www.nature.com/npjvaccines ARTICLE OPEN An engineered HIV-1 Gag-based VLP displaying high antigen density induces strong antibody-dependent functional immune responses Ferran Tarrés-Freixas1, Carmen Aguilar-Gurrieri1, María Luisa Rodríguez de la Concepción1, Victor Urrea 1, Benjamin Trinité 1, Raquel Ortiz1, Edwards Pradenas 1, Pau Blanco2, Sílvia Marfil1, Luis Manuel Molinos-Albert 1,7, Ana Barajas1, Anna Pons-Grífols 1 , Carlos Ávila-Nieto1, Ismael Varela1, Laura Cervera3, Sònia Gutiérrez-Granados3, María Mercedes Segura3, Francesc Gòdia3, Bonaventura Clotet1,4, Jorge Carrillo1,5,6 ✉ and Julià Blanco 1,4,5,6 ✉ Antigen display on the surface of Virus-Like Particles (VLPs) improves immunogenicity compared to soluble proteins. We hypothesised that immune responses can be further improved by increasing the antigen density on the surface of VLPs. In this work, we report an HIV-1 Gag-based VLP platform engineered to maximise the presence of antigen on the VLP surface. An HIV-1 gp41-derived protein (Min), including the C-terminal part of gp41 and the transmembrane domain, was fused to HIV-1 Gag. This resulted in high-density MinGag-VLPs. These VLPs demonstrated to be highly immunogenic in animal models using either a homologous (VLP) or heterologous (DNA/VLP) vaccination regimen, with the latter yielding 10-fold higher anti-Gag and anti-Min 1234567890():,; antibody titres. Despite these strong humoral responses, immunisation with MinGag-VLPs did not induce neutralising antibodies. Nevertheless, antibodies were predominantly of an IgG2b/IgG2c profile and could efficiently bind CD16-2. Furthermore, we demonstrated that MinGag-VLP vaccination could mediate a functional effect and halt the progression of a Min-expressing tumour cell line in an in vivo mouse model. npj Vaccines (2023)8:51 ; https://doi.org/10.1038/s41541-023-00648-4 INTRODUCTION rabbits) and different formulations are evaluated: nucleic Human Immunodeficiency Virus-1 (HIV-1) has developed several acids21–23, purified VLPs24–27 or heterologous strategies28–30. strategies to impair the development of protective immune Although these studies have demonstrated that retroviral Gag- responses. Among them, the low incorporation of envelope based VLPs are able to induce potent immune responses31, a glycoproteins (Env) on the viral surface may result in reduced limitation remains since HIV-1 Env is poorly incorporated on viral antibody avidity, which may hamper the development of potent particles and VLPs32. neutralising Env-specific humoral immune responses1,2. The Strategies to increase antigen density on the surface of VLPs delivery of antigen at high-density on multivalent platforms is include the incorporation of multimerization tags33, the modifica- considered an important mean to induce potent B-cell responses tion of the Env cytoplasmic tail34 or its substitution by those from both in natural infection or during vaccination3–6. Therefore, these other viral proteins23. In this work, we describe a high-density types of strategies are progressively reaching the human vaccine antigen-displaying HIV-1 Gag-based VLP platform generated by field, with one recent example being the Novavax nanoparticle- the fusion of an extracellular antigen to HIV-1 Gag via a based subunit vaccine against SARS-CoV-2 (NVX-CoV2373)7. Other transmembrane domain. A small HIV-1 gp41-derived antigen strategies currently in development are based on synthetic containing a fragment of the HR2 domain, the membrane nanoparticles, such as liposomes and Virus-like Particles (VLPs), proximal external region (MPER) and the gp41 transmembrane or DNA/RNA delivery systems that are able to present a high domain was selected as model antigen35. This antigen improves number of membrane-bound antigens to naive B cells, improving the exposure of the MPER36, which is one of the most conserved their priming and supporting antibody maturation in germinal HIV-1 Env regions. In addition, anti-MPER neutralising antibodies centres8–17. In this sense, HIV-1 Gag-based enveloped VLPs are a (NAbs) are among the antibodies with the broadest neutralising promising vaccine platform17,18. activity (i.e., 10E8) described so far. Therefore, the MPER is an Enveloped Gag-VLPs are non-infectious and non-replicative viral attractive target for HIV-1 vaccine development35. Theoretically, in particles. Gag-VLPs are assembled at the cell membrane by our fusion-protein VLPs, the number of antigens displayed would oligomerisation of the HIV-1 p55Gag polyprotein releasing to the be stoichiometrically equivalent to Gag (2500 Gag proteins/VLP)19 extracellular space particles that mimic virion structural fea- and far superior to the expected number of Env glycoproteins on tures18–20. VLPs are currently being tested as HIV-1 vaccine the surface of HIV-1 virions (4–20 Env/virion1,2). Our VLPs induced candidates in preclinical animal models (mice, macaques and a non-neutralising but potent and functional humoral immune 1 IrsiCaixa AIDS Research Institute, Can Ruti Campus, 08916 Badalona, Catalonia, Spain. 2Comparative Medicine and Bioimage Centre of Catalonia (CMCiB), Germans Trias i Pujol Research Institute (IGTP), Can Ruti Campus, 08916 Badalona, Catalonia, Spain. 3Grup d’Enginyeria Cel•lular i Bioprocessos, Department of Chemical, Biological and Environmental Engineering, Escola d’Enginyeria, Universitat Autònoma de Barcelona, Campus de Bellaterra, 08913 Cerdanyola del Vallès, Catalonia, Spain. 4University of Vic–Central University of Catalonia (UVic-UCC), 08500 Vic Barcelona, Spain. 5Germans Trias i Pujol Research Institute (IGTP), Can Ruti Campus, Badalona, 08916 Barcelona, Spain. 6CIBER Enfermedades Infecciosas (CIBERINFEC), Instituto de Salud Carlos III, Madrid, Spain. 7Present address: ISGlobal, Hospital Clínic-Universitat de Barcelona, Barcelona, Spain. ✉email: jcarrillo@irsicaixa.es; jblanco@irsicaixa.es Published in partnership with the Sealy Institute for Vaccine Sciences F. Tarrés-Freixas et al. 2 response that could mediate a protective effect when used as a of gp4136. This experimental setting mimics natural antigen vaccine platform. incorporation to VLPs. A nude Gag-VLP was also produced (Fig. 2c, and Supplementary Fig. 2). Western blot analysis showed a lower content of Min antigen in standard VLPs compared to MinGag and RESULTS Min(RA)Gag-VLPs (Fig. 2d). Quantitative analysis confirmed that MinGag-VLPs display similar morphology and composition as the Min density in MinGag and Min(RA)Gag-VLPs was 10- and 14- Gag-VLPs fold higher, respectively, than for standard VLPs (Fig. 2e). Plasmids encoding HIV-1 Gag or the fusion-protein MinGag were transiently transfected into Expi293F cells to produce Gag-VLPs Induction of a Th1-like humoral response by MinGag-VLPs and MinGag-VLPs, respectively (Fig. 1a). Min antigen was Immunisation strategies combining heterologous formulations of efficiently detected by the anti-MPER 10E8 antibody on the the same antigen have proven to elicit superior immune surface of cells (Fig. 1b), while intracellular co-staining with an responses39 and have reached massive testing in humans during anti-p24 Gag antibody (KC57-FITC) confirmed the co-expression of COVID-19 pandemics40. Therefore, we assessed the immunogeni- Gag. Cells transfected with gag, and mock cells were used as city of MinGag-VLPs and Min(RA)Gag-VLPs in C57BL/6JOlaHsd controls (Fig. 1b). Gates are shown in Supplementary Fig. 1 mice following two different strategies: a homologous regimen Gag-VLPs and MinGag-VLPs were successfully recovered from using four doses of purified VLPs (VVVV) or a combined/ the supernatant of transfected cells, but p24 Gag concentration heterologous regimen priming with two doses of DNA and was 10-fold lower for the latter (Fig. 1c), suggesting a lower boosting with two doses of purified VLPs (DDVV). Gag-VLPs and MinGag-VLP production. SDS-PAGE and WB protein analyses PBS were included as controls. showed that Gag-VLPs were only detected using anti-p24 Overall, both strategies induced both anti-Gag and anti-Min IgG antibodies (Fig. 1d and Supplementary Fig. 2), whereas MinGag antibody