A potent interdomain epitope-targeting antibody protects against SFTSV in mice and non-human primates

Jinhao Bi; Ziniu Dai; Xiaoci Hong; Yuanyuan Zhang; Shuaiyao Lu; Qiujing Wang; Yunshu Liu; Xinyi Li; Mingxi Li; Guangyuan Song; Qing Chen; Wenhai Yu; Yun Yang; Yingqiu Ma; Huasheng Liu; Xiangzhi Meng; Zhaoqian Wang; Dan Liu; Ren Sun; Qiangming Sun; Shibo Li; Qiang Zhou; Dapeng Li

doi:10.15302/vita.2025.12.0002

Vita > Article > DOI: 10.15302/vita.2025.12.0002

Vita Published:

A potent interdomain epitope-targeting antibody protects against SFTSV in mice and non-human primates

Jinhao Bi ¹^,²^,³^,^# ,
Ziniu Dai ¹^,³^,^# ,
Xiaoci Hong ¹^,³^,^# ,
Yuanyuan Zhang ⁴^,⁵^,⁶^,⁷^,^# ,
Shuaiyao Lu ⁹^,^# ,
Qiujing Wang ⁸^,^# ,
Yunshu Liu ¹^,³ ,
Xinyi Li ¹^,³ ,
Mingxi Li ¹^,³ ,
Guangyuan Song ¹^,³ ,
Qing Chen ¹^,³ ,
Wenhai Yu ⁹ ,
Yun Yang ⁹ ,
Yingqiu Ma ⁸ ,
Huasheng Liu ⁸ ,
Xiangzhi Meng ¹^,³ ,
Zhaoqian Wang ¹^,³^,⁵ ,
Dan Liu ³^,⁶^,⁷ ,
Ren Sun ¹^,³^,⁵ ,
Qiangming Sun ⁹ ,
Shibo Li ⁸ ,
Qiang Zhou ⁴^,⁵^,⁶^,⁷ ,
Dapeng Li ¹^,³^,⁵

Author information +

¹. Zhejiang Key Laboratory of Multi-Omics in Infection and Immunity, Center for Infectious Disease Research, Westlake University, Hangzhou, Zhejiang, China

². Affiliated Hangzhou First People’s Hospital, Westlake University, Hangzhou, Zhejiang, China

³. School of Medicine, Westlake University, Hangzhou, Zhejiang, China

⁴. Zhejiang Key Laboratory of Structural Biology, Westlake University, Hangzhou, Zhejiang, China

⁵. School of Life Sciences, Westlake University, Hangzhou, Zhejiang, China

⁶. Westlake Laboratory of Life Sciences and Biomedicine, Hangzhou, Zhejiang, China

⁷. Institute of Biology, Westlake Institute for Advanced Study, Hangzhou, Zhejiang, China

⁸. Department of Infectious Diseases, Zhoushan Hospital, Wenzhou Medical University, Zhoushan, Zhejiang, China

⁹. National Kunming High-Level Biosafety Primate Research Center, Institute of Medical Biology, Chinese Academy of Medical Sciences and Peking Union Medical College, Kunming, Yunnan, China

qsun@imbcams.com.cn

lsb0398@126.com

zhouqiang@westlake.edu.cn

lidapeng@westlake.edu.cn

Show less

History +

Vita () Cite this article

PDF (13808KB)

ABSTRACT

Severe fever with thrombocytopenia syndrome virus (SFTSV) poses a growing global health threat with substantial mortality and no effective treatments. We report the discovery of ZS1C5, a human antibody that neutralizes SFTSV with subnanomolar potency by targeting a previously unknown interdomain epitope on glycoprotein Gn, mediated by an elongated CDRH3. A single dose of ZS1C5 conferred robust protection in both murine and, for the first time, non-human primate SFTSV infection models, demonstrating rapid viral control and immune restoration. Importantly, we applied a structure-guided mining approach to human B cell repertoires and identified germline-encoded ZS1C5-like antibodies, underscoring the potential for rapid recall of protective immunity. Together, these findings establish ZS1C5 as a promising clinical candidate and provide a blueprint for developing therapeutics and vaccines targeting interdomain epitopes in emerging bunyaviruses.

登录浏览全文

4963

注册一个新账户忘记密码

INTRODUCTION

Severe fever with thrombocytopenia syndrome virus (SFTSV) is a tick-borne virus in the family Phenuiviridae of the order Hareavirales (formerly Bunyavirales), causing hemorrhagic fever and thrombocytopenia. Identified in 2011¹, SFTSV has rapidly spread across Asia over the past decade, with mortality rates ranging from 16.2% to 30%^2-4. Transmission primarily occurs through tick bites, with documented cases of human-to-human transmission, and multiple animals serve as natural reservoirs^5,6. Therefore, SFTSV has been listed as one of the priority pathogens with high Public Health Emergency of International Concern risk by the World Health Organization (WHO). Currently, there is no effective antiviral treatment or vaccine, and treatment remains supportive. Critically ill patients require intensive care such as plasma exchange therapy⁷, highlighting the urgent need for potent therapeutic antibodies to reduce mortality and minimize the public health burden caused by SFTSV.

The SFTSV genome consists of three negative-sense RNA segments: L, M, and S. The M segment encodes a membrane protein precursor that undergoes proteolytic maturation to produce two glycoproteins, Gn and Gc, which form heterodimers and play crucial roles in viral entry. Gn is composed of the N-terminal Gn head region and the C-terminal Gn stem region. The N-terminal Gn head consists of three domains⁸: Domain I (or Domain A⁹), which is fully exposed and may directly interact with host receptors; Domain II (or β-connector⁹), which is mainly β-Strands and forms the major interface for interacting with Gc; and Domain III (or Domain B⁹), which caps the fusion loops of Gc^8,9. Gn heads cluster together to form a crown on top of Gc, which is a class II viral fusion protein, rendering Gc less accessible and only accessible during the intermediate states of membrane fusion⁹. Thus, the Gn head is a more potent target for neutralizing antibodies (NAbs).

Several NAbs targeting SFTSV Gn head have been developed. MAb4–5, the first documented NAb, targets helices α6 in Domain III of Gn but has limited in vivo protection^8,10. Ab10, a single-chain variable fragment (scFv) antibody, binds to Domain II and the stem region of Gn¹¹. Mouse mAbs 40C10 and S2A5^12-14, as well as human mAb SF5 and JK-2/JK-8^15,16, recognize epitopes on Gn Domain I and have shown protective effects in immunodeficient mouse models. Despite these advances, our understanding of SFTSV NAb epitopes, especially those with sufficient potency for therapeutic use, remains limited. Further characterization of novel neutralizing epitopes is the key to discovering more potent NAbs and designing effective vaccines. Additionally, it is crucial to test therapeutic NAbs in immunocompetent animal models, particularly non-human primates (NHPs), where SFTSV infection closely mimics human infection¹⁷.

In the present study, we identified a human NAb ZS1C5 that neutralizes SFTSV with a subnanomolar neutralization titer. Structural analysis revealed that ZS1C5 recognizes a novel interdomain epitope that spans Domain I and Domain III of the Gn head. A single dose of ZS1C5 demonstrated robust protection in immunodeficient mice against lethal SFTSV challenge and in rhesus macaques, reducing viremia and tissue viral RNA load while suppressing hyperinflammation and preserving immune homeostasis, as shown by transcriptomic profiling. Using the structural insight, computational B cell receptor (BCR) mining uncovered ZS1C5-like antibodies within the human repertoire capable of neutralizing SFTSV. These findings highlight the therapeutic potential of ZS1C5 and provide insights to inform SFTSV vaccine design.

RESULTS

Characterization of SFTSV glycoprotein Gn/Gc-reactive BCRs from convalescent donors

To select convalescent donors for NAbs isolation, we obtained plasma samples from individuals who had suffered symptomatic laboratory-confirmed SFTSV infection during both acute (n = 12) and convalescent (n = 9) phases. We screened these samples for SFTSV-reactive antibody binding activities (Supplementary Fig. S1a) and neutralization against authentic SFTSV (Fig. 1a). Based on these results, five donors were selected for the subsequent NAbs isolation steps, in which peripheral blood mononuclear cells (PBMCs) collected in the convalescent phase were sorted based on antigen specificity for single-cell PCR or single-cell BCR sequencing (BCR-seq) (Fig. 1b, c).

Transcriptomic analysis indicated that the SFTSV Gn/Gc-specific, IgM^–/IgD^– B cells were mostly a cluster of cells that fell into the memory B cell subset (Fig. 1c; Supplementary Fig. S1b–e). To characterize the gene usage of the antigen-specific BCRs, we analyzed all the Gn/Gc-specific B cells with paired heavy and light chain sequences (n = 2,678) and compared with a reference human antibody repertoire¹⁸. Over 16% of the heavy chain variable region (V_H) genes were VH3-33 in Gn/Gc-specific BCRs, whereas only 2.4% of reference B cells using VH3-33 (Fig. 1d). Light-chain variable regions (V_κ/V_λ) of the Gn/Gc-specific BCRs were not skewed (Fig. 1d). Importantly, Gn/Gc-specific BCRs showed a trend to have longer heavy-chain complementarity determining region 3 (CDRH3) lengths than reference antibodies, particularly ultra-long CDRH3 exceeding 24 amino acids (aa) in length, while no difference was observed for light-chain complementarity determining region 3 (CDRL3) (Fig. 1e, f). Next, we compared the Gn/Gc-specific BCRs with a reference human antibody repertoire containing both naïve and antigen-experienced antibodies¹⁹. Both V_H and V_κ/V_λ genes exhibited antigen-experienced features, with average somatic hypermutation rates (SHM%) of 3.2% and 2.1%, respectively (Fig. 1g). Thus, SFTSV Gn/Gc-specific BCRs were derived from memory B cells, displayed skewed V_H gene usage, and had long CDRH3.

Identification of SFTSV-specific antibody clonotypes in PBMC and lymph node (LN)

To understand how SFTSV Gn/Gc-specific B cell clones expand after SFTSV infection, we performed single-cell and bulk BCR-seq on PBMCs collected from five convalescent donors during both acute and convalescent phases (Fig. 2a). After clonal normalization based on heavy-chain genes, we obtained a non-antigen-specific BCR repertoire with 84,025 clonotypes (Fig. 2b). For each donor, we identified shared clones present in both acute and convalescent stages, as well as clonotypes with highly mutated (SHM% > 10%) V_H genes (Fig. 2b), indicating these clones could develop from the early infection phase to the convalescent phase. The SFTSV Gn/Gc antigen-specific BCR pairs obtained from the previous FACS experiment (Fig. 1b, c) consisted of 2,451 clonotypes, including 314 clonotypes that were also observed in the non-specific BCR repertoire during both acute and convalescent phases (Fig. 2b; Supplementary Fig. S1f). To further validate the BCR specificities, out of these 2,451 clonotypes, we synthesized selected BCR pairs with certain criteria (CDRH3 length > 17 aa and SHM% of V_{_H}> 6%; or with differentially mutated BCRs in one clonotype) to generate recombinant IgG antibodies. After SFTSV Gn/Gc binding screening using Enzyme-Linked Immunosorbent Assay (ELISA), we discovered 25 Gn-specific and 24 Gc-specific antibodies, some of which belong to clonotypes that expanded across both acute and convalescent phases (Fig. 2c). Notably, we observed IgM and IgG with high V_H mutation rates (> 5%) in the acute phase (Fig. 2c; Supplementary Table S1), which may suggest previous SFTSV exposure history.

