三模块猫传染性腹膜炎病毒纳米颗粒疫苗在雌猫中提供完全保护且无可检测的抗体依赖性增强 机翻

Article | Article in Presshttps://doi.org/10.1038/s41467-026-78999-4 A tri-modular feline infectious peritonitis virus nanoparticle vaccine confers complete protection in female cats without detectable antibody-dependent enhancement A list of authors and their affiliations appears at the end of the paper Antibody-dependent enhancement (ADE) complicates coronavirus vaccine development. Feline infectious peritonitis virus (FIPV) is a classic model of coronavirus vaccine-mediated ADE. Here, we develop ferritin-based FIPV nanoparticle vaccines displaying nucleocapsid (N) antigen, antigen-presenting cell (APC)-targeting ligands, and immunostimulatory factors, and evaluate their protective efficacy, serum-mediated ADE risk, and associated immune features. Compared with soluble N protein, the single-module FIPV N nanoparticle vaccine (sfNnps) enhances FIPV N-specific T-cell responses yet confers only 25.0% (2/8) protection.Dual-module vaccines incorporating either the APC-targeting ligand HCV E2 (dfN/Hnps) or the immunostimulatory factor IFNγ(dfN/Inps) further boost N-specific T-cell responses and confer 50.0% (4/8) and 62.5% (5/8) protection,respectively. The tri-modular vaccine co-displaying N, HCV E2, and IFNγ(tfN/HInps) elicits the most robust N-specific T-cell responses and affords 100% (8/8) protection in female cats, while showing no detectable serum-mediated ADE under the tested conditions and reducing peripheral-blood viral RNA levels by 73.0%. Immune profiling associates tfN/HInps with increasedcDC1 representation and antigenpresentation gene expression,aswellasashifttowardCD8⁺effector-memory T cells and a FOXO1-centered program linked to transcriptional features of increased proliferation, memory formation, and cytotoxicity. Overall, this engineered FIPV nano‑display vaccine could provide a framework for cellular‑immunity‑oriented vaccine development against ADE‑prone coronaviruses. Coronaviruses continue to threaten human and animal health due to their high mutation rates, broad host range, and capacity for crossspecies transmission 1–6 . Although vaccination remains the most effective approach for long-term control, antibody-dependent enhancement (ADE) has complicated the development of safe and broadly protective coronavirus vaccines 7–12 . ADE, triggered by non-neutralizing or sub-neutralizing antibodies, facilitates viral entry into Fc receptor–bearing cells and drives aberrant immune activation, thereby worsening disease outcomes and compromising vaccine efficacy 13,14 . Feline infectious peritonitis virus (FIPV) provides one of the most wellestablished models of coronavirus vaccine–induced ADE. ADE in FIPV infection can provoke severe inflammatory pathology and high mortality, yet despite decades of research, no vaccine has achieved reliable protection without ADE risk 12,15–19 . This challenge underscores a broader need for vaccine strategies capable of inducing protective immunity while avoiding antibody-mediated disease enhancement. Received: 9 January 2026 Accepted: 7 October 2026 Check for updates e-mail:mingjunsu@zafu.edu.cn;dongbosun@126.com Nature CommunicationsArticle in Press 1234567890():,;1234567890():,;

Although the coronavirus spike (S) protein is the predominant target of neutralizing antibody–based vaccines, its association with ADE renders it suboptimal for FIPV and other ADE-prone coronaviruses 9,11,20–22 . To circumvent ADE risk, researchers have begun exploring other viral proteins as vaccine antigens. The nucleocapsid (N) protein, a major structural component of coronaviruses, plays key roles in the viral life cycle, including genomic RNA packaging, stabilization of the replication complex, and regulation of virion assembly and budding 23–26 . Due to its high-level expression early in infection and high sequence conservation among different strains, it has been primarily used in pathogen detection and rapid diagnostic platforms 27–30 . As understanding of coronavirus immunity deepens, the N protein is being investigated as a complementary antigen alongside the S protein in vaccine design. In various coronavirus models, including SARS-CoV-2, this strategy enhances CD8 + T-cell activity and cytokine secretion, consistent with synergistic induction of cellular immunity 31,32 . Recent SARS-CoV-2 vaccine studies further show that combined S and N protein immunization can reduce viral loads in the lung and brain and alleviate associated tissue damage 32–35 . Notably, although the N protein itself does not typically induce neutralizing antibodies, it is enriched for MHC class I–restricted T-cell epitopes, enabling Th1-associated immune programming and promoting clearance of virus-infected cells 36–38 . Leveraging this property, N protein has also been evaluated in the high-ADE-risk FIPV setting. Across multiple platforms, including recombinant raccoon poxvirus vectors, baculovirus, and adenovirus vectors, N-based immunization delayed disease progression and reduced viral load 39,40 . Although the protection remained partial, ADE was not observed. Collectively, despite limitations in standalone protective efficacy, the N protein’s capacity to elicit cellular immunity and its favorable safety profile in ADE-prone coronavirus contexts support its continued development as a vaccine antigen. When used alone, the N protein often confers suboptimal protection because antigen presentation is inefficient and overall immunogenicity is limited. These constraints motivate the incorporation of molecular adjuvants to elicit robust cellular immunity. Among immune-enhancement strategies, antigen-presenting cell (APC)-targeting ligands can deliver vaccine antigens directly to APCs, improving uptake and presentation. For instance, CD40L, HCV E2, and HIV Gp120 bind receptors on APC surfaces and can enhance antigen uptake, processing, and presentation 41–44 . Immunostimulatory factors provide an additional route to strengthen cellular responses. IFNγcan augment APC activity,flagellin activates innate immunity via TLR5 signaling, and IL-2 supports T cell expansion and memory formation. In animal models, these factors enhance the magnitude, functional breadth, and durability of CD4 + and CD8 + T-cell responses 45–49 . Ferritin nanoparticles, with their defined geometry and high-density antigen display, provide a suitable scaffold for modular co-display of antigens and immune potentiators, enabling rational design of multi-functional vaccines 50,51 . In our previous work, we successfully used this platform to construct nanoparticle vaccines displaying multiple antigens, achieving significantly enhanced immunoprotection 52 . In this work, we report a rationally engineered ferritin nanoparticle vaccine integrating three functional modules: the FIPV N antigen, the APC-targeting ligand HCV E2, and the immune-stimulatory cytokine IFNγ(Fig.1). Among the tested nanoparticle formulations, the tri-modular tfN/HInps formulation elicits the strongest FIPV N-specific functional T-cell responses, coinciding with 100.0% (8/8) survival following the tested FIPV challenge model and no detectable serummediated ADE under the tested conditions. Immune profiling further identifies two coordinated features associated with the observed protection: cDC1 enrichment and activation-related changes, and a FOXO1-associated transcriptional program in CD8 + T cells consistent with effector-memory differentiation. Together, thesefindings support an immunologically informed strategy for achieving protection while minimizing serum-mediated ADE risk under the tested conditions and offer a potentially useful framework for designing T-celloriented vaccines against ADE-prone coronaviruses and other pathogens. Results Construction and evaluation of a single-module FIPV N-ferritin protein nanoparticle vaccine Given the potential of the coronavirus N protein to activate host cellular immune responses while minimizing the risk of antibodydependent enhancement (ADE), we constructed a single-module FIPV N-ferritin protein nanoparticle vaccine (sfNnps, displaying the N antigen alone) using our established ferritin nanoparticle platform 52 . We then systematically assessed its antigen uptake, immunogenicity, serum-mediated ADE risk, and protective efficacy. To display the FIPV N protein (Npro) on ferritin nanoparticles, we generated a fusion construct encoding Npro and ferritin monomer (N- Fer, Supplementary Fig. 1a, b). Owing to the self-assembly properties of ferritin, N-Fer assembled into uniform 24-mer nanoparticles (Fig.2a). SDS–PAGE revealed a single band at approximately 78 kDa, consistent with the expected molecular weight of N-Fer, and the purified protein was specifically recognized by anti-His and anti-FIPV antibodies (Supplementary Fig. 1c, d). TEM and DLS analyses confirmed the formation of spherical nanoparticles with a hydrodynamic diameter of 21 nm, which was larger than that of the ferritin-only control (13.5 nm), indicating display of Npro on the ferritin nanoparticle surface (Fig.2b). To assess antigen uptake, DC2.4 dendritic cells were incubated with FITC- labeled sfNnps or Npro. Compared with Npro, sfNnps increased the percentage of FITC⁺DC2.4 cells from 39.6 to 69.4% (Fig.2c, d). sfNnps also significantly increased the surface expression of MHC I, MHC II, and CD86, whereas CD80 expression remained unchanged (Fig.2e). Thesefindings indicate that sfNnps were efficiently internalized and promoted dendritic cell maturation and antigen-presenting activity. We next investigated whether nanoparticle presentation altered antigen persistence and tissue distribution in vivo using FITC-labeled sfNnps or soluble Npro.The results showed that sfNnps exhibited strongerfluorescence signals than Npro for up to 48 h after injection (Fig.2f). At 48 h,fluorescence intensities in both the draining lymph nodes and spleen were significantly higher in the sfNnps group than in the Npro group (Fig.2g), indicating that ferritin nanoparticle presentation prolonged antigen persistence in immune-relevant tissues, supporting sustained immune activation. We then evaluated the immunogenicity of sfNnps in cats following the immunization (Fig.3a). Following ex vivo restimulation with an overlapping peptide pool spanning the FIPV N protein, sfNnps significantly increased the number of IFN-γspot-forming units relative to Npro (SFU/10 6 cells; Fig.3b). Consistently, the frequencies of TNF-α⁺ N-specificCD4⁺and CD8⁺T cells were 1.83- and 1.79-fold higher, respectively, in the sfNnps group than in the Npro group (Fig.3c). These results indicate that sfNnps enhanced FIPV N-specific cellular immune responses in cats. To assess whether sera elicited by sfNnps immunization mediate ADE, we evaluated their effects on FIPV attachment, viral replication, and FIPV infection-associated inflammatory responses. In virus attachment assays using feline PBMC-derived adherent monocyte/macrophage-enriched cells, serum from both sfNnps- and Npro-immunized cats did not increase FIPV attachment, and the cell-associated viralfluorescence intensities were significantly lower than those observed with the positive sera control group (Fig.3d). Consistently, neither sfNnps- nor Npro-immune sera increased intracellular FIPV RNA levels across the tested serum dilutions (Fig.3e). Cytokine analysis further revealed that TNF-α,IL-1β,IL-6, and IL-10 levels in cultures treated with sfNnps- or Npro-immune sera were comparable to those in the FIPV-only group and were significantly lower than those in the positive sera control group (Fig.3f). These results indicate that immunization with sfNnps or Npro did not Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

induce detectable serum-mediated ADE under the tested conditions. We next determined the protective efficacy of sfNnps in a feline FIPV challenge model. Cats in the unimmunized control group developed severe clinical signs, and the three cats succumbed at 7, 12, and 15 days post-infection (dpi), respectively. Cats immunized with Npro exhibited partially alleviated clinical symptoms but ultimately all died at 13, 17, and 22 dpi, respectively. In contrast, sfNnps-immunized cats showed further attenuation of disease, with death occurring at 24 and 31 dpi and a survival rate of 33.33% (1/3; Fig.3g, h). Consistent with the clinical outcomes, viral nucleic acid levels were significantly reduced in the spleen and kidney, but not in the liver, in the sfNnps group relative to the Npro group (Fig.3i), in agreement with histopathologicalfindings (Supplementary Fig. 3). Collectively, these results indicate that sfNnps enhanced FIPV N-specific cellular immune responses and provided partial protection against FIPV challenge, while showing no detectable serum-mediated ADE under the tested conditions. Screening of APC-targeting modules to enhance antigen presentation efficiency by sfNnps Although sfNnps delayed disease progression in FIPV-challenged cats andshowednodetectableserum-mediatedADEunderthetested conditions, its protective efficacy remained limited. To further improve vaccine efficacy, we sought to identify an APC-targeting module that could enhance the antigen-presenting capacity of sfNnps. Three candidate modules—HIV Gp120, HCV E2, and CD40L—were therefore individually displayed on sfNnps and comparatively evaluated in murine in vitro and in vivo models, with the selected formulation further validated in feline monocyte-derived dendritic cells. Using the SpyCatcher/SpyTag system, we individually displayed each factor on the surface of sfNnps to generate three dual-module APC-targeting screening formulations (dfN/Vnps, dfN/Hnps, and dfN/ Cnps were engineered to co-display the N antigen with HIV Gp120, HCV E2, and CD40L, respectively; Fig.4a). Protein expression and Western blot analysis confirmedthatHIV-Gp120-SC,HCV-E2-SC,and CD40L-SC migrated at the expected molecular weights of approximately 42.8 kDa, 71.3 kDa, and 59.6 kDa, respectively, and were specifically recognized by an anti-His tag antibody. The SpyTag-fused N-Fer scaffold (ST-N-Fer) was also prepared (Supplementary Fig. 5a–d). Following conjugation, dfN/Vnps, dfN/Hnps, and dfN/Cnps exhibited the expected mobility-shifted bands at approximately 126 kDa, 154 kDa, and 143 kDa, respectively, whereas additional exploratory formulations carrying multiple helper factors exhibited the corresponding multiband patterns (Supplementary Fig. 5e). TEM and DLS analyses further confirmed that all formulations assembled into spherical Construction of multi-modular FIPV N-ferritin nanoparticles ×24 N-Fer monomer FIPV N(G 2 S) 3 Ferritin monomer sfNnps tfN/HInps enhance the long-term cytotoxic efficacy of effector T cells CD8 TnCD8 TemtfN/HInps DC cellscDC1 cellstfN/HInps Accelerate antigen processing Improve antigen presentation antigen Tcell tfN/HInps enhance cDC1-associated antigen presentation TcellCD8TemcDC 1 Proliferation Memory tfN/HInps FOXO1 Cytotoxicity Transcriptional control of MHC I Transcriptional control of MHC II Co-stimulatory molecules & cytokine axis JAK–STAT–SOCS axis Endolysosomal antigen processing Vesicle-mediated antigen crosspresentation Proteasome-dependent crosspresentation Peptide transport & editing for MHC I tfN/HInps confers complete protection against FIPV challenge without detectable serum-mediated ADE under the tested conditions tfN/HInps No detectable ADE Survival (100%; 8/8) FIPV No detectable ADE Survival (25.0%; 2/8) No detectable ADE Survival (50.0%; 4/8) No detectable ADE Survival (62.5%; 5/8) ×2×6×4×4×5×3 sfNnpsdfN/HnpsdfN/Inps Promote antigen uptake Pattern-recognition receptors Complement receptors Scavenger receptors IgG Fc receptors C-type lectin receptors Single-module nanoparticle tfN/HInps + HCV E2-SC IFNγ-SC sfNnps Trir a l u d o m nanoparticle ++ Dual-module nanoparticle APC-targeting module T-cell-activation module sfNnps T-cell activation dfN/Inps dfN/Fnps dfN/Lnps Antigen uptake dfN/Vnps dfN/Cnps dfN/Hnps IFNγ-SC IL-2-SC Flagellin-SC HCV E2-SC CD40L-SC HIV Gp120-SC Fig. 1 | Schematic illustration of an engineered FIPV nano-display vaccine platform and immune features associated with complete protection in cats. Single-module (sfNnps), dual-module (dfN/Hnps or dfN/Inps), and tri-modular (tfN/ HInps) ferritin nanoparticles were constructed. Following lethal FIPV challenge, survival rates were 25.0% (2/8), 50.0% (4/8), 62.5% (5/8), and 100.0% (8/8), respectively. No detectable serum-mediated ADE was observed in the N protein nanoparticle vaccine groups under the tested conditions. Mechanistically, tfN/ HInps immunization was associated with cDC1 enrichment, enhanced antigen cross-presentation-related features, and a shift of CD8 + T cells toward an effectormemory phenotype accompanied by a FOXO1-centered transcriptional program. Figure 1 was partially created in BioRender. Feiyu Zhao. (2026)https://BioRender. com/2xev4lk. The nanoparticle illustrations are original and were created by the authors using Autodesk 3ds Max 2020. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

ba 0 2 040 5 10 15 20 Size (d.nm) Size distribution by intensity (%) 200 nm sfNnps 02040 5 10 15 20 Size (d.nm) Size distribution by i n t e nsity (%) 02040 5 10 15 20 Size (d.nm) Size distribution by i n te n s i t y(%) cd e 0 12243648 0 1 2 3 Fluorescence efficacy (×10 10 ) Time (h) ns ✱ ✱✱ ns ns g Mock Npro sfNnps 0 5 10 15 20 Percen t age of MHC-I/ MHC-II positive cells (%) MHC IMHC II ✱✱✱✱ ✱✱✱✱ 0 20 40 60 CD80CD86 ns ✱✱ Per centage o f CD80/CD86 positive cells (%) sfNnpsNproMock Mo c k N p ro s fNnps 0 25 50 75 100 Percentage of DC - Np rop o sit iv e cell s( %) ✱✱✱✱ 100 μm100 μm100 μm FSC + 0.67%27.30% 49.70% FITC + sfNnpsNproMocksfNnps Ferritin 200 nm NproMock M ock Npro s fNnps 0 25 50 75 100 Percentage of DC-Npro-po s itive cell s (%) ✱✱✱ 0 5 10 15 F lu or es c en ce e f f ic acy ( ×1 0 7 ) Spleen Lymph Node ✱✱✱✱✱ sfNnpsNproMock f FIPV N(G 2 S) 3 Ferritin monomer FIPV N-Fer monomer sfNnps (Single-module FIPV N nanoparticles) ˜24 FIPV N-Fer monomer Control 200 nm 2.8˜10 10 2.6˜10 10 2.4˜10 10 2.2˜10 10 1.8˜10 10 2.0˜10 10 sfNnps Npro Mock 0 h12 h24 h36 h48 h 10 8 10 7 10 6 10 5 010 2 10 4 10 8 10 6 10 8 10 7 10 6 10 5 10 8 10 7 10 6 10 5 010 2 10 4 10 8 10 6 010 2 10 4 10 8 10 6 0.34% 0.66% 1.25% 1.82%9.34% 11.60% MHC II + I C H M + sfNnpsNproMock 10 6 10 4 10 2 0 10 6 10 4 10 2 0 10 6 10 4 10 2 0 010 2 10 4 10 6 010 2 10 4 10 6 010 2 10 4 10 6 0.02% 0.16% 14.10%15.40% 41.40%29.00% CD86 + CD80 + 010 2 10 4 10 6 010 2 10 4 10 6 010 2 10 4 10 6 10 6 10 4 10 2 0 10 6 10 4 10 2 0 10 6 10 4 10 2 0 sfNnpsNproMock Npro sfNnps Mock Spleen Lymph Node 2.8˜10 10 2.6˜10 10 2.4˜10 10 2.2˜10 10 1.8˜10 10 2.0˜10 10 Fig. 2 | Construction and characterization of the single-module FIPV N–ferritin nanoparticle vaccine (sfNnps). aSchematic of sfNnps design and self-assembly. The FIPV N protein is genetically fused to ferritin (N-Fer), which self-assembles into 24-mer nanoparticles.awas partially created in BioRender. Feiyu Zhao. (2026) https://BioRender.com/npwbtng. The nanoparticle illustrations are original and were created by the authors using Autodesk 3ds Max 2020.bTEM images of ferritin-only control and N-Fer nanoparticles, and DLS measurements of hydrodynamic size distributions. Scale bars, 200 nm. The TEM images are representative of results obtained from at least three independent experiments with similar outcomes.cAntigen uptake by DC2.4 dendritic cells treated with FITC-labeled soluble N protein (Npro) or FITC-labeled sfNnps, assessed byfluorescence microscopy.F(2, 6) = 417.6,p<0.0001,η 2 = 0.9929; sfNnps vs Npro, mean difference = 29.79, 95% CI [22.89-36.69],p<0.0001.dAntigen uptake assessed byflow cytometry.F(2, 6) = 117.7,p<0.0001,η 2 = 0.9752; sfNnps vs Npro, mean difference = 31.00, 95% CI [20.48-41.52],p=0.0003.eAfter treatment of BMDCs with Npro or sfNnps, surface expression of MHC I/II and the costimulatory molecules CD80/CD86 was analyzed byflow cytometry to assess dendritic-cell maturation and activation. Significant treatment × marker interactions were observed for MHC I/II (F(2, 12) = 9.621, p= 0.0032,η p 2 = 0.6159) and CD80/CD86 (F(2, 12) = 33.65,p<0.0001, η p 2 = 0.8486). For sfNnps vs Npro, mean differences were 12.02 and 6.357 for MHC I and MHC II, with 95% CI [9.499-14.53] and [3.839-8.874], respectively; both p< 0.0001. Mean differences were−1.903 and 12.93 for CD80 and CD86, with 95% CIs [−9.734-5.928] and [5.102-20.76], respectively;p= 0.7722 andp=0.0026, respectively.fFollowing subcutaneous administration of FITC-labeled Npro or sfNnps, whole-bodyfluorescence imaging in mice (f)andquantification offluorescence signals at the injection site. A significant treatment × time interaction was observed by two-way repeated-measures ANOVA,F(8, 24) = 14.55,p<0.0001, η p 2 = 0.8290. For sfNnps vs Npro, mean differences at 0, 12, 24, 36, and 48 h were 0.2867, 0.9133, 0.7800, 0.4033, and 0.5733, with 95% CIs [−1.044-1.617], [−0.2249- 2.052], [0.09362-1.466], [−0.08139-0.8881], and [0.3492-0.7975], respectively; p= 0.6126, 0.0778, 0.0335, 0.0807, and 0.0021, respectively.gQuantification of fluorescence signals in the spleen and draining lymph nodes. A significant treatment × tissue interaction was observed,F(2, 12) = 4.132,p=0.0431,η p 2 =0.4078; sfNnps vs Npro, mean differences = 4.730 and 7.293 for the spleen and draining lymph nodes, with 95% CIs [1.302-8.158] and [3.865-10.72], respectively;p= 0.0089 andp= 0.0004, respectively. Data are presented as mean ± SD.c,dwere analyzed by one-way ANOVA,eandgby two-way ANOVA, andfby two-way repeated-measures ANOVA with time treated as the within-subject factor, followed by two-sided Dunnett’s multiple-comparisons test where applicable. Pairwise comparisons are reported as mean differences with 95% CIs and Dunnett-adjustedpvalues; effect sizes are reported asη 2 for one-way ANOVA and partialη 2 (η p 2 ) for two-way and twoway repeated-measures ANOVA. For (c–e),n= 3 independent experiments; for (f,g), n= 3 biologically independent mice per group. NS, not significant; *p<0.05; **p< 0.01; ***p< 0.001; ****p< 0.0001. Source data are provided as a Source Datafile. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