responses (Fig. 3a, b). Vaccination with purified VLPs did fusion-protein was detected by both anti-p24 and anti-MPER not induce potent antibody responses against Gag (Fig. 3a). antibodies. MinGag protein had the expected apparent molecular Interestingly, DDVV vaccination induced a 10-fold higher IgG weight of 72 kDa. However, some degradation products were concentration in serum at endpoint compared to VVVV for both detected by both anti-p24 and anti-MPER antibodies, suggesting 1234567890():,; the anti-Gag and anti-Min responses (Fig. 3a, b). Anti-Gag partial fragmentation or processing of the fusion-protein (Fig. 1d). responses in DDVV and VVVV (Fig. 3a), as well as anti-Min Despite the lower concentration of MinGag-VLPs compared to responses in VVVV (Fig. 3b), reached the maximal IgGs concentra- Gag-VLPs in the supernatant, both were effectively purified by a tion in serum 3 weeks after the first immunisation. In comparison, two-step protocol involving either TFF + UC or TFF + AEX (see concentration of anti-Min IgGs in animals vaccinated with methods). Considering both procedures, TFF + UC yielded super- MinGag-VLPs and Min(RA)Gag-VLPs in the DDVV regimen ior recoveries, but for MinGag-VLPs, TFF + AEX resulted in a 10- increased after each DNA shot and after the first VLP immunisa- fold increase of sample purity as assessed by the ratio of p24 tion. However, significant differences compared to the VVVV protein over the total protein content of these preparations regimen were only observed for the MinGag-VLP group (Fig. 3b) (Fig. 1d, and Supplementary Table 1). As expected, Cryo-EM due to a high dispersion in the Min(RA)Gag-VLP group. Min(RA) analysis of purified Gag- and MinGag-VLPs revealed a morphology Gag induced a slightly superior, although non-significant, anti-Min of round-shaped particles with a lipid membrane containing a IgG concentration than MinGag in the DDVV regimen. While no p55Gag electrodense core (Fig. 1e, f). However, compared to significant differences were detected between females and males, Gag-VLPs, the diameter of MinGag-VLPs was significantly reduced a trend towards higher IgG concentrations was observed in the (Fig. 1g, 124.7 ± 13.6 nm vs. 62.9 ± 13.0 nm; p < 0.0001). former (Supplementary Fig. 3a–d). As VLPs were produced in a human cell line, human host Impact of transmembrane and linker domains on antigen proteins were expected to be incorporated into the VLPs. Indeed, exposure and density mice immunised with purified VLPs elicited antibody responses To optimise antigen exposure on the surface of VLPs, different against human proteins (Fig. 3c). In the VVVV regimen, IgG titres MinGag transmembrane and linker modifications were tested. against human proteins progressively increased, reaching a Such variations included the substitution of the transmembrane plateau after the third dose. In contrast, in the DDVV regimen, domain of gp41 for the transmembrane domain of the human no anti-Expi293F protein responses were detected after DNA CD44 protein, which mostly localises in cholesterol-rich micro- immunisations, since VLPs were produced directly in vivo by domains37. Furthermore, we also evaluated the effect of murine cells (Fig. 3c), while similar kinetics of anti-Expi293F substituting the arginine at position 696 in the gp41 wildtype antibodies were observed after subsequent VLP immunisation. transmembrane domain by an alanine (R696A). This arginine has Epitope mapping showed that anti-Min antibodies preferen- been described to mediate conformational changes during HIV-1 tially targeted the N-terminus region of Min, which contains a fusion with the host cell38, and hence it could impact the antigen fragment of the HR2 domain (Supplementary Fig. 4a–c). However, display. Besides the modifications in the transmembrane domain, in the DDVV regimen, anti-Min IgG antibodies also targeted the the original GGGS flexible linker between Min and Gag was also C-terminus of the HR2 domain (Supplementary Fig. 4b). removed (no linker, NL) to evaluate its impact on VLP antigenicity. Detection of the IgG subclass profile demonstrated that VLP- Of note, binding of 10E8 on the surface of VLP-producing cells induced Gag-specific antibodies were heterogeneous (IgG1, IgG2b demonstrated that the construct with the R696A substitution and IgG2c), while Min-specific antibodies were mainly IgG2c (Min(RA)Gag) and Min(CD44tm)Gag led to a better antigenic (Fig. 4a), regardless of the vaccination regimen. IgG subclasses are exposure. In contrast, the removal of the linker had no effect on relevant for Fc-mediated functions, since murine IgG2c and, to a antigenicity (Fig. 2a). Accordingly, MinGag and Min(RA)Gag lesser extent, IgG2b can bind to the NK cell activating receptor constructs were selected since they only displayed HIV-1 derived mFcγRIV (CD16-2)41. Consistently, anti-Min antibodies bound to sequences and provided two different levels of antigenicity CD16-2 receptor (Fig. 4b), indicating that they could mediate Fc- (Fig. 2b). effector functions such as antibody-dependent cell cytotoxicity In addition, we also evaluated the antigenic density of final VLP (ADCC)42,43. preparations. For comparative purposes, we produced standard Neutralisation capacity of VLP-induced antibodies was tested VLPs by co-transfection of Expi293 cells with plasmids encoding in vitro against 3 subtype B HIV-1 pseudoviruses (NL4.3, AC10.0.29 Gag protein and a Min protein containing the intracellular domain and TRO.11), an HIV-2/HIV-1 MPER chimera (7312 A/C1), and an npj Vaccines (2023) 51 Published in partnership with the Sealy Institute for Vaccine Sciences F. Tarrés-Freixas et al. 3 Fig. 1 Characterisation of high-density HIV-1 Gag-based VLP production in Expi293F cells. a Schematic representation of the constructs and hypothetical structure of Gag-VLPs and high-density MinGag-VLPs. Min is a gp41-derived protein that contains a fragment of the HR2, MPER and TM domains. b FACS dot plots for the staining of non-transfected (left), gag-transfected (middle) and mingag-transfected (right) Expi293F cells stained with the anti-p24 KC57-FITC-conjugated mAb (X-axis) and the anti-MPER 10E8 antibody (Y axis). c Quantification of p24 by ELISA on harvested supernatants of gag-transfected and mingag-transfected Expi293F cells. Data represented as mean ± SD. Significant differences on production were found using a Mann–Whitney U test (****p < 0.0001). d Characterisation of HIV-1 Gag-VLPs and MinGag-VLPs from harvested supernatants and after purification with either sucrose-cushion ultracentrifugation (TFF + UC) or anion exchange chromatography (TFF + AEX) by Coomassie blue staining (left), anti-p24 WB (middle) and anti-2F5 WB (right). Both western blots correspond to the same gel. Uncropped images are shown at Supplementary Fig. 2a. e Cryo-EM of purified Gag-VLPs. Scale bar = 200 nm. f Cryo-EM of purified MinGag-VLPs. Scale bar = 200 nm. g Diameter comparison of Gag-VLP (n = 17) and MinGag-VLP (n = 27). Significant differences on size were found using a Mann–Whitney U test (****p < 0.0001). Data represented as mean ± SD. Published in partnership with the Sealy Institute for Vaccine Sciences npj Vaccines (2023) 51 F. Tarrés-Freixas et al. 4 Fig. 2 Optimisation of Min antigenicity on the surface of VLP-producing Expi293F cells. a Dot plots of Expi293F cells transfected with mingag constructs bearing different transmembrane or linker variants: original (MinGag; same as shown in Fig. 1b) no linker (Min(NL)Gag), gp41 R696A mutation (Min(RA)Gag), CD44tm (Min(CD44)Gag). Cells were extracellularly stained with an anti-MPER antibody (10E8) and intracellularly stained with an anti-p24 antibody (KC57-FITC). b Dose-response curves of 10E8-binding to Min on the surface of Expi293F cells transfected with transmembrane and linker mingag variants. c Schematic representation of the strategies used to produce Gag-VLPs, low- density Minfull + Gag-VLPs or high-density MinGag- and Min(RA)Gag-VLPs. d Western Blot analysis of the same gel loaded with VLPs purified from supernatant of Expi293 cells transfected with Minfull + Gag, MinGag or