Mouse cytokine results don't carry over: what Yingzhong's Q3 disclosure means for your experiment

If you are pricing out an experiment with SV-Delivery™ kits and the cargo is a cytokine (IL-2, IL-12, IL-15, IFN-α, etc.), there is a question on the 2026-05-15 panel transcript that should change your experimental design. An audience member asked Yingzhong Li: "Are you planning to test in a model where you test human versions of those constructs?" His answer, in plain terms, was: not yet — current efficacy data uses mouse cytokine sequences, and human-construct preclinical testing is the necessary next step.

This is more important than it might sound. Cytokines are not cleanly cross-species. Mouse IL-12 and human IL-12 share about 70% sequence identity at the protein level for the p35 subunit and roughly similar for p40 — but the receptor binding interfaces have selected divergences that mean mouse IL-12 is poorly active on human IL-12 receptors and vice versa. So if you design an experiment around SunVax's published mouse efficacy numbers (the 525-fold IL-12 result in B16F10 melanoma; the 94.1% mouse CD4 T transfection from slide 3), and you transcribe the same cargo into a human-cytokine construct, your downstream pharmacology will be different. Sometimes very different.

What the published data covers vs. what it doesn't. The peer-reviewed CATP paper (PMID 40851032) is a mouse-cargo, mouse-host result: mouse-IL-12 sequence delivered into mouse B16F10 melanoma in mouse, with 525× amplification measured by mouse-cytokine-specific assays. The slide-3 in vivo numbers (52.3%/52.2% Balb/c CD4 T at 2/10 μg) are mouse-cargo, mouse-host. These are real, peer-reviewable, and informative. They are NOT a prediction of what happens when you switch the cargo to a human cytokine sequence and deliver it (a) to human primary cells ex vivo or (b) to a human host in any clinical or pre-clinical setting.

What to do if your experiment requires human cytokine cargo. Three concrete moves. First, do not extrapolate the slide-3 percentages or the 525-fold number to human-construct cargos as if they were upper bounds. They are mouse-construct results, and the cross-species drop can be 5× to 100× depending on cytokine and assay. Second, when you order a kit (SV102 or SV106 for human cells), order one extra dose-titration arm specifically for the human-construct cargo — do not assume the dose curve calibrated on mouse-construct material transfers. Third, when you talk to SunVax partnership (partners@sunvaxmrna.com), specifically ask whether they have ANY internal data on human-cytokine-construct delivery efficacy. They may. The panel statement is about published data; internal pilots may exist under NDA.

What this does NOT mean. It does not mean the SunVax platform "doesn't work in human." The delivery chemistry (ionizable lipid + LNP formulation) is the same regardless of cargo; the kit-versus-cell interaction is reasonably species-agnostic at the membrane level. What changes is the downstream pharmacology of the cytokine itself once it's expressed inside the targeted cell. That's a cargo problem, not an LNP problem.

What honest framing looks like for buyers. If you are presenting downstream results to your team and you used a SunVax kit with a human-cytokine cargo, your write-up should explicitly note: "Delivery efficacy was 80%+ (or whatever the kit's published mouse number was) per the SunVax slide-3 data; activity at the human cytokine level was [your measurement]." Do not implicitly imply that the SunVax efficacy and your cytokine activity are the same number. They are not. Yingzhong said as much, on the record.

The right next click. If you want the kit, order.sunvaxmrna.com is unchanged. If you want to ask SunVax for any human-construct internal data, partners@sunvaxmrna.com is the right channel. If you want to read more about cytokine engineering for cancer immunotherapy in general, the open literature is large; PMID 40851032 is a good entry point into the SunVax-specific direction.