Structure-function relationship of S protein cleavage in cowpea mosaic virus intratumoral immunotherapy

Abstract

Translational studies are underway for cowpea mosaic virus (CPMV) as a candidate for intratumoral immunotherapy. To further strengthen the argument for translation, here we demonstrate that a structural deviation in CPMV's small (S) coat protein does not negatively impact its antitumor efficacy, immune cell uptake, or cytokine profile.

---

The structure of cowpea mosaic virus (CPMV) was first solved in 1999 via X-ray diffraction (XRD) and is listed on the Protein Data Bank (PDB) under ID 1NY7.¹ CPMV comprises 60 identical copies each of a large (L, 42 kDa) and small (S, 24 kDa) CP forming an ∼30 nm icosahedral virus with pseudo-T = 3 symmetry. CPMV encapsidates bipartite positive-sense single-stranded (ss) RNAs (1 = 6.0 knts and 2 = 3.5 knts). The same year, a region of 24 amino acids forming the C-terminus of the CPMV S protein was mapped and investigated for its role in virus replication.² In the subsequent ∼15 years, it was discovered that this C-terminal peptide is vital to CPMV's viral replication cycle and plays a central role in RNA packaging, capsid assembly, and viral movement in plants, but once the particle matures, it is proteolytically cleaved.^2–4 Although it was not known at the time of publication, the method of preparing CPMV for XRD in the 1999 study induced cleavage of the C-terminal portion of the S protein. Later, virus-like particles (VLPs) of CPMV devoid of RNA (termed eCPMV) were produced that cleaved slower than the wild-type (WT) CPMV, allowing the structure of the CPMV with full-length S protein to be solved first by cryo-electron microscopy in 2015, and then by XRD in 2016; these structures are listed on the PDB under IDs 5FMO and 5A33, respectively.^5,6Fig. 1A shows the structure of CPMV, highlighting the S protein and its C-terminal fragment.

Fig. 1 Characterization differences between typical, uncleaved, and cleaved CPMV. (A) The S protein of CPMV experiences proteolytic cleavage of a 24 amino acid sequence (highlighted in yellow) in its c-terminus, which is shown in these surface representations as conversion from the uncleaved, full length S protein in CPMV PDB structure 5FMO to the cleaved CPMV structure in 1NY7. Images were created using UCSF Chimera version 1.16. Note that 5FMO is an empty CPMV (no RNA) while 1NY7 contains RNA, but there is no PDB ID representing cleaved, RNA-containing CPMV with which to create a structure on Chimera, so 5FMO is used here for visualization purposes. (B) Denaturing (left) and native (right) gel electrophoresis methods reveal differences in electrophoretic mobility of the S protein (24 kDa) between the three cleavage states of CPMV. L-protein (42 kDa) remains unchanged. (C) Structural characterization studies via DLS, SEC, and TEM show no significant differences between CPMV, CPMV-S_FL, and CPMV-S_C.

CPMV is a member of the Secoviridae family within the Picornavirales order, and although it closely resembles mammalian picornaviruses, it does not infect or replicate in mammalian cells.⁷ In 2016, we discovered that CPMV is, however, a potent immunomodulator when administered intratumorally.⁸ As an intratumoral immunotherapy agent, CPMV is taken up by tumour-resident immune cells to initiate a cascade to remodel the tumour microenvironment (TME) and reactivate the immune system through multiple mechanisms. The repetitive nature of its capsid proteins presents as pathogen-associated molecular patterns (PAMPs) enabling recognition by pattern recognition receptors (PRRs) including toll-like receptors (TLRs) 2 and 4; furthermore, its ssRNA is recognized by TLR 7. Innate immune cell activation by CPMV releases inflammatory cytokines and chemokines, including type I interferons.⁹ The inflammatory response drives the infiltration of innate immune cells (M1 macrophages, N1 neutrophils, and dendritic cells) and natural killer (NK) cells into the TME. CPMV also engages with the complement system by presentation of T helper epitopes and induces production of antiviral T cells, thus further strengthening the immune response.¹⁰ It is the combination of these recruited and activated cells that causes tumour cell death and revamps the cancer immunity cycle to eliminate tumours.