Min(RA)Gag. Left panel: anti-p15 antibody. Right panel: 2F5 antibody (anti-MPER). Uncropped images are displayed at Supplementary Fig. 2b. e Quantification of Min/Gag ratios from the Western Blot as a measure of Min incorporation into VLPs. Data represented as mean ± SD, and values indicate fold change. npj Vaccines (2023) 51 Published in partnership with the Sealy Institute for Vaccine Sciences F. Tarrés-Freixas et al. 5 Fig. 3 Humoral responses induced by homologous and heterologous immunisation regimens. C57BL/6JOlaHsd mice were immunised at weeks 0, 3, 6, and 9 with a homologous (VVVV; dashed line) or heterologous (DDVV; solid line) regimens consisting on Gag-VLPs (grey rounds; n = 10), MinGag-VLPs (gold squares; n = 10) or Min(RA)Gag-VLPs (blue triangles; n = 10) and a pVAX1/PBS control or PBS control (black triangles; n = 4). a Concentration of anti-Gag IgG in mouse sera determined by ELISA. b Concentration of anti-Min IgG in mouse sera determined by ELISA. c Concentration of IgGs against Expi293F host cell proteins analysed by FACS. Data are presented as mean ± 95%CI for the ELISA results and geo-MFI ± SD for the FACS results. Statistical differences at week 12 were found using a Kruskal–Wallis test with Dunn’s correction (*p < 0.05; **p < 0.01; ***p < 0.001). HIV-2 control (7312 A) showing no significant neutralising activity albeit no significant differences were found between VLP (Supplementary Fig. 5a, b). constructs (Supplementary Fig. 6a, b). Moreover, no T-cell T-cell responses against Min and Gag proteins were evaluated responses were detected against the MPER44,45, whereas weak by IFN-γ ELISpot using cryopreserved murine splenocytes. As responses were elicited against Min in MinGag-VLP-immunised anticipated, T-cell responses were mostly directed against Gag, animals in the heterologous regimen (Supplementary Fig. 6a, b). Published in partnership with the Sealy Institute for Vaccine Sciences npj Vaccines (2023) 51 F. Tarrés-Freixas et al. 6 Fig. 4 Functional properties of VLP-induced humoral responses. Serum samples at week 12 from C57BL/6JOlaHsd mice vaccinated with heterologous (left panels) or homologous (right panels) regimens consisting of Gag-VLPs (grey; n = 10), MinGag-VLPs (gold; n = 10), Min(RA) Gag-VLPs (blue; n = 10) and pVAX1/PBS or PBS controls (black, n = 4). a Heatmaps of the predominant murine IgG subclasses (IgG1, IgG2b, IgG2c, or IgG3) for both anti-Gag and anti-Min humoral responses in both vaccination regimens. Each line represents one animal. b Binding to mFcγRIV/CD16-2 by Min-specific antibodies in serum from VLP-immunised mice collected at week 12. Data represented as mean ± SD of relative binding, this being the absorbance values of CD16-2 binding of each serum sample against Gp41-MinTT referred to its absorbance against BSA. Significant differences were found using a Kruskal–Wallis test with Dunn’s comparison (*p < 0.05; **p < 0.01; ***p < 0.001). In vivo functional protection assay test a more conventional approach with a shorter vaccination To evaluate whether VLPs could induce a functional immune schedule. Two weeks after the second immunisation, mice were response in vivo, C57BL/6JOlaHsd mice were electroporated with inoculated with a B16F10 melanoma clone that stably expressed plasmids encoding Luciferase (pVAX1-FLuc:pVAX1 at 1:1 ratio in the Min protein on its surface (Supplementary Fig. 7a–c, Fig. 6a). the right hindlimb) or Luciferase+Min(RA)Gag (pVAX1-FLuc:p- The B16F10 cell line was chosen because of its poor immuno- VAX1-min(RA)gag at a 1:1 ratio in the left hindlimb) (Fig. 5a). A genicity46. After tumour inoculation, animals were monitored for specific effector immune response against Min was expected to up to 80 days. Unvaccinated mice started to show detectable induce a faster clearance of the luciferase expression in Min(RA) tumour growth approximately 10 to 15 days after B16F10Min cell inoculation (Supplementary Fig. 8a). In comparison, the detection Gag-immunised animals compared to control animals as a of tumour growth in animals vaccinated with either Gag-VLPs and consequence of the elimination of Min-expressing cells. Biolumi- Min(RA)Gag-VLPs was delayed (20–25 days post-tumour inocula- nescence analysis demonstrated, on the one hand, a stable tion) (Supplementary Fig. 8a). Importantly, Min(RA)Gag-VLP production of luciferase over time (up to 77 days after vaccination significantly impacted tumour progression (Fig. 6b, electroporation) in control animals, indicating that anti-luciferase p = 0.0414) compared to Gag-VLP-vaccinated animals, and tumour immunogenicity was minimal and/or did not affect its expression. growth was detected only in one out of eight Min(RA)Gag- On the other hand, Luciferase+Min(RA)Gag co-electroporated vaccinated mice, compared to six out of nine Gag-vaccinated muscles showed a decrease in luminescence that was evident animals that reached the humane endpoint. 3 weeks after electroporation (Fig. 5b). Interestingly, the decrease Immune profiling of the IgG responses against Gag (Supple- in luciferase expression was faster after a second co- mentary Fig. 8b), Min (Fig. 6c) and Expi293-derived proteins electroporation with Luciferase+Min(RA)Gag (p = 0.0012). Taken (Supplementary Fig. 8c) and the cellular responses (Fig. 6d) together, these data suggest that pVAX1-min(RA)gag could induce induced in Gag-VLP and Min(RA)Gag-VLP vaccinated, and tumour- an effector immune response that cleared VLP-producing cells. inoculated animals matched with our previous vaccination results To functionally characterise anti-Min antibody responses, (Fig. 3). Concentration of anti-Min IgG was not significantly C57BL/6JOlaHsd mice were vaccinated with two doses of Gag- boosted after tumour inoculation (Fig. 6b). As anticipated, no VLPs or Min(RA)Gag-VLPs. In this setting, a homologous VLP significant neutralisation activity was detected in immunised and regimen was preferred over a heterologous DNA/VLP regimen to tumour-inoculated mice (Supplementary Fig. 8d), and anti-Min npj Vaccines (2023) 51 Published in partnership with the Sealy Institute for Vaccine Sciences F. Tarrés-Freixas et al. 7 expected to be densely covered with it, since it would theoretically display as many antigens as Gag molecules (about 2500 molecules of Gag/VLP)20. This hypothesis was supported by the fact that MinGag and Min(RA)Gag-VLPs showed 10- to 14-fold higher Min/Gag ratios compared to VLPs obtained by co- transfection of Min and Gag in separate plasmids, confirming the positive impact on antigen density of our approach. Furthermore, FACS and quantitative WB analyses showed that the antigen was properly displayed on the surface of VLP- producing cells and that the MinGag fusion-protein was recognised by both anti-p24 and anti-gp41 antibodies. Besides demonstrating that MinGag fusion-protein was properly expressed in transfected cells, we also showed that MinGag-VLP morphology is similar to Gag-VLPs but with a reduced diameter and production yield. These differences could be a consequence of a premature association of Gag to the lipid membranes in VLP- producing cells. This premature membrane association of the MinGag nascent protein could promote intracellular accumulation that negatively impacts the release of VLPs. Thus, the concentra- tion of MinGag-VLPs in the supernatant of VLP-producing cells was considerably lower than that of Gag-VLPs. Structural elements of the fusion proteins (transmembrane domain and linker) were evaluated in order to improve the immunogen exposure on the VLP membrane. Interestingly, the introduction of a single point mutation (R696A) into the gp41 transmembrane domain, which is thought to stabilise the transmembrane domain38, increased the antigen exposure on the surface of the VLP-producing cells, improving the MPER recognition by the 10E8 antibody. Immunogenicity studies performed in C57BL/6JOlaHsd mice Fig. 5 In vivo bioluminescence kinetics of DNA co-electroporation demonstrated that Min(RA)Gag-VLPs were immunogenic in vivo in mice. C57BL/6JOlaHsd mice (n = 5) were electroporated with when administered as DNA vaccine or as