nanoparticles, with the three dual-module formulations exhibiting peak hydrodynamic diameters of 50.7 nm, 142 nm, and 106 nm, respectively, and multi-component particles showing matching size distributions. Additional exploratory formulations also formed particulate assemblies with distinct size distributions (Supplementary Fig. 5f). These results confirmed the generation of an sfNnps-based nanoparticle library displaying distinct APC-targeting modules for antigen-presentation screening. To assess antigen-presenting efficiency, we measured nanoparticle uptake of FITC-labeled formulations by DC2.4 dendritic cells. Among the three dual-module APC-targeting screening formulations, the percentage of FITC + DC2.4 cells was highest in the dfN/Hnps group, increasing from 62.2% in the sfNnps group to 75.9%. In contrast, dfN/ Vnps and dfN/Cnps showed no significant improvement in uptake compared with sfNnps (Fig.4b), indicating that the HCV-E2-containing formulation was associated with enhanced dendritic-cell uptake. We then evaluated the cellular immune responses induced by these formulations in a mouse model. Compared with sfNnps, dfN/Vnps, dfN/ Hnps, and dfN/Cnps all significantly increased IgG2a levels, with the highest IgG2a level observed in the dfN/Hnps group (Supplementary Fig. 5h). Consistently, serum IFN-γconcentrations were numerically highest in the dfN/Hnps group (Fig.4c). Flow cytometric analysis further revealed that all three dual-module formulations increased the frequency of CD3 + CD4 + T cells, again with dfN/Hnps showing the largest numerical increase (Fig.4d). Notably, dfN/Hnps was the only formulation with a significantly higher frequency of CD3 + CD8 + Tcells than sfNnps (Fig.4d). Moreover, we evaluated the uptake-enhancing effect of dfN/Hnps in feline peripheral blood monocyte-derived dendritic cells. Consistent with thefindings in DC2.4 cells, dfN/Hnps significantly increased the proportion of FITC + feline dendritic cells from 32.0% with sfNnps to 42.4% (Fig.4f). Collectively, these results showed that dfN/Hnps (sfNnps displaying HCV E2) exhibited the strongest overall profile for dendritic-cell uptake and Th1-associated cellular immune readouts among the tested APC-targeting formulations, supporting its selection for subsequent vaccine development (Table1 and Fig.4e). Screening of immunostimulatory modules that enhance T-cell activation In parallel with the screening of APC-targeting modules, we sought to identify an immunostimulatory module that could further strengthen the cellular immune responses induced by sfNnps. Three candidate modules—flagellin, IFNγ,andIL-2—were therefore individually displayed on sfNnps and comparatively evaluated for their effects on dendritic-cell activation and T-cell-associated immune responses, with the selected formulation further validated in feline monocyte-derived dendritic cells. 0204060 0 3 6 Mock Cli nical symp tom score 0204060 0 25 50 75 Probability of s u rvival 100 Days post-infection a d 50 μm50 μm50 μm MockFIPVNormal saline+FIPV 50 μm50 μm50 μm Npro+FIPVsfNnps+FIPV Positive sera+FIPV Mock FIPV+Normal saline FIPV+Npro FIPV+sfNnps 0204060 36 39 42 Mock 0204060 36 39 42 FIPV+Normal saline 0204060 36 39 42 FIPV+Npro 0204060 36 39 42 FIPV+sfNnps 0204060 50 100 150 200 Initial weight (% ) Mock 0 204060 50 100 150 200 FIPV+Normal saline 0204060 50 100 150 200 FIPV+Npro 0204060 50 100 150 200 FIPV+sfNnps 0204060 0 3 6 FIPV+Normal saline 0204060 0 3 6 FIPV+Npro 0204060 0 3 6 FIPV+sfNnps gh b No rmal sa li n e N pro sfNn p s 0.0 0.5 1.0 1.5 Percent s of CD4 + TNFα + Tcells(%) ✱✱✱✱ N o rmal saline Np ro s fNn ps 0.0 0.2 0.4 0.6 P ercen t sof C D8 + T NF α + Tc e ll s (% ) ✱✱✱✱ TNF-αIL-1βIL-6IL-10 0.0 0.5 1.0 R elative d e n si ty valu e ✱✱✱✱✱✱✱✱✱✱ 1:21: 4 1 : 8 1:16 1:3 2 1:6 4 1: 12 8 1:256 0 3 6 9 Log 10 RNA copies/mg Mock FIPV Normal saline+FIPV Npro+FIPV sfNnps+FIPV Positive sera+FIPV Serum dilution ef Npro+FIPVsfNnps+FIPV FIPV Normal saline+FIPV Mock Positive sera+FIPV Normal saline Npro c Normal salineNprosfNnps CD4 + TNFα + FSC CD8 + TNFα + Normal salineNprosfNnps 0.00% 0.60%1.13% 0.00% 0.22%0.39% N o rm a ls a l ine Npro s f N np s 0 25 50 75 100 IFN-γ spotf orming un i ts ( S FU / 1 0 6 ce l l s ) ✱✱✱✱ Normal saline (n=3) Npro (n=6) sfNnps (n=6) Blood was collected 0 21426364 Continue to observe for 60 days Immunization FIPV (2 10 5.5 TCID 50 ) Pathologic al Clinical symptoms 124 (Day) 10 4 010 6 10 2 10 8 10 4 010 6 10 2 10 8 10 4 010 6 10 2 10 8 10 7 10 6 10 4 10 5 10 7 10 6 10 4 10 5 10 7 10 6 10 4 10 5 10 7 10 6 10 4 10 5 10 7 10 6 10 4 10 5 10 7 10 6 10 4 10 5 10 5 10 -1 10 8 10 2 10 11 10 5 10 -1 10 8 10 2 10 11 10 5 10 -1 10 8 10 2 10 11 Days post-infection Days post-infection Days post-infection sfNnps LiverSpleenKidney 0 2 4 6 8 Log 10 R N A c o pie s/m g ns i Mock FIPV+Normal saline FIPV+Npro FIPV+sfNnps M ock FIPV N o r m al saline +F I P V N pro+ FI P V sf N nps+F IP V Posi t ive ser a + F I PV 5 15 25 Mean fluor e scence intensity Body temperature ( ) Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

Using the SpyCatcher/SpyTag system,flagellin, IFNγ,orIL-2was individually displayed on the surface of sfNnps to generate three dualmodule formulations, dfN/Fnps, dfN/Inps, and dfN/Lnps, respectively (Fig.4g). Protein expression and Western blot analysis confirmed that Flagellin-SC, IFNγ-SC, and IL-2-SC migrated at the expected molecular weights of approximately 82.2 kDa, 50.5 kDa, and 48.5 kDa, respectively, and were recognized by an anti-His tag antibody (Supplementary Fig. 6a, b). Following conjugation, SDS-PAGE showed the expected mobility-shifted bands for dfN/Fnps, dfN/Inps, and dfN/Lnps at approximately 165 kDa, 133 kDa, and 131 kDa, respectively, whereas additional exploratory formulations carrying multiple helper factors exhibited the corresponding multiband patterns (Supplementary Fig. 6c). TEM and DLS analyses confirmed that all formulations assembled into spherical nanoparticles, with the three dual-module formulations exhibiting peak hydrodynamic diameters of 142 nm, 68 nm, and 58 nm, respectively. Additional exploratory formulations also formed particulate assemblies with distinct size distributions (Supplementary Fig. 6d). These results established the construction of three sfNnps-based dual-module immunostimulatory screening formulations displaying distinct immunostimulatory modules. We next assessed whether these formulations enhanced cellular features associated with T-cell priming. Relative to sfNnps, MHC I expression was significantly higher in the dfN/Inps group, whereas MHC II levels were not significantly different among the tested formulations (Fig.4h; Supplementary Fig. 6e). For costimulatory molecules, CD80 expression was significantly increased in dfN/Inps and dfN/Lnps, with dfN/Inps showing the largest numerical increase. CD86 expression showed an increasing trend but did not reach statistical significance (Fig.4h and Supplementary Fig. 6e). The immune responses induced by these formulations were further evaluated in mice. Serum IgG2a levels were significantly elevated in all formulations, with the highest IgG2a level observed in the dfN/Inps group (Supplementary Fig. 6f). Serum IFN-γconcentrations were significantly higher in the dfN/Fnps and dfN/Inps groups than in the sfNnps group (p< 0.05), with dfN/Inps exhibiting the highest response (Fig.4i). Flow cytometric analysis further showed that the frequencies of both CD3 + CD4 + and CD3 + CD8 + T cells were significantly higher in the dfN/Inps group than in the sfNnps group (Fig.4j). Moreover, we evaluated the effects of dfN/Inps on dendritic-cell activation and antigen presentation using feline peripheral blood monocyte-derived dendritic cells. Consistent with thefindings from the murine cells, RT-qPCR analysis showed thatCD80andMHC ImRNA expression levels were significantly higher in dfN/Inps-treated cells than in sfNnps-treated cells (Fig.4l). Collectively, these results showed that dfN/Inps (sfNnps displaying IFNγ) exhibited the strongest overall profile across T-cellpriming-associated readouts and increased CD4⁺and CD8⁺T-cell responses among the tested immunostimulatory formulations, supporting its selection for subsequent vaccine development (Table1;Fig.4k). Construction and evaluation of a tri-modular FIPV N-ferritin nanoparticle vaccine The above results indicated that formulations containing HCV E2 or IFNγwere associated with distinct dendritic-cell uptake-, activation-, and T-cell-response-associated profiles. We therefore combined HCV E2 and IFNγon the sfNnps scaffold to generate a tri-modular FIPV N-ferritin nanoparticle vaccine (tfN/HInps) and systematically evaluated its immunogenicity, serum-mediated ADE risk, and protective efficacy in cats. To assemble tfN/HInps, purified HCV E2-SC and IFNγ-SC were conjugated to sfNnps using an equal molar input ratio (Supplementary Fig. 7a). SDS-PAGE revealed two major bands at approximately 154 kDa and 133 kDa, consistent with the expected molecular weights of the corresponding conjugated species (N-Fer–HCV E2 and N-Fer–IFNγ, respectively; Supplementary Fig. 7b). TEM and DLS analyses further confirmed spherical particulate structures and a bimodal hydrodynamic-size distribution, with peaks at 68 and 142 nm (Supplementary Fig. 7c). These results confirmed the generation of tfN/ HInps co-displaying HCV E2 and IFNγ. The immunogenicity of tfN/ HInps was then evaluated in cats following the immunization schedule shown in Fig.5a. After the third immunization, serum FIPV N-specific IgG levels in the tfN/HInps group were significantly higher than those in the sfNnps group (Fig.5b). Cellular immune responses induced by tfN/HInps were assessed using an IFN-γELISpot assay. Following Fig. 3 | Immune responses induced by sfNnps in cats and protection against FIPV infection. aCat immunization, challenge, and sampling scheme. Following intraperitoneal FIPV challenge, after all cats in the challenge control group (FIPV + normal saline) succumbed, three cats from the FIPV + Npro group and three cats from the FIPV + sfNnps group were randomly selected for humane euthanasia, and liver, spleen, and kidney tissues were collected for histopathological examination and viral load quantification.awas partially created in BioRender. Feiyu Zhao. (2026)https://biorender.com/gyvfykn. The nanoparticle illustrations are original and were created by the authors using Autodesk 3ds Max 2020.bFIPV N-specific IFN-γspot-forming units after ex vivo stimulation of peripheral blood cells with an overlapping N peptide pool following the third immunization. sfNnps vs Npro: Welch’st(9.969) = 15.76,p<0.0001,η 2 = 0.9614, mean difference = 50.83, 95% CI [43.64-58.02].cFrequencies of TNF-α⁺CD4⁺and TNF-α⁺CD8⁺TcellsamongFIPVN peptide-stimulated peripheral blood lymphocytes in cats after the third immunization were quantified by multicolorflow cytometry to evaluate sfNnps-induced N- specific cellular immune responses. For TNF-α⁺CD4⁺Tcells,sfNnpsvsNpro: Welch’st(9.576) = 8.825,p<0.0001,η 2 = 0.8905, mean difference = 0.5583, 95% CI [0.4165-0.7002]. For TNF-α⁺CD8⁺Tcells,Welch’st(7.055) = 6.517,p= 0.0003, η 2 = 0.8575, mean difference = 0.1717, 95% CI [0.1095-0.2339].d–fIn vitro assessment of serum-mediated ADE using feline PBMCs exposed to FIPV pre-incubated with post–third immunization sera. Viral attachment (d), viral replication (e), and pro-inflammatory cytokine responses (f)werequantified. Ford,F(5, 12) = 45.05, p<0.0001,η 2 = 0.9494; FIPV-positive sera + FIPV vs Npro + FIPV and sfNnps + FIPV, mean differences = 10.57 and 5.457, with 95% CIs [6.576-14.56] and [1.462-9.451], respectively; Dunnett-adjustedp<0.0001andp= 0.0077, respectively. Fore,a significant treatment × serum-dilution interaction was observed,F(35, 96) = 9.419, p<0.0001,η p ² = 0.7744. Forf,asignificant treatment × cytokine interaction was observed,F(15, 48) = 2.729,p=0.0043,η p 2 = 0.4603. Compared with FIPV-positive sera + FIPV, mean differences for Npro + FIPV were 0.1933, 0.2900, 0.1600, and 0.1767 for TNF-α, IL-1β, IL-6, and IL-10, with 95% CIs [0.08200-0.3047], [0.1787- 0.4013], [0.04866-0.2713], and [0.06533-0.2880], respectively; Dunnett-adjusted p= 0.0002, <0.0001, = 0.0023, and = 0.0007, respectively. The corresponding mean differences for sfNnps + FIPV were 0.2033, 0.3400, 0.1667, and 0.2133, with 95% CIs [0.09200-0.3147], [0.2287-0.4513], [0.0526605533-0.2780], and [0.1020- 0.3247], respectively; Dunnett-adjustedp< 0.0001, <0.0001, = 0.0014, and <0.0001, respectively.g,hProtective efficacy following FIPV challenge is shown as clinical scores (g) and survival curves (h).iViral RNA loads in liver, spleen, and kidney following challenge were quantified by RT-qPCR to assess the impact of sfNnps-induced immunity on viral clearance in vivo. A significant treatment effect was observed,F(3, 24) = 124.0,p<0.0001,η p 2 = 0.9394. sfNnps vs Npro, mean differences =−0.3867,−1.287, and−1.857 for liver, spleen, and kidney, with 95% CIs [−1.658-0.8848], [−2.558 to−0.01520], and [−3.128 to−0.5822], respectively; p= 0.7879, 0.0469, and 0.0034, respectively. Data are presented as mean ± SD. Comparisons between sfNnps and Npro in (b)and(c) were performed using twotailed Welch’s unpairedt-tests.dwas analyzed by one-way ANOVA, ande,f,iby twoway ANOVA, followed by two-sided Dunnett’s multiple-comparisons test where applicable. Pairwise comparisons are reported as mean differences with 95% CIs and Dunnett-adjustedpvalues. Effect sizes are reported asη 2 for Welch’st-tests and one-way ANOVA and partialη 2 (η p 2 ) for two-way ANOVA. In (b)and(c),n=3biologically independent cats for the normal saline group andn= 6 biologically independent cats for the Npro and sfNnps groups; in (d–f)andi,n= 3 biologically independent cats per group. Survival curves in (h) were generated using the Kaplan-Meier method and compared using the log-rank (Mantel-Cox) test. NS, not significant; *p<0.05;**p< 0.01; ***p<0.001;****p< 0.0001. Source data are provided as a Source Datafile. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

ex vivo stimulation with an overlapping peptide pool spanning the FIPV N protein, tfN/HInps induced significantly higher N-specificIFN-γ ELISpot responses than dfN/Hnps, dfN/Inps, and sfNnps (Fig.5cand Supplementary Fig. 7e). Similarly, the frequency of TNF-α⁺N-specific CD4⁺T cells in the tfN/HInps group was significantly higher than that in the dfN/Hnps, dfN/Inps, and sfNnps groups (Fig.5d and Supplementary Fig. 7f). Thesefindings indicated that tfN/HInps was associated with higher IgG responses and Th1-associated cellular immune responses in cats. We further used the inactivated vaccine group as an ADE reference to evaluate whether tfN/HInps-elicited sera could mediate ADE in feline cells. FIPV attachment in cells exposed to tfN/HInps immune sera was comparable to that in the FIPV group but significantly lower than that in the inactivated vaccine and positive sera control groups (Fig.5e). Across the tested serum dilutions, tfN/HInps immune sera did not enhance viral replication (Fig.5f). Additionally, TNF-α,IL-1β,IL-6, and IL-10 levels were significantly lower in the tfN/HInps group than those in the inactivated vaccine and positive sera control groups (Fig.5g). To determine whether the observed enhancement was associated with serum IgG, serum IgG was depleted using Protein A/G before viral attachment, replication, and inflammatory responses were reassessed. IgG depletion attenuated the enhanced viral attachment, replication, and inflammatory responses observed in positive sera and sera from the inactivated vaccine group. By contrast, sera from all N-protein nanoparticle vaccine groups showed no detectable ADE- related enhancement either before or after Protein A/G treatment (Supplementary Fig. 8). These data indicated that tfN/HInps immunization did not induce detectable serum-mediated ADE under the tested conditions. In the feline FIPV challenge experiment, sfNnps conferred 25.0% (2/8) survival, whereas dfN/Hnps and dfN/Inps achieved 50.0% (4/8) and 62.5% (5/8) survival, respectively. tfN/HInps achieved 100% (8/8) survival over the 60-day observation period (Fig.5h, i). In contrast, all cats in the Freund’s adjuvant, inactivated vaccine, and normal saline groups succumbed within 7-19 days post-infection. Furthermore, among the non-surviving cats in the sfNnps, dfN/Hnps, and dfN/Inps groups, the earliest deaths occurred at 27, 28, and 24 days post-challenge, respectively, representing delays of 20, 21, and 17 days relative to the inactivated vaccine group. Thus, these groups did not exhibit the accelerated disease progression and earlier mortality typically associated with ADE (Fig.5h, i). Consistent with survival outcomes, peripheral-blood viral RNA loads were numerically lowest in the tfN/ HInps group at days 12 and 19 post-challenge. By day 19, tfN/HInps reduced peripheral-blood viral RNA levels by 73.0% relative to the normal-saline challenge control group (Fig.5j). Collectively, these results indicated that tfN/HInps (sfNnps incorporating HCV E2 and IFNγ) elicited enhanced Th1-associated cellular immunity, achieved 100% (8/8) survival in the tested FIPV challenge model, and showed no detectable serum-mediated ADE under the tested conditions, supporting its potential as an optimized tri-modular N-ferritin nanoparticle vaccine candidate (Table2). Global changes in the splenic immune-cell landscape following tfN/HInps immunization To characterize changes in the splenic immune-cell landscape associated with tfN/HInps immunization in cats, we performed 10x Genomics single-nucleus RNA sequencing (snRNA-seq) on splenic samples from tfN/HInps-immunized cats, cats receiving the inactivated vaccine, and unimmunized controls. Unsupervised clustering using Seurat, followed by annotation based on canonical marker genes, identified nine major cell populations (Supplementary Fig. 9a). Relative to the unimmunized controls, tfN/HInps immunization was associated with changes in splenic cellular composition. We observed lower relative proportions of monocytes-DC, endothelial cells, and B cells, accompanied by higher relative Table 1 | Comparative screening of APC-targe ting and immunostimulatory dual-module sfNnps formulations in mouse cells and mice Formulation Composition Antigen uptak e MHC I expression CD80 expression IFN- γ response CD3 ⁺ CD4 ⁺ T-cell response CD3 ⁺ CD8 ⁺ T-cell response Normal saline / / / / ○○○○○○○○○○○○○○○ Single-module Formulation (sfNnps) N antigen ●○○○○ // ●○○○○●○○○○●○○○○ Dual-module APC-targeting formulation (dfN/Vnps) N antigen, HIV Gp120 ●○○○○ // ●●◐○○●●○○○◐○○○○ Dual-module APC-targeting formulation (dfN/Hnps) N antigen, HCV E2 ●●○○○ // ●●◐○○●●●○○●●●○○ Dual-module APC-targeting formulation (dfN/Cnps) N antigen, CD40L ●○○○○ // ●●○○○●●○○○○○○○○ Dual-module immunostimulatory formulation(dfN/Fnps) N antigen, Flagellin / ●●○○○●●◐○○●●○○○●●●○○●◐○○○ Dual-module immunostimulatory formulation(dfN/Inps) Nantigen,IFN γ / ●●●●◐●●●●●●●◐○○●●●●◐●●●○○ Dual-module immunostimulatory formulation(dfN/Lnps) Nantigen,IL-2 / ●●◐○○●●●●◐●○○○○●●●●○●○○○○ Relative performance of the different formulations across the evaluated parameters was expressed using a sfNnps-referenced relative scoring syst em. The experimentally measured value of sfNnps for each corresponding parameter was used as the reference, and the score was calculated according to the formula Y = 20% + [(X − A)/A] × 100%, where X represents the measured value of the formulation being evaluated and A represents the corresponding measured value of the sfNnps group. Accordingly, the sfNnps group was assigned a score of 20% for each parameter evaluated. Scores were constrained to a range of 0-100%; values below 0% were set to 0%, whereas valu es above 100% were set to 100%. The resulting scores were further converted into a multilevel circlebased scale for intuitive vi sualization as follows: Y ≤ 0, ○○○○○ ;0<Y ≤ 10%, ◐○○○○ ;10%<Y ≤ 20%, ●○○○○ ;20%<Y ≤ 30%, ●◐○○○ ; and 30% <Y ≤ 40%, ●●○○○ . Thereafter, each additional 10-percentage-poi nt increase in Y was represented by an additional halffi lled circle ( ◐ ). A baseline score of 20% was assigned to sfNnps to enable decreases relative to sfNnps to be visually represented within the scoring scale. This scori ng system was used solely to facilitate visual comparison of the relative performance and overall trends among formulations across different parameters and does not indicate statistical signi fi cance between groups. The original measured values and correspond ing statistical analyses for eac h parameter are provided in Fig. 4 . “ / ” indicates that the corresponding parameter was not evaluated for that formulation. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