To track BCR clonal dynamics induced by SFTSV infection in different locations, we obtained a submandibular LN from one hospitalized patient during acute SFTSV infection, sequenced and performed clonal analysis together with peripheral B cells collected in both acute and convalescent phases (Fig. 2d). Among the 151 BCR clonotypes identified from this donor, 131/151 were composed of B cell clones from different stages, and 120/151 contained at least one BCR from the LN (Supplementary Fig. S1g). We expressed selected LN BCRs as recombinant IgGs and found two representative SFTSV Gn-specific clonotypes, Clone 29124 and Clone 77209. While the unmuated ancestor 29124-germline (29124-GL) antibody failed to bind Gn, the LN-derived antibody 29124-1 and PBMC-derived antibody 29124-2 equivalently bound to Gn in ELISA, despite being from different stages of infection and exhibiting different V_H mutation rates (Fig. 2e; Supplementary Fig. S1h). In Clone 77209, the LN-derived antibody 77209-1 was an unmutated germline antibody with no binding activity to Gn. However, antibody 77209-2 from the convalescent phase underwent SHM and acquired mutations essential for recognizing Gn (Fig. 2e; Supplementary Fig. S1h). These clonal analysis data, together with the identification of Gn/Gc-binding antibodies, indicated that some SFTSV-reactive B cells may already exist and were rapidly recalled and expanded upon infection.

Isolation of SFTSV neutralizing antibodies from convalescent donors

Among the Gn/Gc-reactive recombinant IgG antibodies identified by ELISA (Supplementary Table S1), we evaluated antibody affinity using bio-layer interferometry (BLI) and screened for NAbs against authentic SFTSV. Of the 25 Gn-reactive antibodies, eight demonstrated potent neutralization with IC₅₀ values below 1 μg/mL. In contrast, Gc-reactive antibodies exhibited weaker neutralization, with only one potent neutralizer (IC₅₀ < 1 μg/mL) and eight weak neutralizers (20 μg/mL < IC₅₀< 1 μg/mL) (Fig. 3a, b; Supplementary Fig. S2a–d). Notably, we identified an ultrapotent Gn-targeting neutralizing antibody, ZS1C5, which displayed exceptionally high neutralizing efficacy against SFTSV, with an IC₅₀ of 0.3 ng/mL (Fig. 3b; Supplementary Fig. S2c). Furthermore, sequence analysis revealed a high SHM rate in ZS1C5 V_H region (SHM% = 8.2%, Supplementary Table S1), indicating that ZS1C5 underwent extensive affinity maturation during B-cell development.

To determine whether these NAbs recognize overlapping epitopes and thus compete for binding, competitive binding assays were performed using BLI. The Gn-targeting NAbs were segregated into three distinct clusters based on their ability to mutually block binding (Fig. 3c; Supplementary Fig. S2e). ZS1C5 was grouped into Gn Cluster I and demonstrated cross-blocking activity against five other potent Gn-targeting NAbs, whose neutralization IC₅₀ values ranged from 0.037 to 0.257 μg/mL. In contrast, antibodies that did not cross-block Cluster I exhibited relatively lower neutralization titers and were classified into Clusters II and III. Unlike Gn NAbs, Gc-targeting NAbs showed weaker cross-competition, segregating into six distinct clusters (Fig. 3c; Supplementary Fig. S2f). Taken together, these findings indicate the presence of multiple neutralizing epitopes on SFTSV Gn and Gc, with ZS1C5 emerging as an exceptionally potent antibody warranting further structural analysis and in vivo experimental studies.

Cryo-electron microscopy (cryo-EM) structure revealed a novel neutralization epitope spanning Domain I and Domain III of Gn

To elucidate the structural basis of the potent neutralizing activity of ZS1C5, we determined the cryo-EM structure of the ZS1C5/SFTSV Gn complex at a resolution of 2.99 Å (Fig. 4a; Supplementary Fig. S3 and Table S2). The refined structure reveals that the ZS1C5 heavy chain occupies a groove formed between Domain I (residues 20–175) and Domain III (residues 249–311) of Gn (Fig. 4a; Supplementary Fig. S6a). Binding is predominantly mediated by the CDRH3 of ZS1C5, which forms multiple interactions with Gn. Specifically, residues E104 and R116 of CDRH3 form salt bridges with K260 and E268 of Gn (Fig. 4b), while Y107, Y109, and Y114 form hydrogen bonds with R53 and E46 of Gn (Fig. 4c). Additionally, Y107 and Y114 engage in hydrophobic interactions with P39, L41, L42, and I55 of Gn (Fig. 4d). Importantly, Y114 also interacts with the N-linked glycosylation site at N63 of Gn through hydrophobic contact with the glycans (Fig. 4d).

The epitope targeted by ZS1C5, as defined by the antibody footprint on the Gn surface, spans 18 residues across Domain I and Domain III, with a total buried surface area of 742.7 Å² (Fig. 4e). Conservation analysis of all known SFTSV variants from GenBank, including genotypes C1–4 and J (Fig. 4f), shows that this interdomain epitope is highly conserved (Fig. 4g), underscoring ZS1C5’s broad-spectrum neutralizing potential. Comparative epitope mapping against previously characterized NAbs (MAb4–5^8,10, 40C10^13,14, S2A5¹², and SF5¹⁵) revealed distinct targeting patterns. Unlike MAb4–5, which exclusively binds to Domain III, or 40C10/S2A5/SF5, which target Domain I, ZS1C5 defines a unique class of NAbs that recognizes a conformational epitope spanning both Domain I and Domain III (Fig. 4h; Supplementary Fig. S3e, f). Notably, ZS1C5 possesses an elongated CDRH3 loop (23 aa), suggesting that its extended structure facilitates access to this inter-domain epitope.

To further explore binding modes within the same NAb cluster (Fig. 3c), we resolved the cryo-EM structures of Gn in complex with two competing NAbs, ZS336 (3.03 Å resolution) and ZS65 (3.50 Å resolution) (Fig. 4i, j; Supplementary Figs. S4, S5 and Table S2). While ZS1C5 binding is predominantly mediated by CDRH3 interactions, both ZS336 and ZS65 utilize additional heavy chain CDRs, as well as light chain residues, for Gn recognition (Supplementary Fig. S6b, c). ZS336 establishes an extensive polar network through interactions involving 10 heavy chain residues (T31, V56, S57, T58, Y103, Y105, D106, S107, S108, V109) and 2 light chain residues (D1, Q27), with hydrogen bonding to Gn glycosylation sites N33 and N63 (Supplementary Fig. S6d–f). Structural π–π stacking interactions are observed between ZS336 light chain Y94, heavy chain H110, and Gn P39 (Supplementary Fig. S6f). In contrast, ZS65 exhibits enhanced hydrophobic complementarity, involving heavy chain residues I54, F55, W104, A105, I109 and Gn residues A27, P29, I30, H31, I55, and Y70, along with polar interactions mediated by seven heavy chain residues (S28, S30, Y32, R65, S102, A106, D108) and one light chain residue (S98) (Supplementary Fig. S6g–i). Epitope comparison identifies seven conserved contact residues (P39, L41, Q50, R53, K260, T263, and A301) shared among ZS1C5, ZS336, and ZS65 (Fig. 4k). Despite minor variations, ZS336 and ZS65 adopt binding angles similar to ZS1C5, resulting in buried surface areas of 979.82 Å² (ZS336) and 914.67 Å² (ZS65) (Supplementary Fig. S6j, k).

Notably, neither ZS336 nor ZS65 exhibited neutralizing potency comparable to that of ZS1C5, highlighting the distinct functional properties of the ZS1C5 epitope. In comparison to ZS336 and ZS65, ZS1C5 possesses the longest CDRH3 loop and forms a more stable complex with the Gn head, as indicated by lower energies calculated using both FoldX and Rosetta (Fig. 4l). These computational results suggest that ZS1C5 engages Gn with higher stability than ZS336 and ZS65. To explore the functional relevance of ZS1C5 binding, we modeled its interactions within the context of Gn–Gc heterodimers on the surface of the SFTSV virion⁹. Our analysis revealed that ZS1C5 binds efficiently to Gn in both hexon and penton configurations without inducing steric hindrance (Fig. 4m). This structural versatility, together with the ability to target an interdomain neutralization epitope and the presence of a long CDRH3 region, likely contributes to the superior neutralization potency of ZS1C5.

ZS1C5 confers robust therapeutic protection in a lethal SFTSV murine model

To assess the therapeutic efficacy of ZS1C5, we employed a lethal SFTSV infection model in STAT1^–/– mice. A single intraperitoneal administration of ZS1C5 at 24 hours (24 h) post-SFTSV-infection demonstrated dose-dependent protection: high (500 μg) and medium (50 μg) doses achieved 100% survival, while the low-dose regimen (5 μg) resulted in 60% survival (Fig. 5a, b). Body weight dynamics inversely correlated with antibody dosage, with full recovery observed in the high-dose group (Fig. 5b). In contrast, the control IgG and reference NAb MAb4–5 exhibited no protective efficacy, exhibiting rapid weight loss and 80–100% mortality (Fig. 5c). Histopathological evaluation by haematoxylin and eosin (H&E) staining and immunochemistry (IHC) staining demonstrated that a single ZS1C5 administration profoundly mitigated SFTSV-induced pathology. Control animals displayed extensive splenic necrosis (Fig. 5d) and elevated viral antigen burden, as evidenced by SFTSV nucleoprotein (NP) expression (Fig. 5e). In contrast, ZS1C5-treated mice displayed significantly reduced histopathology scores in the spleen, liver, lung, and kidney (Fig. 5d, f; Supplementary Fig. S7a, b), with near-physiological tissue architecture. IHC analysis revealed dose-dependent suppression of viral replication. High-dose ZS1C5 treatment (500 μg) resulted in near-undetectable viral antigen expression across all examined organs, while medium (50 μg) and low (5 μg) doses achieved statistically significant reductions in viral antigen burden compared to control IgG (Fig. 5e, g; Supplementary Fig. S7a, b). These findings correlate with the dose-dependent survival outcomes (Fig. 5b, c), confirming that ZS1C5 effectively restricts viral replication and limits end-organ damage.

To delineate the therapeutic window, a single 500 μg dose of ZS1C5 was administered at 48, 72, or 96 h post-infection (Fig. 5h). Treatment at 48 h post-infection maintained 100% survival (Fig. 5i) and reduced viral antigen to undetectable levels, with significantly less severe SFTSV-induced tissue damage compared to the control antibody group (Fig. 5j, k; Supplementary Fig. S7c). In contrast, only 4/10 and 1/10 mice in each group survived when receiving ZS1C5 treatment at 72 and 96 h post-infection, respectively (Fig. 5i). Late treatment at 72 and 96 h also showed less effectiveness in preventing SFTSV viral antigen expression and histopathological alterations across multiple organs (Fig. 5j, k; Supplementary Fig. S7c). These results demonstrated that a single therapeutic dose of ZS1C5, when delivered within 48 h post-infection, confers complete protection against lethal SFTSV challenge.