purified VLPs. 20 µg of DNA of both a 1:1 mix of pVAX1:pVAX1-Fluc at the right Comparison of a homologous VLP regimen (VVVV) and the limb (black rounds) at day 0 and a 1:1 mix of pVAX1-Fluc:pVAX1- heterologous DNA/VLP regimen (DDVV) showed that the com- min(RA)gag at the left limb (blue triangles) at day 0 and 45. bined regimen induced stronger anti-Gag and anti-Min humoral a Representative bioluminescence images of co-electroporated mice at 48 h (left), 7 days (middle) and 14 days (right) post- responses, in line with recent observations in humans for SARS- electroporation. b Bioluminescence follow-up of electroporated CoV-2 vaccination combining different vaccines50,51. Anti-Gag and mice. Data represented as mean ± SD. Significant differences anti-Min humoral responses in the VVVV regimen did not increase between the first and the second electroporation (days 0–35 and after each vaccination, but in the DDVV regimen these responses 46–77, respectively) at the left limb (pVAX1-Fluc:pVAX1-min(RA)gag) were properly boosted. The inferior performance of the VVVV were assessed comparing nested mixed effect models and regimen could be explained by the fact that, after the first dose, considering time as a categorical factor (**p = 0.0012; likelihood VLP administration mainly boosted humoral responses against ratio test). human proteins derived from VLP-producing cells. In the DDVV regimen, these “anti-vector” responses were delayed, and anti-Min antibodies presented predominantly an IgG2c subclass (Supple- responses were boosted more efficiently. Furthermore, the fact mentary Fig. 8e) that could bind to the CD16-2 receptor (Fig. 6e). that purified VLPs, both in the homologous regimen or as a Furthermore, anti-Min antibodies were able to bind B16F10Min booster in the heterologous regimen, induced lower anti-Gag cells (Fig. 6f). Interestingly, the Min(RA)Gag-vaccinated mouse that responses compared to anti-Min responses further supports that had to be euthanised due to uncontrolled tumour progression MinGag-VLPs were properly formed in a particulate manner. showed the lowest anti-Min antibody titre against B16F10Min cells Overall, the potent humoral response induced by MinGag-VLPs and the lowest binding to CD16-2 at endpoint (Fig. 6e, f). against Min was remarkable, especially considering that the Altogether, these results showed that Min(RA)Gag-VLP could C57BL/6 mouse model is biased towards Th1 immune induce an in vivo functional immune response mediating responses52. protection, possibly through antibody-dependent effector Besides the differences in antibody magnitude, the profiles of functions. humoral responses were similar between the two regimens. Both targeted the N-terminus part of Min, which contains the HR2 peptide. This HR2 dominance could explain the absence of DISCUSSION neutralising responses, and indeed, we could not induce Strategies to potentiate immune responses against weak immu- antibodies against the MPER which is the primary target of nogens include antigen modifications47, intrastructural help48, and gp41-specific broadly neutralising antibodies (bNAbs)35. Notwith- multivalent antigen display involving complex structures such as standing, anti-Min antibodies were mainly IgG2c, which is VLPs8,49. Enveloped HIV-1 Gag-based VLPs mimic the virus functionally equivalent to human IgG1, and efficiently bound to morphology and therefore may present immunogens in a more the murine CD16-2 receptor (mFcγRIV)41,43. Fc binding to CD16-2 native fashion9. Multiple strategies to increase immunogen in mice, as well as binding to CD16A in humans, activates NK cells density on the VLPs’ surface have been developed23. Here, we and promotes ADCC. Therefore, our results suggest that VLP- present a strategy based on the fusion of an HIV-1 gp41-based induced anti-Min antibodies could mediate protective responses immunogen to HIV-1 Gag via the gp41 transmembrane domain. beyond neutralisation. In contrast, there was no predominant IgG By fusing the immunogen to Gag, the surface of the VLP was subclass among anti-Gag antibodies in both immunisation Published in partnership with the Sealy Institute for Vaccine Sciences npj Vaccines (2023) 51 F. Tarrés-Freixas et al. 8 Fig. 6 In vivo protection assay to identify vaccine-induced protection. a Schematic representation of the experimental model. C57BL/ 6JOlaHsd mice were immunised twice with Gag-VLPs (grey rounds; n = 9), Min(RA)Gag-VLPs (blue triangles; n = 8), or PBS controls (black inverted triangles; n = 3) at weeks 0 and 3. At week 5, animals were inoculated with a 100,000 B16F10Min cells and followed up for a period of 80 days after tumour inoculation. b Kaplan–Meier curves of tumour-free survival in VLP vaccinated and tumour-inoculated mice. Animals were euthanised for humane criteria when the tumour reached a size bigger than 1000 mm3. Significance was detected between Gag-VLP and Min(RA)Gag-VLP groups with a Mantel–Cox test (*p = 0.0414). c Anti-Min IgG concentration in the serum of VLP-immunised and tumour- inoculated mice determined by ELISA. d Cellular immune response against pools of Gag, Min, and MPER using splenocytes from vaccinated and tumour-inoculated mice and collected at endpoint. Data represented as mean ± SD. No significant differences were found in panels c–e between vaccinated groups using a Kruskal–Wallis test. e Relative binding to FcγRIV/CD16-2 of anti-Min antibodies referred to a BSA control. Serum was collected at week 5. Data represented as mean ± SD. Significant differences were found with a Kruskal–Wallis test with multiple comparison correction (*p ≤ 0.05). f Binding capacity to B16F10Min cells of antibodies induced by VLP-induced by VLP vaccination in serum from week 4. Data represented as mean ± SD of relative MFI of each serum sample binding to B16F10Min cells referred to their respective B16F10 control. Significant differences were found in using a Kruskal-Wallis test with Dunn’s comparison (*p < 0.05, **p < 0.01). regimens, suggesting that antigens displayed on the surface or construct. Second, since mice cannot get infected with HIV-1, and within the VLP could induce qualitatively distinct antibody humanised mice do not have a fully competent immune system, responses. we developed an indirect mouse model to investigate the To investigate whether VLP-induced responses could mediate functionality of the immune response generated after vaccination protection, we performed two in vivo experiments. First, with Min(RA)Gag-VLPs. This model was based on the inoculation mice were electroporated with a plasmid that coded for luciferase of a Min-expressing tumour cell line derived from B16F10 alone or combined with a plasmid that coded for Min(RA)Gag. melanoma cells after VLP immunisation. These Min-expressing Luciferase was detected for almost 3 months when the plasmid cells would act as a surrogate of HIV-1 infected cells by expressing was electroporated alone. Conversely, when both plasmids were Min proteins on its surface, and functional immune responses co-administered, luciferase signal progressively declined, suggest- could delay tumour growth. B16F10 cells were chosen for their ing the loss of the co-transfected cells over time. Interestingly, this poorly immunogenic profile and because they are resistant to decay was accelerated after a second electroporation, suggesting T-cell responses46; thus, any effect on tumour growth would be that a specific immune response directed against Min(RA)Gag was more likely due to a Min-specific antibody-dependent effect. With mediating this clearance, instead of a direct toxic effect by the this model we demonstrated that Min(RA)Gag-VLP-vaccinated npj Vaccines (2023) 51 Published in partnership with the Sealy Institute for Vaccine Sciences F. Tarrés-Freixas et al. 9 animals significantly halted tumour progression compared to Gag- 136-098; Jackson ImmunoResearch, West Grove, PA, USA) at a VLP-vaccinated animals. Min(RA)Gag-VLP-vaccinated animals 1:500 dilution and intracellularly stained with KC57-PE (6604667; developed anti-Min antibodies that bound to Min on the surface Beckman