proportions offibroblasts, plasma cells, T cells, granulocytes, and mast cells. Compared with the inactivated vaccine group, tfN/HInps induced a more pronounced enrichment of T cells, while monocytes-DC, fibroblasts, and endothelial cells also exhibited increasing trends. In contrast, B cells and plasma cells showed modestly lower relative proportions (Fig.6a). These data suggested that tfN/HInps immunization was associated with a splenic immune-cell composition consistent with a stronger cellular-immunity-oriented profile. Given the antigen-uptake- and presentation-associated effects observed in feline dendritic-cell assays (Fig.4f), we further resolved the monocytes-DC cluster intofiner subsets. Re-clustering identifiedfive subpopulations, including cDC1, cDC2, pDC, monocytes, and moDC (Supplementary Fig. 9b). Compared with the inactivated vaccine group, the tfN/HInps groupshowedahigherrelativeproportion of cDC1 but lower relative proportions of cDC2, pDC, monocytes, and moDC. In the analyzed dataset, cDC1 increased by 177 cells, while the other four subsets decreased by a total of 97 cells (Fig.6b). Thesefindings indicated that tfN/HInps immunization was associated with a higher relative representation of cDC1, a pattern consistent with stronger crosspresentation-related immune features. We further examined whether the increased T-cell representation was accompanied by changes in specific T-cell subsets. Re-clustering of the T cell compartment identified eight functional subsets, including CD8 + naïve T cells (CD8 Tn), CD8 + stem cell memory T cells (CD8 Tscm), CD8 + central memory T cells (CD8 Tcm), CD8 + effector memory T cells (CD8 Tem), CD8 + effector T cells (CD8 Teff), CD4 + central memory T cells (CD4 Tcm), Th1 cells, and T follicular helper cells (Tfh; Supplementary Fig. 9c). Relative to the inactivated vaccine group, the tfN/HInps group showed higher relative proportions of multiple T-cell subsets, including CD8 Tn, CD8 Tcm, CD8 Tem, CD8 Teff, CD4 Tcm, and Th1, whereas CD8 Tscm showed a modest reduction (Fig.6c). Together, these results indicated that tfN/HInps immunization was associated with higher relative representation of both effector and memory T-cell compartments and with stronger Th1-associated immune features, consistent with a more diversified distribution of CD8⁺effector and memory populations. Immune features associated with enhanced antigen uptake, processing, and presentation following tfN/HInps immunization Given the higher relative representation of cDC1 observed following tfN/HInps immunization, we further characterized DC state distributions and cDC1 transcriptional features associated with antigen recognition, uptake, processing, and presentation, as well as predicted cDC1-T-cell communication. To examine DC state distributions associated with tfN/HInps immunization, we reconstructed DC trajectories using pseudotime ac ) L m / g p 2 0 1 × ( n o i t a r t n e c n o c - N F I m u r e S 5.3 0.3 5.2 0.2 5.1 0.1 5.0 0.0 s p n C / N f d s p n H / N f d s p n V / N f d s p n N f s e n i l a s l a m r o N sfNnpsdfN/Vnps dfN/HnpsdfN/Cnps )%(sllecevitisopfostnecreP 07 530 Normal saline sfNnps dfN/Hnps f DCs (Feline) of FITC + s n )%(sllecevitisopfostnecreP s n 061080 Mock sfNnps dfN/Vnps dfN/Hnps dfN/Cnps DC2.4 of FITC + s n s n )%(sllecevitisopfostnecreP 041070 Normal saline sfNnps dfN/Vnps dfN/Hnps dfN/Cnps s n s n )%(sllecevitisopfostnecreP 05520 Normal saline sfNnps dfN/Vnps dfN/Hnps dfN/Cnps CD3 + CD8 + T lymphocytes b d e BMDCs of CD80 + s n )%(sllecevitisopfostnecreP 05520 Normal saline sfNnps dfN/Fnps dfN/Inps dfN/Lnps sn ) L m / g p 2 0 1 × ( n o i t a r t n e c n o c - N F I m u r e S 4 2 0 s p n L / N f d s p n I / N f d s p n F / N f d s p n N f s e n i l a s l a m r o N enileF(noisserpxeANRmevitaleR enilaslamroN spnNfs spnI/Nfd 210 )enileF(noisserpxeANRmevitaleR enilaslamroN spnNfs spnI/Nfd 210 sfNnps dfN/Fnps dfN/InpsdfN/Lnps g s n s n )%(sllecevitisopfostnecreP 05520 Normal saline sfNnps dfN/Fnps dfN/Inps dfN/Lnps BMDCs of MHC I + Normal saline sfNnps dfN/Fnps dfN/Inps dfN/Lnps CD3 + CD4 + T lymphocytes Normal saline sfNnps dfN/Fnps dfN/Inps dfN/Lnps CD3 + CD8 + T lymphocytes h MHC I CD80 ijkl CD3 + CD4 + T lymphocytes )%(sllecevitisopfostnecreP 021060 s n s n ) %(sllecevitisopfostnecreP 05520 ST-N-Fer HCV E2-SC HIV Gp120-SC CD40L-SC SpyCatcher /SpyTag dfN/VnpsdfN/HnpsdfN/Cnps ST-N-Fer IFNγ-SC Flagellin-SC IL-2-SC dfN/FnpsdfN/InpsdfN/Lnps SpyCatcher /SpyTag Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

analysis. In unimmunized controls, only a limited number of cells were distributed across the pseudotime trajectory, with most cells residing in the early stage and predominantly annotated as cDC1 (104 cells). In the inactivated vaccine group, the early pseudotime was dominated by pDC (116 cells) and cDC1 (100 cells), the middle stage primarily contained cDC1 (50 cells), and the late stage consisted of cDC1 (19 cells) and cDC2 (35 cells). In contrast, although the overall trajectory structure in the tfN/HInps group was similar to that of the inactivated vaccine group, cDC1 cells were more frequently represented at late pseudotime in the tfN/HInps group, reaching 229 cells (Fig.7a). These data indicated that tfN/HInps immunization was associated with a cDC1-enriched late-pseudotime state. We then examined the expression of receptors associated with antigen recognition and uptake within cDC1. Compared with the inactivated vaccine group, cDC1 from tfN/HInps-immunized cats exhibited increased expression of multiple pattern-recognition receptors (includingTLR8andNLRP3), complement-associated receptors (e.g.,ITGB2), scavenger receptors (MSR1andSCARB1), IgG Fc receptors (FCGR2B), and C-type lectin receptors (CLEC7A;Fig.7b). This coordinated upregulation was consistent with transcriptional features related to enhanced antigen sensing and internalization in cDC1 following tfN/HInps immunization. We further assessed pathways supporting intracellular antigen processing and MHC loading. In cDC1 from the tfN/HInps group, key components of the JAK-STAT-SOCS regulatory axis (includingSOCS3andSTAT3), the MHC class I transactivatorNLRC5, and MHC class II regulatory factors (CIITA,RFX5,andRFXAP) were upregulated (Fig.7c). In parallel, endo-lysosomal processing genes (includingATP6V1A,CTSB,CTSD, and related factors) were elevated, consistent with enhanced transcriptional features related to proteolytic and vesicular processing. Genes implicated in cross-presentation were also expressed at higher levels, spanning vesicle-mediated pathways (RAB11A,RAB14,RAB27A, etc.), proteasome-associated components (PSMB1,PSMB2,and PSMB3), and MHC I peptide transport/editing machinery (ERAP1, LNPEP,SEC61A1,andTAP1). Moreover, transcripts linked to costimulatory and cytokine signaling (CD40,CD70,IL10RA,andIL15) were increased, consistent with transcriptional features associated with T-cell priming (Fig.7c). These signatures indicated that tfN/HInps immunization was associated with a coordinated transcriptional program related to antigen processing and cross-presentation in cDC1. To assess whether the transcriptional remodeling of cDC1 was associated with predicted changes in cDC1-T-cell communication, we performed cell–cell communication analysis. Relative to the inactivated vaccine group, cDC1 in the tfN/HInps group exhibited higher predicted outgoing signaling to both CD4⁺and CD8⁺T cells. For cDC1- CD4⁺T-cell interactions, the predicted interaction strengths of VCAM1-ITGA4, ICOSLG-ICOS, and CD48-CD244 increased by more than 0.05, accompanied by higher expression of VCAM1, ITGA4, ICOSLG, and ICOS, whereas CD48 and CD244 were not markedly upregulated (Fig.7d). For cDC1-CD8⁺T-cell interactions, ICOSLG-ICOS, ICAM2-ITGAL, and VCAM1-ITGA4 interactions were strengthened by over 0.05;VCAM1,ITGA4,ICOSLG,andICOSshowed higher expression, whileICAM2showed no clear transcriptional induction (Fig.7e). These data were consistent with stronger predicted cDC1-mediated communication with T-cell subsets following tfN/HInps immunization, especially for the cDC1–CD8⁺T-cell axis. Collectively, these results indicated that tfN/HInps immunization was associated with increased representation of cDC1 at late pseudotime and a broad transcriptional program consistent with enhanced antigen uptake, processing, and cross-presentation. Additionally, tfN/HInps was associated with stronger predicted cDC1-T-cell interaction networks, consistent with the enhanced cellular immune profile observed in tfN/HInpsimmunized cats. Immune features associated with enhanced CD8+T-cell responses following tfN/HInps immunization Given the increased representation of multiple CD8⁺effector and memory subsets and the stronger predicted cDC1-CD8⁺T-cell communication observed following tfN/HInps immunization, we further characterized CD8⁺T-cell differentiation states, functional features, and transcriptional regulatory programs. To examine CD8⁺T-cell state distributions associated with tfN/ HInps immunization, we reconstructed CD8⁺T-cell trajectories using pseudotime analysis. In unimmunized controls, most cells were in the early pseudotime stage, mainly CD8 Tn (188 cells), with few cells represented at late pseudotime. In the inactivated vaccine group, Fig. 4 | Construction and screening of sfNnps-based dual-module formulations displaying APC-targeting or immunostimulatory modules. a,gSchematics of SpyCatcher/SpyTag-mediated assembly of sfNnps-based APC-helper formulations displaying HIV Gp120-SC, HCV E2-SC, or CD40L-SC (a), and T-cell-helper formulations displayingflagellin-SC, IFN-γ-SC, or IL-2-SC (g). The nanoparticle illustrations are original and were created by the authors using Autodesk 3ds Max 2020.bFlowcytometric analysis of the uptake of FITC-labeled APC-helper formulations by DC2.4 dendritic cells.F(4, 10) = 272.9,p<0.0001,η 2 = 0.9909; compared with sfNnps, mean differences were 0.9000 for dfN/Vnps, 13.70 for dfN/Hnps, and 2.067 for dfN/Cnps, with 95% CIs [−6.572-8.372], [6.228-21.17], and [−5.405-9.538], respectively;p= 0.9893, 0.0012, and 0.8382, respectively.c, i.SerumIFN-γconcentrations in mice after the third immunization with the indicated APC-helper (c) or T-cell-helper (i) formulations. Forc,F(4, 10) = 566.3,p<0.0001,η 2 = 0.9956; dfN/ Hnps vs sfNnps, mean difference = 0.6733, 95% CI [0.4942-0.8525],p< 0.0001; fori, F(4, 10) = 295.4,p<0.0001,η 2 = 0.9916; compared with sfNnps, mean differences were 0.2999 for dfN/Fnps and 0.7069 for dfN/Inps, with 95% CIs [0.01279-0.5870] and [0.4198-0.9940], respectively;p= 0.0405 and 0.0001, respectively. d,jFrequencies of CD3⁺CD4⁺and CD3⁺CD8⁺T cells among splenic lymphocytes after the third immunization with the indicated APC-helper (d)orT-cell-helper(j) formulations. Ford,F(4, 10) = 75.56,p<0.0001,η 2 = 0.9680 for CD4⁺T cells and F(4, 10) = 141.5,p<0.0001,η 2 = 0.9826 for CD8⁺T cells; dfN/Hnps vs sfNnps, mean differences = 21.49 and 6.147, 95% CI [9.149-33.84] and [3.167–9.126], respectively; p= 0.0018 andp= 0.0005, respectively. Forj,F(4, 10) = 320.6,p<0.0001, η 2 = 0.9923 for CD4⁺T cells andF(4, 10) = 163.1,p<0.0001,η 2 = 0.9849 for CD8⁺ T cells; dfN/Inps vs sfNnps, mean differences = 32.12 and 6.277, 95% CI [25.78-38.47] and [3.373-9.180], respectively;p< 0.0001 andp= 0.0003, respectively.e,kRadar plots summarizing the normalized screening readouts for the APC-helper (e;b–d) and T-cell-helper (k;h–j) formulations. For each readout, sfNnps was assigned a baseline score of 20%, and the scores for the other formulations were calculated as 20% + [(value of the indicated formulation−value of sfNnps) / value of sfNnps] × 100%. Scores were not capped at 100%.f,lValidation in feline peripheral blood monocyte-derived dendritic cells, including uptake of FITC-labeled nanoparticles quantified byflow cytometry (f)andMHC IandCD80mRNA expression quantified by RT-qPCR after treatment with the indicated formulations (l). Forf,F(2, 12) = 2645,p<0.0001,η 2 = 0.9977; dfN/Hnps vs sfNnps, mean difference = 10.47, 95% CI [8.949-11.99],p< 0.0001. Forl,F(2, 12) = 263.2,p<0.0001,η 2 =0.9777for MHC I andF(2, 12) = 358.6,p<0.0001,η 2 = 0.9835 for CD80; dfN/Inps vs sfNnps, mean differences = 0.4769 and 0.2672, 95% CI [0.3704-0.5833] and [0.1847- 0.3496], respectively; bothp<0.0001.hFlow-cytometric analysis of MHC I and CD80 surface expression in bone marrow-derived dendritic cells treated with the indicated T-cell-helper formulations.F(4, 10) = 29.99,p<0.0001,η 2 =0.9231for MHC I andF(4, 10) = 31.38,p<0.0001,η 2 = 0.9262 for CD80; For MHC I, dfN/Inps vs sfNnps, mean difference = 10.00, 95% CI [2.927-17.08],p= 0.0073. For CD80, compared with sfNnps, mean differences were 13.88 for dfN/Inps and 10.63 for dfN/ Lnps, with 95% CIs [5.284-22.48] and [2.034-19.23], respectively;p= 0.0030 and 0.0164, respectively. Data are presented as mean ± SD.b–d,f,h–j,lwere analyzed by one-way ANOVA followed by two-sided Dunnett’s multiple-comparisons test, with sfNnps as the reference group. Pairwise comparisons are reported as mean differences with 95% CIs and Dunnett-adjustedpvalues; effect sizes are reported asη 2 . For (b,h),n= 3 independent experiments per group; for (c,d) and (i,j),n= 3 biologically independent mice per group; for (f)and(l),n= 5 independent experiments per group. NS, not significant; *p< 0.05; **p< 0.01; ***p<0.001; ****p< 0.0001. Source data are provided as a Source Datafile. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

early-to-intermediate pseudotime was dominated by CD8 Tscm and CD8 Tcm, whereas late pseudotime was enriched for CD8 Tcm (552 cells) and CD8 Tem (231 cells). The tfN/HInps group showed a similar trajectory pattern but with more CD8 Tcm throughout and a marked increase in the representation of CD8 Tem at late pseudotime (320 cells; Fig.8a). These data indicated that tfN/HInps immunization was associated with greater relative representation of central- and effectormemory CD8⁺T-cell states than that observed following inactivatedvaccineimmunization.Wenextquantified functional programs associated with proliferation, memory, and cytotoxicity along the differentiation continuum. Functional module scores were comparable among groups at early and middle pseudotime. However, at late pseudotime, CD8⁺T cells from tfN/HInps-immunized cats exhibited higher scores across all three functional dimensions (Fig.8b). Additionally, subset-level analysis revealed that late-pseudotime CD8 Tem in the tfN/HInps group exhibited higher proliferation and memory scores than other CD8⁺subsets (Fig.8c), suggesting that tfN/HInps immunization was associated with higher proliferative potential and memory-related features in CD8 Tem. To characterize transcriptional regulatory features associated with the enhanced CD8 Tem phenotype, we inferred transcription factor (TF) regulatory activity within CD8 Tem. Relative to the controls, both the tfN/HInps and inactivated vaccine groups showed higher inferred activity of TBX21, RUNX2, NFATC1, and IRF1, while showing lower inferred activity of STAT3 and NFKB1. The tfN/HInps group abd FIPV+Normal saline FIPV+sfNnps FIPV+dfN/Hnps Mock FIPV+dfN/Inps FIPV+tfN/HInps FIPV+Inactivated vaccine hi ef g 0 2142 6364124 FIPV (2˜ 10 5.5 TCID 50 ) Blood was collected Continue to observe for 60 days Three immunizations Clinical symptoms (Day) Normal saline (n=5) Inactivated vaccine (n=8) sfNnps (n=8) dfN/Hnps (n=8) dfN/Inps (n=8) tfN/HInps (n=8) Freund’s adjuvant (n=5) sfNnps+FIPVdfN/Hnps+FIPVdfN/Inps+FIPVtfN/HInps+FIPV Positive sera+FIPVInactivated vaccine+FIPV FIPV Normal saline+FIPVMockFreund’s adjuvant+FIPV FIPV+Freund’s adjuvant j FIPV+Normal saline FIPV+sfNnps FIPV+dfN/Hnps Mock FIPV+dfN/Inps FIPV+tfN/HInps FIPV+Inactivated vaccine FIPV+Freund’s adjuvant Days post-infection Days post-infection Days post-infection Mock FIPV Freund’s adjuvant + FIPV Inactivated vaccine+FIPV Normal saline + FIPV sfNnps + FIPV Positive sera + FIPV Inactivated vaccine +FIPV Mock 50 μm N/DAPI dfN/Hnps +FIPV 50 μm N/DAPI FIPV 50 μm N/DAPI dfN/Inps +FIPV 50 μm N/DAPI Normal saline +FIPV 50 μm N/DAPI tfN/HInps +FIPV 50 μm N/DAPI Freund’s adjuvant +FIPV 50 μm N/DAPI 50 μm N/DAPI sfNnps +FIPV 50 μm N/DAPI Positive sera +FIPV 50 μm N/DAPI c dfN/Hnps + FIPV dfN/Inps + FIPV tfN/HInps + FIPV Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