After the initial discovery of CPMV's potential as an intratumoral immunotherapy agent, we have delineated its mechanism of action^9,11 and proven its efficacy in multiple murine tumour models^12,13 and canine cancer patients^14–16 as a solo therapy, neoadjuvant therapy, and in combination therapy. These developments have prompted interest in translational studies. As part of that endeavour, we aim to elucidate which components of CPMV make it uniquely efficacious and ensure that its production method can provide a consistently effective drug candidate. While eCPMV shows potent anti-tumour efficacy,⁸ RNA-laden CPMV is more potent than eCPMV or inactivated CPMV (with crosslinked ssRNA),^17,18 which is explained by ssRNA providing signalling through TLR 7, a key driver of type-I interferon release. Of note, there is no difference in efficacy when comparing CPMV containing RNA-1 vs. RNA-2, which indicates that RNA identity is not critical.¹⁹ In a recent study detailing longitudinal storage stability of CPMV, we found that long-term refrigeration (6+ months at 4 °C) caused the anti-tumour potency to decline.²⁰ The only discernible difference between long-term refrigerated CPMV vs. fresh or frozen CPMV was the composition of the S protein: fresh and frozen CPMV contained a mixture of full-length and cleaved S, while the long-term refrigerated CPMV contained cleaved S.

To investigate whether the C-terminal peptide of the S protein is critical for its immunomodulatory properties and anti-tumour potency, here we have compared the efficacy of distinct CPMV preparations (Fig. 1A):

• CPMV: a typical CPMV preparation with a mixture of CPMV-S_C (C = cleaved) and CPMV-S_FL (FL = full-length);

• CPMV-S_FL: a CPMV preparation from an early harvest, which contains primarily uncleaved, full-length S protein; and

• CPMV-S_C: CPMV containing cleaved S protein obtained through enzymatic processing using established methods.³

Purified CPMV preparations were characterized to assay the state of the S protein: denaturing and native gel electrophoresis indicated that typical CPMV preparations (harvested 14 days post infection, dpi) comprised a mixture of ∼30% S_FL and ∼70% S_C, as evidenced by CPMV particles with slow and fast electrophoretic mobility (Fig. 1B, lane 1). In CPMV-S_FL preparations (harvested 7 dpi), the majority of – albeit not all – particles contained the full-length S with a ratio of ∼74% S_FLvs. ∼26% S_C, Fig. 1B, lane 2. We tested whether the ratio of full-length to cleaved S could be further increased through harvesting 3 or 5 dpi; however, this did not substantially increase the ratio of S_FL : S_C but rather decreased the overall yield (Fig. S1). Therefore, we continued with preparations harvested 7 days post infection. Finally, α-chymotrypsin treatment resulted in 100% CPMV-S_C (Fig. 1B, lane 3).

To confirm the structural integrity and purity of the CPMV preparations, the samples were characterized via dynamic light scattering (DLS), size exclusion chromatography (SEC), and transmission electron microscopy (TEM). All three preparations had the typical characteristics of CPMV: DLS confirmed a hydrodynamic diameter of 30–35 nm, SEC confirmed an elution volume of 11.4–11.6 mL from the Superose 6 Increase column, and TEM showed visibly intact and monodisperse particles of ∼30 nm diameter. A subtle difference, however, was noted: CPMV-S_C appeared slightly smaller with a hydrodynamic radius of 31.85 nm (PDI = 0.052) vs. 34.55 nm (PDI = 0.125) for CPMV and 35.56 nm (PDI = 0.141) for CPMV-S_FL – the increase in PDI for the latter is consistent with a mixture of particles. This was reflected by SEC showing elution at 11.62 mL for CPMV-S_Cversus 11.44 and 11.39 mL for CPMV and CPMV-S_FL, respectively. In a thermal shift assay to evaluate structural stability via melting point, there was an additional subtle difference: CPMV-S_FL had the highest melting point, followed by CPMV and then CPMV-S_C, aligning from the most intact S protein to the least (Fig. S2). We also evaluated the ssRNA integrity encapsulated within each CPMV preparation by extracting the RNA and analysing it through agarose gel electrophoresis (Fig. S3). All three CPMV preparations had 260/280 ratios over 2.0, indicating pure RNA (measured through a NanoDrop UV–Vis), and the agarose gel showed intact ssRNA-1 (6.0 knts) and ssRNA-2 (3.5 knts) for each version of CPMV.