Coulter, Brea, CA, USA) at a 1:200 dilution. Samples were of B16F10Min cells, and also bound to CD16-2 (mFcγRIV) via the fixed and permeabilised with Fix&Perm (ThermoFisher Scientific) Fc, potentially mediating protection. To further support these and analysed using a BD FACSCelesta™ Flow Cytometer (Becton observations, Min(RA)Gag-VLPs induced a weak anti-Min and anti- Dickinson (BD), Franklin Lakes, NJ, USA). Gating strategies are Gag cellular immune response. displayed in Supplementary Fig. 1. Taken together, these results show that our engineered fusion- VLPs were concentrated from cell-cultured supernatant and protein generated high-density MinGag-VLPs that induced a buffer exchanged to PBS by tangential flow filtration (TFF) with a potent and functional immune response against a gp41-derived 300,000 MWCO cassette (Pall Laboratories, Port Washington, NY, protein, in the absence of adjuvant and at low VLP dose. Finally, USA). Concentrated VLPs were purified by two chromatography VLP-induced antibodies potentially mediate antibody-dependent steps using: (1) ligand-activated core chromatography HiTrap effector functions through binding to CD16-2. Induction of CaptoCore700 column (Cytiva, Marlborough, MA, USA)54, and, (2) antibodies with effector function capabilities is relevant, since in anion exchange (AEX) HiTrap Q XL column (Cytiva)55. The ÄKTA the RV114 clinical trial the presence of anti-HIV-1 IgGs that start chromatography system (Cytiva) was used for protein mediated antibody-dependent responses correlated with protec- purification. This purification protocol was compared to sucrose- tion53. Further studies will help unveiling the versatility of this VLP cushion ultracentrifugation (UC). Finally, VLPs were concentrated platform to present more complex immunogens, such as larger at 1 mg/mL of total protein and sterilised by filtration using PVDF antigens or high ordered structural antigens (full HIV-1 gp160 0.22 µm filter (Merck & Co., Kenilworth, NJ, USA). envelope, for instance) and their ability to induce potent Purified VLPs were quantified by ELISA (INNOTEST HIV antigen neutralising antibody responses. mAb, Fujirebio, Tokyo, Japan) and total protein was quantified using the Pierce BCA Protein Assay (ThermoFisher Scientific). Purified VLPs were routinely characterised by Western Blot (WB) METHODS using anti-MPER antibodies 2F5 (AB001; Polymun Scientific, Plasmids Klosterneuburg, Austria) and 10E8 (NIH HIV Reagent Program) at The mingag fusion protein was generated by fusing a codon- 4 µg/mL and 1 µg/mL, respectively, and an anti-HIV-1 p24 optimised HIV-1 subtype B gp41 miniprotein (“min”) containing a (subsequently referred as p24) antibody [clone 39/5.4 A] fragment of the HR2 domain, the membrane proximal external (ab9071; Abcam, Cambridge, UK) at 1 µg/mL. Western blot analysis region (MPER) and the transmembrane domain (TM) (HXB2: 8057- was performed on TransBlot® Turbo™ Membrane (Bio-Rad, 8345 bp)36 with a codon-optimised HIV-1 subtype B gagHXB2. A Hercules, CA, USA) using a rabbit polyclonal anti-p15 antibody flexible GGGS linker was inserted between the C-terminal end of (ab66951; Abcam) at a 1:1000 dilution and the human antibody Min (behind the transmembrane domain) and the N-terminal end 2F5 (Polymun) at 4 µg/mL, for the detection of HIV-1 Gag and of Gag (Fig. 1a). For the production of regular HIV-1 VLPs, a MPER, respectively. An IRDye® 680RD-labelled goat anti-rabbit (IgG plasmid coding for HIV-1 Gag (HXB2 isolate) and a plasmid coding Fc) (926-68071; LI-COR, Lincoln, NE, USA) at a 1:20,000 dilution and for a version of the Min antigen containing the intracellular an HRP-labelled goat anti-human IgG (H + L) (109-035-088; domain of gp41 (HXB2: 8057–8792 bp, Minfull) were also Jackson ImmunoResearch) at a 1:20,000 dilution were used as designed36. For the generation of MinGag variants, the transmem- secondary antibodies. Western blot images were obtained in a brane domain of the mingag construct was replaced by the Chemidoc™ MP Imaging System (Bio-Rad). Precision Plus Protein™ transmembrane region of CD44 or an R696A mutated version of Dual Xtra Prestained Standard (Bio-Rad) was used as a molecular the gp41 transmembrane domain (GeneArt). Also, a linker-free ladder. The band intensity corresponding to Gag, Minfull, and construct that joints directly the TM domain of Min with the MinGag was calculated with the Image Lab 6 software (Bio-Rad). N-terminal region of Gag was assayed. The ratio of Min (2F5 staining) and Gag (p15 staining) signals were Codon-optimised sequences were generated by GeneArt calculated as a relative measure of antigen density in VLPs. (ThermoFisher Scientific, Waltham, USA) and were cloned into The VLP morphology was assessed by cryogenic electron both pcDNA3.1 and pVAX1 vectors (ThermoFisher Scientific) using microscopy (cryo-EM) on a carbon-coated copper grid that was the KpnI and XhoI restriction enzymes and the T4-DNA ligase prepared using a Leica EM GP workstation (Leica, Wetzlam, (ThermoFisher Scientific). pcDNA3.1- and pVAX1-derived vectors Germany). VLPs were analyzed using a Jeol JEM-2011 electron were used for in vitro VLP production or in vivo DNA microscope (JEOL Ltd, Tokyo, Japan), equipped with a CCD 895 immunisation, respectively. Endotoxin-free plasmids were purified USC4000 camera (Gatan, Pleasanton, CA, USA). with HiSpeed Plasmid Maxi or Mega EF Kits (QIAGEN, Hilden, Germany) and filtered at 0.22 µm before use. Animal procedures All animal work was performed at the Centre for Comparative VLP production, purification, and characterisation Medicine and Bioimage (CMCiB) under the approval of the Expi293F cells (ThermoFisher Scientific) were grown in serum-free Committee on the Ethics of Animal Experimentation of the (SF) and animal-derived component free (ADCF) conditions using Germans Trias i Pujol Research Institute (IGTP) and the authorisa- Expi293F medium (ThermoFisher Scientific). Cells were transiently tion of Generalitat de Catalunya (codes: 9525 and 9943). All transfected with pcDNA3.1-gag, pcDNA3.1-mingag or pcDNA3.1- procedures are in accordance with the 3 R principle and prioritise animal welfare. min(RA)gag encoding vectors or a mix of pcDNA3.1-minfull and pcDNA3.1-gag plasmids using ExpiFectamine293 Transfection kit and following manufacturer’s protocol (ThermoFisher Scientific). In vivo electroporation and bioluminescence Supernatant was harvested 48 h post-transfection and clarified by C57BL/6JOlaHsd (Envigo, Horst, Netherlands) mice were intramus- centrifugation at 3000 × g for 5 min and 0.22 µm filtration. cularly injected in the hindlimbs. For the bioluminescence assay, Transfected Expi293F cells were characterised by Fluorescence- we prepared a 1:1 mix of a firefly luciferase encoding vector Activated Cell Sorting (FACS). Briefly, cells were stained extra- (pVAX1-Fluc) and pVAX1 empty vector and a 1:1 mix of pVAX1-Fluc cellularly with the anti-MPER antibody 10E8 (ARP-12294; NIH HIV and pVAX1-min(RA)gag. Twenty micrograms of each plasmid mix Reagent Program, Bethesda, MD, USA) at 1 µg/mL plus a were injected at the right hindlimb and left hindlimb, respectively. secondary APC-conjugated AffiniPure Goat anti-human IgG (109- Immediately after each DNA injection, electroporation was Published in partnership with the Sealy Institute for Vaccine Sciences npj Vaccines (2023) 51 F. Tarrés-Freixas et al. 10 performed at the injection site using NEPA21 electroporator toxoid (TT) at C-terminus16,36) proteins were produced in One- (NepaGene, Chiba, Japan). Two different electroporation protocols Shot™ BL21(DE3) cells (ThermoFisher Scientific) and extracted were used. For the immunisation studies, we used an electro- from inclusion bodies with an 8 M urea buffer. Protein was purified poration protocol optimised for animal welfare consisting of two by ion-metal affinity chromatography with 1 mL of Ni2+-Sephar- high-voltage poring pulses (200 V and 0.1 milliseconds (ms) ose® High Performance beads (Cytiva) and eluted with 0.5 M positive pulses with a 20% decay separated by a 10 ms interval) imidazole in 8 M urea buffer. Protein was