showed higher inferred FOXO1 regulon activity than the inactivated vaccine group, andFOXO1expression was highest in the tfN/HInps group (Fig.8d, e). Consistent with this, FOXO1 downstream targets were broadly upregulated in tfN/HInps-immunized cats relative to bothcontrolsandtheinactivatedvaccinegroup,includinggenes linked to proliferation (e.g.,CENPF,TPX2,BCL11B,NR4A3,KLF10,and FKBP5), memory-associated programs (e.g.,BCL11B,GATA3,CD69, NR4A3,andKLF10), and cytotoxicity-related functions (e.g.,NCK2, WASF2,BCL11B,CENPF,andTPX2;Fig.8f). Collectively, these results indicated that tfN/HInps immunization was associated with increased representation of CD8 Tem at late pseudotime and a FOXO1-centered transcriptional program consistent with enhanced proliferative, memory-associated, and effector-related features. Thesefindings provided a transcriptional correlate of the enhanced CD8⁺T-cell response observed following tfN/HInps immunization. Discussion The development of traditional vaccines against highly pathogenic coronaviruses such as FIPV has been persistently constrained by ADE, posing a major challenge to achieving robust efficacy without compromising safety. Vaccines that target the surface S protein induce high-titer neutralizing antibodies but may not fully eliminate the risk of ADE 12,13,15,22 . In contrast, approaches based on conserved internal antigens, such as the N protein may reduce reliance on potentially ADE- associated antibody responses, but they often induce cellular responses that are insufficient to confer complete protection. To address this, we established a ferritin nanoparticle vaccine platform centered on the FIPV N protein and rationally incorporated two functional immune-enhancement modules: HCV E2, selected for its potential to enhance antigen uptake, and IFNγ, selected to promote Th1-associated cellular immunity. The combined display of these functional modules on the FIPV N-only ferritin scaffold generated a trimodular nanoparticle vaccine (tfN/HInps). In a lethal FIPV challenge model, tfN/HInps conferred 100% (8/8) survival and showed no detectable serum-mediated ADE under the tested conditions. These findings indicate that tfN/HInps combined high protective efficacy with no detectable serum-mediated ADE under the tested conditions and provide a framework for dissecting the cellular mechanisms underlying protection and for extending this multimodule nanoparticle strategy to related coronavirus vaccine applications. Current vaccine strategies aimed at mitigating ADE can be broadly grouped into two conceptual approaches, each with inherent limitations. Thefirst strategy focuses on structural optimization and epitope reshaping of surface antigens such as S or E proteins. This approach is intended to reduce the exposure of ADE-associated epitopes and focus immune responses on neutralizing determinants. Related examples include stabilized E-protein dimers designed to preserve neutralizing quaternary epitopes inflavivirus vaccine development and RBD- focused subunit vaccines such as ZF2001 for SARS-CoV-2 53–55 .These approaches concentrate immune responses on dominant neutralizing epitopes and can enhance humoral immunity. However, protection remains predominantly antibody-dependent, and a residual risk of ADE may persist, particularly following viral antigenic drift or waning antibody titers. The second strategy shifts the immune focus away from dominant viral surface antigens toward conserved non-surface antigens, such as coronavirus N proteins orflavivirus NS1. This approach aims to elicit T-cell-mediated immunity, as exemplified by DENV NS1-based vaccine candidates and vaccines based on baculovirus-expressed FIPV N protein 40,56,57 . These approaches may reduce reliance on potentially ADE-associated antibody responses but often exhibit insufficient immunogenicity, providing only disease mitigation or partial improvements in survival—a point corroborated by the 25.0% (2/8) survival rate of the single-module vaccine (sfNnps) Fig. 5 | Immunogenicity, serum-mediated ADE assessment, and protective efficacy of tfN/HInps in cats. aCat immunization, challenge, and sampling scheme. Following intraperitoneal FIPV challenge, when all cats in either the challenge control group (FIPV + normal saline) or the inactivated vaccine group had succumbed, peripheral blood was collected from all groups for viral load analysis. awas partially created in BioRender. Feiyu Zhao. (2026)https://biorender.com/ 320us6k. The nanoparticle illustrations are original and were created by the authors using Autodesk 3ds Max 2020.bSerum FIPV N-specific IgG levels were quantified by ELISA after the third immunization.W(4, 16.71) = 15.07,p<0.0001, ω 2 = 0.722. tfN/HInps vs sfNnps, mean difference = 0.6236, 95% CI [0.2605-0.9868], p=0.0010.cFIPV N-specific IFN-γspot-forming units in cats after the third immunization were quantified by ELISpot to assess tfN/HInps-induced Th1- associated cytokine responses.W(4, 16.95) = 368.1,p<0.0001,ω 2 = 0.985; tfN/ HInps vs sfNnps, dfN/Hnps, and dfN/Inps, mean differences = 73.88, 33.63, and 14.13, with 95% CIs [64.97–82.78], [23.30–43.95], and [7.676–20.57], respectively; p< 0.0001, <0.0001, and = 0.0001, respectively.dFrequencies of TNF-α⁺cells within CD4⁺and CD8⁺T-cell populations among FIPV N peptide-stimulated peripheral blood lymphocytes in cats after the third immunization were quantified by multicolorflow cytometry to evaluate tfN/HInps-induced N-specific cellular immune responses. For TNF-α⁺CD4⁺Tcells,W(4, 16.85) = 995.8,p<0.0001, ω 2 = 0.995; tfN/HInps vs sfNnps, dfN/Hnps, and dfN/Inps, mean differences = 4.290, 3.561, and 2.591, with 95% CIs [3.976-4.604], [3.276–3.847], and [2.291–2.891], respectively; allp< 0.0001. For TNF-α⁺CD8⁺Tcells,W(4, 16.86) = 1044,p<0.0001, ω 2 = 0.995; tfN/HInps vs sfNnps, dfN/Hnps, and dfN/Inps, mean differences = 2.196, 1.269, and 0.9125, with 95% CIs [2.051-2.342], [1.131-1.406], and [0.7453-1.080], respectively; allp<0.0001.e–gIn vitro assessment of serum-mediated ADE using feline PBMCs exposed to FIPV pre-opsonized with post–third immunization sera. Viral attachment (e), viral replication (f), and pro-inflammatory cytokine responses (g)werequantified. Fore,F(9, 64) = 2673,p<0.0001,η 2 =0.9973;tfN/HInps+FIPV vs FIPV, inactivated vaccine + FIPV, and positive sera + FIPV, mean differences = −0.3387,−40.59, and−33.06, with 95% CIs [−1.582-0.9043], [−41.83 to−39.35], and [−34.31 to−31.82], respectively;p= 0.9824, <0.0001, and <0.0001, respectively. Treatment × serum-dilution (f;F(63, 512) = 17.19,p<0.0001,η p 2 = 0.6790) and treatment × cytokine (g;F(27, 256) = 8.914,p<0.0001,η p 2 = 0.4846) interactions were significant. Forg, tfN/HInps + FIPV vs inactivated vaccine + FIPV showed mean differences =−0.5387,−0.5200,−0.5163, and−0.4938 for TNF-α, IL-1β, IL-6, and IL- 10, with 95% CIs [−0.5983 to−0.4792], [−0.5795 to−0.4605], [−0.5758 to−0.4567], and [−0.5533 to−0.4342], respectively; allp< 0.0001. Corresponding comparisons with positive sera + FIPV showed mean differences =−0.3562,−0.3775,−0.1225, and −0.3150, with 95% CIs [−0.4158 to−0.2967], [−0.4370 to−0.3180], [−0.1820 to −0.06295], and [−0.3745 to−0.2555], respectively; allp<0.0001.h,iProtective efficacy following FIPV challenge is shown as clinical scores (h) and survival curves (i).jPeripheral-blood viral RNA loads were quantified by RT–qPCR at two analysis endpoints corresponding to the time by which all cats in the inactivated vaccine group (12 d) and normal saline group (19 d) had reached the humane endpoint. For cats euthanized earlier, the last blood sample collected immediately before euthanasia was used. Viral RNA reduction rate (%) = [(V control -V immunized )/V control ]× 100%. At 12 d, Kruskal-WallisH(7) = 53.61,p<0.0001,ε 2 = 0.9322, with tfN/HInps vs normal saline,p= 0.0152; at 19 d,H(6) = 45.04,p<0.0001,ε 2 = 0.9079, tfN/HInps vs normal saline,p= 0.0015. Data are presented as mean ± SD.b–dwere analyzed by Welch’s ANOVA followed by two-sided Dunnett’s T3 multiple-comparisons test, with tfN/HInps as the reference group.ewas analyzed by one-way ANOVA, and f,gby two-way ANOVA, followed by two-sided Dunnett’s multiple-comparisons test where applicable. Survival curves iniwere analyzed using the Kaplan-Meier method and log-rank (Mantel-Cox) test.jwas analyzed separately at each time point using the Kruskal-Wallis test followed by two-sided Dunn’s multiple-comparisons test. Pairwise comparisons are reported as mean differences or mean-rank differences, with 95% CIs and multiplicity-adjustedp-values where applicable. Effect sizes are reported asω 2 for Welch’s ANOVA,η 2 for one-way ANOVA, partialη²(η p 2 )fortwoway ANOVA, andε 2 for Kruskal-Wallis tests. Forb–d, statistical analyses included the sfNnps, dfN/Hnps, dfN/Inps, tfN/HInps, and inactivated vaccine groups (n=8 biologically independent cats per group). For (e–g,j) at 12d,n= 5 biologically independent cats in the normal saline and Freund’s adjuvant groups andn=8 biologically independent cats in each vaccine group; at 19 d in (j), the same sample sizes were used except that the inactivated vaccine group was not included. NS, not significant; *p<0.05;**p< 0.01; ***p<0.001;****p< 0.0001. Source data are provided as a Source Datafile. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

in this study. Accordingly, a central unresolved challenge in vaccine development is how to endow non-surface-antigen-based vaccines with high-level protective efficacy while maintaining a low ADE liability. To address this, we combined HCV E2 and IFNγon the sfNnps scaffold to generate the tri-modular formulation tfN/HInps, which achieved 100% (8/8) survival in the tested FIPV challenge model. Our study supports a complementary design principle: maintain a conserved internal antigen as the immunological core while incorporating two distinct enhancement modules. The HCV E2-containing formulation was selected based on its association with enhanced dendritic-cell uptake, whereas the IFNγ-containing formulation showed stronger readouts associated with dendritic-cell activation and T-cell responses. Their integration into tfN/HInps was associated with enhanced Th1- related cellular immune responses and high-level protection in the tested FIPV challenge model. Improving the protective efficacy of N-protein–based vaccines likely requires the integration of complementary immuneenhancement strategies rather than reliance on a single functional module. Prior studies have shown that targeting APC receptors such as CLEC9A or XCR1 can facilitate antigen delivery to cDC1. For example, XCL1–antigenfusionconstructsselectivelytargetXCR1⁺DCs and have achieved near-complete protection in lethal influenza challenge models in mice. However, thesefindings are largely limited to mouse models and a few pathogens, and these approaches have not been systematically validated for achieving 100% survival in natural-host settings with lethal coronavirus infection 58 . In parallel, cytokine codelivery—such as IL-2 or IFN-γadministered with vaccination—also increases vaccine-induced T-cell abundance and function in infectious bursal disease virus (IBDV) and hepatitis B virus (HBV) models. Yet, protection is often only partial 59,60 . In this study, screening of helper factors revealed distinct immunological profiles associated with the tested formulations. The HCV E2-containing formulation showed enhanced dendritic-cell uptake, whereas the IFNγ-containing formulation exhibited stronger dendritic-cell activation- and T-cellresponse-associated readouts. Consistent with these roles, dualmodule vaccines dfN/Hnps and dfN/Inps improved survival to 50.0% (4/8) and 62.5% (5/8), respectively, without detectable serummediated ADE under the tested conditions; however, neither formulation achieved complete survival in the tested challenge model. Thesefindings support the notion that a single enhancement axis is insufficient for maximal efficacy. In contrast, combining HCV E2 and IFNγin tfN/HInps further increased survival to 100% (8/8). The combined formulation was also associated with stronger features related to antigen uptake and cross-presentation and enhanced Th1-associated cellular immune responses. Collectively, thesefindings support the potential value of integrating functionally complementary modules to enhance cellular immunity and protective efficacy in the lethal FIPV challenge model. The cDC1 subset is a central driver of antiviral CD8⁺T-cell immunity. Accumulating evidence indicates that XCR1⁺/CLEC9A⁺cDC1 are uniquely specialized for antigen cross-presentation and play a critical role in priming effective antiviral and antitumor CD8⁺T-cell responses 61,62 .Batf3-deficient mice, which have a marked reduction in cDC1, exhibit significantly impaired virus- or tumor-specificCD8⁺ T-cell responses and clearance in models of West Nile virus infection and highly immunogenic tumors, further underscoring the importance of cDC1 63 . Furthermore, studies have shown that targeted antigen delivery to cDC1, whether in influenza-vaccine mouse models or in human CD141⁺cDC1, significantly enhances cross-presentation efficiency and antigen-specificCD8⁺T-cell expansion, thereby affirming its status as a key cellular target for vaccines 58,62 .Inthisstudy,singlenucleus transcriptomic pseudotime analysis revealed that tfN/HInps immunization was associated with increased cDC1 representation in the spleen at late pseudotime, accompanied by coordinated Table 2 | Comparative evaluation of feline dendritic-cell responses, vaccine-induced immune responses, protective efficacy, and serum-mediated ADE across experimental groups FormulationCompositionAntigen uptake MHC I expression CD80 expression N-specific IgG IFN-γSFUs TNF-α⁺ CD4⁺ Tcells TNF-α⁺ CD8⁺ T cells ADE evidence Survival Normal saline////○○○○ ○○○○ ○○○○ ○○○○ ○○○○ ○○○○ ○○○○ ○○○○ Not detected 0.0% (0/5) Freund’s adjuvant////○○○○ ○○○○ ○○○○ ○○○○ ○○○○ ○○○○ ○○○○ ○○○○ Not detected 0.0% (0/5) Single-module formulation (sfNnps) Nantigen●○○○ ○○○○ ●○○○ ○○○○ ●○○○ ○○○○ ●○○○ ○○○○ ●○○○ ○○○○ ●○○○ ○○○○ ●○○○ ○○○○ Not detected 25.0% (2/8) Dual-module APC-targeting formulation (dfN/Hnps) Nantigen, HCV E2 ●●●○ ○○○○ //●●◐○ ○○○○ ●●●○ ○○○○ ●●●● ○○○○ ●●●● ●●●● Not detected 50.0% (4/8) Dual-module immunostimulatory formulation (dfN/Inps) Nantigen, IFNγ /●●●● ○○○○ ●●●○ ○○○○ ●●○○ ○○○○ ●●●◐ ○○○○ ●●●● ●●●◐ ●●●● ●●●● Not detected 62.5% (5/8) Tri-modular formulation (tfN/HInps) Nantigen, HCV E2, IFNγ // /●●◐○ ○○○○ ●●●● ◐○○○ ●●●● ●●●● ●●●● ●●●● Not detected 100.0% (8/8) Inactivated vaccine////●●◐○ ○○○○ ○○○○ ○○○○ ◐○○○ ○○○○ ◐○○○ ○○○○ Detected0.0% (0/8) Relative performance of the different vaccine formulations across the quantitative parameters was expressed using an sfNnps-referenced relativescoring system. For parameters with corresponding experimental measurements, the measured value of the sfNnps group for each parameter was used as the reference, and the relative score was calculatedaccording to the formula Y = 20% + [(X−A)/ A] × 100%, where X represents the measured value of the formulation being evaluated and A represents the corresponding measured value of the sfNnps group. Accordingly, sfNnps was assigned a baseline score of 20% for each parameter evaluated. Scores were constrained to a range of 0-160%; values below 0% were set to 0%, whereas values above 160% were set to 160%. The resulting scores were further converted into a multilevel circle-based scale for intuitive visualization as follows: Y≤0,○○○○○○○○;0<Y≤10%,◐○○○○○○○;10%<Y≤20%,●○○○○○○○;20%<Y≤30%, ●◐○○○○○○; and 30% <Y≤40%,●●○○○○○○. Thereafter, each additional 10-percentage-point increase in Y was represented by an additional half-filled circle (◐). A baseline score of 20% was assigned to sfNnps to allow decreases relative to sfNnps to be visually represented within the scoring system. This scoring system was used solely to facilitate visual comparison of the relative performance and overall trends among formulations across different parameters and does not indicate statistical significance between groups. ADE outcomes were reported directly according to whether serum-mediated enhancement of infection was detected in the corresponding assays and were not included in the relative scoring system. Survival was expressed as the proportion of animals surviving to the end of the observation period, with the number of surviving animals/total number of animals shown in parentheses, and was likewise not included in the relative score calculation. The original measured values and corresponding statistical analyses for each parameter are provided in the relevantfigures.“/”indicates that the corresponding parameter was not evaluated for that formulation. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

upregulation of genes involved in cross-presentation and MHC class I antigen processing. Genes associated with vesicular transport, costimulation, and immunological synapse formation were also highly expressed in cDC1. Together, these cDC1-associated transcriptional features were consistent with the enhanced CD8⁺T-cell responses observed following tfN/HInps immunization and suggested that cDC1- associated changes might contribute to more efficient initiation of downstream CD8 + T-cell responses. Achieving high-level protection may depend not only on efficient priming of T-cell responses, but also on the establishment of highquality memory T-cell compartments. CD8⁺effector memory T cells (CD8 Tem) rapidly execute effector functions upon antigen reencounter. They are widely recognized as key contributors to durable protective immunity 64,65 . For instance, mucosal vaccination strategies that induce SARS-CoV-2-specific memory responses enriched in tissue-associated CD8 Tem within the respiratory tract have been associated with reduced lung viral loads and attenuated pathology 66,67 . In this study, tfN/HInps immunization was associated with a higher relative representation of CD8 Tem cells and with increased proliferative potential and enhanced memory-associated features within All groups 01020 Mast cells Granulocytes Tcells Plasma cells Bcells Endothelial cells Fibroblasts Monocytes-DC 1500300040008000 Cell type count a Control Inactivated vaccine tfN/HInps 10075 50 25 Inactivated vaccine Control tfN/HInps 0 Relative proportions of cell type (%) 6 4 5 1 1 7 9 3 2 8 10 Cell Type (All cell) Monocytes-DC Fibroblasts Endothelial cells B cells Plasma cells T cells Granulocytes Mast cells Erythrocyte Unknown 1 2 3 4 5 6 7 8 9 10 UMAP -2 UMAP-1 Control UMAP- 2 UMAP- 1 Inactivated vaccine UMAP- 2 UMAP- 1 tfN/HInps UMAP- 2 UMAP- 1 All groups 1 5 3 4 6 3 2 04080 pDC Monocyte moDC cDC2 cDC1 100200300 Cell type count Cell Type (Monocytes-DC) cDC1 cDC2 pDC Monocyte moDC Unknown 1 2 3 4 5 6 UMAP -2 UMAP-1 Control UMAP- 2 UMAP- 1 Inactivated vaccine UMAP- 2 UMAP- 1 tfN/HInps UMAP- 2 UMAP- 1 10075 50 25 Inactivated vaccine Control tfN/HInps 0 Relative proportions of cell type (%) Control Inactivated vaccine tfN/HInps b All groups 04080 Th1 CD4 Tcm CD8 Teff CD8 Tem CD8 Tcm CD8 Tscm CD8 Tn 10002000 Cell type count 7 1 2 6 3 8 4 5 9 Cell Type (T cell) UMAP- 2 UMAP-1 Control UMAP- 2 UMAP- 1 Inactivated vaccine UMAP- 2 UMAP- 1 tfN/HInps UMAP- 2 UMAP- 1 Control Inactivated vaccine tfN/HInps CD8 Tn CD8 Tscm CD8 Tcm CD8 Tem CD8 Teff CD4 Tcm Th1 Tfh Unknown 1 2 3 4 5 6 7 8 9 10075 50 25 Inactivated vaccine Control tfN/HInps 0 Relative proportions of cell type (%) c Fig. 6 | Single-nucleus RNA sequencing defines splenic immune-cell landscapes following tfN/HInps immunization in cats. aUMAP visualization of snRNA-seq profiles from feline spleens showing major immune and stromal cell populations (top). Group-wise comparisons of cell-type frequencies and absolute cell numbers (bottom) were used to assess global remodeling of the splenic cellular composition following tfN/HInps immunization.bUMAP of the monocytes-DC compartment after subsetting (top), with quantification of subset frequencies and absolute cell numbers across experimental groups (bottom) to evaluate tfN/HInps-associated changes in the monocyte/dendritic-cell lineage.c. UMAP of the T-cell compartment after subsetting (top), with group-wise quantification of T-cell subset frequencies and absolute cell numbers (bottom) to assess how tfN/HInps reshapes splenic T-cell composition and responses. Data in (a–c) were obtained fromn= 2 biologically independent cats per group. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