To test whether the subtle structural differences impacted the anti-tumour efficacy of CPMV, we tested the various formulations against dermal B16F10 melanoma and A20 murine diffuse large B cell lymphoma. All animal procedures were performed in accordance with the Guidelines of Care and Use of Laboratory Animals of the University of California, San Diego (UCSD), and approved by the Animal Ethics Committee of UCSD's Institutional Animal Care and Use Committee (IACUC) under protocol S18021. Treatments began when intradermal tumours reached a volume of ∼20 mm³, on day 8 for B16F10 or day 10 for A20. In the B16F10 model, we tested a dose of 100 μg CPMV vs. CPMV-S_Cvs. CPMV-S_FL once per week for a total of 3 weeks. By day 15, there was an evident divergence in tumour growth among CPMV treatment groups leading to statistically significant reductions in tumour growth and burden: PBS **** vs. CPMV (p < 0.0001), *** vs. CPMV-S_C (p = 0.0002), and * vs. CPMV-S_FL (p = 0.0398), calculated by 2-way ANOVA (Fig. 2A). While tumour growth in the CPMV-S_FL group appeared slightly accelerated compared to that in CPMV/CPMV-S_C, the differences were statistically insignificant. Ultimately, B16F10 tumours were not eliminated, but survival was prolonged vs. PBS with no significant differences compared to the CPMV formulations (Fig. S4).

Fig. 2 Antitumor activity of CPMV in different S protein cleavage states. 200 000 B16F10 murine melanoma cells (A) or A20 murine diffuse large B cell lymphoma cells (B) were inoculated intradermally in female C57BL/6 or Balb/c mice, respectively. (A) Mice were treated intratumorally starting on day 8 with 100 μg of CPMV, CPMV-S_FL, or CPMV-S_C (B16) or on day 10 with 20 μg or 100 μg of CPMV-S_FL or CPMV-S_C (A20). For both models, mice were subjected to treatments once per week for 3 weeks and were euthanized when tumor volumes exceeded 1000 mm³. Volumes are shown as means ± standard deviation (n = 10) and growth curves end when there are <3 mice remaining in each group. Individual tumor curves and survival curves show the full outcome in Fig. S2 and S3. Statistical analysis of growth curves was conducted using two-way ANOVA and survival curve significance was calculated using the Mantel-Cox log rank test, with **** = p < 0.0001, *** = p < 0.001, ** = p < 0.01, * = p < 0.05, and ns = p > 0.05. To reduce the number of mice that had to be sacrificed, groups in the A20 model (PBS control and CPMV-S_FL) were shared with another study, which is reported elsewhere: Opdensteinen et al.²¹

Using the A20 tumour model, we tested 20 μg and 100 μg doses of either CPMV-S_C or CPMV-S_FL. In numerous prior studies, we established 100 μg of CPMV as a suitable dose for mouse tumour studies. While the 100 μg dose of CPMV eliminates A20 tumors,¹² this dose is generally not curative in the more aggressive B16F10 model. Here, we opted for the 20 μg and 100 μg dose levels to ensure that differences could be observed in both tumour mouse models. PBS control mice reached the endpoint by day 31, at which point CPMV-treated groups had significantly smaller tumours (****, p < 0.0001 from 2-way ANOVA, Fig. 2B) – but there were no differences in tumour sizes when comparing CPMV formulations, independent of dose (Fig. S5). 4–5 mice survived at the study endpoint for CPMV-S_C (20 μg and 100 μg dose) and CPMV-S_FL (100 μg). Only 1 mouse survived at the lower dose of CPMV-S_FL (20 μg) – however, there was no difference in the probability of survival (vs. CPMV-S_C 20 μg, p = 0.1490). There were also no differences in survival between CPMV-S_C and CPMV-S_FL at 100 μg or between low and high doses of CPMV-S_C.