ZS1C5 protected rhesus macaques from SFTSV-induced symptoms

To further evaluate ZS1C5 in a model closely resembling human physiology, we performed in vivo protection study in rhesus macaques (Macaca mulatta), which exhibit symptoms of mild SFTSV infection similar to those in humans^17,20. Macaques (n = 4 per group) were subcutaneously challenged with 1 × 10⁷ PFU SFTSV and intravenously administered a single dose of ZS1C5 (30 mg/kg) or control mAb within 6 h post-infection (Fig. 6a). Serum neutralization titers (NT₅₀) in ZS1C5-treated animals reached 1,152 on Day 1 and remained high levels (NT₅₀ = 512) at Day 7, whereas no neutralization activity was observed in the control mAb group (Fig. 6b). Clinical monitoring revealed marked differences between the two groups: all the four macaques in the control mAb group developed lethargy and anorexia within two days, with two subjects experiencing severe complications such as limb stiffness or diarrhea lasting over three days. Conversely, ZS1C5-treated animals showed only transient mild anorexia on Day 2 post-infection (Fig. 6c; Supplementary Table S3). Both groups displayed comparable low-grade fevers two days after SFTSV challenge, with no significant differences in body weight or temperature trajectories (Supplementary Fig. S8a and Table S3).

Complete blood count (CBC) tests showed that ZS1C5 reversed SFTSV-induced thrombocytopenia, restoring platelet counts (PLT) and plateletcrit (PCT) to normal ranges, significantly higher than those in the control mAb group (Fig. 6d; Supplementary Fig. S8b and Table S3). Coagulation dysfunction, evidenced by prolonged activated partial thromboplastin time (APTT), was observed exclusively in the control mAb group (Fig. 6e; Supplementary Fig. S8c and Table S3). Additionally, comprehensive metabolic panel (CMP) tests revealed dramatically elevated ferritin levels in control monkeys, peaking on Day 6 post-infection and resulting in higher blood iron levels (Supplementary Fig. S8d and Table S3). We also noted a sharp decline in blood calcium levels in SFTSV-infected monkeys, which returned to normal in the ZS1C5 group but remained low in the control mAb group (Supplementary Fig. S8d and Table S3). Therefore, ZS1C5 antibody treatment alleviated a series of symptoms caused by SFTSV infection.

ZS1C5 treatment suppressed SFTSV viremia and tissue viral replication

To evaluate viral replication dynamics across groups, SFTSV genomic RNA levels were quantified in serum and tissues. In the control mAb-treated animals, severe viremia was observed, with average serum viral loads reaching 1.14 × 10⁶ copies/mL on Day 1 post-infection, progressively increasing to a peak of 1.06 × 10⁸ copies/mL by Day 5. In contrast, ZS1C5 treatment significantly reduced serum viral RNA levels by ~ 90% (averaging 1.27 × 10⁵ copies/mL) on Day 1 and reached undetectable levels (< 500 copies/mL) in all the four monkeys by Day 3 (Fig. 6f). At necropsy (Day 7), control mAb-treated animals exhibited extensive tissue viral replication, with the highest viral RNA loads detected in lymphoid and gastrointestinal tissues, including the spleen (5.6 × 10⁹ copies/g), axillary LNs (1.7 × 10⁹ copies/g), inguinal LNs (2.1 × 10⁸ copies/g), rectum (4.3 × 10⁷ copies/g), and duodenum (3.7 × 10⁶ copies/g). ZS1C5 treatment markedly suppressed viral replication across all organs, reducing RNA levels to undetectable ranges in the lung, liver, duodenum, and rectum, while achieving statistically significant reductions in the heart, kidney, pancreas, spleen, and inguinal LNs (Fig. 6g).

Although H&E staining showed no obvious tissue histopathological changes on Day 7 (Supplementary Fig. S8e), IHC analysis of SFTSV NP expression revealed ubiquitous antigen presence in tissues from the control mAb group. In contrast, ZS1C5 treatment nearly abolished antigen presence. Quantitative comparisons demonstrated significant reductions in viral antigen levels in the lung, heart, kidney, duodenum, rectum, and spleen (Fig. 6h, i). In tissues such as the brain, liver, pancreas, and inguinal LNs, inter-group differences in antigen levels did not reach statistical significance, consistent with the baseline inherently low viral antigen expression in these tissues (Fig. 6i; Supplementary Fig. S8f). These data collectively demonstrate that a single-dose ZS1C5 treatment rapidly eliminates circulating virus and potently suppresses tissue-specific viral replication.

Single-cell transcriptomic profiling reveals ZS1C5 treatment restores immune homeostasis and reduces inflammation

To elucidate how ZS1C5 treatment modulates systemic immunity during SFTSV infection, we performed single-cell RNA sequencing (scRNA-seq) on spleen cells and PBMCs from macaques at necropsy (Day 7 post-infection). Unsupervised clustering and comparative analysis revealed profound differences in immune cell composition and transcriptional programs between the ZS1C5 and control mAb groups (Supplementary Fig. S9a–c). Control animals exhibited pronounced expansion of plasmablasts and plasma cells (PB/PC) in both spleen and PBMCs (Fig. 7a, b), likely a consequence of uncontrolled viral replication. While spleen germinal center (GC) B cells showed a decreasing trend in control animals (P = 0.057), ZS1C5 treatment preserved GC B cell populations, implying that ZS1C5 may mitigate SFTSV-associated lymphoid follicle disruption^21,22. Cellular immunity also exhibited significant differences between groups: compared to ZS1C5-treated animals, control animals showed elevated levels of effector CD8⁺ T cells (CD8⁺ Teff) in both spleen and PBMCs (Fig. 7c, d) and higher cytotoxicity scores in CD8⁺ Teff, tissue-resident memory T cells (Ttrm), NK cells, and NKT cells (Fig. 7e, f; Supplementary Table S4), indicative of hyperactivated cytotoxic profiles associated with prolonged viremia and tissue damage.

ZS1C5 treatment influenced megakaryocytes, which play essential roles in thrombopoiesis and immune regulation. While megakaryocyte numbers exhibited a decreasing trend in control animals (not statistically significant, Supplementary Fig. S10a), Gene Set Variation Analysis (GSVA) revealed enrichment of antiviral immune response pathways in megakaryocytes from the control group (Supplementary Fig. S10b). These megakaryocytes predominantly functioned as signal senders to other immune cells while receiving weaker signals. In contrast, megakaryocytes from ZS1C5-treated animals primarily acted as signal receivers (Supplementary Fig. S10c, d).

Monocytes, a key immune cell type targeted by SFTSV, exhibited distinct transcriptional profiles between treatment groups. While no significant difference was observed in intermediate monocyte (CD14⁺CD16⁺) levels (Supplementary Fig. S11a, b), which are associated with unresolved inflammation in SFTSV-infected patients²³, GSVA showed that monocytes from ZS1C5-treated macaques were enriched in negative regulators of inflammation, suggesting suppression of inflammatory responses (Fig. 7g). Consistently, these monocytes displayed significantly higher anti-inflammatory transcript signature scores, along with reduced apoptosis and interferon-stimulated gene (ISG) scores (Fig. 7h; Supplementary Table S4). ZS1C5 treatment also altered cell–cell interaction signals in key pathways, including IL-1, TNF, TGF-β, and IL-2, across monocytes, dendritic cells (DCs), and macrophages, further indicating suppression of hyperinflammatory states (Supplementary Figs. S11c–j). Serum cytokine profiling corroborated these findings: control animals exhibited sustained elevations of IFN-γ and IL-1RA, whereas ZS1C5-treated animals restored cytokine levels by Day 3 post-infection. Control animals also displayed dysregulated inflammatory cytokine expression, including suppressed TGF-α, elevated IL-2, and increased levels of Th17-associated cytokines (IL-17α and IL-23) (Fig. 7i; Supplementary Table S3). Collectively, scRNA-seq and cytokine profiling demonstrated that ZS1C5 treatment effectively restored immune homeostasis, suppressed infection-driven hyperinflammation, and preserved critical lymphoid and myeloid interactions during SFTSV infection.

Computation-assisted identification of ZS1C5-like NAbs from human BCR repertoire

Potent NAbs such as ZS1C5 were rare in humans but critical for protection against SFTSV infection. Based on the ZS1C5/Gn structure, we sought to determine whether ZS1C5-like antibodies are present in the human B cell repertoire and whether these antibodies have Gn-binding and SFTSV neutralization functions similar to those of ZS1C5. To systematically identify structural homologs of the inter-domain epitope-targeting antibody ZS1C5, we established a multi-step computational mining pipeline combining sequence analysis, structural modeling, and functional validation (Fig. 8a). From the dataset of 87,600 antibody clonotypes established in this study by single-cell and bulk RNA-seq (Fig. 2a), we prioritized candidates based on genetic signatures — specifically, IGHV3-15/IGHJ4 pairing and CDRH3 lengths of 17 residues or more — reducing the pool to 121 antibodies, a manageable set for detailed computational prediction. We then implemented a scoring system incorporating nine distinct metrics, including AlphaFold2 and RoseTTAFold2, to rank these candidates by their structural similarity to ZS1C5. The normalized average scores generated a comprehensive ranking of the 121 antibodies (Fig. 8b; Supplementary Table S1).

To evaluate the functional relevance of our computational approach, we synthesized, expressed, and tested the top 30 ranked antibodies for SFTSV Gn binding by ELISA. Their binding activities showed a significant correlation with the computational scores (P = 0.0082) (Fig. 8c). Notably, two of these 30 antibodies demonstrated SFTSV neutralizing activity (IC₅₀ = 2.9 and 2.8 μg/mL, respectively) (Fig. 8d). These two NAbs, Ab12842 and Ab38988, ranked 1^st and 5^th among all 121 candidates (Fig. 8b), demonstrating the reliability of our computational scoring system. These results confirm the effectiveness of our structure-guided computational pipeline for high-throughput discovery of epitope-specific NAbs.

Remarkably, the Ab12842 heavy chain is a fully germline-encoded IgM, lacking mutations in the variable (V_H) and joining (J_H) regions, suggesting the presence of ZS1C5-like germline or precursor antibodies capable of SFTSV neutralization. To further investigate this, we mined BCR repertoires from scRNA-seq data of six healthy individuals from the UK²⁴, who are highly unlikely to have been exposed to SFTSV. Applying the same genetic filters (IGHV3-15/IGHJ4 pairing and CDRH3 ≥ 17 residues) and restricting to IgM/IgD antibodies with 0% V_H or J_H mutations (Fig. 8e), we identified 33 candidates from a total of 83,835 heavy chain clonotypes for subsequent synthesis and validation (Supplementary Table S1). Among these 33 germline candidates, one antibody — Ab168b30, a germline-encoded IgD — demonstrated specific binding to SFTSV Gn (Fig. 8f) and dose-dependent neutralization of authentic SFTSV, albeit with modest potency (IC₅₀ = 20.8 μg/mL) (Fig. 8g). The identification of germline-encoded IgM/IgD antibodies with SFTSV neutralizing activity (Fig. 8h) indicates that the human antibody repertoire contains precursors capable of neutralizing SFTSV prior to somatic hypermutation, highlighting the feasibility of eliciting such antibodies through vaccination.