concentrated and buffer followed by six low-voltage transfer pulses (3 pulses of 60 V, 20 ms exchanged to PBS + 0.1% SDS with an Amicon Ultra 15 mL and positive polarity with a 20% decay separated by a 50 ms Centrifugal Filter (Merck & Co.)16. interval, followed by three pulses with the same parameters but inverted polarity). For the in vivo bioluminescence assays, we used ELISA an electroporation protocol optimised in house for maximal efficiency and consisted of eight positive pulses of 60 V lasting Concentration of antibodies against Gag and Min in mouse sera 20 ms with a 1 s interval. samples was determined by ELISA. Nunc MaxiSorp 96-well plates For in vivo bioluminescence analysis, D-luciferin (150 mg/kg, (ThermoFisher Scientific) were coated with either 50 ng of Biovision, Milpitas, CA) was injected intraperitoneally to mice. recombinant Gag or MinTT in PBS or 500 ng of overlapping Min Then, mice were anaesthetised using 5% isoflurane at 1 L/min for peptides in 0.2 M carbonate/bicarbonate buffer at pH = 9.6 and induction and maintained in the imaging system with 3% incubated overnight at 4 °C. After coating, the plate was washed isoflurane. Images were taken with 30 s of exposure time and with 1× PBS + 0.05% Tween20 and blocked with 1% BSA + 0.05% medium binning using an IVIS Lumina III In Vivo Imaging System Tween20. Then, samples were incubated overnight. Increasing (Perkin Elmer, Waltham, MA, USA). concentrations of mouse monoclonal anti-HIV-1 p24 antibody (clone 39/5.4 A; Abcam) or an anti-gp41 antibody (clone D50, NIH HIV Reagent Program) were used to generate standard curves for Immunisation Gag and Min, respectively. Results were expressed as p24 or D50 VLP immunogenicity was assessed in C57BL/6JolaHsd (Envigo) equivalents16. Total IgG was determined with a secondary HRP- mice following two different approaches to assess the effect of conjugated AffiniPure Goat anti-mouse IgG (115-036-071; Jackson combined regimens: (a) a homologous regimen consisting of four ImmunoResearch) at a 1:10,000 dilution. IgG subclasses were doses of purified VLPs (VVVV; 90 ng p24/dose in PBS), or (b) a determined using Biotin-conjugated AffiniPure goat anti-mouse heterologous regimen consisting of two doses of electroporated IgG1, IgG2b, IgG2c, and IgG3 antibodies (1106205, 115-065-207, DNA (20 µg DNA in PBS) followed by two doses of purified VLPs 115-065-208 and 115-065-209, respectively; Jackson ImmunoR- (DDVV). Purified VLPs were injected subcutaneously into the esearch) at a 1:5,000 dilution and an HRP-conjugated Streptavidin hock56. Each immunisation was performed with a 3-week interval. (N100; ThermoFisher Scientific) at a 1:6000 dilution. Finally, after Experimental groups were Gag-, MinGag- or Min(RA)Gag-immu- five washes, the assay was developed using o-phenylenediamine nised animals and controls were pVAX1/PBS or PBS-injected (OPD; ThermoFisher Scientific) for 10 min and stopped using 4 N animals. H2SO4. Before each immunisation, a small blood sample was obtained from the facial vein. Serum was recovered from whole blood by coagulation and centrifugation, and then were heat-inactivated at Flow cytometry for the detection of specific humoral 56 °C. Spleens were collected at endpoint and disaggregated responses against Expi293F, B16F10 and B16F10Min cells through a 70 µm mesh (Greiner Bio-One, Kremsmünster, Ger- Concentration of murine antibodies targeting proteins on the many). Splenocytes were cryopreserved in Fetal Bovine Serum surface of Expi293F, B16F10, or B16F10Min cells was determined (FBS) with 10% dimethyl sulfoxide (DMSO) in liquid nitrogen until by flow cytometry. Cells were incubated with mouse sera samples further use. for 30 min at room temperature (RT), washed with PBS + 10%FBS, and then incubated for 15 min at RT with a secondary In vivo protection assay AlexaFluor647-labelled goat anti-mouse IgG Fc (115-605-071; A C57BL/6JolaHsd mouse model inoculated with a poorly Jackson ImmunoResearch). Cells were acquired in a FACSCelesta™ immunogenic, Min-expressing, syngeneic tumour cell line (Sup- Flow Cytometer (BD) and mean fluorescence intensity (MFI) was plementary Fig. 7a–c) was employed to functionally assess analysed with FlowJo_v10.6.1 (BD). To evaluate the specific Min(RA)Gag-VLP-induced immune responses. Briefly, a B16F10 response induced against Min, the MFI ratio of B16F10Min and melanoma cell line (ATCC, Manassas, VA, USA) was stably B16F10 cells was calculated. transfected with pcDNA3.1-Min and cells were selected with DMEM supplemented with 2 mg/mL geneticin and 10% heat- In vitro Neutralisation assay inactivated FBS (ThermoFisher Scientific). Geneticin-resistant Min- Antibody neutralisation capacity of mouse sera was analysed by expressing cells were stained with the 10E8 antibody (NIH HIV an in vitro pseudoviral neutralisation assay using four subtype B Reagent Program) at 1 µg/mL and single-cell sorted with a BD HIV-1 Env (Bal.01, NL4-3, AC10.0.29, TRO.11) (NIH HIV Reagent FACSAriaIII Cell Sorter (BD) (Supplementary Fig. 7). Program), an HIV-2/HIV-1 MPER chimaera (7312 A/C1) and an C57BL/6JolaHsd mice were immunised twice with purified Gag- HIV-2 Env control (7312 A) plasmids58. Pseudoviruses were VLPs or MinGag-VLPs. Two weeks after the last immunisation, mice generated by co-transfection of Expi293F using Expifecta- were subcutaneously inoculated with 100,000 B16F10Min- mine293 co-transfected with each Env-expressing plasmid and expressing cells in the right flank (Fig. 6a). Tumour length and pSG3Δenv (NIH HIV Reagent Program). Pseudovirus-containing width was measured periodically with a calliper and tumour supernatants were harvested 48 h post-transfection59. For the volume was calculated as follows57: 0.52*(length*(width2)). neutralisation assay, serum samples were diluted 1/100 in DMEM Humane endpoint was reached when tumour size was bigger supplemented with 10% FBS and incubated for 60 min with than 1 cm3. pseudoviruses at 37 °C. After that, 10,000 TZM-bl cells/well (NIH HIV Reagent Program) were added and cultured for 48 h in the Production of recombinant proteins presence of the pseudoviruses/mouse sera mix and 7.5 µg/mL of Recombinant Gag and MinTT (MinTT contained a fragment of HR2, DEAE-dextran for 48 h. Readout was obtained using BriteLite the MPER, the gp41 transmembrane domain and the tetanus plus Reporter Gene Assay System (Perkin Elmer)16,60 and npj Vaccines (2023) 51 Published in partnership with the Sealy Institute for Vaccine Sciences F. Tarrés-Freixas et al. 11 bioluminescence detected with an Ensight Multimode Plate REFERENCES Reader (Perkin Elmer). 1. Amitai, A., Chakraborty, A. K. & Kardar, M. The low spike density of HIV may have evolved because of the effects of T helper cell depletion on affinity maturation. PLoS Comput. Biol. 14, e1006408 (2018). Binding to CD16-2 2. Klein, J. S. & Bjorkman, P. J. Few and far between: how HIV may be evading Determination of anti-Min antibody binding to mouse CD16-2 antibody avidity. PLoS Pathog. 6, e1000908 (2010). (mFcγRIV) receptor was assessed by ELISA. Nunc MaxiSorp plates 3. Bachmann, M. F. et al. The influence of antigen organization on B cell respon- were coated overnight at 4 °C with 500 ng/well of MinTT in 0.2 M siveness. Science 262, 1448–1451 (1993). carbonate/bicarbonate buffer at pH = 9.6; background was 4. Batista, F. D. & Neuberger, M. S. Affinity dependence of the B cell response to determined in uncoated wells. Then, plates were blocked for 2 h antigen: a threshold, a ceiling, and the importance of off-rate. Immunity 8, at RT with 1% BSA and diluted sera samples from immunised mice 751–759 (1998). 5. Kato, Y. et al. Multifaceted effects of antigen valency on B cell response com- were added and incubated overnight at 4 °C. After washing, plates position and differentiation in vivo. Immunity 53, 548–563.e8 (2020). were incubated overnight at 4 °C with 1 µg/mL of biotinylated 6. Viant, C. et al. Antibody affinity shapes the choice between memory and germinal mouse CD16-2 recombinant protein (AcroBiosystems, Newark, DE, center B cell fates. Cell 183, 1298–1311.e11 (2020). USA). CD16-2 binding to Min antibodies was detected using HRP- 7. Heath, P. T. et al. Safety and efficacy of NVX-CoV2373 Covid-19 vaccine. N. Engl. J. Streptavidin (ThermoFisher Scientific) for 30 min at RT and TMB Med. 385, 1172–1183 (2021). substrate. Reaction was stopped with 100 µl of 1 N H2SO4. 