While differences in anti-tumour efficacy were not apparent, we sought to investigate whether differences could be detected on a cellular level and assayed cell uptake rates and cytokine release profiles. First, we used RAW264.7 macrophages to study cell uptake because CPMV uptake in this cell line is well-documented;⁷ here, we added 10⁵ or 10⁶ CPMV particles (CPMV, CPMV-S_C, and CPMV-S_FL)/RAW264.7 macrophages and evaluated cell uptake by flow cytometry. Across two separate trials (Fig. 3A and Fig. S6), minor differences were noted: overall, all formulations showed high cell uptake rates, with CPMV showing slightly increased rates compared to CPMV-S_C and CPMV-S_FL (based on MFI). This is interesting because the CPMV sample, structurally, sits in between the CPMV-S_C (0%, S_FL, 100% S_C) and CPMV-S_FL (>74% S_FL, 26% S_C) samples with 30% S_FL and 70% S_C. It is possible that subtle differences in concentration measurements or additional processing steps and purification of the CPMV-S_C sample impact cell uptake – it is also possible that there are differences in antibody recognition as a function of S protein cleavage; nevertheless, there appears to be no emerging trend from this data. A limitation of this study is that RAW264.7 macrophages do not fully represent the complex tumour microenvironment. However, because there were also no differences when testing human PBMCs (see Fig. 3B and C and Fig. S7), we opted not to test additional immune cells.

Fig. 3 In vitro assays to probe cell uptake and interferon-stimulating abilities between CPMV cleavage states. (A) S cleavage may minimally affect CPMV uptake by RAW 264.7 macrophages. CPMV formulations (10⁵ or 10⁶ particles per cell) were incubated with macrophages for 6 h, then stained with α-CPMV antibodies and analysed by flow cytometry. Statistical significance was calculated using 1-way ANOVA with Tukey's multiple comparison test. Minor differences were noted between this trial and an additional trial in Fig. S6, likely attributed to slight variations in laser strength and gating. (B and C) Human PBMCs were incubated with a PBS negative control (NC), 5 μg mL⁻¹ ODN2216 positive control (PC), or 20 μg mL⁻¹ CPMV, CPMV-S_FL, or CPMV-S_C for 8, 24, 48, and 72 h. Cytokine activation was measured using a 4-plex interferon panel. (B) Heat map showing the mean response of two independent samples (n = 2) tested in PBMC cultures of 2 donors (I3L4 and Q3G6) at four time points for four cytokines: IFNα, IFNβ, IFNγ, and IFNω. (C) Individual graphs of the four cytokines from (B), showing consistent induction between all CPMV samples for IFNα, IFNγ, and IFNω, and slightly lower induction by CPMV-S_FL of IFNβ. Each bar shows the mean response ± SD of two donor cultures (n = 2). Dots represent the mean response of two independent samples (n = 2) tested in individual donor cultures (black = Donor I3L4, blue = Donor Q3G6).

Next, we assayed cytokines – focusing on interferon responses, which were identified as a key signalling pathway for CPMV efficacy.^9,11 The research donor blood was obtained from healthy donor volunteers under the NCI-Frederick IRB-approved protocol OH99CN046 D, NCT # NCT00339911. This protocol is designed to establish and operate a Research Donor Program that meets the requirements for the protection of human subjects from research risks, as detailed in the NIH Multiple Projects Assurance with the OPRR and is in compliance with the Code of Federal Regulations, Public Welfare, 45 CFR 46: Protection of Human Subjects, Office for Protection from Research Risks, U.S. Government Printing Office, Washington, DC, 1997. Informed consent was obtained from all human participants of the NCI-Frederick Research Donor Program. Human peripheral blood mononuclear cells (PBMCs) were stimulated with CPMV formulations and no differences in IFNα, IFNλ, or IFNω levels were noted (Fig. 3B and C). Some differences were noted for IFNβ levels with CPMV-S_FL stimulation, resulting in slightly lower levels – however, considering all data, these differences do not elucidate any trends and are likely within experimental variation. Conversely, data instead suggest that IFN response varies between donors and depends on the type of IFN rather than the S protein cleavage state (Fig. 3B and C). For all samples, IFNα and IFNβ response initiates after 8 h, whereas IFNω and IFNλ appear later at 24 h. While all IFN responses peak at 24 h, IFNβ disappears at later time points and IFNα, IFNω, and IFNλ continue through 72 h. A lower level of IFNs was consistently produced in PBMCs from Donor Q3G6 than from Donor I3L4, but there were no discernible differences in relation to the S protein cleavage state. These findings were confirmed in PBMC cultures from 3 additional donors incubated with CPMV, CPMV-S_FL, and CPMV-S_C for 24 h (Fig. S7).