late-stage CD8⁺T cells. Transcriptional regulatory network inference identified higher inferred FOXO1 regulon activity and enrichment of FOXO1-linked memory-associated programs in CD8 Tem following tfN/HInps immunization. FOXO1 is known as a key regulator supporting memory T-cell self-renewal, functional persistence, and maintenance of a less exhausted state. Prior infection and vaccination studies have demonstrated that FOXO1 is required for sustaining the long-term function of memory-like CD8⁺T cells. Preservation or augmentation of FOXO1 activity can also constrain differentiation toward terminal effector or exhausted phenotypes in chronic stimulation settings 68–70 . Recent work on CAR T cells also suggests that preserving FOXO1-related memory programs enhances T-cell persistence and resistance to exhaustion under chronic stimulation 70,71 . Therefore, tfN/ CD4 T Control State1 (Early) State2 (Middle) State3 (Late) a e n i c c a v d e t a v i t c a n I tfN/HInps cDC 1 cDC2 pDC 50 100 150 Major cell type count cD C 1 cDC 2 p D C 50 100 150 Major c ell t ype cou n t cD C 1 cDC 2 p D C 50 150 250 Major cell type count State1 State2 State3 State1 (Early) State2 (Middle) State3 (Late) State1 (Early) State2 (Middle) State3 (Late) State1 State2 State3 State1 State2 State3 Pattern-recognition receptors Family tfN/HInps d e t a v i t c a n I vaccine Complement-associated receptors Scavenger receptorsIgG Fc receptors C-type lectin receptors 1 -1 0 b d 0.4 0.3 0.2 0.1 cDC1-Sender CD86 CD80 ICAM1 CXCL10 CD48 CD47 CCL21 ICAM2 ICAM3 IFNG VCAM1 ICOSLG BTLA IL12B TNFSF9 CCL19 FASLG IL12A IL15 IL15RA IL18 CD86 CD80 ICAM1 CXCL10 CD48 CD47 CCL21 ICAM2 ICAM3 IFNG VCAM1 ICOSLG BTLA IL12B TNFSF9 CCL19 FASLG IL12A IL15 IL15RA IL18 CD4 T-Receiver CD28 ITGAL IL12RB2 CD244 IL2RG CCR7 IL18R1 ITGA4 TNFRSF14 CXCR3 TNFRSF9 SIRPA IL2RB IFNGR1 ICOS CTLA4 CD28 ITGAL IL12RB2 CD244 IL2RG CCR7 IL18R1 ITGA4 TNFRSF14 CXCR3 TNFRSF9 SIRPA IL2RB IFNGR1 ICOS CTLA4 tfN/HInpsInactivated vaccine 0.0 0.1 0.2 0.0 Sum Lr_score per ligand 0.4 0.8 1.6 1.2 Mean Lr_score CD8 T 0.1 0.2 0.0 Sum Lr_score per ligand 0.4 0.8 1.6 1.2 Mean Lr_score 0.4 0.3 0.2 0.1 cDC1-Sender CD86 CD80 ICAM1 CXCL10 CD48 CD47 CCL21 ICAM2 ICAM3 IFNG VCAM1 ICOSLG BTLA IL12B TNFSF9 CCL19 FASLG IL12A IL15 IL15RA IL18 CD86 CD80 ICAM1 CXCL10 CD48 CD47 CCL21 ICAM2 ICAM3 IFNG VCAM1 ICOSLG BTLA IL12B TNFSF9 CCL19 FASLG IL12A IL15 IL15RA IL18 CD8 T-Receiver CD28 ITGAL IL12RB2 CD244 IL2RG CCR7 IL18R1 ITGA4 TNFRSF14 CXCR3 TNFRSF9 SIRPA IL2RB IFNGR1 ICOS CTLA4 CD8 T-Receiver CD28 ITGAL IL12RB2 CD244 IL2RG CCR7 IL18R1 ITGA4 TNFRSF14 CXCR3 TNFRSF9 SIRPA IL2RB IFNGR1 ICOS CTLA4 Inactivated vaccine 0.0 Inactivated vaccinetfN/HInpsInactivated vaccinetfN/HInps e CD4 T-Receiver cDC1cDC1 CD8 TCD4 T tfN/HInps cDC1 pDC cDC2 cDC1 pDC cDC2 cDC1 pDC cDC2 SOCS1 STAT1 STAT3 SOCS3 NLRC5 CIITA RFXANK RFXAP RFX5 ATP6V1A CTSS CTSB CTSD IFI30 RAB11A RAB43 RAB14 RAB27A SEC22B SNAP23 STX7 STX11 STX4 VAMP3 VAMP7 WDFY4 PSMB9 PSMB10 PSMB8 PSMB1 PSMB2 PSMB3 LNPEP ERAP1 SEC61A1 TAP1 TAP2 CD70 CD40 IL10RA IL12B IL15 1 0-0.6 tfN/HInps Inactivated vaccine JAK–STAT–SOCS regulatory axis Transcriptional control of MHC I Transcriptional control of MHC II Endo-lysosomal antigen processing (MHC II) Vesicle-mediated antigen crosspresentation Proteasomedependent crosspresentation Peptide transport/ editing for MHC I Co-stimulatory and cytokine signaling c Expression (Log 1p-CRM ) 2 3 4 5 6 7 JAK–STAT–SOCS signaling axis Transcriptional control of MHC I Module score Module score Transcriptional control of MHC II Endo-lysosomal antigen processing (MHC II) Module score Module score Vesicle-mediated antigen crosspresentation Proteasomedependent crosspresentation Module scoreModule score Peptide transport/ editing for MHC I Co-stimulatory and cytokine signaling Module score Module score Inactivated vaccinetfN/HInps 0 2 4 0 2 4 2 3 4 5 6 7 2 4 6 8 10 2 3 4 3 6 9 2 3 4 5 6 Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

HInps immunization was associated with a larger CD8 Tem compartment and a FOXO1-centered transcriptional program consistent with enhanced memory potential, proliferative capacity, and effector function. These immune characteristics may have contributed to the reduced viral burden and 100.0% (8/8) survival observed in the tested lethal FIPV challenge model. However, because functional perturbation experiments, such as cDC1 or CD8⁺T-cell depletion, were not performed, the causal roles of these cell populations in tfN/HInpsmediated protection remain to be established. In young animals, ongoing immune maturation and agedependent differences in dendritic-cell antigen presentation, costimulatory capacity, and T-cell priming may constrain the magnitude and durability of vaccine-induced protective immunity 72–74 .In particular, in kittens aged 1–3 months, dendritic-cell maturation, antigen-presentation capacity, T-cell activation, and memory formation remain under development, which may partly account for the limited protective efficacy of vaccination in this age group 75–77 .Inthis study, cats aged 3–5 months, an age group with a relatively high incidence of FIP, were selected as experimental animals 78 . The immunization regimen was established with reference to the vaccination intervals recommended in the guidelines of the World Small Animal Veterinary Association (WSAVA), and the immunogenicity, protective efficacy, and ADE risk of tfN/HInps were systematically evaluated 77 .The results showed that tfN/HInps immunization was associated with a higher relative representation of splenic cDC1, transcriptional features related to antigen uptake and presentation, together with enhanced transcriptional and functional features associated with CD8 + effectormemoryTcells.GiventhecriticalroleofcDC1inantigencrosspresentation and the initial priming of CD8⁺T cells, together with the capacity of CD8⁺effector-memory T cells to mount rapid effector responses upon antigen re-encounter, these immune features are relevant to key immunological limitations that may restrict vaccine protection in young kittens. Thus, the enhanced features related to antigen presentation and cellular immunity elicited by tfN/HInps suggest that this formulation may help improve the relatively limited cellular immune activation capacity of young animals. However, the immunogenicity and protective efficacy of tfN/HInps have not been directly evaluated in kittens aged 1–3 months. Therefore, this inference should be interpreted cautiously and cannot substitute for direct assessment in young kittens. In addition, the developmental stage of the immune system, maternally derived antibody levels, and natural exposure conditions may all influence vaccine-induced immune responses in this population. Future systematic studies in kittens of different ages will therefore be required to further evaluate the safety, immunogenicity, and protective potential of tfN/HInps in young animals. The outcome of this study may be related, at least in part, to the rational integration of functionally complementary immuneenhancement modules rather than the simple accumulation of multiple factors. Previous studies, together with thefindings of the present study, suggest that combining molecules acting through similar pathways does not necessarily produce additive immune enhancement and may even result in diminished responses because of signaling redundancy or antagonism 73,79 . For example, in aListeria monocytogenessubunit vaccine model, Lee et al. found that adding the TLR9 agonist CpG or the TLR4 agonist GLA to the nanoemulsion adjuvant ADJ did not further enhance antigen uptake or initial T-cell activation; instead, these combinations reduced memory CD8⁺T-cell numbers and protective efficacy 80 . Consistent with these observations, among the dual-module screening formulations evaluated in this study, dfN/Hnps showed the strongest profile related to APC uptake, whereas dfN/Inps showed the strongest immunostimulatory profile. By contrast, combining multiple factors with overlapping functions did not produce a progressive or additive improvement in immune responses and, in some formulations, was associated with reduced dendritic-cell uptake or maturation and weaker T-cell-related readouts (Supplementary Fig. 5, 6). Thesefindings suggest that selecting modules with distinct and potentially complementary functions may be more important than simply increasing the number of incorporated factors. In the present system, the HCV E2-containing formulation was associated with enhanced antigen-uptake-related readouts, whereas the IFNγ-containing formulation showed stronger features associated with APC activation and T-cell responses. Their combination in tfN/ HInps was associated with enhancement of a broader range of immune readouts and with greater protective efficacy than either dual-module formulation alone. However, these immunological differences cannot be attributed solely to the identities of the incorporated functional modules. Nanoparticle size can independently influence lymphatic drainage, APC uptake, intracellular trafficking, and subsequent immune activation 81,82 . The formulations evaluated in this study differed in hydrodynamic size, and tfN/HInps exhibited a bimodal DLS distribution, suggesting the presence of particle populations with different sizes or assembly states. Although the overall response patterns were broadly consistent with the expected functions of the incorporated modules, strictly size-matched controls were not included. The distinct particle populations within tfN/HInps were also not separated and evaluated independently. Therefore, the respective contributions of particle size, assembly state, module composition, and surface loading cannot currently be resolved. Accordingly, the immune enhancement and protection associated with tfN/HInps should be interpreted as the integrated effect of the complete nanoparticle formulation rather than as the isolated action of HCV E2 or IFNγ. Ferritin is an attractive scaffold for vaccine delivery and antigen display because of its structural stability, self-assembly capacity, and amenability to molecular engineering. Its naturally self-assembled 24- mer nanocage provides a highly ordered scaffold that supports multivalent and repetitive presentation of antigens on the particle surface 50,51,83 . This architecture may also facilitate antigen retention, lymphatic trafficking, and accumulation in secondary lymphoid tissues, thereby supporting enhanced immune recognition 51,84,85 .The ferritin nanoparticle platform offers a practical advantage in this setting: vaccine antigens can be incorporated through genetic fusion or Fig. 7 | tfN/HInps immunization is associated with cDC1 enrichment and enhanced antigen-presentation-related programs in late-stage cDC1. aPseudotime trajectory reconstruction of dendritic-cell subsets (cDC1, cDC2, and pDC) showing developmental/activation state of different dendritic cell subsets after tfN/HInps immunization. Cells are colored by type, and the direction of the inferred pseudotime is indicated by the arrow.bHeatmap of genes encoding antigen recognition and uptake-associated receptors in State 3 (late pseudotime) cDC1, used to assess remodeling of antigen sensing and internalization capacity after tfN/HInps immunization.cHeatmap showing the expression of genes associated with antigen processing and presentation pathways in State 3 cDC1 (left), together with box plots of the corresponding pathway module scores (right). For the box plots, boxes represent the 25–75th percentiles, centre lines indicate the median, and whiskers extend to the most extreme values within 1.5 × the interquartile range. Individual points represent cells.dCell-cell communication analysis between State 3 cDC1 and CD4⁺T cells shown as ligand-receptor interaction chord plots (top) and corresponding ligand/receptor gene-expression heatmaps (bottom), to analyze intercellular signaling and helper T cell-activating potential between cDC1 cells and CD4⁺T cells after tfN/HInps immunization.eCell-cell communication analysis between State 3 cDC1 and CD8⁺T cells shown as ligand–receptor interaction chord plots (top) and corresponding ligand/receptor gene-expression heatmaps (bottom), highlighting reinforcement of the cDC1–CD8⁺ T-cell interaction network after tfN/HInps immunization. Data in (a–e)were obtained fromn= 2 biologically independent cats per group. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

modular conjugation, facilitating iterative design and optimization 50,51 . In addition, ferritin nanoparticles are compatible with scalable expression and purification systems, including prokaryotic platforms for suitable constructs, which may simplify manufacturing and facilitate the timely updating of candidate antigens in response to emerging viral variants. For example, Kanekiyo et al. fused influenza hemagglutinin to ferritin to generate hemagglutinin-displaying nanoparticles, resulting in substantially higher hemagglutination-inhibition titers than those elicited by a licensed inactivated vaccine 83 .Yangetal. used SpyTag/SpyCatcher-mediated covalent coupling to display distinct receptor-binding domains of PEDV S1 on ferritin nanoparticles, thereby enhancing protective immunity 52 . Taken together, these FOXO1 TBX21 STAT3 IRF1 NFKB1 RUNX2 NFATC1 900 600 300 0 CD8 TnCD8 TscmCD8 Tcm CD8 TemCD8 Teff Late Middle Early Number of cells b Early 4 3 2 1 0 4 3 2 1 0 4 3 2 1 0 MiddleLate ) n o i t a r e f i l o r P (

e r o S Control CD8 TscmCD8 TcmCD8 TemCD8 Teff ProliferationMemoryCytotoxic c d Up-regulated DEGs ͧtfN/HInps vs. Controlͨ Down-regulated DEGs ͧtfN/HInps vs. Controlͨ Regulated TFs in CD8 Tem f Expression pattern of FOXO1 regulated geneset of CD8 Tem Up-regulated DEGs ͧInactivated vaccine vs. Controlͨ Down-regulated DEGs ͧInactivated vaccine vs. Controlͨ Regulated TFs in CD8 Tem Control Inactivated vaccine tfN/HInps 1.5 0.0 -1.5 e 1 0 -1 FOXO1 TBX21 STAT3 IRF1 NFKB1 RUNX2 NFATC1 Control Inactivated vaccine tfN/HInps Inactivated vaccine CD8 TscmCD8 TcmCD8 TemCD8 Teff ProliferationMemoryCytotoxic tfN/HInps CD8 TscmCD8 TcmCD8 TemCD8 Teff ProliferationMemoryCytotoxic The relative regulon activity score of core regulatory TFs Regulon activity score a CD8 Teff CD8 Tscm CD8 Tem CD8 Tn CD8 Tcm Control Pseudotime 20 0 Component 1 2

t n e n o p m o C Early Middle Late CD8 Teff CD8 Tscm CD8 Tem CD8 Tn CD8 Tcm Inactivated vaccine Pseudotime 20 0 Component 1 Component 2 Early Middle Late CD8 Teff CD8 Tscm CD8 Tem CD8 Tn CD8 Tcm tfN/HInps Pseudotime 20 0 Component 1 Component 2 Early Middle Late ) y r o m e M (

e r o S ) c i x o t o t y C (

e r o S Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

studies illustrate that ferritin nanoparticles permit theflexible integration of antigenic and immunomodulatory components and support the replacement or combination of antigens for platform-based vaccine development. Using this platform, we incorporated HCV E2 and IFNγinto an FIPV N-based nanoparticle to generate the tri-modular vaccine tfN/HInps. In a type II FIPV challenge model, tfN/HInps conferred 100% (8/8) survival and showed no detectable serum-mediated ADE under the tested conditions, supporting the potential utility of this platform for the development and optimization of multicomponent FIPV nanoparticle vaccines. However, because protective efficacy was evaluated against only a single homologous type II FIPV strain, the breadth and generalizability of protection against heterologous strains remain to be determined. Although the N protein is generally conserved across FCoV serotypes, biotypes, and strains, with amino acid sequence identities of approximately 90%-99.5% (Supplementary Fig. 10), sequence differences remain and may limit the ability of an N protein derived from a single strain to provide sufficient crossprotection against other serotypes or heterologous strains. The modular nature of the platform may help address this limitation by permitting the N antigen to be updated according to circulating sequences or by incorporating N proteins from multiple representative strains to generate strain-matched or multivalent candidates. Nevertheless, the present study evaluated protective efficacy against only a single homologous type II FIPV strain. Future studies using heterologous clinical isolates or genetically distinctfield strains will therefore be required to determine the breadth of protection and generalizability of tfN/HInps. Similarly, an FECV-derived N protein could be incorporated into the same platform to explore multicomponent nanoparticle vaccines targeting enteric FCoV infection. Such candidates could be evaluated in persistent-infection models for their ability to reduce viral burden and shedding and to determine whether limiting FECV persistence or replication influences the subsequent risk of FIP development. Vaccine-induced antigen-specific functional T-cell responses are an important component of antiviral immune protection 86,87 .Inthis study, compared with soluble recombinant N protein, the singlemodule vaccine sfNnps significantly enhanced FIPV N-specificIFN-γ responses as well as TNF-α⁺CD4⁺and CD8⁺T-cell responses, indicating that the N protein itself is capable of eliciting antigen-specific functional T-cell responses and that its display on the surface of ferritin nanoparticles can further enhance its cellular immunogenicity. Consistent with this immunological enhancement, sfNnps conferred protection in 25.0% (2/8) of cats following FIPV challenge. Comparison among the different vaccine formulations showed that protection increased progressively as FIPV N-specific functional T-cell responses became stronger. Addition of either HCV E2 or IFNγto sfNnps enhanced N-specificIFN-γresponses and TNF-α⁺N-specificCD4⁺and CD8⁺T-cell responses to varying degrees; correspondingly, the protection rate increased from 25.0% (2/8) with sfNnps to 50.0% (4/8) with dfN/Hnps and 62.5% (5/8) with dfN/Inps. Building on this, the trimodular vaccine tfN/HInps, incorporating both HCV E2 and IFNγ, elicited the strongest N-specific functional T-cell responses among all tested formulations and achieved complete protection (100%, 8/8). Comparison of the three N-specific functional T-cell readouts measured in this study showed particularly pronounced increases in TNF- α⁺N-specificCD4⁺and CD8⁺T-cell responses as additional vaccine modules were incorporated. Compared with the dual-module vaccine, tfN/HInps increased these two responses by 85.0% and 56.8%, respectively, suggesting that stronger TNF-α⁺N-specificCD4⁺and CD8⁺T-cell responses may be closely associated with enhanced vaccine protection. By contrast, the increase in the N-specificIFN-γ response was relatively modest across vaccine formulations; nevertheless, its overall magnitude increased sequentially from sfNnps through the dual-module vaccines to tfN/HInps, in parallel with the progressive increase in protection. Spearman correlation analysis further showed a positive association between the N-specificIFN-γ response and vaccine protection (Spearman’sρ= 1.0), supporting a positive relationship between stronger N-specificIFN-γresponses and improved vaccine protective efficacy. Overall, the progressive enhancement of N-specific functional T-cell responses induced by the different N protein nanoparticle vaccines was accompanied by a corresponding increase in protection against FIPV infection, indicating a close association between the magnitude of N-specific functional T-cell immunity and vaccine protective ef ficacy. Several ferritin nanoparticle-based vaccine candidates have entered clinical evaluation in humans, providing important precedents for the further translation of this platform. For example, the H2HA- ferritin, H1ssF, and H10ssF influenza vaccine candidates have completed phase 1 clinical evaluation, demonstrating favorable safety and tolerability while eliciting antigen-specific binding and, to varying degrees, neutralizing antibody responses 88–90 .TheSARS-CoV-2spikeferritin nanoparticle vaccine SpFN/ALFQ also showed favorable tolerability and broad immunogenicity in a phase 1 clinical study 91 .These advances demonstrate the feasibility of advancing ferritin nanoparticle antigen-display platforms from preclinical development to human evaluation and provide platform-level translational support for the further development of tfN/HInps. However, these clinical advances primarily support the translational feasibility of the ferritin platform itself and do not directly resolve the challenges associated with compositional uniformity and process control in multicomponent conjugated nanoparticles. For multicomponent ferritin nanoparticles constructed using the SpyTag/SpyCatcher system, the actual conjugation ratios, surface occupancy, single-particle composition, and batch-to-batch consistency of the functional components remain critical factors affecting process scale-up and translational evaluation. In this study, wefixed the input ratios of ST-N-Fer and the two functional proteins, as well as the reaction and purification conditions, with HCV E2-SC and IFNγ-SC added to the reaction mixture at an equimolar ratio. However, because HCV E2-SC and IFNγ-SC compete for the same type of SpyTag conjugation sites and differ in molecular size, steric hindrance, and reaction accessibility, thefinal formulation may contain particle populations with different levels of surface occupancy and component ratios. Accordingly, the current SDS–PAGE results only demonstrate the simultaneous presence of N-Fer–HCV E2 and N- Fer–IFNγcovalent conjugates in the bulk formulation, whereas TEM supports the retention of particle morphology and DLS characterizes Fig. 8 | tfN/HInps immunization is associated with a FOXO1-related CD8 Tem program linked to proliferation- and memory-associated features. aPseudotime trajectory reconstruction of CD8 + T cells showing differentiation from early states to effector and memory compartments, highlighting tfN/HInpsassociated shifts in late-stage populations.bBox plots showing module score dynamics for proliferation-, memory-, and cytotoxicity-associated gene programs across early, middle, and late CD8⁺T-cell pseudotime states. Boxes represent the 25th-75th percentiles, centre lines indicate the median, and whiskers extend to the most extreme values within 1.5 × the interquartile range. Individual points represent cells.cSubset-level comparison of proliferation, memory, and cytotoxicity module scores at late pseudotime, shown as radar plots to visualize functional bias among terminal-stage CD8⁺T-cell subsets.dInferred transcription factor regulatory network in late-pseudotime CD8 Tem cells, highlighting key transcriptional regulators and their interactions associated with CD8 Tem functional programs.eHeatmap of inferred regulatory activity scores for core transcription factors in late-pseudotime CD8 Tem cells across experimental groups.fHeatmap of FOXO1 downstream target gene expression in late-pseudotime CD8 Tem cells, used to assess tfN/HInpsassociated induction of FOXO1-linked programs related to proliferation, memory maintenance, and effector function. Data in (a–f) were obtained fromn= 2 biologically independent cats per group. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