8. Brouwer, P. J. M. et al. Enhancing and shaping the immunogenicity of native-like Absorbance was read at 450 nm with background subtraction at HIV-1 envelope trimers with a two-component protein nanoparticle. Nat. Com- 620 nm using an Ensight Plate Reader (Perkin Elmer). mun. 10, 4272 (2019). 9. Hardy, G. J. et al. HIV-1 antibodies and vaccine antigen selectively interact with lipid domains. Biochim. Biophys. Acta 1838, 2662–2669 (2014). IFN-γ ELISpot assay 10. Jardine, J. G. et al. HIV-1 broadly neutralizing antibody precursor B cells revealed Splenocytes from vaccinated animals were seeded at 0.4 × 106 by germline-targeting immunogen. Science 351, 1458–1463 (2016). cells/well in ELISpot white PVDF plates (Merck & Co.) precoated 11. Krebs, S. J. et al. Multimeric scaffolds displaying the HIV-1 envelope MPER induce with 0.2 µg/well of anti-mouse IFN-γ antibodies (AN18; BioLegend, MPER-specific antibodies and cross-neutralizing antibodies when co-immunized with gp160 DNA. PLoS ONE 9, e113463 (2014). San Diego, CA, USA) and blocked with 10% FBS-supplemented 12. Sliepen, K. et al. Presenting native-like HIV-1 envelope trimers on ferritin nano- RPMI (R10). Cells were stimulated overnight at 37 °C and 5% CO2 particles improves their immunogenicity. Retrovirology 12, 82 (2015). with either a pool of all overlapping Min peptides (Supplementary 13. Tokatlian, T. et al. Enhancing humoral responses against HIV envelope trimers via Fig. 4a; 15 a.a. long overlapped by 11 residues; Covalab, Bron, nanoparticle delivery with stabilized synthetic liposomes. Sci. Rep. 8, 16527 France) or a pool of 20 overlapping Gag peptides (18 a.a. long (2018). overlapped by 9 residues covering p24, p2 and p7) at a 14. Wieczorek, L. et al. Comparable antigenicity and immunogenicity of oligomeric concentration of 14 µg/mL per peptide. Concanavalin A (ConA; forms of a novel, acute HIV-1 subtype C gp145 envelope for use in preclinical and 7 µg/mL; Merck & Co.) and R10 + 0.5% DMSO were used as clinical vaccine research. J. Virol. 89, 7478–7493 (2015). positive and negative controls, respectively. IFN-γ secretion was 15. Ingale, J. et al. High-density array of well-ordered HIV-1 spikes on synthetic liposomal nanoparticles efficiently activate B cells. Cell Rep. 15, 1986–1999 (2016). detected with an anti-mIFN-γ biotinylated-mAb (R4-6A2; BioLe- 16. Molinos-Albert, L. M. et al. Proteoliposomal formulations of an HIV-1 gp41-based gend) at a 1:2000 dilution and streptavidin-AP (Mabtech, miniprotein elicit a lipid-dependent immunodominant response overlapping the Cincinnati, OH, USA) and developed with AP Conjugate Substrate 2F5 binding motif. Sci. Rep. 7, 40800 (2017). Kit (Bio-Rad) at a 1:2000 dilution. 17. Zhang, P. et al. A multiclade env–gag VLP mRNA vaccine elicits tier-2 HIV-1- neutralizing antibodies and reduces the risk of heterologous SHIV infection in Statistical analysis macaques. Nat. Med. 27, 2234–2245 (2021). 18. Cervera, L. et al. Production of HIV-1-based virus-like particles for vaccination: Statistical analyses were performed using Prism 9.0 (GraphPad achievements and limits. Appl. Microbiol. Biotechnol. 103, 7367–7384 (2019). software Inc., CA, USA) and R-3.6.3 (R Foundation for Statistical 19. Carlson, L.-A. et al. Three-dimensional analysis of budding sites and released virus Computing). Multiple comparisons were performed using suggests a revised model for HIV-1 morphogenesis. Cell Host Microbe 4, 592–599 Kruskal–Wallis tests with Dunn’s comparison or Tukey’s multiple (2008). comparison test. Unpaired data were compared using a non- 20. Lavado-García, J. et al. Characterization of HIV-1 virus-like particles and deter- parametric Mann–Whitney test. Bioluminescence assay curves mination of Gag stoichiometry for different production platforms. Biotechnol. were compared using nested mixed effect models and considering Bioeng. 118, 2660–2675 (2021). 21. Buchbinder, S. P. et al. Immunogenicity of a novel Clade B HIV-1 vaccine com- time as a categorical factor with a likelihood ratio test. Survival in a bination: results of phase 1 randomized placebo controlled trial of an HIV-1 GM- Kaplan–Meier table was analysed using a Mantel–Cox test. A two- CSF-expressing DNA prime with a modified vaccinia Ankara vaccine boost in sided p value below 0.05 was considered statistically significant. healthy HIV-1 uninfected adults. PLoS ONE 12, e0179597 (2017). 22. Perdiguero, B. et al. A novel MVA-based HIV vaccine candidate (MVA-gp145-GPN) Reporting summary co-expressing clade C membrane-bound trimeric gp145 Env and Gag-induced virus-like particles (VLPs) triggered broad and multifunctional HIV-1-specific T cell Further information on research design is available in the Nature and antibody responses. Viruses 11, 160 (2019). Research Reporting Summary linked to this article. 23. Chapman, R. et al. Immunogenicity of HIV-1 vaccines expressing chimeric envelope glycoproteins on the surface of Pr55 Gag virus-like particles. Vaccines 8, 54 (2020). DATA AVAILABILITY 24. Gonelli, C. A. et al. Immunogenicity of HIV-1-based virus-like particles with All sequences used are properly described in the material and methods section. Data increased incorporation and stability of membrane-bound Env. Vaccines 9, 239 are available upon request to the authors. (2021). 25. Beltran-Pavez, C. et al. Guiding the humoral response against HIV-1 toward a MPER adjacent region by immunization with a VLP-formulated antibody-selected CODE AVAILABILITY envelope variant. PLoS One 13, e0208345 (2018). All statistical analyses displayed in Figs. 1–6 were performed with publicly available R 26. Crooks, E. T. et al. Effects of partially dismantling the CD4 binding site glycan packages using R-3.6.3 (R Foundation for Statistical Computing). fence of HIV-1 envelope glycoprotein trimers on neutralizing antibody induction. Virology 505, 193–209 (2017). 27. Tohidi, F., Sadat, S. M., Bolhassani, A., Yaghobi, R. & Larijani, M. S. Induction of a Received: 6 May 2022; Accepted: 17 March 2023; robust humoral response using HIV-1 VLP(MPER-V3) as a novel candidate vaccine in BALB/c mice. Curr. HIV Res. 17, 33–41 (2019). Published in partnership with the Sealy Institute for Vaccine Sciences npj Vaccines (2023) 51 F. Tarrés-Freixas et al. 12 28. Xiao, P. et al. Parainfluenza virus 5 priming followed by SIV/HIV virus-like-particle 58. Montefiori, D. C. Measuring HIV neutralization in a luciferase reporter gene assay. boosting induces potent and durable immune responses in nonhuman primates. Methods Mol. Biol. 485, 395–405 (2009). Front. Immunol. 12, 162 (2021). 59. Sánchez-Palomino, S. et al. A cell-to-cell HIV transfer assay identifies humoral 29. Storcksdieck genannt Bonsmann, M., Niezold, T., Hannaman, D., Überla, K. & responses with broad neutralization activity. Vaccine 29, 5250–5259 (2011). Tenbusch, M. The improved antibody response against HIV-1 after a vaccination 60. Li, M. et al. Human immunodeficiency virus type 1 env clones from acute and based on intrastructural help is complemented by functional CD8+ T cell early subtype B infections for standardized assessments of vaccine-elicited responses. Vaccine 34, 1744–1751 (2016). neutralizing antibodies. J. Virol. 79, 10108–10125 (2005). 30. Huang, X. et al. In vivo electroporation in DNA-VLP prime-boost preferentially enhances HIV-1 envelope-specific IgG2a, neutralizing antibody and CD8 T cell responses. Vaccine 35, 2042–2051 (2017). ACKNOWLEDGEMENTS 31. Storcksdieck genannt Bonsmann, M. et al. Enhancing the quality of antibodies to We are grateful to Y. Rosales from CMCiB for her excellent technical help with the HIV-1 envelope by GagPol-specific Th cells. J. Immunol. 