Together, data suggest that the status of the S protein – cleaved C-terminus or not – does not significantly – statistically or biologically – impact the anti-tumour efficacy of CPMV. Thus, differences that were noted in our earlier study comparing long-term refrigerated CPMV vs. fresh or frozen CPMV may be attributed to other structural changes that occur during long-term storage.

Conclusions

This study suggests that the state of the C-terminus of the S protein of CPMV does not impact its immunomodulatory activity and differences in potency were not observed as a function of S cleavage. This result helps define the drug product specification of the CPMV drug candidate.

Author contributions

Andrea Simms: investigation, methodology, data analysis, visualization, and writing – original draft. Jessica Fernanda Affonso de Oliveira: investigation and methodology. Narek Minasov and Edward Cedrone: investigation. Marina A. Dobrovolskaia: formal analysis and writing – review & editing. Nicole F. Steinmetz: conceptualization, funding acquisition, resources, and writing – review & editing.

Conflicts of interest

The authors declare the following competing financial interest(s): Dr Steinmetz is a co-founder and CEO of, has equity in, and has a financial interest in PlantiosX Inc. Dr Steinmetz is a co-founder of, has equity in, and has a financial interest in Mosaic ImmunoEngineering Inc. Dr Steinmetz is a co-founder and manager of Pokometz Scientific LLC, under which she is a paid consultant to Flagship Labs 95 Inc. The other authors declare no potential COI.

Data availability

The data supporting this article are available in the cancer Nanotechnology Laboratory (caNanoLab) data repository portal (https://cananolab.cancer.gov/#/).

Supplementary information: Detailed methods, additional CPMV characterization (early harvest SDS-PAGE and western blot), thermal shift assay, B16F10 survival and individual tumor growth curves, A20 individual tumor growth curves, and additional RAW264.7 cell uptake. See DOI: https://doi.org/10.1039/d5bm00969c.

Acknowledgements

This work was supported in part by the NIH (R01CA274640, R01CA224605, R01CA253615) as well as the American Cancer Society and F.M. Kirby Foundation Inc. Mission Boost Grant, MBGI-23-1030244-01-MBG. The study was funded in part (M.A.D. and E.C.) by federal funds from the National Cancer Institute, National Institutes of Health, under contract 75N91019D00024. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.

The authors would like to thank the University of California, San Diego – Cellular and Molecular Medicine Electron Microscopy Core (UCSD-CMM-EM Core, RRID:SCR_022039) for equipment access and technical assistance. The UCSD-CMM-EM Core is supported in part by the NIH (S10OD023527).

Molecular graphics and analyses were performed with UCSF Chimera, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from NIH (P41GM103311).

We would also like to thank Patrick Opdensteinen and Jamie Gatus for their assistance in monitoring animals for tumour treatment studies.