DISCUSSION

The development of effective therapeutics against SFTSV remains an urgent global health priority, given the high mortality rate, lack of approved treatments, and expanding geographic reach. This study identifies ZS1C5, a human NAb with subnanomolar potency, which recognizes a novel conformational epitope spanning Gn Domains I and III. Beyond in vitro ultrapotency, ZS1C5 demonstrated in vivo efficacy, achieving 100% survival in immunodeficient mice and mitigating viremia, thrombocytopenia, and tissue damage in rhesus macaques. Transcriptomic profiling further revealed its capacity to dampen hyperinflammation and restore immune homeostasis. Structural insights from ZS1C5 facilitated the computational identification of germline-encoded neutralizing antibodies, highlighting its dual role as a therapeutic candidate and as a foundation for rational vaccine design.

ZS1C5 employs a novel interdomain neutralizing mechanism, distinct from previously described antibodies that target isolated Gn domains. By simultaneously engaging Domain I, which mediates receptor attachment, and Domain III, which supports Gc fusion loop activity^8,9, ZS1C5 achieves ultrapotent neutralization. This mechanism is facilitated by its long 23-aa CDRH3 loop, which forms a "clamp-like" structure to bridge and disrupt the functional interplay between the two domains. This structural configuration likely disrupts viral entry through two potential mechanisms: 1) steric blockade of Domain I-mediated receptor binding and 2) disruption of Domain III interactions that are necessary for the Gn–Gc heterodimer conformational changes required for membrane targeting. In contrast, single-domain targeting antibodies such as MAb4–5^8,10 (which binds Domain III) or 40C10^13,14 (which binds Domain I) lack this dual-targeting capability and therefore do not exhibit the same dual mechanism of action. Nevertheless, these proposed mechanisms remain speculative and require further experimental validation. Future studies will be essential to confirm this dual-action model, including identifying SFTSV entry receptors, structurally characterizing Gn–Gc heterodimer dynamics during viral entry, and elucidating the role of Gn in the pre- to post-fusion transition.

Conservation analysis of all published SFTSV glycoprotein sequences shows that the interdomain epitope recognized by ZS1C5 is highly conserved across diverse human SFTSV variants, supporting its broad-spectrum neutralizing activity and therapeutic potential. In contrast, ZS336 and ZS65, though targeting overlapping epitopes, are less potent, likely due to shorter CDRH3 loops and reliance on auxiliary light-chain interactions, restricting their inter-domain engagement. Consistent with this, computational modeling using tools such as FoldX and Rosetta indicates that ZS336 and ZS65 form less stable antigen–antibody complexes compared to ZS1C5. Additional factors, such as glycan-mediated steric hindrance (in the case of ZS336) and weaker hydrophobic complementarity (in ZS65), may further reduce their neutralization efficacy. The interdomain targeting strategy employed by ZS1C5 represents a novel paradigm for SFTSV neutralization, with promising implications for the development of pan-phlebovirus therapeutic antibodies and epitope-focused vaccines. Nonetheless, continued surveillance and further studies are important to monitor emerging mutations that could affect antibody efficacy.

Pre-existing immunity to SFTSV remains poorly understood. Our findings revealed that SFTSV-reactive B cells with high V_H mutation rates and IgG subtypes were present not only during the convalescent phase but also during acute infection. Specifically, Gn/Gc-specific BCRs with high V_H mutation rates (> 5%) and IgG isotypes were detected in PBMCs as early as 4–7 days post-symptom onset. Furthermore, analysis of submandibular LN samples from an acute-phase patient identified Gn-specific antibodies, such as 29124-1, with SHM rates of 7.0% in the V_H region. Given the short time frame, it is highly unlikely for B cells to undergo extensive SHM and class-switch to IgG during the acute phase of infection. Instead, these findings strongly suggest a rapid recall and expansion of pre-existing memory B cells, possibly primed by prior, subclinical exposure to SFTSV or antigenically related bunyaviruses. This is particularly evident given that four out of five convalescent donors were over 65 years old and had a history of tick bites.

In parallel, our computational identification of ZS1C5-like germline antibodies demonstrates that the human antibody repertoire harbors genetic precursors with the capacity for SFTSV neutralization, independent of prior antigen exposure. Germline IgM and IgD antibodies, such as Ab12842 and Ab168b30, which lack V_H and J_H mutations, nevertheless exhibited neutralizing activity against SFTSV. While this raises the possibility that the interdomain epitope may be accessible to antibodies without extensive SHM, further structural studies are required to determine whether these germline antibodies indeed engage the same epitope as ZS1C5. These findings have important implications for vaccine design, highlighting the potential to develop immunogens that activate germline antibodies capable of targeting key neutralizing epitopes. Moreover, structure-guided antibody discovery frameworks may be broadly applicable for the rapid identification of therapeutic candidates against emerging viral threats.

The ability of ZS1C5 to confer robust protection in both murine and NHP models highlights its therapeutic potential. In STAT1^–/– mice, a single dose of ZS1C5 achieved 100% survival against lethal SFTSV challenge, with dose-dependent reductions in viral antigen expression and tissue damage. Histopathological analyses confirmed the capacity of ZS1C5 to suppress viral replication and mitigate end-organ injury. In rhesus macaques, which exhibit mild SFTSV infection similar to humans^17,20, ZS1C5 significantly reduced viremia, preserved platelet counts, and alleviated coagulation dysfunction. Despite the high viral dose used for infection (5 × 10⁷ TCID₅₀) that is vastly exceeding natural exposure levels through tick bites, ZS1C5 effectively suppressed symptoms and viral loads, reducing peak viremia by 90% within 24 h and eliminating detectable serum RNA by day 3. Importantly, the use of NHP models ensures the translational relevance of these findings, providing critical insights into the safety and efficacy of ZS1C5 in a biological context closely resembling human physiology.

Single-cell transcriptomic profiling demonstrated that ZS1C5 treatment effectively mitigates SFTSV-induced immune dysregulation, preserving immune homeostasis and restoring essential cellular functions. In alignment with patient data²³, uncontrolled viral replication in the control mAb-treated monkeys led to: 1) expansion of PB/PC; 2) upregulated cytotoxicity signatures in CD8⁺ T, NK and NKT cells; and 3) hyperactivation of monocytes, as evidenced by altered inflammatory-related signatures and inflammatory pathway cell–cell interactions. In contrast, ZS1C5 neutralized the circulating virus rapidly, thereby preventing these dysregulated immune phenotypes. Specifically, ZS1C5 treatment exhibited enrichment of anti-inflammatory pathways and suppression of hyperactivated immune states, which were further validated at the protein level by Luminex assay. In addition to phenotypes paralleling severe human cases, decreased GC B cells and increased CD8⁺ Teff cells were observed in the control group. While intermediate monocyte expansion was noted in human studies, it was only observed in one control monkey, likely due to differences in sampling time points. Interestingly, megakaryocytes, rarely seen in peripheral blood and therefore have not been analyzed in human SFTSV cases, were observed in monkeys. Transcriptional analyses of spleen megakaryocytes revealed functional alterations linked to SFTSV-induced thrombocytopenia and coagulation dysfunction.

While this study highlights ZS1C5 as a promising therapeutic candidate, several limitations remain to be addressed. First, our stringent selection criteria during NAb discovery — requiring long CDRH3 (> 17 aa) and high SHM rates (> 6%) — may have inadvertently excluded potent neutralizing anti-bodies with shorter CDRH3 loops or lower SHM. Although ZS1C5 was identified based on these parameters, additional studies are needed to determine whether antibodies targeting other valuable neutralizing epitopes may also hold therapeutic potential. Second, the efficacy of ZS1C5 in treating late-stage infections remains unclear, as most patients seek medical care several days after symptom onset. Future studies should focus on optimizing treatment regimens and evaluating long-term safety and efficacy in clinical settings. Third, the high viral dose used in the NHP challenge model does not fully replicate natural infection dynamics, where SFTSV transmission typically occurs via tick bites and involves only small amounts of virus. Notably, ZS1C5 demonstrated efficacy even against high-dose viral challenges in macaques, suggesting that under real-world conditions with much lower viral loads, ZS1C5 could be even more effective. This enhanced effectiveness may enable a larger time window for post-exposure treatment, which is critical for addressing delays in medical care often seen in SFTSV patients. Finally, while the computational pipeline for identifying ZS1C5-like antibodies proved reliable, further refinement is needed to improve predictive accuracy and scalability for broader applications. Despite these limitations, ZS1C5 represents a significant advance in therapeutic antibody development for SFTSV. The ability to target a novel interdomain epitope, combined with strong in vivo protective and immunomodulatory effects, provides a solid foundation for clinical translation.

MATERIALS AND METHODS

Human subjects and PBMC isolation

Convalescent blood samples were collected from 4 donors (3 males and 1 female) who had been previously infected with SFTSV between 2016 and 2022. In another group of donors, both acute- and convalescent-phase samples were collected from 12 individuals (3 males and 9 females) who were hospitalized for treating SFTSV infection in the summer of 2023. Peripheral blood was collected during both the acute (Days 4–7 after the onset of the symptoms) and convalescent phases (2 months after discharge) based on informed consent. One of the donors received submandibular LN dissection on Day 5 after the onset of the symptoms to have one of the submandibular LNs removed, from which we obtained LN B cell samples. The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Review Committees of Zhoushan Hospital (2020071) and Westlake University (20230414LDP001). Informed consent to participate was obtained from all individual participants involved in the study.

Fresh anticoagulated whole blood from donors was collected, and PBMCs were isolated using lymphocyte separation tubes (DAKEWE, Cat# 7922112) according to the manufacturer's instructions. Briefly, 10 mL of blood containing EDTA was centrifuged at 1,200× g for 10 min. The upper plasma was transferred to a new tube and stored at –80 °C. The lower cell layer was carefully transferred to a new tube, mixed with an equal amount of D-PBS (Gibco, Cat# C14190500BT), and centrifuged again. The middle layer, consisting of white PBMCs, was carefully transferred to a new tube and washed twice with a large volume of D-PBS containing 2% FBS (ExCellBio, Cat# FSP500) at 300× g for 8 min. Finally, the cells were resuspended in 500 µL of 2 % FBS D-PBS and counted. The entire blood processing procedure was completed in a BSL-2 laboratory at Westlake University (BSL20225712116).

Recombinant protein expression and purification

The codon-optimized SFTSV Gn head domain (aa 20–340, Genbank: AQX34652.1) and Gc head domain (aa 563–996, Genbank: AQX34652.1) were individually cloned into the pCAGGS with N-terminal Kozak sequence and IL-2 signal sequence, and with a C-terminal 10× His-tag. The recombinant glycoproteins were expressed transiently by transfecting Expi293F cells using PEI MAX (Polysciences, Cat# 24765). Five days post-transfection, the supernatant was collected by centrifugation and filtered through a 0.22-µm filter before purification. Cell supernatants were purified using a Ni²⁺ chromatography column (SunResin, Cat# A453201) on the AKTA system.