the corresponding hydrodynamic size distribution. These data do not yet permit quantitative resolution of surface occupancy at the singleparticle level or demonstrate that individual particles contain uniform or defined stoichiometries of the functional components. Building on the identification of HCV E2-SC and IFNγ-SC as functional modules in this study, different strategies could be used to more precisely control their display ratio and spatial distribution on the nanoparticle surface and to identify an appropriate combination range. For example, two mutually orthogonal Tag/Catcher systems could be introduced into the nanoparticle scaffold, allowing HCV E2 and IFNγto be specifically conjugated to predefined attachment sites. Alternatively, genetic fusion and programmable multicomponent self-assembly could be used to assemble different functional subunits on the nanoparticle surface at predefined ratios and spatial positions. Previous studies have shown that the SpyTag/SpyCatcher and SnoopTag/SnoopCatcher systems can be used to conjugate different antigens or functional proteins separately onto the same nanoparticle 92,93 . In addition, I53-50 icosahedral protein nanoparticles, which are assembled from multiple structural components, can support multivalent antigen display with predefined copy numbers and spatial organization while maintaining a high degree of structural consistency with the designed model 94,95 . In summary, we implemented a rational design integrating an antigen (FIPV N protein), a carrier (ferritin), and two functional modules (HCV E2 and IFN-γ). This approach led to the development of the tri-modular nanoparticle vaccine, tfN/HInps. In a lethal FIPV challenge model, tfN/HInps conferred 100.0% (8/8) protection. It showed no evidence of serum-mediated ADE under the tested conditions. Mechanistically, tfN/HInps was associated with coordinated enhancement of cDC1-associated cross-presentation and a FOXO1-linked CD8 Tem memory module. Thesefindings provide a plausible cellular basis for the strong, lasting protection observed. Beyond its translational relevance for FIPV control, ourfindings support a broader vaccine design principle for pathogens with ADE risk: rather than relying predominantly on surface antigens and attempting to minimize ADE- prone antibody responses, conserved internal antigens can be leveraged as the immunological core. Structured nanoparticle carriers and complementary functional modules can then amplify targeted, safe T-cell immunity. This framework may also be adaptable to other ADE- associated pathogens, including dengue virus and SARS-CoV-2 variants. Here, durable cellular immunity could provide an additional layer of protection when antibody-focused strategies are constrained by safety concerns. Methods Ethics All animal experiments were approved by the Institutional Animal Care and Use Committee and the Laboratory Animal Ethics Committee of Heilongjiang Bayi Agricultural University (approval nos. DWKJXY2022004,DWKJXY2023122,DWKJXY2024068, DWKJXY2024069,DWKJXY2024093,DWKJXY2025003,and DWKJXY2026072) and were performed in accordance with relevant guidelines for the care and use of laboratory animals. Female BALB/c inbred mice (Mus musculus; 7 weeks old, 20-25 g) were purchased from Liaoning Changsheng Biotechnology Co., Ltd. (China) and maintained in specific pathogen-free (SPF) barrier facilities with a 1-week acclimation period prior to experimentation. Female mice were selected to reduce aggression-related stress during group housing. The SPF health-monitoring report provided by the supplier included routine screening for murine coronaviruses, and the results were negative. Animals were maintained under controlled conditions (20 °C, 50% relative humidity) on a 12-h light/12-h dark cycle with ad libitum access to standard chow and water. Female domestic cats (Felis catus; 3-5 months old) were used for feline immunization and challenge experiments. The cats were purchased from Jiaxiang Huarong Cat Farm (Jining, China). Only female cats were used to reduce sexrelated variability across experiments. Upon arrival, cats underwent physical examination and body-weight assessment to confirm values within the expected range for age. On the day of arrival, oral and/or rectal swabs were collected and screened by virus-specificPCRorRT- qPCR for feline coronavirus (FCoV), feline immunodeficiency virus (FIV), feline panleukopenia virus (FPV), feline herpesvirus (FHV), feline leukemia virus (FeLV), and SARS-CoV-2; all animals tested negative. Serum samples were also collected on the day of arrival and screened for FCoV N-specific IgG by ELISA. Only cats negative for FCoV RNA and FCoV N-specific IgG, and negative for the other screened feline pathogens, were considered eligible for enrollment. Cats were acclimated for 1 month before study initiation. At the end of the acclimation period and before thefirst immunization, FCoV screening was repeated using RT-qPCR and FCoV N-specific IgG ELISA. Only cats that continued to meet both nucleic-acid and serological inclusion criteria were included in the subsequent immunization and challenge experiments. During experiments, each cat was housed individually in a cage in an animal room maintained at approximately 25 °C, with ad libitum access to a complete commercial feline diet andfiltered drinking water. Husbandry was performed daily by trained personnel. Animal health and welfare were assessed at least twice daily, including general appearance and behavior, appetite, excretion, and cage hygiene. Cats were closely monitored for clinical abnormalities and received veterinary care as needed. All procedures were performed under the supervision of a laboratory animal veterinarian, and all reasonable measures were taken to minimize discomfort and ensure animal welfare throughout the study. Cell lines, reagents, and viruses DC2.4 dendritic cells (SCC142; Merck Millipore, Germany) and Crandell–Rees feline kidney (CRFK) cells (CCL-94; ATCC, USA) were kindly provided by the Division of Swine Digestive System Infectious Diseases, State Key Laboratory of Veterinary Biotechnology, Harbin Veterinary Research Institute, Chinese Academy of Agricultural Sciences. DC2.4 cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) and 100 U/mL penicillinstreptomycin. CRFK cells were cultured in Dulbecco’smodified Eagle’s medium (DMEM) containing 10% FBS and 100 U/mL penicillinstreptomycin. All cells were grown at 37 °C with 5% CO 2 . Cells were not authenticated. Reagents used in this study included: isopropylβ-D-1- thiogalactopyranoside (IPTG; Catalog #I8070, Solarbio, China); FBS (Catalog #A15-101, PAA, Austria); recombinant murine granulocytemacrophage colony-stimulating factor (GM-CSF; Catalog #HY-P7361A, MedChemExpress, USA); recombinant feline GM-CSF (Catalog #KFB- RP1323F-005, Kingfisher Biotech, USA); recombinant feline IL-4 (Catalog #984-FL-025, R&D Systems, USA); RPMI-1640 Medium (Catalog #12633020, Thermo Fisher Scientific, USA); isoflurane (Catalog #R510- 22-10, RWD, China), and DAPI solution withfluorescence quencher (Catalog #P0131-25 mL, Beyotime, China); BD GolgiPlugTM Protein Transport Inhibitor (Catalog #555029, BD Biosciences, USA); and eBioscienceTM Cell Stimulation Cocktail (500X; Catalog #00-4970-93, Thermo Fisher Scientific, USA); Pierce Protein A/G Magnetic Beads (Catalog #88802, Thermo Fisher Scientific, USA). Antibodies were obtained from the following sources (name; catalog no.; dilution): Biosharp (China): HRP-conjugated goat anti-mouse IgG (Catalog #BL001A; 1:10,000); Jackson ImmunoResearch (USA): HRP-conjugated goat anti-cat IgG (Catalog #102-035-003; 1:10,000); BioLegend (USA): APC-Cy7 anti-mouse CD11c, clone N418 (Catalog #117324; 1:100), FITC anti-mouse H-2K d /H-2D d ,clone34-1-2S(MHCClassI;Catalog#114706; 1:100), PE/Cy7 anti-mouse I-A b , clone AF6-120.1 (MHC Class II, Catalog #116420; 1:200), PE anti-mouse CD80, clone 16-10A1 (Catalog #104708; 1:200), APC anti-mouse CD86, clone GL-1 (Catalog #105012; 1:100), anti-mouse CD16/CD32 (clone 93; Catalog #101302; 1:200), FITC anti-mouse CD3, clone 17A2 (Catalog #100203; 1:200), PE antimouse CD4, clone RM4-4 (Catalog #116005; 1:200), APC anti-mouse Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

CD8, clone 53-6.7 (Catalog #100711; 1:200); Boster Biological Technology (China): Cy3-conjugated AffiniPure goat anti-mouse IgG (H + L; Catalog #BA1031; 1:100); Invitrogen (USA): 6x-His Tag Monoclonal Antibody, clone HIS.H8 (Catalog #MA1-21315; 1:1,500), FITC anti-feline CD4, clone 3-4F4 (Catalog #MA5-28776; 1:100), PE anti-feline CD8, clone fCD8 (Catalog #MA5-28793; 1:100); BD Biosciences (USA): BD FastImmuneTM APC Mouse Anti-Human TNF-α, clone 6401.1111 (Catalog #340534; 5μL/test); Servicebio (China): Alexa Fluor 488- conjugated goat anti-mouse IgG (H + L) secondary antibody (Catalog #GB25301; 1:500). Cat-derived polyclonal antibodies against whole FIPV and mouse-derived polyclonal antibodies against the FIPV N protein were generated in-house. The FIPV DQ2023 strain (GenBank accession no. OR711916.1) was propagated in Crandell-Rees feline kidney (CRFK) cells. Cells at 80-90% confluence were inoculated with FIPV and cultured at 37 °C with 5% CO 2 until evident cytopathic effects (CPE) developed. Cells and supernatants were then collected, subjected to freeze–thaw treatment, clarified by centrifugation, aliquoted, andstoredat−80 °C. Viral titers were determined in CRFK cells by a TCID 50 assay using tenfold serial dilutions and calculated by the Reed- Muench method. An overlapping peptide library spanning the fulllength N protein of FIPV strain DQ2023 was synthesized by GenScript (Jiangsu, China). The library consisted of 62 synthetic 15-mer peptides overlapping by nine amino acids, each with a purity of≥90%. Construction of recombinant plasmid The recombinant expression plasmid pET32a-N-Fer was generated by gene synthesis using codon-optimized sequences encoding the FIPV N protein from the reference strain DQ2023 andHelicobacter pyloriferritin (GenBank accession no. WP_000949190; amino acids 5-167), joined by aflexible peptide linker. To enable SpyTag/SpyCatcher conjugation, SpyTag was fused to the N terminus of the N-Fer cassette by fusion PCR using KOD One™PCR Master Mix (KMM-101, TOYOBO), generating pET32a-ST-N-Fer. For expression of soluble N protein, the N coding sequence was amplified from pET32a-N-Fer using gene-specific primers and cloned into theBamH I (FD0054, Thermo Fisher Scientific) andHindIII (FD0505, Thermo Fisher Scientific) sites of the pET32a vector to produce pET32a-N. For helper-factor expression, codon-optimized gene cassettes encoding HIV Gp120 (GenBank accession no. Y14420.1), HCV E2 (GenBank accession no. JN870282.1), CD40L (GenBank accession no. NM_001009298.1),flagellin (GenBank accession no. AY353358.1), IFN-γ(GenBank accession no. X86972.1), and IL-2 (GenBank accession no. NM_001043337.1) were designed and cloned into pET32a to generate pET32a-HIV Gp120-SC, pET32a-HCV E2-SC, pET32a-CD40L-SC, pET32a-Flagellin-SC, pET32a-IFNγ-SC, and pET32a-IL-2-SC, respectively. In each construct, the antigen was fused to SpyCatcher via aflexible linker, and all sequences were optimized for expression inE. coli. All synthesized gene cassettes and recombinant plasmids were produced by GenScript (Nanjing, China). Construct identities were confirmed by Sanger sequencing (Sangon Biotech, Changchun, China). Gene-specific primer sequences are listed in Table S2 (Supplementary Information). Expression and purification of recombinant proteins pET32a-N-Fer, pET32a-N, pET32a-Ferritin (ferritin-only control), pET32a-Flagellin-SC, pET32a-IFNγ-SC, pET32a-IL-2-SC, pET32a-HIV Gp120-SC, pET32a-HCV E2-SC, and pET32a-CD40L-SC were individually transformed intoE. coliBL21 (DE3). Expression was induced at an OD 600 of 0.6-0.8 by IPTG to afinal concentration of 1 mM, followed by incubation at 16 °C with shaking (160 rpm) for 16 h. Cells were harvested and resuspended in lysis buffer (25 mM Tris-HCl, 50 mM NaCl, 10% (v/v) glycerol, 0.1% Tween-20, pH 8.0), and lysed by ultrasonication. Lysates were clarified by centrifugation (12,000 × g, 20 min, 4 °C). Supernatants werefiltered (0.22μm) and applied to gravity-flow columns packed with Ni-NTA agarose resin, followed by incubation at 4 °C for 1 h. Columns were washed with 10 column volumes of wash buffer (25 mM Tris-HCl, 50 mM NaCl, 10% (v/v) glycerol, 0.1% Tween-20, 20 mM imidazole, pH 8.0), and bound proteins were eluted using elution buffer (25 mM Tris-HCl, 50 mM NaCl, 10% (v/v) glycerol, 0.1% Tween-20, 250 mM imidazole, pH 8.0). Eluted fractions were analyzed by 10% SDS–PAGE to verify the target proteins. Protein concentrations were measured using a BCA protein assay kit (P0010, Beyotime). Endotoxin was removed using the ToxinEraseTM Endotoxin Removal Kit (GenScript), and endotoxin levels in all purified protein preparations were <0.1 EU/mL. SDS-PAGE and Western blotting Recombinant protein purity was assessed by SDS-PAGE. For western blotting, proteins separated by SDS-PAGE were transferred onto nitrocellulose membranes (10600004, Sartorius). Membranes were incubated for 2 h at room temperature with cat sera or commercial primary antibodies diluted in PBS containing 0.05% Tween-20 (PBST) to detect the indicated proteins. After washing, membranes were probed with HRP-conjugated goat anti-mouse IgG or HRP-conjugated goat anti-cat IgG, as appropriate, following the manufacturers’ instructions. Immunoreactive bands were visualized using an enhanced chemiluminescence (ECL) detection system. Characterization of nanoparticle vaccine Nanoparticle morphology was examined by transmission electron microscopy (TEM) using an H-7650 instrument (Hitachi, Japan). Protein nanoparticles were adjusted to 0.2 mg/mL, applied to TEM grids, and negatively stained with uranyl acetate prior to imaging. Particle size distribution was measured by dynamic light scattering (DLS) using a Zetasizer Nano Z system (Malvern Instruments, UK). Nanoparticles were diluted in PBS (pH 7.4) to 0.1-0.5 mg/mL before measurement. Generation of mouse BMDCs and feline monocyte-derived dendritic cells BMDCs were generated from bone marrow cells isolated from the femurs and tibias of BALB/c mice as follows. Mice maintained under SPF conditions were euthanized by CO 2 inhalation followed by cervical dislocation. Femurs and tibias were aseptically collected, and bone marrow wasflushed with complete culture medium. After red blood cell lysis, bone marrow cells were cultured in RPMI 1640 supplemented with 10% FBS and recombinant murine GM-CSF (20 ng/mL). Fresh medium containing GM-CSF at the same concentration was added on days 3 and 7. Cells were harvested on day 8 for subsequent experiments. Feline monocyte-derived dendritic cells were generated from peripheral blood mononuclear cells isolated from healthy cats using a feline peripheral blood mononuclear cell isolation kit (P4500, Solarbio). Briefly, isolated PBMCs were resuspended in complete RPMI 1640 medium containing 10% FBS and seeded into culture plates to allow monocyte adherence. After incubation, non-adherent cells were removed, and adherent monocytes were cultured in complete RPMI 1640 medium supplemented with recombinant feline GM-CSF (20 ng/ mL) and recombinant feline IL-4 (20 ng/mL). Fresh cytokine-containing medium was replaced on days 3 and 5. Feline monocyte-derived dendritic cells were harvested on day 7 and used for nanoparticle uptake and activation assays. Nanoparticle uptake by DC2.4 cells or feline monocyte-derived dendritic cells Nanoparticles were labeled withfluorescein isothiocyanate (FITC) and incubated with DC2.4 dendritic cells or feline monocyte-derived dendritic cells for 24 h. Cells were then washed with PBS, and intracellular fluorescence was examined using a Nikonfluorescence microscope. Forflow cytometric quantification, DC2.4 cells or feline monocytederived dendritic cells incubated with FITC-labeled nanoparticles were Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

collected, washed with PBS, and analyzed on a BD LSRFortessa SORP flow cytometer to determine the frequency of FITC-positive cells. In vitro activation of mouse BMDCs and feline monocytederived dendritic cells BMDCs were seeded in 24-well plates at 1.0 × 10 6 cells per well in 1 mL of complete RPMI 1640 and incubated at 37 °C for 24 h. Cells were then stimulated with the indicated nanoparticle formulations for 8 h. Following stimulation, cells were harvested in prechilled PBS and stained withfluorochrome-conjugated antibodies: APC-Cy7–anti-mouse CD11c, FITC–anti-mouse H-2K d /H-2D d (MHC class I), PE-Cy7–antimouse MHC class II, PE–anti-mouse CD80, and APC–anti-mouse CD86. Staining was performed for 30 min at 4 °C in the dark. Cells were washed with PBS and analyzed byflow cytometry. For feline monocyte-derived dendritic-cell activation analysis, day 7 feline monocyte-derived dendritic cells were seeded in 24-well plates at 1.0 × 10 6 cells per well in 1 mL of complete RPMI 1640 and incubated at 37 °C for 24 h. Cells were then stimulated with the indicated nanoparticle formulations for 8 h. After stimulation, total RNA was extracted from the cells and reverse-transcribed into cDNA. The mRNA levels of feline CD80 and MHC I were quantified by RT-qPCR, with GAPDH used as the internal reference gene. Relative gene expression was calculated using the 2 −ΔΔCt method. Primer sequences are provided in Table S3 (Supplementary Information). In vivofluorescence imaging and biodistribution of sfNnps FITC-labeled sfNnps (50μg) or an equimolar amount of soluble FIPV N protein was administered by subcutaneous injection into the right hind limb of BALB/c mice. Prior to injection, thefluorescence intensity of each preparation was measured in vitro to normalize inputfluorescence across groups. Following injection,fluorescence at the injection site was monitored at 12-h intervals. For each imaging time point, mice were anesthetized with isoflurane to minimize movement. Whole-bodyfluorescence signals were acquired using a small-animal in vivo imaging system (AniView600 SE, Boluteng) to track antigen retention in real time. At 48 h post-injection, mice were humanely euthanized in accordance with the Guidelines for the Care and Use of Laboratory Animals issued by the Ministry of Science and Technology of China. Draining lymph nodes from the injected side and spleens were collected for ex vivo biodistribution analysis. Tissuefluorescence intensities were quantified using the manufacturer’s imaging software. Mouse immunization Nanoparticle vaccines were emulsified 1:1 (v/v) with Freund’s adjuvant (complete Freund’s adjuvant for the prime immunization and incomplete Freund’s adjuvant for boosts) and administered intraperitoneally to BALB/c mice on days 0, 14, and 28. Blood samples were collected on days 13, 27, and 41, and serum IgG2a and IFN-γlevels were measured by ELISA. On day 42, mice were humanely euthanized in accordance with the Guidelines for the Care and Use of Laboratory Animals issued by the Ministry of Science and Technology of China, and spleens were collected for T-cell phenotyping. Cat immunization and FIPV challenge Cats were immunized subcutaneously in the neck region on days 0, 21, and 42 with the indicated vaccine formulations emulsified 1:1 (v/v) with Freund’s adjuvant. Complete Freund’s adjuvant was used for the prime immunization, and incomplete Freund’s adjuvant was used for booster immunizations. For the sfNnps/Npro comparison experiment, cats received 100μg sfNnps or an equimolar dose of Npro. For experiments involving the selected nanoparticle formulations, cats received antigen-matched doses of the indicated formulations. The Freund’s adjuvant control group received adjuvant alone, and the normal-saline control group received saline. The inactivated FIPV vaccine was prepared by formaldehyde treatment. FIPV suspensions were treated with formaldehyde at afinal concentration of 0.05% for 24 h at 37 °C. After inactivation, residual formaldehyde was removed, and the treated virus was inoculated onto CRFK cells. Complete inactivation was confirmed by monitoring for cytopathic effects (CPE) and viral replication during cell culture. Injection sites were monitored after each immunization for local reactions, including swelling, induration, ulceration, exudation, and other gross skin lesions. Peripheral blood was collected 20 days after each immunization for serum preparation and isolation of peripheral blood lymphocytes. On day 22 after the third immunization, cats were challenged intraperitoneally with 2 mL of FIPV suspension (10 5.5 TCID 50 /mL). A group of healthy, uninfected cats served as mock controls. Following challenge, clinical signs, survival, body weight, and rectal temperature were monitored daily for 60 days. Disease severity was assessed using a clinical scoring system in which one point was assigned for each observed sign, including lethargy, anorexia, ataxia, fever (> 39 °C), jaundice, and marked weight loss (> 5%) 96 . To minimize suffering, cats exhibiting severe respiratory distress and/or inability to stand were humanely euthanized in accordance with the Guidelines for the Care and Use of Laboratory Animals issued by the Ministry of Science and Technology of China. Peripheral blood samples were collected immediately prior to euthanasia. Measurement of serum antibody and cytokine responses Serum samples from mice or cats collected at the indicated time points were diluted to appropriate working concentrations for analysis. FIPV N-specific IgG levels in mouse and cat sera were measured by indirect ELISA using purified recombinant FIPV N protein as the coating antigen. Briefly, 96-well ELISA plates were coated overnight at 4 °C with recombinant FIPV N protein at 2μg/mL. Plates were washed with PBST and blocked with 5% skim milk for 2 h at 37 °C. After washing with PBST, serially diluted mouse or cat sera were added and incubated at 37 °C for 1 h. Plates were washed again with PBST, and bound antibodies were detected using HRP-conjugated goat anti-mouse IgG for mouseseraorHRP-conjugatedgoatanti-catIgGforcatsera.After incubation with the corresponding secondary antibody, plates were washed with PBST and developed with TMB substrate. The reaction was stopped with stop solution, and absorbance was measured at 450 nm. Mouse serum IgG2a and IFN-γlevels were quantified using commercial ELISA kits specific for mouse IgG2a (MM-43785M1, Meimian, China) and mouse IFN-γ(MM-0182M1, Meimian, China) according to the manufacturers’instructions. FIPV N-specificIFN-γELISpot assay FIPV N-specificIFN-γ-secreting cells were quantified using an ELISpot Plus: Cat IFN-γ(HRP) kit (Mabtech, Catalog #3122-4HPW-2) 97 .Peripheral blood lymphocytes were isolated from immunized cats 20 days after the third immunization using a feline peripheral blood lymphocyte isolation kit (P5740, Solarbio). Pre-coated ELISpot plates were preconditioned with complete RPMI 1640 medium containing 10% FBS and then seeded with 5 × 10 5 cells per well. Cells were stimulated overnight at 37 °C with 5% CO 2 using an FIPV N 15-mer overlapping peptide pool at afinal concentration of 2μg/mL. Medium containing the corresponding peptide solvent was used as the negative control, and eBioscience TM Cell Stimulation Cocktail (500X) was used as the positive control. After incubation, plates were washed with PBS and sequentially incubated with biotinylated detection antibody MT114 (0.5μg/mL) and streptavidin-HRP (1:1000). Spots were developed using ready-to-use TMB substrate, and the reaction was stopped by extensive washing with deionized water. Plates were air-dried, and spots were counted using an ELISpot reader. Results were expressed as spot-forming units (SFU) per 10 6 cells after subtraction of background spots from solventtreated control wells. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