References

1. T. Lin, et al., The Refined Crystal Structure of Cowpea Mosaic Virus at 2.8 Å Resolution, Virology, 1999, 265, 20–34,  DOI:10.1006/viro.1999.0038.
2. K. M. Taylor, V. E. Spall, P. J. Butler and G. P. Lomonossoff, The cleavable carboxyl-terminus of the small coat protein of cowpea mosaic virus is involved in RNA encapsidation, Virology, 1999, 255, 129–137,  DOI:10.1006/viro.1998.9567.
3. F. Sainsbury, K. Saunders, A. A. Aljabali, D. J. Evans and G. P. Lomonossoff, Peptide-controlled access to the interior surface of empty virus nanoparticles, ChemBioChem, 2011, 12, 2435–2440,  DOI:10.1002/cbic.201100482.
4. Y. Meshcheriakova and G. P. Lomonossoff, Amino acids at the exposed C-terminus of the S coat protein of cowpea mosaic virus play different roles in particle formation and viral systemic movement, J. Gen. Virol., 2019, 100, 1165–1170,  DOI:10.1099/jgv.0.001285.
5. E. L. Hesketh, et al., Mechanisms of assembly and genome packaging in an RNA virus revealed by high-resolution cryo-EM, Nat. Commun., 2015, 6, 10113,  DOI:10.1038/ncomms10113.
6. N. T. Huynh, et al., Crystal Structure and Proteomics Analysis of Empty Virus-like Particles of Cowpea Mosaic Virus, Structure, 2016, 24, 567–575,  DOI:10.1016/j.str.2016.02.011.
7. A. O. Omole, et al., Cellular fate of a plant virus immunotherapy candidate, Commun. Biol., 2024, 7, 1382,  DOI:10.1038/s42003-024-06982-0.
8. P. H. Lizotte, et al., In situ vaccination with cowpea mosaic virus nanoparticles suppresses metastatic cancer, Nat. Nanotechnol., 2016, 11, 295–303,  DOI:10.1038/nnano.2015.292.
9. C. Mao, V. Beiss, J. Fields, N. F. Steinmetz and S. Fiering, Cowpea mosaic virus stimulates antitumor immunity through recognition by multiple MYD88-dependent toll-like receptors, Biomaterials, 2021, 275, 120914,  DOI:10.1016/j.biomaterials.2021.120914.
10. M. F. Bachmann and G. T. Jennings, Vaccine delivery: a matter of size, geometry, kinetics and molecular patterns, Nat. Rev. Immunol., 2010, 10, 787–796,  DOI:10.1038/nri2868.
11. A. O. Omole, et al., Comparative analyses for plant virus-based cancer immunotherapy drug development, Cell. Biomater., 2025, 100095,  DOI:10.1016/j.celbio.2025.100095.
12. J. F. Affonso de Oliveira, et al., Plant Virus Intratumoral Immunotherapy with CPMV and PVX Elicits Durable Antitumor Immunity in a Mouse Model of Diffuse Large B-Cell Lymphoma, Mol. Pharm., 2024, 21, 6206–6219,  DOI:10.1021/acs.molpharmaceut.4c00507.
13. A. Misiaszek, et al., Intratumoral Immunotherapy with Cowpea Mosaic Virus and Interleukin-12 Eliminates Established Treated Tumors and Generates Robust Systemic Immunity to Suppress the Growth of Untreated Metastatic Tumors, Mol. Pharm., 2025, 22(7), 3747–3756,  DOI:10.1021/acs.molpharmaceut.4c01539.
14. D. Alonso-Miguel, et al., Neoadjuvant in situ vaccination with cowpea mosaic virus as a novel therapy against canine inflammatory mammary cancer, J Immunother Cancer, 2022, 10(3) DOI:10.1136/jitc-2021-004044.
15. P. Sergent, et al., An Abscopal Effect on Lung Metastases in Canine Mammary Cancer Patients Induced by Neoadjuvant Intratumoral Immunotherapy with Cowpea Mosaic Virus Nanoparticles and Anti-Canine PD-1, Cells, 2024, 13(17) DOI:10.3390/cells13171478.
16. G. Valdivia, et al., Neoadjuvant Intratumoral Immunotherapy with Cowpea Mosaic Virus Induces Local and Systemic Antitumor Efficacy in Canine Mammary Cancer Patients, Cells, 2023, 12(18) DOI:10.3390/cells12182241.
17. C. Wang, V. Beiss and N. F. Steinmetz, Cowpea Mosaic Virus Nanoparticles and Empty Virus-Like Particles Show Distinct but Overlapping Immunostimulatory Properties, J. Virol., 2019, 93(21) DOI:10.1128/jvi.00129-19.
18. P. L. Chariou, V. Beiss, Y. Ma and N. F. Steinmetz, In situ vaccine application of inactivated CPMV nanoparticles for cancer immunotherapy, Mater. Adv., 2021, 2, 1644–1656,  10.1039/D0MA00752H.
19. E. Jung, Y. H. Chung, C. Mao, S. N. Fiering and N. F. Steinmetz, The Potency of Cowpea Mosaic Virus Particles for Cancer In Situ Vaccination Is Unaffected by the Specific Encapsidated Viral RNA, Mol. Pharm., 2023, 20, 3589–3597,  DOI:10.1021/acs.molpharmaceut.3c00214.
20. A. Simms, Z. Zhao, E. Cedrone, M. A. Dobrovolskaia and N. F. Steinmetz, Cowpea mosaic virus intratumoral immunotherapy maintains stability and efficacy after long-term storage, Bioeng. Transl. Med., 2024, 9, e10693,  DOI:10.1002/btm2.10693.
21. P. Opdensteinen, Toward Translation of Cowpea Mosaic Virus Intratumoral Immunotherapy With a Scalable Production Process, Plant Biotechnol. J., 2025 DOI:10.1111/pbi.70281.

---