Antigen-specific B cell sorting

Antigen-specific B cells were sorted from freshly isolated PBMC samples. Specifically, PBMCs were stained with anti-human IgD conjugated FITC (BioLegend, Cat# 348206), anti-human IgM conjugated PerCP-Cy5.5 (BD Biosciences, Cat# 561285), anti-human CD3 conjugated PE-Cy5 (BioLegend, Cat# 300310), anti-human CD27 conjugated PE-Cy7 (BioLegend, Cat# 302838), anti-human CD38 conjugated AF700 (BioLegend, Cat# 303524), anti-human CD19 conjugated APC-Cy7 (Biolegend, Cat# 363010), anti-human CD14 conjugated BV605 (Biolegend, Cat# 301834), anti-human CD16 conjugated BV785 (BioLegend, Cat# 302046), Zombie Aqua (BioLegend, Cat# 423102) for live/dead, at 4 °C for 30 min in the dark. To probe SFTSV-Gn/Gc-specific B cells, the his-tagged Gc and Gn were incubated with PE anti-His (BioLegend, Cat# 362603) and APC anti-His (BioLegend, Cat# 362605). Bait proteins were subsequently incubated with PBMCs in separate tubes. After fluorescent surface staining, the PBMCs were loaded onto a CytoFLEX SRT (Beckman) for analysis and sorting. SFTSV-Gn/Gc-specific B cells were gated as CD14^–CD16^–CD3^–CD19⁺IgM^–IgD^–PE⁺ and CD14^–CD16^–CD3^–CD19⁺IgM^–IgD^–APC⁺, respectively.

For antibody gene PCR and cloning, antigen-specific B cells were single-cell sorted into 96-well PCR plates containing 20 µL per well of lysis buffer (2.5 µL of 10× RT buffer, 2.5 µL of 5× gDNA wiper mix (Vazyme, Cat# R312), 0.0625 µL of IGEPAL (Sigma-Aldrich, Cat# I8896), 0.5 µL of Carrier RNA (Sangon Biotech, Cat# B518271-0100), and 14.4375 µL dH₂O). The plates were briefly centrifuged and then stored at –80 °C. For single-cell transcriptome and BCR-seq, antigen-specific B cells were sorted into D-PBS containing 2% FBS and counted for subsequent sequencing analysis. For bulk RNA-seq, non-antigen-specific B cells were collected into a tube and lysed with 1 mL of Trizol (TaKaRa, Cat# 9109) for RNA extraction and subsequent sequencing analysis.

Single B cell PCR and antibody cloning

The antibody gene PCR and cloning were performed as previously described²⁵. Briefly, cDNA is generated from sorted single B cell RNA by HiScript III 1^st strand cDNA Synthesis Kit (Vazyme, Cat# R312). Correspondingly, HC and LC were amplified with 2 rounds of PCR, where the first-round primer sets target 5’ leader region and immunoglobulin (Ig) constant regions to amplify different Ig fragments, the second-round primer sets perform a nested-PCR for targeted amplification of HC and LC, as well as to incorporate the homologous end of the expression vectors. The resulting PCR products were visualized on 1% agarose gels, and bands at ~500 bp and ~450 bp were sent for sequencing as HC and LC sequences, respectively. The sequenced DNA was analyzed on IMGT/V-QUEST. The identified HC and LC pairs from the second-round PCR were digested and cloned (Vazyme, Cat# C116) into mammalian expression vectors.

Human single-cell transcriptome and BCR-seq

PBMC RNA was extracted, and quality control was performed (Agilent, Cat# 4200). The RNA samples were then analyzed by high-throughput sequencing of IGH with the BCR profiling system at a deep level. Briefly, a 5' RACE unbiased amplification protocol was used. This protocol utilizes unique molecular barcodes (UMBs) introduced during cDNA synthesis to control bottlenecks and minimize PCR and sequencing errors. Sequencing was performed on an Illumina NovaSeq system using the PE150 mode (Illumina, X plus). The UMBs attached to each raw sequence read were applied to correct PCR and sequencing errors and to remove PCR duplicates.

Human sequencing data processing and cell type identification

For the single-cell transcriptome sequencing data obtained from patient PBMCs, we applied the Cell Ranger (v7.2) workflow for the assembly and annotation of the raw sequencing data. The reference genome used was the GRCh38 human reference genome downloaded from the 10× Genomics official website. The expression matrix processed by Cell Ranger was further analyzed using the Seurat R package for downstream dimensionality reduction, clustering, and cell annotation. For the integrated object, cells with nFeature values greater than 200 and less than 8,000, and nCount values less than 70,000, were filtered for subsequent analysis. A total of 45,654 cells were included in the study after filtering. After removing batch effects, we performed dimensionality reduction and clustering on the data. The dims value was set to 1:25, and the resolution value was set to 0.6. We identified the differentially expressed genes in cell clusters using the FindAllMarkers function and annotated cell types and subtypes using both the SingleR and SCSA methods. The marker gene database for SCSA was the human blood marker gene set downloaded from the CellMarker 2.0 database.

Antibody gene sequence and clonal analysis

The antibody sequences involved in this study were derived from single-cell BCR-seq data from 11 samples of 5 patients, bulk BCR-seq data from 3 patient samples, and single-cell BCR-seq data from 1 pooled sample. Single-cell BCR-seq data were processed and annotated using the VDJ module of Cell Ranger (v7.2), with the GRCh38 Human V(D)J reference (GRCh38-alts-ensembl-7.1.0) downloaded from the 10× Genomics website as the reference genome. Bulk BCR-seq data were assembled and annotated using TRUST4 (v1.0.13), with the reference genome provided by TRUST4 (hg38_bcrtcr.fa) and the detailed V/D/J/C gene reference file downloaded from the IMGT database.

The identification of antibody clonal lineages was performed using the Change-O workflow. The workflow employed the IMGT human IG reference for processes such as IgBLAST and germline reconstruction. For single-cell BCR data, annotations from the Cell Ranger VDJ module were additionally incorporated. During clone definition, nucleotide Hamming distance was used as the substitution model for calculating sequence distances. The clone definition distance threshold was determined using the distToNearest function in the SHazaM R package. Ultimately, antibodies sharing the same V and J genes and with inter-sequence distances of less than 0.1100311 were identified as belonging to the same clone.

Production of monoclonal antibodies

For antibody plasmids derived from single B cell PCR and cloning, purified PCR products were used for overlapping PCR to generate linear human IgG expression cassettes as previously described²⁵. The expression cassettes were transfected into 3 mL Expi293F cells using PEI MAX. The supernatant samples containing recombinant IgGs were used for IgG quantification and preliminary ELISA binding screening. For antibody genes obtained through single-cell BCR-seq, HC and LC genes were synthesized and cloned into mammalian expression vectors. Plasmids were transfected into 10 mL Expi293F cells, harvested after 4 days, and then purified using protein A magnetic bead (GenScript, Cat# L00695) enrichment for preliminary ELISA binding screening and neutralization assay. Antibodies that showed positive ELISA results and visibly reduced virus foci were selected for large-scale production. To produce antibodies JK-8¹⁶, MAb4-5^8,10, 40C10^13,14, S2A5¹², and SF5¹⁵, the HC and LC variable regions were synthesized and cloned into human IgG expression cassettes for subsequent large-scale production.

For large-scale production of the down-selected antibodies, HC and LC plasmids were expressed transiently by transfecting Expi293F cells with a 1:1 ratio using PEI MAX. After 5 days, the supernatant was collected by centrifugation and filtered through a 0.22-µm filter before purification. Cell supernatants were purified using protein A affinity chromatography column (SunResin, Cat# A4093101) on the AKTA system. This was eluted with 100 mM glycine at pH 2.5 and neutralized with 1/10 volume of 1 M Tris-HCl. The eluted mAbs were buffer-exchanged into PBS.

ELISA

To assess antibody–antigen binding by ELISA, 96-well high-binding ELISA plates were coated overnight with specific glycoproteins at 0.5 µg/mL. Plates were washed three times in wash buffer (PBST, 0.1% Tween-20 in D-PBS) and blocked with 100 µL/well of blocking buffer (3% BSA in PBS) for 2 h at 37 °C. Plates were washed with wash buffer, and antibodies diluted to 5 µg/mL in blocking buffer were added to the first well, followed by 3-fold serial dilutions across subsequent wells. Following 2 h incubation, the plates were washed, and a 1:5,000 dilution of anti-human IgG-HRP (Promega, Cat# W4031) was added and incubated for 1 h. The plates were washed three times in the washing buffer. TMB Chromogen solution (Beyotime, Cat# P0209) was added at 100 µL per well and developed for 10 min at room temperature in the dark. After which, the reaction was stopped with 50 µL 1M sulfuric acid per well. Absorbance was measured at 450 nm. All mAbs were tested in duplicates.

BLI

BLI kinetic measurements were acquired at 30 °C with an Octet protein analysis system (Sartorius, Octet R8). For binding kinetics, antibodies (Basic buffer: 0.1% BSA in D-PBS) were captured with Protein A sensors (Sartorius, Cat# 18-5010) for 300 s as loading. After loading, the baseline was recorded for 200 s in the basic buffer before association. For association, the sensors were immersed in the antigen solution with serial 3-fold dilutions for 300 s. Following that, the sensors were dipped into a basic buffer for 600 s to facilitate dissociation. The sensors were regenerated with Glycine-HCl to start a new round of binding. The binding curves were fit to a 1:1 binding model using the Octect Analysis Studio 13.0.2.46.

Cross-blocking BLI was established by loading the Gn/Gc protein onto HIS1K biosensors (Sartorius, Cat# 18-5120), followed by association with the first antibody. Subsequently, the antibody–antigen complexes were incubated with the same or a second antibody. The experimental conditions were the same as previously described. The shifted binding signal between the second mAb and the first, resulting from different binding domains, was captured and reported as RLU. Data were analyzed using the Epitope Binning model in Octet Analysis Studio.

Neutralization assay

Antibodies were diluted to a starting concentration of 1 µg/well, followed by serial 2-fold dilutions. Authentic SFTSV (WCH/97/HN/China/2011 Strain) at 10² TCID₅₀/well was incubated with diluted antibodies for 1 h at 37 °C. 1.5 × 10⁶ cells/well of Vero cells were added to the antibody-virus mixture in a 96-well plate and incubated for 48 h at 37 °C. To perform immunofluorescence, the infectious media were discarded, and cells were fixed and permeabilized with pre-cooled 80% acetone at room temperature for 30 min. Plates were then washed twice with PBST, followed by incubation with the primary antibody, anti-SFTSV-NP polyclonal antibody (1:200), for 2 h at 37 °C. Then, the plates were washed and incubated with 2,000-fold diluted goat anti-rabbit IgG, conjugated to AF488 (Thermo Fisher Scientific, Cat# A-11008), for 1 h at 37 °C in the dark. Infected cells were visualized under the fluorescence microscope. IC₅₀ values for each antibody were determined from log response curves with variable slopes and four parameters using nonlinear regression analysis (GraphPad Prism 10).