195, 4861–4872 (2015). immunisation protocols. We acknowledge M. de Cabo from the UAB Electron 32. Crooks, E. T. et al. Vaccine-elicited tier 2 HIV-1 neutralizing antibodies bind to Microscopy service for his help with cryo-EM. We would also like thank M. A. quaternary epitopes involving glycan-deficient patches proximal to the CD4 binding site. PLoS Pathog. 11, e1004932 (2015). Fernández for providing his expertise in FACS single-cell sorting. We especially thank 33. Escolano, A. et al. Immunization expands B cells specific to HIV-1 V3 glycan in F. Cunyat for providing insightful knowledge on CD16-2 and ADCC techniques. The mice and macaques. Nature 570, 468–473 (2019). following reagents were obtained through the NIH AIDS Reagent Program, Division 34. Stano, A. et al. Dense array of spikes on HIV-1 virion particles. J. Virol. 91, of AIDS, NIAID, NIH: Anti-HIV-1 gp41 Monoclonal (D50) (ARP-11393), anti-HIV-1 gp41 e00415–e00417 (2017). mAb (10E8) (ARP-12294), TZM-bl cells (ARP-8129), HIV-1 SG3Δenv non-infectious 35. Molinos-Albert, L. M., Clotet, B., Blanco, J. & Carrillo, J. Immunologic insights on molecular clone (ARP-11051) and subtype B envelopes (NL4-3 – ARP-114; BaL.01 – the membrane proximal external region: a major human immunodeficiency virus ARP-11445; TRO.11 – ARP-11023; AC10.0.29 – ARP-11024). HIV-2 7312 A and HIV-2/ type-1 vaccine target. Front. Immunol. 8, 1154 (2017). HIV-1 MPER chimera p7312A-C1 Env plasmids were kindly provided by G.M. Shawn 36. Molinos-Albert, L. M. et al. Anti-MPER antibodies with heterogeneous neu- (University of Pennsylvania). We also thank Foundation Dormeur for equipment tralization capacity are detectable in most untreated HIV-1 infected individuals. support. This work has been funded by the Ministerio de Economía y Competitividad of Retrovirology 11, 44 (2014). the Spanish Government, co-financed with the European Regional Development 37. Oliferenko, S. et al. Analysis of CD44-containing lipid rafts: recruitment of annexin funds (FEDER), by the Instituto de Salud Carlos III through the projects PI14/01307, II and stabilization by the actin cytoskeleton. J. Cell Biol. 146, 843–854 (1999). PI17/01518, PI20/00093 and by the CERCA Programme/Generalitat de Catalunya 2017 38. Kim, J. H., Hartley, T. L., Curran, A. R. & Engelman, D. M. Molecular dynamics SGR 252. EP was supported by a doctoral grant from National Agency for Research studies of the transmembrane domain of gp41 from HIV-1. Biochim. Biophys. Acta. and Development of Chile (ANID): Grant 72180406. PB was supported by La Caixa 1788, 1804–1812 (2009). 39. Excler, J.-L. & Kim, J. H. Novel prime-boost vaccine strategies against HIV-1. Expert Foundation (ID 100010434), under the agreement LCF/PR/GN17/5300003. CA-N and Rev. Vaccines 18, 765–779 (2019). AP-G were supported by a predoctoral grant from Secretaria d’Universitats de la 40. Atmar, R. L. et al. Homologous and heterologous Covid-19 booster vaccinations. Generalitat de Catalunya and the European Social Funds (2020 FI_B 00742 and 2022 N. Engl. J. Med. 386, 1046–1057 (2022). FI_B 00698, respectively). The funders had no role in the study design, data collection 41. Lux, A. & Nimmerjahn, F. Of mice and men: the need for humanized mouse and analysis, the decision to publish, or the preparation of the manuscript. This work models to study human IgG activity in vivo. J. Clin. Immunol. 33, 4–8 (2013). has also been sponsored partly by Grífols (Indivac). 42. Watzl, C. & Long, E. O. Signal transduction during activation and inhibition of natural killer cells. Curr. Protoc. Immunol. 90, 11.9B.1–11.9B.17 (2010). 43. Collins, A. M. IgG subclass co-expression brings harmony to the quartet model of AUTHOR CONTRIBUTIONS murine IgG function. Immunol. Cell Biol. 94, 949–954 (2016). F.T.-F., J.C. and J.B. conceived and designed the experiments in this manuscript. 44. Verkoczy, L. et al. Rescue of HIV-1 broad neutralizing antibody-expressing B cells L.M.M.-A., B.C., J.C. and J.B. designed and developed the platform. M.L.R.-C., S.M., in 2F5 VH x VL knockin mice reveals multiple tolerance controls. J. Immunol. 187, L.M.M.-A., L.C., S.G.-G., M.M.S., F.G., B.C., J.C. and J.B. designed and performed 3785–3797 (2011). 45. Verkoczy, L. et al. Induction of HIV-1 broad neutralizing antibodies in 2F5 knock-in preliminary studies with the platform. F.T.-F., C.A.-G., M.L.R.-C., B.T., R.O.-L., A.B., C.A.-N., mice: selection against membrane proximal external region-associated auto- J.C. and J.B. optimised the platform. F.T.-F., C.A.-G., B.T., R.O., E.P., P.B., A.P.-G., C.A.-N., reactivity limits T-dependent responses. J. Immunol. 191, 2538–2550 (2013). I.V. and J.C. contributed with the animal procedures, sample processing and 46. Wang, J., Saffold, S., Cao, X., Krauss, J. & Chen, W. Eliciting T cell immunity against experimental analysis. F.T.-F., V.U., J.C. and J.B. built the figures and performed the poorly immunogenic tumors by immunization with dendritic cell-tumor fusion statistical analysis. F.T.-F., J.C. and J.B. wrote the manuscript. All authors corrected the vaccines. J. Immunol. 161, 5516 LP–5524 (1998). manuscript and approved the submitted version. 47. Graham, B. S., Gilman, M. S. A. & McLellan, J. S. Structure-based vaccine antigen design. Annu. Rev. Med. 70, 91–104 (2019). 48. Elsayed, H. et al. Intrastructural help: harnessing T helper cells induced by COMPETING INTERESTS licensed vaccines for improvement of HIV Env antibody responses to virus-like Unrelated to the submitted work, J.C., J.B. and B.C. are founders and shareholders of particle vaccines. J. Virol. 92, e00141–18 (2018). AlbaJuna Therapeutics, S.L.; B.C. is founder and shareholder of AELIX Therapeutics, 49. Moyer, T. J., Zmolek, A. C. & Irvine, D. J. Beyond antigens and adjuvants: for- S.L.; J.B. reports institutional grants from HIPRA, NESAPOR Europe and MSD. The mulating future vaccines. J. Clin. Invest. 126, 799–808 (2016). 50. Borobia, A. M. et al. Immunogenicity and reactogenicity of BNT162b2 booster in authors declare no other competing conflicts of interest. This work is protected by ChAdOx1-S-primed participants (CombiVacS): a multicentre, open-label, rando- intellectual property rights through the patent WO/2018/020324, developed at the mised, controlled, phase 2 trial. Lancet (Lond., Engl.) 398, 121–130 (2021). IrsiCaixa AIDS Research Institute and published by Dr. J. Blanco, Dr. J. Carrillo and Dr. 51. Normark, J. et al. Heterologous ChAdOx1 nCoV-19 and mRNA-1273 vaccination. L.M. Molinos-Albert (co-authors of this work). N. Engl. J. Med. 385, 1049–1051 (2021). 52. Watanabe, H., Numata, K., Ito, T., Takagi, K. & Matsukawa, A. Innate immune response in Th1- and Th2-dominant mouse strains. Shock 22, 460–466 (2004). ADDITIONAL INFORMATION 53. Rerks-Ngarm, S. et al. Vaccination with ALVAC and AIDSVAX to prevent HIV-1 Supplementary information The online version contains supplementary material infection in Thailand. N. Engl. J. Med. 361, 2209–2220 (2009). available at https://doi.org/10.1038/s41541-023-00648-4. 54. Weigel, T. et al. A flow-through chromatography process for influenza A and B virus purification. J. Virol. Methods 207, 45–53 (2014). Correspondence and requests for materials should be addressed to Jorge Carrillo or 55. Steppert, P. et al. Purification of HIV-1 gag virus-like particles and separation of Julià Blanco. other extracellular particles. J. Chromatogr. A 1455, 93–101 (2016). 56. Kamala, T. Hock immunization: a humane alternative to mouse footpad injec- Reprints and permission information is available at http://www.nature.com/ tions. J. Immunol. Methods 328, 204–214 (2007). reprints 57. Mei, K.-C., Bai, J., Lorrio, S., Wang, J. T.-W. & Al-Jamal, K. T. Investigating the effect of tumor vascularization on magnetic targeting in vivo using retrospective design Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims of experiment. Biomaterials 106, 276–285 (2016). in published maps and institutional affiliations. npj Vaccines (2023) 51 Published in partnership with the Sealy Institute for Vaccine Sciences F. Tarrés-Freixas et al. 13 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http:// creativecommons.org/licenses/by/4.0/. © The Author(s) 2023 Published in partnership with the Sealy Institute for Vaccine Sciences npj Vaccines (2023) 51