved in the transcription of the proinflammatory cytokine TNF-α. These results indicated that the PD-L1-SARS peptide could induce the expression of IFN-γ and TNF-α and to promote their activities in immune activation and inflammation in the TME.

We computationally profiled the 14-amino-acid SARS-derived peptide by differentiating its sequence, predicting structure, and performing reverse docking to prioritize candidate host receptors. Figure 6A prioritizes eight candidate proteins predicted to interact with the overlapping 6-aa subfragments; the corresponding predicted binding index (unitless) are summarized in Table S4. These in silico scores are hypothesis-generating and do not represent experimental affinities. Among these, ACE2 (previously reported to bind SARS-CoV-2 spike) and MHC class I (predicted by IEDB to present the peptide) were recovered as internal/face-validity controls, supporting the plausibility—but not validating—the computational screen.

Figure 6B shows representative docking poses between the full peptide and the shortlisted proteins. Based on confidence scores, cathepsin D and interleukin-1 receptor antagonist were deprioritized from the initial set. Literature mapping (see Table S5) indicates that several candidate receptors are mechanistically connected to dendritic cell, macrophage, B-cell, or T-cell activation and have been associated with IL-17 and/or IFN-γ biology, providing contextual coherence with our in vitro/in vivo cytokine, IHC, and RNA-seq findings. As an initial, exploratory cross-check consistent with this framing, a plate-based binding ELISA using a biotinylated SARS-derived peptide and immobilized receptors (APJ, APN/CD13, AGTR1, NMBR) showed dose-dependent OD₄₅₀ signals, with APJ and APN generally higher than AGTR1/NMBR (Fig. S6). Because plate-coating assays are subject to orientation/avidity artifacts, these data are qualitative and do not establish affinity or physical binding; accordingly, Fig. 6 remains hypothesis-generating.

Figure 6C shows hypothesis-generating working model for how a PD-L1–SARS peptide may re-condition the TME and why it synergizes with anti-PD-L1. The PD-L1 peptide supplies antigenic epitopes for MHC-I/II presentation, whereas the SARS-derived motif may provide adjuvant-like priming. In silico docking nominates candidate receptor/pathway interfaces (e.g., APN/CD13, AGTR1, APJ, NMBR) that could link the motif to innate sensing. Together, the vaccine program may enhance antigen presentation and myeloid polarization (↑CD86/CD163), followed by Th1/Th17-skewed adaptive responses (↑IFN-γ, ↑IL-17), B-cell activation/anti–PD-L1 antibodies, and CD8⁺ CTL cytotoxicity, consistent with tumor control. In combination therapy, the therapeutic anti-PD-L1 mAb (10 F.9G2) releases PD-1/PD-L1 inhibition on T cells; CTL killing and antibody-mediated opsonization promote tumor-cell apoptosis and release of PD-L1 antigen, which is captured by APCs, reinforcing a positive feedback loop of antigen uptake, presentation, and effector activation.

All animal procedures were approved by the Institutional Animal Care and Use Committee of National Yang Ming Chiao Tung University (NYCU-IACUC-111023) and conducted in accordance with the ARRIVE guidelines.

Not applicable.

The authors declare no competing interests.

All data generated or analyzed during this study are available from the corresponding author upon reasonable request. In addition, all supporting next-generation sequencing (NGS) data have been deposited in the Gene Expression Omnibus (GEO) database under accession number PRJNA1007364.