Cryo-EM sample preparation and data collection

The Gn was incubated with antibodies at a molar ratio of about 1:3 for 2 h. Then the mixture was subject to size-exclusion chromatography (Cytiva, Superose Increase 10/300 GL) in buffer containing 25 mM Tris (pH 8.0), 150 mM NaCl. The peak fractions were collected and concentrated for EM analysis. The complexes of Gn bound with ZS1C5, 336 and 65 were concentrated to ~3.5 mg/mL and applied to the grids, respectively. Aliquots of the protein (3.5 μL) were deposited on glow-discharged holey carbon grids (Quantifoil Au R1.2/1.3). The grids were blotted for a period of 4 s and flash-frozen in liquid ethane cooled by liquid nitrogen with a Vitrobot (Thermo Fisher Scientific, Mark IV). The prepared grids were transferred to a Titan Krios operating at 300 kV, which is equipped with a Gatan K3 detector and a GIF Quantum energy filter. The movie stacks were automatically collected using the EPU software (Thermo Fisher Scientific), with a slit width of 20 eV on the energy filter and a defocus range from ~1.2 µm to ~2.2 µm in super-resolution mode at a nominal magnification of 81,000×. Each stack was exposed for 2.56 s, with an exposure time of 0.08 s per frame, resulting in a total of 32 frames per stack. The total dose rate was ~ 50 e^–/Å² for each stack.

Cryo-EM data processing

The movie stacks were motion corrected with MotionCor2²⁶ and binned twofold, resulting in a pixel size of 1.087 Å/pixel. Meanwhile, dose weighting was performed²⁷. Then the preprocessing movie stacks were imported to CryoSPARC (v4.6.0, Structure Biotechnology) for CTF estimation and downstream processing^28,29. Particles of Gn–Abs complex were automatically picked using CryoSPARC²⁹. After 2D classification, the particles with clear secondary structure features were selected and subjected to ab initio reconstruction to obtain the initial models, then multi-heterogeneous refinement without symmetry was performed to select good particles using CryoSPARC. The selected particles were subjected to non-uniform refinement, global CTF refinement and local refinement with C1 symmetry, resulting in the 3D reconstruction of the whole structure. However, the ZS65 complex failed to yield a high-resolution density map after the above-mentioned processing steps. Therefore, the obtained particles were proceeded to transfer to Relion^30-33 for further 3D classification. Finally, the best classes were selected and subjected to 3D refinement to get a superior overall map. The resolution was estimated with the gold-standard Fourier shell correlation 0.143 criterion³⁴ with high-resolution noise substitution³⁵. Refer to Supplementary Table S2 for details of data collection and processing.

Model building and structure refinement

In the construction of the ZS1C5 model, the predicted atomic model generated by AlphaFold 2^36,37 was employed as a template. That was then subjected to molecular dynamics flexible fitting³⁸ and manual adjustment with Coot³⁹ to create the atomic model of the antibody in its various states. Each residue was manually checked with the chemical properties taken into consideration during model building. For the complex, the model building was performed on the basis of the focused refinement. For the rGn, the atomic model (PDB ID: 5Y10) was used as a template. Each residue was manually checked against the chemical properties considered during model building. Several segments whose corresponding densities were not visible were not modelled. Structural refinement was performed in Phenix⁴⁰ with secondary structure and geometry constraints to prevent overfitting. To monitor for potential overfitting, the model was refined against one of the two independent half maps from the gold standard 3D refinement approach. The refined model was then tested against the other map. Statistics associated with data collection, 3D reconstruction, and model building of other antigen-antibody complex structures are provided in Supplementary Table S2.

Conservation analysis

All 1,331 SFTSV M segment sequences used for conservation analysis were downloaded from the NCBI GenBank database (Taxonomy IDs:1003835 and 2748958), and screened to ensure sequence integrity and accurate translation. The aa sequences for the Gn protein were derived from the nucleotide sequences, with residues 23–340 (matching the structural model) extracted for further analysis.

Phylogenetic trees were constructed using the Neighbor-Joining method implemented in MEGA X, based on the 1,331 complete M segment sequences. Tree confidence was assessed with 1,000 bootstrap replications. Evolutionary distances were computed with the Kimura 2-parameter method and are expressed as the number of base substitutions per site.

For conservation analysis, conservation scores were calculated using Jalview⁴¹, with scores ranging from 0 (non-conserved) to 11 (highly conserved). A custom Python script assigned these conservation scores to the B-factor field of the Gn structure file, and epitope conservation was visualized using ChimeraX.

Energy calculation

We evaluated the structural stability of the antigen–antibody complex using both FoldX⁴² and Rosetta⁴³. FoldX computes an empirical folding free energy in kcal/mol using a fast physics-based energy function, while Rosetta reports a total energy score in Rosetta Energy Units (REU) derived from a combination of physical and statistical potentials. Although the units are not directly comparable, both methods provide relative metrics for structural stability and were used for cross-validation.

To perform the energy calculation, all antigen–antibody complexes were truncated to ensure consistent antigen sequence length (residues 23–340 of Gn), and antibodies were trimmed to retain only the variable region. For FoldX, the input structures were first preprocessed using the RepairPDB command to correct structural errors. Then, the Stability command was used to calculate the folding free energy (ΔG_folding) of the entire complex. For Rosetta, we employed the rosetta_relax to optimize each complex structure and evaluate its total energy score. Each structure was subjected to three independent runs in parallel.

Computational identification of structurally related antibodies

We employed multiple methods to assess the structural similarity between candidate antibodies and the target antibody. The crystal structures of all candidate antibodies were predicted as monomeric structures using AlphaFold v2.3 and RoseTTAFold v2. Structural similarity searches between candidate antibodies and the target antibody were conducted using the align function in PyMOL v2.5.7 and the easy-search module in Foldseek v8.ef4e960.

To evaluate the binding capability of candidate antibodies to the target antigen epitope, we used three docking prediction tools: ZDOCK v3.0.2, HDOCK v1.1, and FTDock v2.0. ZDOCK and HDOCK were utilized to directly assess binding to the antigen epitope, while FTDock employed SCscore to evaluate the similarity between candidate and target antibodies.

The final scoring for candidate antibodies comprised nine components: the mean pLDDT of AlphaFold2 monomeric predictions, the mean pLDDT of RoseTTAFold2 monomeric predictions, HDOCK scores for RoseTTAFold2-predicted structures, HDOCK scores for AlphaFold2-predicted structures, ZDOCK scores for RoseTTAFold2-predicted structures, ZDOCK scores for AlphaFold2-predicted structures, RMSD values from PyMOL align, Pidentity scores from Foldseek easy-search, and SCscore from FTDock. Each of the nine scores was normalized to a 1–10 scale, and the final candidate antibody score was calculated as the average of these normalized values.

Mouse challenge

The dose-dependent efficacy of ZS1C5 was evaluated in STAT1^–/– mice (Jax# 012606, B6.129S(Cg)-Stat1^tm1Dlv/J; n = 10 per group) infected intraperitoneally with 10 LD₅₀ of SFTSV (WCH/97/HN/China/2011 Strain). Three doses of ZS1C5 (high: 500 µg, medium: 50 µg, low: 5 µg) were administeredintraperitoneally 24 h post-infection, alongside a control human antibody⁴⁴ (500 µg). Survival rates, body weight changes, and tissue pathology (liver, spleen, lung, kidney) were monitored for 18 days. At the end of the study, mice were euthanized, and tissues were fixed in 4% paraformaldehyde for histopathological analysis using H&E staining and IHC staining with a rabbit anti-SFTSV NP polyclonal antibody (developed in-house). Representative tissue sections were analyzed using ImageJ software. For H&E sections, the healthy tissue area ratio was quantified using the color deconvolution plugin in "vectors=H&E" mode. For IHC sections, the infected tissue area ratio was determined by splitting channels, isolating the layer specifically recognized by anti-SFTSV NP polyclonal antibodies, and performing quantitative analysis.

A separate study was conducted to evaluate time-dependent efficacy. ZS1C5 (500 µg) was administered at 48, 72, and 96 h post-infection with 10 LD₅₀ SFTSV. All experiments were conducted in an ABSL-2 facility, where mice were provided with adequate food and water throughout the experiment. Ethical approval for the animal challenge study was obtained from the Biosafety Committee of Westlake University (AP#22-069-2-LDP).

NHP challenge

The SFTSV HBMC5 strain (genotype C3, GenBank ID: KY440769-KY440771), purified through two rounds of plaque isolation, was used to infect eight male rhesus monkeys (Macaca mulatta) aged 3–5 years (4–6 kg). Animals were randomly divided into two groups and administered multiple subcutaneous injections of 5 × 10⁷ TCID₅₀ SFTSV in the hind limb inner surface. 6 h post-infection, monkeys received intravenous injections of either 30 mg/kg ZS1C5 (test antibody) or S2P6⁴⁵ (control anti-SARS-CoV-2 spike antibody) at 2 mL/min. Both antibodies were produced via transient transfection in Expi293F cells (10 L fermentation scale) and subjected to endotoxin adsorption (Thermo Fisher Scientific, Cat# 88274). Endotoxin levels were confirmed below 5 EU/kg/h during infusion using a chromogenic assay (Thermo Fisher Scientific, Cat# A39552).

Clinical scores were obtained by trained observers who were blinded to the vaccine groups from 0 to 7 days post-infection. Two observers recorded symptoms for 30 min each morning for each group. Clinical symptoms included spontaneous movement, feeding behavior, coat hair, muscle stiffness, urine color, faeces, reaction to tactile stimulation, and abdominal palpation. These indicators are further categorized into two levels: Healthy (0 point) and Sick (1 point). Scores are assigned based on actual observations. The sum of all scores is the total score. Specific evaluation criteria and symptom severity grading are detailed in Supplementary Table S3.

Body weight and temperature were monitored daily. Blood samples were collected continuously from 0–7 days post-infection for hematological profiling, viral load quantification, and serum-based analyses (biochemistry, coagulation, ELISA, neutralization assays, and cytokine Luminex assays). All procedures, including blood sampling, body weight, temperature measurement and euthanasia, were conducted under respiratory anesthesia.

Euthanasia was performed at 7 day post-infection, followed by complete necropsy. Heart, liver, spleen, lungs, kidneys, brain, pancreas, rectum, duodenum, axillary LNs, and inguinal LNs were collected for H&E staining, IHC staining, and viral load quantification.

The NHP experiments were conducted at the National Kunming High-level Biosafety Primate Research Center (ABSL-3). All rhesus macaques had a microbial background that met the national standard (GB 14922-2022). During the breeding and experimental periods, all animals were provided with readily accessible and adequate food, water, and veterinary care. The experiments were approved by the Ethics Committee of the Chinese Academy of Medical Sciences and Peking Union Medical College (DWSP202410014).

CBC, Basic metabolic panel (BMP), and coagulation test

During the entire monitoring period, from 0 to 7 days post-infection, rhesus monkeys were subjected to daily respiratory anesthesia for the collection of peripheral blood. Serum was collected using clot activator tubes and centrifuged at 3,000 rpm for 30 min at 4 °C. Whole blood was collected using EDTA anticoagulant tubes, and plasma samples were collected after additional centrifugation to obtain the supernatant. CBC were performed using a five-part differential veterinary automatic hematology analyzer (Mindray, BC-5000Vet) with fresh whole blood collected daily. The parameters tested included total white blood cell count (WBC), absolute neutrophil and lymphocyte counts (Neu#, Lym#), percentages of neutrophils, lymphocytes, and monocytes (Neu%, Lym%, Mon%), red blood cell count (RBC), hemoglobin (HGB), hematocrit (HCT), mean corpuscular hemoglobin concentration (MCHC), PLT, and PCT.