Flow cytometric analysis of lymphocyte subsets and N-specific functional lymphocyte responses Lymphocytes were isolated from immunized mice and cats using a mouse spleen lymphocyte isolation kit (P8860, Solarbio) and a feline peripheral blood lymphocyte isolation kit (P5740, Solarbio), respectively. For surface staining, mouse lymphocytes were pre-incubated with anti-mouse CD16/CD32 antibody and subsequently labeled with FITC-conjugated anti-mouse CD3, PE-conjugated anti-mouse CD4, and APC-conjugated anti-mouse CD8 antibodies. Feline lymphocytes were stained with FITC-conjugated anti-feline CD4 and PE-conjugated antifeline CD8 antibodies. All staining steps were performed on ice for 30 min. Cells were then washed with PBS (2.5 mL), centrifuged at 200 × g for 5 min, and resuspended in PBS (250μL). An equal volume of 3% paraformaldehyde (250μL) was added to achieve afinal paraformaldehyde concentration of 1.5% forfixation. Samples were acquired on an LSRFortessaflow cytometer (BD Biosciences) using FACSDiva (BD Biosciences), and data were analyzed with FlowJo (v10; FlowJo LLC). For N-specific functional T-cell analysis, feline peripheral blood lymphocytes collected 20 d after the third immunization were resuspended in complete RPMI 1640 medium and seeded into 96-well U- bottom plates at 2 × 10 6 cells per well. Cells were stimulated with an FIPV N 15-mer overlapping peptide pool at afinal concentration of 2μg/mL. Medium containing the corresponding amount of peptide solvent was used as the negative control, and eBioscience TM Cell Stimulation Cocktail (500X) was used as the positive control according to the manufacturer’s instructions. After stimulation for 2 h at 37 °C with 5% CO 2 , BD GolgiPlug TM Protein Transport Inhibitor was added according to the manufacturer’s instructions, and cells were further incubated overnight. The next day, cells were harvested and stained with FITC- conjugated anti-feline CD4 and PE-conjugated anti-feline CD8 antibodies. After surface staining, cells werefixedandpermeabilizedusing the BD Cytofix/Cytoperm TM Fixation/Permeabilization Kit (Catalog #554714, BD Biosciences) according to the manufacturer’sinstructions. Intracellular cytokine staining was then performed using BD FastImmune TM APC Mouse Anti-Human TNF-αantibody 97 .Afterwashing, samples were acquired on an LSRFortessaflow cytometer (BD Biosciences) using FACSDiva software, and data were analyzed with FlowJo v10. The frequencies of TNF-α⁺CD4⁺and TNF-α + CD8 + lymphocytes were calculated after subtraction of the corresponding unstimulated background. ADE assay FIPV attachment assay in feline PBMC-derived adherent monocyte/ macrophage-enriched cells. The healthy donor cats used for preparation of these adherent monocyte/macrophage-enriched cells were obtained from the same animal source. Feline peripheral blood mononuclear cells (PBMCs) were isolated from heparinized blood collected from healthy cats using a feline PBMC isolation kit (P4500, Solarbio), followed by red blood cell lysis and cell counting. PBMCs were seeded into culture plates and allowed to adhere overnight. Nonadherent cells were removed, and the adherent monocyte/macrophage-enriched cells were used for subsequent ADE assays. To determine whether the effects of immune sera on FIPV infection were mediated by serum IgG, aliquots of FIPV-positive serum and immune sera collected 20 days after the third immunization were treated with Protein A/G. Briefly, serum samples were incubated with Protein A/G agarose beads according to the manufacturer’sinstructions at 4 °C with gentle rotation. After incubation, the samples were centrifuged to remove the Protein A/G beads and bead-bound immunoglobulins, and the resulting supernatants were collected as Protein A/G-treated sera. Untreated aliquots of the corresponding sera were retained for parallel comparison. Protein A/G-treated and untreated sera were subsequently tested under identical experimental conditions. FIPV (MOI = 1) was mixed 1:1 (v/v) with FIPV-positive serum or immune sera collected 20 days after the third immunization from each vaccine group, and the mixtures were preincubated at 37 °C for 1 h. PBMCs were then incubated with the virus–serum mixtures at 4 °C for 2 h to allow viral attachment. Cells were washed with PBS andfixed in 4% (v/v) paraformaldehyde in PBS at room temperature. Afterfixation, cells were permeabilized with 2% (v/v) Triton X-100 and stained with a mouse polyclonal antibody against the FIPV N protein, followed by an Alexa Fluor 488-conjugated goat anti-mouse IgG (H + L) secondary antibody. Nuclei were counterstained with DAPI for 5 min at room temperature, andfluorescent images were acquired using a Nikon fluorescence microscope. Effect of immune serum dilution on FIPV replication in feline PBMCs. PBMCs were seeded in 12-well plates and allowed to adhere overnight. FIPV-positive serum and immune sera from each group were serially diluted (from 1:2 to 1:256) and mixed 1:1 (v/v) with FIPV (MOI = 1), followed by preincubation at 37 °C for 1 h. The virus–serum mixtures were added to PBMCs and incubated at 37 °C for 2 h. After removal of the inoculum, cells were washed and cultured in fresh medium for 48 h at 37 °C. Cells were then harvested, and viral replication was quantified by RT-qPCR targeting the FIPV N gene. Cytokine responses in PBMCs exposed to virus–serum mixtures. PBMCs were seeded in 12-well plates and cultured overnight. FIPV (MOI = 1) was mixed 1:1 (v/v) with FIPV-positive serum or immune sera from each group and preincubated at 37 °C for 1 h. Virus-serum mixtures were added to PBMCs and incubated for 2 h at 37 °C, after which the inoculum was removed and cells were cultured in fresh medium for an additional 48 h. Following treatment with the virus–serum mixtures, cells were collected and total RNA was extracted and subsequently reverse-transcribed into cDNA. The relative transcript levels of felineTNF,IL1B,IL6,andIL10were quantified by real-time quantitative PCR (RT-qPCR). RT-qPCR was performed using SYBR Green qPCR reagents on a QuantStudio 3 Real-Time PCR System (Applied Biosystems, Thermo Fisher Scientific).GAPDHwas used as the reference gene, and relative transcript levels were calculated using the 2 −ΔΔCt method. Primer sequences for these cytokine genes and the reference geneGAPDHwere based on those reported by Takano et al. 98 .and were synthesized by Sangon Biotech (Shanghai) Co., Ltd. The primer sequences are provided in Supplementary Table S2. Histopathology and immunohistochemistry Formalin-fixed, paraffin-embedded liver, spleen, and kidney tissues were sectioned at 3μm and stained with hematoxylin and eosin (H&E) for histopathological evaluation or processed for immunohistochemistry (IHC). For IHC, sections were incubated for 2 h at room temperature with a mouse polyclonal antibody against the FIPV N protein, washed with 0.1% PBST (PBS containing 0.1% Tween-20), and then incubated for 40 min at room temperature with a Cy3-conjugated goat anti-mouse secondary antibody. After afinal wash with 0.1% PBST, nuclei were counterstained with DAPI solution containing antifade reagent. Viral load quantification To generate a quantitative standard, the FIPV N gene was cloned into the pET-32a vector to generate a reference plasmid for construction of a standard curve. Viral RNA was extracted from collected samples using a viral RNA extraction kit (YDP315-R, Tiangen), and cDNA was synthesized using a one-step reverse transcription kit (KR118-03, Tiangen). RT-qPCR was performed on a Thermo Fisher real-time PCR system using a SYBR Green PCR kit (0521961, Novoprotein) to quantify FIPV N cDNA alongside serial dilutions of the reference plasmid. Viral RNA copy numbers were calculated from the standard curve and normalized to total RNA input, and results are presented as log 10 Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

(copies/mg total RNA). Primer sequences and standard-plasmid quality control information are provided in Tables S2 and S3 (Supplementary Information). Conjugation of helper factors to ST-N-Fer via SpyTag/ SpyCatcher For dual-module screening formulations, purified ST-N-Fer was mixed with an individual SpyCatcher-fused helper factor (APC-targeting ligand or immunostimulatory factor) at a 1:1 molar ratio calculated on the basis of ST-N-Fer monomers, and incubated overnight at 4 °C to allow Spy- Tag/SpyCatcher-mediated covalent coupling. Conjugation efficiency was verified by 10% SDS–PAGE. For formulations displaying more than one functional module, purified ST-N-Fer was mixed with the indicated combinations of SpyCatcher-fused helper factors such that the combined molar amount of the helper factors was equal to the molar amount of ST-N-Fer monomers, with the individual helper factors added at equimolar proportions. Reactions were incubated overnight at 4 °C, and covalent coupling was assessed by 10% SDS–PAGE. Sample collection and preparation of nuclei suspensions Six cats were randomly assigned into three groups: Control, Inactivated vaccine, and tfN/HInps. Cats in the immunization groups were immunized subcutaneously on days 0, 21, and 42 with either tfN/HInps or inactivated whole FIPV emulsified with Freund’s adjuvant. On day 63 after the primary immunization, cats were humanely euthanized in accordance with the Guidelines for the Care and Use of Laboratory Animals issued by the Ministry of Science and Technology of China, and spleens were collected for preparation of nuclei suspensions. Briefly, fresh spleen tissue was minced in 1 mL of nuclear lysis buffer (NST) containing 0.1% NP-40, 10 mM Tris-HCl, 146 mM NaCl, 1 mM CaCl 2 ,21mMMgCl 2 , and 40 U/mL RNase inhibitor, and incubated for 7 min to allow nuclear release. Completion of lysis and nuclear release was confirmed by trypan blue staining under a light microscope. Following lysis, 1 mL of ST wash buffer (10 mM Tris-HCl, 146 mM NaCl, 1 mM CaCl 2 ,21mMMgCl 2 , 0.01% BSA, and 40 U/mL RNase inhibitor) was added, and the suspension was passed through a 40μmcell strainer (BD Biosciences) into a 15 mL centrifuge tube. The strainer was rinsed with additional ST wash buffer, and thefiltrates were combined. Nuclei were pelleted by centrifugation at 500 × g for 5 min at 4 °C, resuspended in 5 mL PBS containing 1% BSA, washed once by centrifugation, andfinally resuspended in 100μLPBSwith1%BSA.Nuclei were counted after trypan blue staining, and the concentration was adjusted to 700-1200 nuclei/μL with PBS containing 1% BSA. Sequence conservation analysis of FCoV N proteins Representative full-length N protein amino acid sequences from type I and type II FCoV/FIPV strains, including FECV- and FIPV-related sequences and the study strain OR711916.1, were retrieved from the NCBI GenBank database (strain information is provided in Supplementary Table S4). For each genome record, the annotated nucleocapsid protein-coding sequence was identified from the GenBank CDS features, and the translated amino acid sequence was extracted. N protein amino acid sequences were aligned using MAFFT v7.526 with the auto strategy and default parameters. Pairwise amino acid identities were calculated from the aligned sequences using a custom Python script and displayed as a heatmap with matplotlib 3.10.9. Position-wise conservation was defined as the percentage of non-gap sequences sharing the consensus amino acid at each aligned position. Columns corresponding to gaps in the reference N protein were excluded from residue numbering, and an 11-amino-acid sliding mean was used to generate the linear conservation profile. Single-nucleus library construction and sequencing Nuclei suspensions were processed using the Chromium Next GEM Single Cell 3ʹReagent Kits v3.1 (10× Genomics, 1000268) to generate 3ʹ gene expression libraries according to the manufacturer’s protocol. Sequencing libraries were prepared with the Chromium Single Cell 3ʹ/5ʹ Library Construction Kit (10× Genomics, 1000020). Final libraries were sequenced on an Illumina NovaSeq 6000 platform using a paired-end 150 bp (PE150) configuration. Single-nucleus transcriptomic data processing and analysis Raw FASTQfiles were processed with Cell Ranger (version 9.0.0, 10 × Genomics) to generate gene-by-cell expression matrices. Downstream analysis of single-nucleus transcriptomic data was performed in R (version 4.4.1) using the Seurat package (version 4.0.0) 99 .Low-quality nuclei were removed according to the following criteria: fewer than 200 detected genes, fewer than 1,000 unique molecular identifiers (UMIs), log10GenesPerUMI < 0.7, >10% mitochondrial RNA content, or >5% hemoglobin RNA content. Putative doublets were identified and excluded using DoubletFinder (version 2.0.3). Gene expression values were normalized using the“LogNormalize”method in Seurat, and 2000 highly variable genes were selected for subsequent analyses. Dimensionality reduction was performed by principal component analysis (PCA), and batch effects among samples were corrected using Harmony (version 1.0). Nuclei were clustered with the FindNeighbors and FindClusters functions based on the leading principal components, and clusters were visualized using UMAP (RunUMAP). Clusterspecific marker genes were identified with FindAllMarkers (test.use = “wilcox”), and differentially expressed genes (DEGs) between groups were detected using FindMarkers, with significance thresholds of *p* <0.05 and |log₂fold change | > 0.58. All analyses were conducted using orthologous genes shared between cat and mouse. Compilation of marker genes and cell-type annotation Classical marker genes for major cell types were curated from published literature and the CellMarker database 100 . Annotation of splenic cell types was based on both the overall transcriptomic similarity to public reference datasets and the expression patterns of canonical marker genes. Briefly, after cross-species conversion to orthologous genes, our snRNA-seq dataset and a reference spleen scRNA-seq dataset were processed as separate Seurat objects. Data integration was performed using the FindIntegrationAnchors and IntegrateData functions in Seurat to obtain a shared embedding space. For each integrated cluster, the proportion of nuclei originating from the reference dataset was calculated, and cell-type labels from the reference metadata were used to determine the composition of reference cell types within each cluster. The reference cell type contributing the highest proportion to a given cluster was initially assigned as the candidate cell identity for that cluster. Final cell type annotations were refined and confirmed by manual inspection of the expression of canonical marker genes and differentially expressed genes within each cluster. After annotation, the absolute numbers and relative frequencies of each cell type were quantified for every sample. Pseudotime analysis with Monocle 2 Pseudotime trajectories of dendritic cells and T cells were inferred using the Monocle 2 algorithm (version 2) 101 . Single cells annotated as dendritic cell subsets (cDC1, cDC2, and pDC) or CD8⁺T-cell subsets were subsetted from the full dataset, and trajectories were constructed separately for each lineage. For dendritic cells, genes with variable expression were identified using the dispersionTable function in Monocle 2, and those with a normalized variance above thefitted curve and a mean expression > 0.001 were selected as ordering genes. For CD8⁺T cells, highly variable genes defined in Seurat (average expression between 0.0125 and 3, and scaled variance > 0.5) were used as input. After dimensionality reduction and cell ordering, cell state transition trajectories were reconstructed using Monocle 2 with default parameters. Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

Expression of key antigen-uptake-related genes in cDC1 cells Cells annotated as cDC1 and assigned to State 3 (late pseudotime) in the Monocle-inferred trajectory were extracted, and the corresponding normalized expression matrix and sample metadata were obtained. Based on previous reports, we curated a panel of dendritic cell receptors involved in antigen uptake, including patternrecognition receptors (TLR7, TLR8, NLRP3), complement receptors (ITGB2, ITGAM), scavenger receptors (MSR1, CD36, SCARB1), IgG Fc receptors (FCGR2B), and C-type lectin receptors (CLEC7A). Only genes detected in the expression matrix were retained for subsequent analysis. For each sample, the mean expression of these genes was calculated across its State 3 cDC1 cells. To facilitate comparisons among samples, gene-wise mean expression values were standardized using z-score normalization with the scale function in R. Module scoring of antigen-presentation programs in latestage cDC1 Cells annotated as cDC1 and assignedto State 3 (late pseudotime) in the Monocle-derived dendritic-cell trajectory were extracted, and the corresponding metadata and normalized expression matrix were retrieved. Based on previous studies, we predefined gene modules related to antigen processing and cross-presentation-associated programs, including: (1) JAK-STAT-SOCS signaling (SOCS1,SOCS3,STAT1,STAT3); (2) transcriptional control of MHC I (NLRC5); (3) regulators of MHC II expression (CIITA,RFX5,RFXANK,RFXAP); (4) Endolysosomal antigen processing (MHC II;ATP6V1A,CTSB,CTSD,CTSS,IFI30); (5) Vesiclemediated antigen cross-presentation (RAB11A,RAB14,RAB27A,RAB43, SEC22B,SNAP23,STX11,STX4,STX7,VAMP3,VAMP7,WDFY4); (6) Proteasome-dependent cross-presentation (PSMB8,PSMB9,PSMB10, PSMB1,PSMB2,PSMB3); (7) Peptide transport and editing for MHC I (ERAP1,LNPEP,SEC61A1,TAP1,TAP2); (8) Co-stimulatory molecules and cytokine axis (CD40,CD70,IL10RA,IL12A,IL12B,IL15, TNFSF4). Only genes detected in the expression matrix were retained, and modules without detectable genes were excluded. For each retained module, a per-cell module score was calculated at State 3 cDC1 by averaging the normalized expression values of all genes within that module. Cell-cell communication analysis Cell-cell communication was analyzed using the CellChat R package, which predicts ligand–receptor (LR) interactions between cell types based on single-nucleus transcriptomic data 102 . A CellChat object was constructed from the Seurat object using createCellChat, and the appropriate CellChatDB was assigned. Signaling genes were retained using subsetData, and overexpressed genes and LR pairs were identified using identifyOverExpressedGenes and identifyOverExpressedInteractions, respectively. Interaction probabilities between cell subsets were computed using computeCommunProb, and interactions supported by small cell numbers were removed usingfilterCommunication. Communication probabilities were thenaggregated to construct a global cell-cell communication network for each group (aggregateNet). Circle plots were generated using netVisual_circle based on interaction weights or counts to visualize interaction strength and frequency among cell populations. Pathway-level communication probabilities were computed using computeCommunProbPathway, and pathway-specific communication patterns were visualized using netVisual_aggregate (and netVisual_heatmap when appropriate). Transcription factor regulatory network analysis Transcription factor regulatory networks were inferred using the SCENIC R package (version 1.3.1) based on single-nucleus RNA-seq data 103 . Differentially expressed gene (DEG) expression matrices from the tfN/HInps vs. Control and Inactivated vaccine vs. Control comparisons were used as input for SCENIC. Gene co-expression networks werefirst reconstructed with GENIE3, followed by cis-regulatory motif analysis and calculation of regulon activity scores. Statistics and reproducibility Unless otherwise indicated, quantitative experimental data are presented as mean ± SD. Statistical analyses were performed using GraphPad Prism 10.0.1. No statistical method was used to predetermine sample size, and no data were excluded from the analyses. Animals were randomly assigned to experimental groups. Investigators were blinded to group allocation during experimental procedures and outcome assessment, including clinical scoring and histopathological evaluation. The number of independent experimental repeats is indicated in the correspondingfigure legends, and similar results were obtained across independent experiments. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Data availability All generated feline snRNA-seq data in this study have been deposited in the Genome Sequence Archive at the National Genomics Data Center, China National Center for Bioinformation, under accession codeCRA037663. All other relevant data supporting thefindings of this study are available within the article and its Supplementary Information. Source data are provided with this paper. GenBank accession numbers and corresponding NCBI database links for the protein sequences used in this study are provided in Supplementary Table S5. Source data are provided with this paper. Code availability The Python scripts used for pairwise amino acid identity and positionwise conservation analyses of representative FCoV/FIPV N proteins are publicly available via Zenodo 104 . References 1.Cui, J., Li, F. & Shi, Z.-L. Origin and evolution of pathogenic coronaviruses.Nat. Rev. Microbiol.17,181–192 (2019). 2.Markov, P. V. et al. The evolution of SARS-CoV-2.Nat. Rev. Microbiol.21,361–379 (2023). 3.Carabelli, A. M. et al. SARS-CoV-2 variant biology: immune escape, transmission andfitness.Nat. Rev. Microbiol.21,162–177 (2023). 4.Hou, Y. J. et al. Host range, transmissibility and antigenicity of a pangolin coronavirus.Nat. Microbiol.8,1820–1833 (2023). 5.Wells, H. L. et al. The coronavirus recombination pathway.Cell Host Microbe31,874–889 (2023). 6.Su, M. et al. Coronavirus porcine epidemic diarrhea virus nucleocapsid protein interacts with p53 to induce cell cycle arrest in S-phase and promotes viral replication.J. Virol.95, e00187–21 (2021). 7.Tao, T. et al. Antibody-dependent enhancement of coronaviruses. Int. J. Biol. Sci.21,1686–1704 (2025). 8.Ao, D., He, X., Liu, J. & Xu, L. Strategies for the development and approval of COVID-19 vaccines and therapeutics in the postpandemic period.Signal Transduct. Target. Ther.8,466(2023). 9.Zhang, A. et al. Beyond neutralization: Fc-dependent antibody effector functions in SARS-CoV-2 infection.Nat. Rev. Immunol.23, 381–396 (2023). 10. Wang, S. et al. Antibody-dependent enhancement (ADE) of SARS- CoV-2 pseudoviral infection requires FcγRIIB and virus-antibody complex with bivalent interaction.Commun. Biol.5, 262 (2022). 11.Arvin, A. M. et al. A perspective on potential antibody-dependent enhancement of SARS-CoV-2.Nature584,353–363 (2020). 12. Olsen,C.W.,Corapi,W.V.,Ngichabe,C.K.,Baines,J.D.&Scott,F. W. Monoclonal antibodies to the spike protein of feline infectious peritonitis virus mediate antibody-dependent enhancement of infection of feline macrophages.J. Virol.66,956–965 (1992). Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