Serum samples were used for BMP testing and underwent heat inactivation (56 °C for 30 min) before testing. BMP testing was performed using an automated biochemical analyzer for animals (Zybio, Cat# EXC-2000), with 39 test indicators: total protein (TP), albumin (ALB), globulin (GLB), albumin/globulin ratio (ALB/GLB), alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), total bilirubin (TBIL), direct bilirubin (DBIL), indirect bilirubin (IBIL), gamma-glutamyl transferase (GGT), uric acid (UA), creatinine (CREA), urea, triglycerides (TG), total cholesterol (CHOL), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), apolipoprotein A1 (ApoA1), apolipoprotein B (ApoB), glucose (GLU), creatine kinase (CK), CK-MB, lactate dehydrogenase (LDH), α-hydroxybutyrate dehydrogenase (α-HBDH), C-reactive protein (CRP), complement C3, complement C4, immunoglobulin A (IgA), immunoglobulin M (IgM), immunoglobulin G (IgG), lipoprotein (a) (Lp(a)), phosphorus (P), magnesium (Mg), calcium (Ca), iron (Fe), α-amylase (α-AMY), lipase (LPS), and ferritin (Fer).

Coagulation tests were conducted to characterize platelet function by monitoring APTT, prothrombin time (PT), fibrinogen (FIB), and thrombin time. The fully automated coagulation analyzer was calibrated and quality-controlled by the manufacturer's recommended protocols (Sysmex, Cat# CA-620). Fresh plasma samples were collected daily for testing.

Histopathology and IHC

On the seventh day post-infection, all rhesus macaques were euthanized. The post-mortem examination was conducted on the heart, liver, spleen, lung, kidney, pancreas, brain, inguinal LNs, axillary LNs, duodenum, and rectal tissues. During the collection of lung tissues, the right and left lobes were distinguished and further divided into upper, middle, and lower lobes. All tissues were immersed in a 4% paraformaldehyde solution (Biosharp, Cat# BL539A) for 24 h to complete fixation. After fixation, the tissues were removed and trimmed to a uniform size under a fume hood.

For tissue embedding: embedding frames containing tissue were placed into the cassette and dehydrated using a gradient of ethanol. In 75% ethanol for 2 h, 85% ethanol for 2 h, 95% ethanol for 3 h, absolute ethanol for 2 h, xylene for 2 h, and melted paraffin for 2 h. The wax-soaked tissue was embedded by using a tissue embedder. Finally, the mold with tissue inside was covered by an embedding frame and then moved to the –20 °C freezing table for cooling. The paraffin block was removed from the mold after the paraffin solidified. The trimmed paraffin block was placed on the microtome to be sliced at a thickness of 4 μm. Sections were floated on 42 °C distilled water to be flattened, and were picked up vertically by using a glass slide. After the tissue was dried, sections were placed in an incubator at 60 °C for 30 min to 1 h. Sections were removed from the incubator and stored at room temperature.

For H&E staining, tissue sections were immersed in xylene for 10 min. Tissue sections were immersed in progressively more dilute ethanol solutions. Then, they rehydrated in the following sequence: absolute ethanol for 5 min, 95% ethanol for 5 min, 85% ethanol for 5 min, and 75% ethanol for 5 min. The sections were rinsed with distilled water for 1 min. They were stained with a hematoxylin solution (BIOSSCI, Cat# BP0211) for 4 min and then washed with distilled water for 2 min until no excess dye was visible on the sections. The Sections were differentiated with 0.8 % hydrochloric acid alcohol for 2 s. They were stained with an eosin solution (alcohol-soluble) for 20 s without water washing. Then, the sections were treated with 95% ethanol for 5 min and dehydrated with absolute ethanol for 4 min. The tissue sections were transparent after being treated with xylene and then mounted with neutral balsam.

For IHC staining, the sections were deparaffinized in an environmentally friendly dewaxing agent for 10 min, followed by hydration through absolute ethanol, 95% ethanol, and 75% ethanol, each for 5 min. The sections were then washed three times with distilled water for 3 min each. Antigen retrieval was performed using high-pressure and high-temperature EDTA (BIOSSCI, Cat# BP0610). After rinsing the sections with distilled water, they were immersed in a 3 % (v/v) H₂O₂ solution (SCR, Cat# 10011218) and incubated at room temperature in the dark for 30 min to block endogenous peroxidase activity. The sections were marked with a histochemical pen and then incubated with normal goat serum (BOSTER, Cat# AR1009) at room temperature for 30 min to block non-specific binding sites. The primary antibody used was an anti-SFTSV-N polyclonal antibody (1:2,000), which was incubated overnight at 4 °C. The following day, the sections were brought to room temperature and washed three times with TBST. The secondary antibody was an HRP-conjugated rabbit polyclonal antibody (1:5,000, Promega, Cat# W4011), which was incubated at 37 °C for 45 min, followed by three washes with TBST. Each section was then incubated with 100 µL of DAB solution (MXB Biotechnologies, Cat# Kit-0038) at room temperature for 5 min to develop color, and the reaction was terminated by rinsing with tap water.

All sections were scanned using a digital pathology system (HAMAMATSU, NanoZoomer S360). The software NDP.View 2 was used for magnification observation. ImageJ software evaluates the sections method in the same way as the mouse challenge.

Luminex assay

All rhesus macaques were euthanized on the 7 day post-infection. Serum collected from 0–7 days post-infection was used for cytokine profile analysis. The serum was obtained by centrifuging whole blood in the clot activator tube at 4 °C, 3,000 rpm for 10 min. For cytokine profiling, undergone heat inactivation at 56 °C 30 min serum samples were measured using a 22-analyte multiplex bead array (Millipore, Cat# PCYTMG-40K-PX22) including IL-2, IL-6, IL-17α, GM-CSF (1.22 –5,000 pg/mL); IFN-γ, VEGF-α, IL-10 (2.44 –10,000 pg/mL); G-CSF, IL-1RA, TNF-α (4.88–20,000 pg/mL); IL-8 (0.05–200 pg/mL); IL-5 (2.93–12,000 pg/mL); IL-4 (12.21–50,000 pg/mL); IL-1β (0.98–4,000 pg/mL); IL-23 (19.53–80,000 pg/mL); IL-15 (6.10–25,000 pg/mL); IL-18 (19.53–80,000 pg/mL); MIP-1α (0.73–3,000); MIP-1β (6.10–25,000 pg/mL); MCP-1 (3.05–12,500 pg/mL); sCD40L (48.83–200,000 pg/mL); TGF-α (0.49–2,000 pg/mL). Samples were prepared according to the manufacturer's instructions, and the Mean Fluorescent Intensity was collected using the Bio-Plex 200 suspension array system, followed by standard curve fitting. Each test well was measured 50 times. The doublet discriminator parameter was set to 8000.

Viral RNA extraction and quantification

SFTSV viral RNA was detected using quantitative polymerase chain reaction (qPCR). Viral RNA was extracted from peripheral blood and Trizol homogenates (Thermo Fisher Scientific, Cat# 10296028) of heart, liver, spleen, lung, kidney, pancreas, brain, inguinal LNs, axillary LNs, duodenum, and rectal tissues using the high-throughput automatic nucleic acid extraction system (Thermo Fisher Scientific, KingFisher Flex). The weight is recorded when taking tissue samples, while the volume is recorded for peripheral blood. A one-step method (Thermo Fisher Scientific, Cat# 4444434) was used for viral RNA quantification, with the SFTSV M segment forward primer: 5'-CAG TGC TAC CCT GCA AAG AA-3', reverse primer: 5'-TGA TGG CAA ACA TTA GCT TC-3', and probe: 5'-FAM/TCA TCC TCC TTG GAT ATG CAG GCC TCA/BHQ1-3'. The qPCR was carried out on a CFX384 Touch Real-Time PCR Detection System (BioRad) using the following thermal cycler parameters: 25 °C for 2 min, heat to 50 °C, hold for 15 min, heat to 95 °C, hold for 2 min, then the following parameters are repeated for 40 cycles: heat to 95 °C, hold for 5 s, cool to 60 °C (collect fluorescence signals) and hold for 31 s. The reaction system consists of 20 µL, including 1 µL of each primer and probe, 5 µL of one-step polymerase, 5 µL of the RNA sample, and 7 µL of nuclease-free water. RNA copies per reaction were interpolated using quantification cycle data and a serial dilution of a highly characterized custom RNA containing the SFTSV M segment sequence.

NHP single-cell sequencing

Single-cell transcriptome libraries were generated following the manufacturer's protocol for the Chromium Next GEM Single Cell 3' Kit (v3.1, 10× Genomics, Cat# 1000268). Briefly, cellular suspensions containing between 10,000 and 20,000 cells per sample were partitioned into Gel Bead-In-EMulsions (GEMs) using the 10X Chromium Controller. Within each GEM, reverse transcription was performed using oligonucleotides that incorporated template-switching functionality and 16-base pair unique molecular identifiers (UMIs). These UMIs were employed to uniquely tag individual mRNA molecules, enabling precise quantification and mitigation of amplification biases and PCR duplicates during downstream analysis.

The resulting cDNA was amplified for 14 PCR cycles, a number optimized to preserve library complexity while minimizing the generation of duplicate reads. Final sequencing libraries were normalized to a concentration of 2 nM and pooled in equimolar amounts. Paired-end sequencing (2 × 150 bp) was performed on an Illumina NovaSeq 6000 (Illumina) instrument equipped with S4 flow cells, yielding a target depth exceeding 50,000 reads per cell. Post-sequencing, the embedded UMIs were used to correct PCR and sequencing errors computationally and to collapse PCR duplicates, ensuring accurate digital gene expression counting.

NHP sequencing data processing and cell type identification

For the single-cell transcriptome sequencing data obtained from 8 rhesus macaque PBMCs and spleens, we applied the Cell Ranger (v7.2) workflow for the assembly and annotation of the raw sequencing data. The reference genome used was the Mmul_10 (GCF_003339765.1) rhesus macaque reference genome downloaded from GenBank. The expression matrix processed by Cell Ranger was further analyzed using the Seurat R package for downstream dimensionality reduction, clustering, and cell annotation. For the integrated object, cells with nFeature greater than 200 and less than 8,000, and nCount less than 90,000 were filtered for subsequent analysis. A total of 98,412 PBMC cells and 65,406 spleen cells were included in the study after filtering. After removing batch effects, we performed dimensionality reduction and clustering on the data. We calculated gene expression across all clusters using the FindAllMarkers and AverageExpression functions, and annotated cell types and subtypes through a hybrid approach combining manual annotation with automated tools. Automated annotation methods included SingleR and SCSA.

Enrichment analysis of differential genes

We used the GSVA package for GSVA and the limma package for differential calculation. The gene set database used for GSVA comprised all gene sets under the BP (Biological Process) subontology of the GO database. For immune-related GSVA differences of interest, we filtered these gene sets using keyword-based criteria. Immune-related gene sets obtained from this filtering were further analyzed to identify differential immune pathway activities.

Intercellular communication analysis

We used the CellChat package to analyze intercellular communication with the "Secreted Signaling" interaction database. We selected "RNA" as the computational assay, set k.min to 10 and nboot to 100 to calculate communication probabilities, and applied min.cell = 10 to filter interactions. Cell–cell communication analyses were performed at multiple classification hierarchies to observe signaling states across different resolutions. For prioritized signaling pathways, we conducted detailed interaction analyses and further examined expression patterns of ligand–receptor pair genes.