13. Wells,T.J.,Esposito,T.,Henderson, I. R. & Labzin, L. I. Mechanisms of antibody-dependent enhancement of infectious disease.Nat. Rev. Immunol.25,6–21 (2025). 14. Liu, Y. et al. An infectivity-enhancing site on the SARS-CoV-2 spike protein targeted by antibodies.Cell184,3452–3466.e18 (2021). 15. Vennema, H. et al. Early death after feline infectious peritonitis virus challenge due to recombinant vaccinia virus immunization.J. Virol.64,1407–1409 (1990). 16. Attipa, C. et al. Feline infectious peritonitis epizootic caused by a recombinant coronavirus.Nature645, 228–234 (2025). 17.Tortorici, M. A. et al. Loss of FCoV-23 spike domain 0 enhances fusogenicity and entry kinetics.Nature645,235–243 (2025). 18. Li, C. et al. Circulation and genetic diversity of Feline coronavirus type I and II from clinically healthy and FIP-suspected cats in China.Transbound. Emerg. Dis.66,763–775 (2019). 19. Dandekar, A. A. & Perlman, S. Immunopathogenesis of coronavirus infections: implications for SARS.Nat. Rev. Immunol.5, 917–927 (2005). 20. Pickering, S. et al. Antibodies to the RBD of SARS-CoV-2 spike mediate productive infection of primary human macrophages. Nat. Commun.15, 10764 (2024). 21. Dai, L. & Gao, G. F. Viral targets for vaccines against COVID-19.Nat. Rev. Immunol.21,73–82 (2021). 22. Lee, W. S., Wheatley, A. K., Kent, S. J. & DeKosky, B. J. Antibodydependent enhancement and SARS-CoV-2 vaccines and therapies.Nat. Microbiol.5, 1185–1191 (2020). 23. Savastano, A., Ibáñez de Opakua, A., Rankovic, M. & Zweckstetter, M. Nucleocapsid protein of SARS-CoV-2 phase separates into RNA-rich polymerase-containing condensates.Nat. Commun.11, 6041 (2020). 24. Favetta, B. et al. Phosphorylation toggles the SARS-CoV-2 nucleocapsid protein between two membrane-associated condensate states.Nat. Commun.16, 7970 (2025). 25. Lu, S. et al. The SARS-CoV-2 nucleocapsid phosphoprotein forms mutually exclusive condensates with RNA and the membraneassociated M protein.Nat. Commun.12, 502 (2021). 26. Malone,B.,Urakova,N.,Snijder,E.J.&Campbell,E.A.Structures and functions of coronavirus replication-transcription complexes and their relevance for SARS-CoV-2 drug design.Nat. Rev. Mol. Cell Biol.23,21–39 (2022). 27. Olsen, D. A. et al. Quantifying SARS-CoV-2 nucleocapsid antigen in oropharyngeal swabs using single molecule array technology. Sci. Rep.11, 20323 (2021). 28. Shan, D. et al. N-protein presents early in blood, dried blood and saliva during asymptomatic and symptomatic SARS-CoV-2 infection.Nat. Commun.12, 1931 (2021). 29. Dao Thi, V. L. et al. A colorimetric RT-LAMP assay and LAMP- sequencing for detecting SARS-CoV-2 RNA in clinical samples. Sci. Transl. Med.12, eabc7075 (2020). 30. Weissleder, R., Lee, H., Ko, J. & Pittet, M. J. COVID-19 diagnostics in context.Sci. Transl. Med.12, eabc1931 (2020). 31. Pinto, P. B. A. et al. Co-immunization with spike and nucleocapsid based DNA vaccines for long-term protective immunity against SARS-CoV-2 Omicron.npj Vaccines9, 252 (2024). 32. Hajnik, R. L. et al. Dual spike and nucleocapsid mRNA vaccination confer protection against SARS-CoV-2 Omicron and Delta variants in preclinical models.Sci. Transl. Med.14, eabq1945 (2022). 33. Dangi, T., Class, J., Palacio, N., Richner, J. M. & Penaloza Mac- Master, P. Combining spike- and nucleocapsid-based vaccines improves distal control of SARS-CoV-2.Cell Rep.36, 109664 (2021). 34. Reguzova, A. et al. A multiantigenic Orf virus-based vaccine efficiently protects hamsters and nonhuman primates against SARS- CoV-2.npj Vaccines9,191(2024). 35. Kim, J. et al. Combined immunization with SARS-CoV-2 spike and SARS-CoV nucleocapsid protects K18-hACE2 mice but increases lung pathology.npj Vaccines10,30(2025). 36. Le Bert, N. et al. SARS-CoV-2-specific T cell immunity in cases of COVID-19 and SARS, and uninfected controls.Nature584, 457–462 (2020). 37. Choy, C. et al. SARS-CoV-2 infection establishes a stable and ageindependent CD8+ T cell response against a dominant nucleocapsid epitope using restricted T cell receptors.Nat. Commun.14, 6725 (2023). 38. Goto, Y. et al. Molecular basis of potent antiviral HLA-C-restricted CD8+ T cell response to an immunodominant SARS-CoV-2 nucleocapsid epitope.Nat. Commun.16,8062(2025). 39. Wasmoen, T. L. et al. Protection of cats from infectious peritonitis by vaccination with a recombinant raccoon poxvirus expressing the nucleocapsid gene of feline infectious peritonitis virus.Adv. Exp. Med. Biol.380, 221–228 (1995). 40. Hohdatsu, T. et al. Vaccine efficacy of a cell lysate with recombinant baculovirus-expressed feline infectious peritonitis (FIP) virus nucleocapsid protein against progression of FIP.Vet. Microbiol. 97,31–44 (2003). 41. Caminschi,I.&Shortman,K.Boostingantibodyresponsesbytargeting antigens to dendritic cells.Trends Immunol.33,71–77 (2012). 42. Elgueta, R. et al. Molecular mechanism and function of CD40/ CD40L engagement in the immune system.Immunol. Rev.229 , 152–172 (2009). 43. Gupta, S. et al. Vaccination with a fusion protein that introduces HIV-1 gag antigen into a multitrimer CD40L construct results in enhanced CD8+ T cell responses and protection from viral challenge by vaccinia-gag.J. Virol.88, 1492–1501 (2014). 44. Liu, X. et al. Expression and immunogenicity of two recombinant fusion proteins comprising foot-and-mouth disease virus structural protein VP1 and DC-SIGN-binding glycoproteins.Biomed. Res. Int.2017, 7658970 (2017). 45. Bhanpattanakul, S. et al. Modulation of MHC expression by interferon-gamma and its influence on PBMC-mediated cytotoxicity in canine mast cell tumour cells.Sci. Rep.14, 17837 (2024). 46. Hajam,I.A.,Dar,P.A.,Shahnawaz,I.,Jaume,J.C.&Lee,J.H. Bacterialflagellin—a potent immunomodulatory agent.Exp. Mol. Med.49,e373(2017). 47. Khim, K. et al. Deimmunization offlagellin for repeated administration as a vaccine adjuvant.npj Vaccines6, 116 (2021). 48. Li, Y. et al. Flagellin co-expression potentiates mRNA vaccineinduced cytotoxic T lymphocyte responses but not anti-tumor immunity.Sci. Rep.15, 37280 (2025). 49. Shimizu, K. et al. Combination of cancer vaccine with CD122-biased IL-2/anti-IL-2 Ab complex shapes the stem-like effector NK and CD8+ T cells against tumor.J. Immunother. Cancer11, e006409 (2023). 50. Tai, W. et al. Development of a ferritin-based nanoparticle vaccine against the SARS-CoV-2 Omicron variant.Signal Transduct. Target. Ther.7, 173 (2022). 51. Nguyen, T. D. et al. A combined adjuvant and ferritin nanocage based mucosal vaccine againstStreptococcus pneumoniae induces protective immune responses in a murine model.Nat. Commun.16,2871(2025). 52. Yang, D. et al. Combination of S1–N-terminal and S1–C-terminal domain antigens targeting double receptor-binding domains bolsters protective immunity of a nanoparticle vaccine against porcine epidemic diarrhea virus.ACS Nano18, 12235–12260 (2024). 53. Rouvinski, A. et al. Covalently linked dengue virus envelope glycoprotein dimers reduce exposure of the immunodominant fusion loop epitope.Nat. Commun.8, 15411 (2017). 54. Yang, S. et al. Safety and immunogenicity of a recombinant tandem-repeat dimeric RBD-based protein subunit vaccine Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

(ZF2001) against COVID-19 in adults: two randomised, doubleblind, placebo-controlled, phase 1 and 2 trials.Lancet Infect. Dis. 21, 1107–1119 (2021). 55. Dai, L. et al. Efficacy and safety of the RBD-dimer-based Covid-19 vaccine ZF2001 in adults.N. Engl. J. Med.386,2097–2111 (2022). 56. Eser, T. M. et al. Nucleocapsid-specific T cell responses associate with control of SARS-CoV-2 in the upper airways before seroconversion.Nat. Commun.14, 2952 (2023). 57. Oliveira, E. R. A. et al. Aspects of T cell-mediated immunity induced in mice by a DNA vaccine based on the dengue-NS1 antigen after challenge by the intracerebral route.PLoS ONE11, e0163240 (2016). 58. Fossum, E. et al. Vaccine molecules targeting Xcr1 on crosspresenting DCs induce protective CD8+ T-cell responses against influenza virus.Eur. J. Immunol.45,624–635 (2015). 59. Hulse,D.J.&Romero,C.H.Partialprotectionagainstinfectious bursal disease virus through DNA-mediated vaccination with the VP2 capsid protein and chicken IL-2 genes.Vaccine22, 1249–1259 (2004). 60. Chow, Y. H. et al. Development of Th1 and Th2 populations and the nature of immune responses to hepatitis B virus DNA vaccines can be modulated by codelivery of various cytokine genes.J. Immunol.160,1320–1329 (1998). 61. Theisen, D. J. et al. WDFY4 is required for cross-presentation in response to viral and tumor antigens.Science362, 694–699 (2018). 62. Schreibelt, G. et al. The C-type lectin receptor CLEC9A mediates antigen uptake and (cross-)presentation by human blood BDCA3+ myeloid dendritic cells.Blood119, 2284–2292 (2012). 63. Hildner,K.etal.Batf3deficiency reveals a critical role for CD8α+ dendritic cells in cytotoxic T cell immunity.Science322, 1097–1100 (2008). 64. Lucas,E.D.etal.CirculatingKLRG1+long-livedeffectormemory T cells retain theflexibility to become tissue resident.Sci. Immunol.9,eadj8356(2024). 65. Olson, J. A., McDonald-Hyman, C., Jameson, S. C. & Hamilton, S. E. Effector-like CD8+ T cells in the memory population mediate potent protective immunity.Immunity38,1250–1260 (2013). 66. Ying, B. et al. Mucosal vaccine-induced cross-reactive CD8+ T cells protect against SARS-CoV-2 XBB.1.5 respiratory tract infection.Nat. Immunol.25,537–551 (2024). 67. Diallo, B. K. et al. Intranasal COVID-19 vaccine induces respiratory memory T cells and protects K18-hACE mice against SARS-CoV-2 infection.npj Vaccines8, 68 (2023). 68. Kim, M. V., Ouyang, W., Liao, W., Zhang, M. Q. & Li, M. O. The transcription factor Foxo1 controls central-memory CD8+ T cell responses to infection.Immunity39,286–297 (2013). 69. Utzschneider, D. T. et al. Active maintenance of T cell memory in acute and chronic viral infection depends on continuous expression of FOXO1.Cell Rep.22,3454–3467 (2018). 70. Doan, A. E. et al. FOXO1 is a master regulator of memory programming in CAR T cells.Nature629,211–218 (2024). 71.Chan, J. D. et al. FOXO1 enhances CAR T cell stemness, metabolic fitness and efficacy.Nature629,201–210 (2024). 72. Pereira,K.H.N.P.etal.Topicsonmaternal,fetalandneonatal immunology of dogs and cats.Vet. Immunol. Immunopathol.266, 110678 (2023). 73. Barman, S. et al. Shaping neonatal immunization by tuning the delivery of synergistic adjuvants via nanocarriers.ACS Chem. Biol. 17,2559–2571 (2022). 74. van Haren, S. D. et al. CAF08 adjuvant enables single dose protection against respiratory syncytial virus infection in murine newborns.Nat. Commun.13, 4234 (2022). 75. Schäfer-Somi, S. Fetal, neonatal, and pediatric immunology.Vet. Clin. North Am. Small Anim. Pract.55,815–828 (2025). 76. Wang, Y. et al. Evaluating triple inactivated vaccine-induced immunity from a large-scale study in feline population.Sci. Rep. 16, 1929 (2026). 77. Squires, R. A., Crawford, C., Marcondes, M. & Whitley, N. 2024 guidelines for the vaccination of dogs and cats–compiled by the Vaccination Guidelines Group (VGG) of the World Small Animal Veterinary Association (WSAVA).J. Small Anim. Pract.65, 277–316 (2024). 78. Pedersen, N. C. A review of feline infectious peritonitis virus infection: 1963–2008.J. Feline Med. Surg.11, 225–258 (2009). 79. Lavelle,E.C.&McEntee,C.P.Vaccine adjuvants: tailoring innate recognition to send the right message.Immunity 57, 772–789 (2024). 80. Lee,W.,Larsen,A.,Kingstad-Bakke, B., Marinaik, C. B. & Suresh, M. Combination adjuvants affect the magnitude of effector-like memory CD8 T cells and protection against listeriosis.Infect. Immun.89,e00768–20 (2021). 81. Roth, G. A. et al. Designing spatial and temporal control of vaccine responses.Nat. Rev. Mater.7,174–195 (2022). 82. Reddy, S. T. et al. Exploiting lymphatic transport and complement activation in nanoparticle vaccines.Nat. Biotechnol.25, 1159–1164 (2007). 83. Kanekiyo, M. et al. Self-assembling influenza nanoparticle vaccines elicit broadly neutralizing H1N1 antibodies.Nature499, 102–106 (2013). 84. Lee, B.-R. et al. Engineered human ferritin nanoparticles for direct delivery of tumor antigens to lymph node and cancer immunotherapy.Sci. Rep.6, 35182 (2016). 85. Zhang, Y.-N. et al. Nanoparticle size influences antigen retention and presentation in lymph node follicles for humoral immunity. Nano Lett.19, 7226–7235 (2019). 86. Fumagalli, V. et al. Antibody-independent protection against heterologous SARS-CoV-2 challenge conferred by prior infection or vaccination.Nat. Immunol.25,633–643 (2024). 87. Sullivan, N. J. et al. CD8⁺cellular immunity mediates rAd5 vaccine protection against Ebola virus infection of nonhuman primates. Nat. Med.17, 1128–1131 (2011). 88. Houser, K. V. et al. Safety and immunogenicity of a ferritin nanoparticle H2 influenza vaccine in healthy adults: a phase 1 trial.Nat. Med.28,383–391 (2022). 89. Widge,A.T.etal.Aninfluenza hemagglutinin stem nanoparticle vaccine induces cross-group 1 neutralizing antibodies in healthy adults.Sci. Transl. Med.15, eade4790 (2023). 90. Casazza, J. P. et al. Phase 1 dose-escalation trial evaluating a group 2influenza hemagglutinin stabilized stem nanoparticle vaccine. npj Vaccines9,171(2024). 91. Ober Shepherd, B. L. et al. SARS-CoV-2 recombinant spike ferritin nanoparticle vaccine adjuvanted with Army Liposome Formulation containing monophosphoryl lipid A and QS-21: a phase 1, randomised, double-blind, placebo-controlled, first-in-human clinical trial.Lancet Microbe5,e581–e593 (2024). 92. Veggiani, G. et al. Programmable polyproteams built using twin peptide superglues.Proc. Natl. Acad. Sci. USA.113, 1202–1207 (2016). 93. Brune, K. D. et al. Dual plug-and-display synthetic assembly using orthogonal reactive proteins for twin antigen immunization.Bioconjug. Chem.28,1544–1551 (2017). 94. Bale, J. B. et al. Accurate design of megadalton-scale two-component icosahedral protein complexes.Science353, 389–394 (2016). 95. Boyoglu-Barnum, S. et al. Quadrivalent influenza nanoparticle vaccines induce broad protection.Nature592,623–628 (2021). 96. Wang, G. et al. Establishment of full-length cDNA clones and an efficient oral infection model forfeline coronavirus in cats.J. Virol. 95,e00745–21 (2021). Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press

97. Brostoff, T. et al. Feline infectious peritonitis mRNA vaccine elicits both humoral and cellular immune responses in mice.Vaccines 12, 705 (2024). 98. Takano, T. et al. A“possible”involvement of TNF-alpha in apoptosis induction in peripheral blood lymphocytes of cats with feline infectious peritonitis.Vet. Microbiol.119,121–131 (2007). 99. Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species.Nat. Biotechnol.36,411–420 (2018). 100. Zhang, X. et al. CellMarker: a manually curated resource of cell markers in human and mouse.Nucleic Acids Res.47, D721–D728 (2019). 101. Qiu, X. et al. Reversed graph embedding resolves complex singlecell trajectories.Nat. Methods14,979–982 (2017). 102. Jin, S. et al. Inference and analysis of cell-cell communication using CellChat.Nat. Commun.12,1088(2021). 103. Aibar, S. et al. SCENIC: single-cell regulatory network inference and clustering.Nat. Methods14,1083–1086 (2017). 104. FeiyuZhao-0806. FIPV-N-protein-sequence-analysis: FCoV/FIPV N-protein sequence analysis v1.0.0. Zenodohttps://doi.org/10. 5281/zenodo.22220087(2026). Acknowledgements The authors have no additional acknowledgements. Author contributions D.B.S., M.J.S., D.Y., and F.Y.Z. designed the study. F.Y.Z., F.B.S., X.L.W., W.Z., W.Y.J., X.X.Y., T.S.L., Z.L., Q.H.Z., and W.Q.X. performed experiments. H.X.L., M.J.S., F.Y.Z., J.Z.C., L.B., and S.S.Q. analyzed data and generatedfigures. J.L.Z. provided suggestions for vaccine design. Y.F.C. and H.X.L. provided suggestions for pathological experimental analysis. D.S. and L.F. provided advice on in vivo experiments. D.B.S., M.J.S., and F.Y.Z. wrote the manuscript. Funding This work was supported by the National Natural Science Foundation of China Regional Innovation and Development Joint Fund Project (No. U23A20236), the National Natural Science Foundation of China (No. 32573354), and the Key Project of Natural Science Foundation of Heilongjiang Province of China (No. ZD2023C006). Competing interests The authors declare no competing interests. Additional information Supplementary informationThe online version contains supplementary material available at https://doi.org/10.1038/s41467-026-78999-4 . Correspondenceand requests for materials should be addressed to Mingjun Su or Dongbo Sun. Peer review informationNature Communicationsthanks Linda Saif, Hadi Yassine and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer reviewfile is available. Reprints and permissions informationis available at http://www.nature.com/reprints Publisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open AccessThis article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicatedotherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visithttp:// creativecommons.org/licenses/by-nc-nd/4.0/. © The Author(s) 2026 This Article in Press is shared early to give you faster access to new research. It is citable and carries a permanent DOI. Thefinal edited version will replace it automatically. Feiyu Zhao 1,6 ,DanYang 1,6 ,HaixinLiu 1,6 ,QingheZhu 1 , Fanbo Shen 1 , Jiongze Cheng 1 ,XinglinWang 1 , Wei Zhou 1 , Weiyue Jiang 1 ,XiaoxiYu 1 , Tianshu Liu 1 , Zhen Li 1 ,WenqiangXu 1 ,LanBi 1 , Shanshan Qi 1 ,DaShi 2 , Jialin Zhang 3 ,LiFeng 2 ,Yung- Fu Chang 4 , Mingjun Su 5 & Dongbo Sun 1 1 College of Animal Science and Veterinary Medicine, Heilongjiang Bayi Agricultural University, Daqing, PR China. 2 State Key Laboratory for Animal Disease Control and Prevention, Harbin Veterinary Research Institute, Chinese Academy of Agricultural Sciences, Harbin, PR China. 3 Fred Hutchinson Cancer Center, Seattle, WA, USA. 4 College of Veterinary Medicine, Cornell University, Ithaca, NY, USA. 5 College of Animal Science and Technology & College of Veterinary Medicine of Zhejiang A&F University, Hangzhou, PR China. 6 These authors contributed equally: Feiyu Zhao, Dan Yang, Haixin Liu. e-mail:mingjunsu@zafu.edu.cn;dongbosun@126.com Articlehttps://doi.org/10.1038/s41467-026-78999-4 Nature CommunicationsArticle in Press