Inflammatory scoring analysis

We evaluated cellular-level activity of four immune responses, cytotoxicity, negative_inflammatory, apoptosis, and ISG, using the AddModuleScore function. Reference gene sets for these immune responses are detailed in Supplementary Table S4.

Quantification and statistical analysis

Data visualization was performed using the R package ggplot2 and GraphPad Prism 10. For inter-group comparisons between experimental and control groups, statistical analyses employed paired t-tests and the limma method, selected based on data type and analytical objectives. Spearman correlation analysis was applied for association testing.

DATA AVAILABILITY

The data and codes that support the findings of this study are available from the corresponding authors on request. The structures of ZS1C5-Gn, ZS336-Gn, ZS65-Gn were deposited to the Protein Data Bank (PDB) with accession numbers 9LOE, 9LOF, and 9LOG, respectively.

[1]	Yu, X.J. et al. Fever with thrombocytopenia associated with a novel bunyavirus in China. N. Engl. J. Med. 364, 1523–1532 (2011).

[2]	Liu, Y. et al. Person-to-person transmission of severe fever with thrombocytopenia syndrome virus. Vector Borne Zoonotic Dis. 12, 156–160 (2012).

[3]	Wang, M., Tan, W.L., Li, J., Fang, L.Q. & Yue, M. The endless wars: severe fever with thrombocytopenia syndrome virus, host immune and genetic factors. Front. Cell. Infect. Microbiol. 12, 808098 (2022).

[4]	Li, H. et al. Epidemiological and clinical features of laboratory-diagnosed severe fever with thrombocytopenia syndrome in China, 2011–17: a prospective observational study. Lancet Infect. Dis. 18, 1127–1137 (2018).

[5]	Cui, H.L. et al. Global epidemiology of severe fever with thrombocytopenia syndrome virus in human and animals: a systematic review and meta-analysis. Lancet Reg. Health West. Pac. 48, 101133 (2024).

[6]	Liu, Z.Y. & Wang, Z.D. Northeastern China, a hotspot for tick-borne viral diseases. Zoonoses 5, 12 (2025).

[7]	Seo, J.W., Kim, D., Yun, N. & Kim, D.M. Clinical update of severe fever with thrombocytopenia syndrome. Viruses 13, 1213 (2021).

[8]	Wu, Y. et al. Structures of phlebovirus glycoprotein Gn and identification of a neutralizing antibody epitope. Proc. Natl. Acad. Sci. USA 114, E7564–E7573 (2017).

[9]	Du, S.W. et al. Cryo-EM structure of severe fever with thrombocytopenia syndrome virus. Nat. Commun. 14, 6333 (2023).

[10]	Guo, X.L. et al. Human antibody neutralizes severe Fever with thrombocytopenia syndrome virus, an emerging hemorrhagic Fever virus. Clin. Vaccine Immunol. 20, 1426–1432 (2013).

[11]	Kim, K.H. et al. An anti-Gn glycoprotein antibody from a convalescent patient potently inhibits the infection of severe fever with thrombocytopenia syndrome virus. PLoS Pathog. 15, e1007375 (2019).

[12]	Ren, X.X. et al. A broadly protective antibody targeting glycoprotein Gn inhibits severe fever with thrombocytopenia syndrome virus infection. Nat. Commun. 15, 7009 (2024).

[13]	Wu, X.L. et al. Identification and characterization of three monoclonal antibodies targeting the SFTSV glycoprotein and displaying a broad spectrum recognition of SFTSV-related viruses. PLoS Negl. Trop. Dis. 18, e0012216 (2024).

[14]	Yang, P.Y. et al. Molecular mechanism and structure-guided humanization of a broadly neutralizing antibody against SFTSV. PLoS Pathog. 20, e1012550 (2024).

[15]	Chang, Z. et al. Bispecific antibodies targeting two glycoproteins on SFTSV exhibit synergistic neutralization and protection in a mouse model. Proc. Natl. Acad. Sci. USA 121, e2400163121 (2024).

[16]	Zhang, S. et al. Discovery and characterization of potent broadly neutralizing antibodies from human survivors of severe fever with thrombocytopenia syndrome. eBioMedicine 111, 105481 (2025).

[17]	Li, Y.H. et al. Longitudinal analysis of immunocyte responses and inflammatory cytokine profiles in SFTSV-infected rhesus macaques. Front. Immunol. 14, 1143796 (2023).

[18]	DeKosky, B.J. et al. In-depth determination and analysis of the human paired heavy- and light-chain antibody repertoire. Nat. Med. 21, 86–91 (2015).

[19]	DeKosky, B.J. et al. Large-scale sequence and structural comparisons of human naive and antigen-experienced antibody repertoires. Proc. Natl. Acad. Sci. USA 113, E2636–E2645 (2016).

[20]	Jin, C. et al. SFTS virus infection in nonhuman primates. J. Infect. Dis. 211, 915–925 (2015).

[21]	Sakai, Y. et al. Lethal severe fever with thrombocytopenia syndrome virus infection causes systemic germinal centre failure and massive T cell apoptosis in cats. Front. Microbiol. 14, 1333946 (2024).

[22]	Kim, D., Lai, C.J., Cha, I. & Jung, J.U. Current progress of Severe Fever with Thrombocytopenia Syndrome Virus (SFTSV) vaccine development. Viruses 16, 128 (2024).

[23]	Li, H. et al. Single-cell landscape of peripheral immune responses to fatal SFTS. Cell Rep. 37, 110039 (2021).

[24]	Lê Quý K. et al. Benchmarking and integrating human B-cell receptor genomic and antibody proteomic profiling. NPJ Syst. Biol. Appl. 10, 73 (2024).

[25]	Li, D.P. et al. In vitro and in vivo functions of SARS-CoV-2 infection-enhancing and neutralizing antibodies. Cell 184, 4203–4219.e32 (2021).

[26]	Zheng, S.Q. et al. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat. Methods 14, 331–332 (2017).

[27]	Grant, T. & Grigorieff, N. Measuring the optimal exposure for single particle cryo-EM using a 2.6 Å reconstruction of rotavirus VP6. Elife 4, e06980 (2015).

[28]	Zhang, K. Gctf: real-time CTF determination and correction. J. Struct. Biol. 193, 1–12 (2016).

[29]	Punjani, A., Rubinstein, J.L., Fleet, D.J. & Brubaker, M.A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14,, 290–296 (2017).

[30]	Kimanius, D., Forsberg, B.O., Scheres, S.H. & Lindahl, E. Accelerated cryo-EM structure determination with parallelisation using GPUs in RELION-2. Elife 5, e18722 (2016).

[31]	Scheres, S.H.W. A Bayesian view on cryo-EM structure determination. J. Mol. Biol. 415, 406–418 (2012).

[32]	Scheres, S.H.W. RELION: implementation of a Bayesian approach to cryo-EM structure determination. J. Struct. Biol. 180, 519–530 (2012).

[33]	Scheres, S.H.W. Semi-automated selection of cryo-EM particles in RELION-1.3. J. Struct. Biol. 189, 114–122 (2015).

[34]	Rosenthal, P.B. & Henderson, R. Optimal determination of particle orientation, absolute hand, and contrast loss in single-particle electron cryomicroscopy. J. Mol. Biol. 333, 721–745 (2003).

[35]	Chen, S.X. et al. High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopy. Ultramicroscopy 135, 24–35 (2013).

[36]	Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).

[37]	Varadi, M. et al. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res. 50, D439–D444 (2022).

[38]	Trabuco, L.G., Villa, E., Mitra, K., Frank, J. & Schulten, K. Flexible fitting of atomic structures into electron microscopy maps using molecular dynamics. Structure 16, 673–683 (2008).

[39]	Bruhn, J.F. et al. Crystal structure of the Marburg virus VP35 oligomerization domain. J. Virol. 91, e01085–16 (2017).

[40]	Adams, P.D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213–221 (2010).

[41]	Waterhouse, A.M., Procter, J.B., Martin, D.M.A., Clamp, M. & Barton, G.J. Jalview Version 2-a multiple sequence alignment editor and analysis workbench. Bioinformatics 25, 1189–1191 (2009).

[42]	Delgado, J., Radusky, L.G., Cianferoni, D. & Serrano, L. FoldX 5.0: working with RNA, small molecules and a new graphical interface. Bioinformatics 35, 4168–4169 (2019).

[43]	Leman, J.K. et al. Macromolecular modeling and design in Rosetta: recent methods and frameworks. Nat. Methods 17, 665–680 (2020).

[44]	Hashiguchi, T. et al. Structural basis for Marburg virus neutralization by a cross-reactive human antibody. Cell 160, 904–912 (2015).

[45]	Pinto, D. et al. Broad betacoronavirus neutralization by a stem helix-specific human antibody. Science 373, 1109–1116 (2021).

RIGHTS & PERMISSIONS

The Author(s) 2026. Published by Higher Education Press. This is an Open Access article distributed under the terms of the CC BY license (https://creativecommons.org/licenses/by/4.0/).

Cite this article

Download citation ▾

Bi, J. et al. A potent interdomain epitope-targeting antibody protects against SFTSV in mice and non-human primates Vita https://doi.org/10.15302/vita.2025.12.0002 ()

AI Summary AI Mindmap

PDF (13808KB)

Supplementary Files

Supplementary Materials

Sections

Figures

References

8776

Accesses

Citation

Detail

Recommended

About the journal

Browse

Authors & reviewers

A potent interdomain epitope-targeting antibody protects against SFTSV in mice and non-human primates

ABSTRACT

INTRODUCTION

RESULTS

Characterization of SFTSV glycoprotein Gn/Gc-reactive BCRs from convalescent donors

Identification of SFTSV-specific antibody clonotypes in PBMC and lymph node (LN)

Isolation of SFTSV neutralizing antibodies from convalescent donors

Cryo-electron microscopy (cryo-EM) structure revealed a novel neutralization epitope spanning Domain I and Domain III of Gn

ZS1C5 confers robust therapeutic protection in a lethal SFTSV murine model

ZS1C5 protected rhesus macaques from SFTSV-induced symptoms

ZS1C5 treatment suppressed SFTSV viremia and tissue viral replication

Single-cell transcriptomic profiling reveals ZS1C5 treatment restores immune homeostasis and reduces inflammation

Computation-assisted identification of ZS1C5-like NAbs from human BCR repertoire

DISCUSSION

MATERIALS AND METHODS

Human subjects and PBMC isolation

Recombinant protein expression and purification

Antigen-specific B cell sorting

Single B cell PCR and antibody cloning

Human single-cell transcriptome and BCR-seq

Human sequencing data processing and cell type identification

Antibody gene sequence and clonal analysis

Production of monoclonal antibodies

ELISA

BLI

Neutralization assay

Cryo-EM sample preparation and data collection

Cryo-EM data processing

Model building and structure refinement

Conservation analysis

Energy calculation

Computational identification of structurally related antibodies

Mouse challenge

NHP challenge

CBC, Basic metabolic panel (BMP), and coagulation test

Histopathology and IHC

Luminex assay

Viral RNA extraction and quantification

NHP single-cell sequencing

NHP sequencing data processing and cell type identification

Enrichment analysis of differential genes

Intercellular communication analysis

Inflammatory scoring analysis

Quantification and statistical analysis

DATA AVAILABILITY

RIGHTS & PERMISSIONS

Cite this article

AI思维导图