Pooling Medicine · English Website

English Site Index › AI4S Case · English

AI4S Case · English

Provec — AI De Novo Design of a Synthetic RNA Delivery Platform

Provec is the latest AI for Science / AI for Medicine case of Pooling Medicine / 100Biotech: starting from the synthetic transfer vehicle STV-C8 (Nature, Sept 2026, Helmholtz Centre), it uses the newest AI protein-design toolchain (RFdiffusion + ProteinMPNN + AlphaFold3 / Chai-1) to systematically redesign every module — full optimization, patent design-around, and shortcoming completion — targeting in-vivo CAR-T, in-vivo T-cell engagers and CRISPR combination therapy.

The RNA delivery "impossible triangle"

Safety, efficacy and accessibility are hard to achieve together. Lentiviral vectors: near-100% ex-vivo CAR-T transduction but integration risk and very high cost. LNP: safe and cheap but >95% of mRNA is trapped in endosomes and liver tropism limits extra-hepatic delivery. STV-C8 proved a "fourth option": a four-module fusion protein budding from HEK293T cells into ~110 nm vesicles that fuse with the target membrane, delivering RNA directly into the cytoplasm — >1,000× the transfection efficiency of LNP and >10,000× less RNA dose.

Nine-dimension optimization (STV-C8 → Provec)

DimensionSTV-C8 (Nature 2026)Provec (Pooling Medicine)
Protein scaffoldHE0690, first-batch RFdiffusion C8 assembly, Tm ~65°CNew C8/D3 assembly library, Tm >80°C, <30% sequence identity, patent design-around
Membrane-binding domainUaPHPLC (black-bear heterologous PH domain)Humanized / human PH domain, immunogenicity −90%
RNA packagingSingle-copy PP7, linear mRNA 3–7 daysMulti-copy PP7, circRNA/saRNA 2–4 weeks, payload 5–10×
TargetingVSV-G mutant + monovalent minibinderMultivalent minibinder array + AND-gate logic gating, precision 100×
Backbone plasmidStandard pCAGOptimized promoter + WPRE, expression 3–5×
Transfection reagentJetOptimus (commercial)In-house poly(β-amino ester): DNA −50%, titer 3–5×, cost −80%
IPProtected by WO2026068725A1Fully independent, autonomous IP
OverallBaselineDelivery efficiency 10–50×, production cost −80%+

These are pre-experimental design targets derived from public literature and AI-assisted research; experimental validation is being planned.

Five differentiated pipelines & roadmap

  1. Dual-target logic-gated in-vivo CAR-T (CLDN18.2 × B7-H3 SynNotch AND-gate).
  2. In-vivo generation of bispecific T-cell engagers (TCE).
  3. Multi-armored CAR-T (CAR + cytokine + checkpoint blockade in one particle).
  4. CRISPR combination therapy (Cas9 mRNA + multiple sgRNAs).
  5. Tumour-microenvironment-responsive delivery (MMP-sensitive masking peptide).

Three stages over ~36 months: basic optimization (6–12 mo), differentiated breakthrough (12–24 mo), full surpassing and IND-enabling (24–36 mo).

Open cooperation — three types of partners

Email a cooperation inquiry Full Chinese case page 100Biotech wet-lab validation services (EN)

Frequently Asked Questions (FAQ)

What is Provec?

An AI for Science redesigned next-generation synthetic RNA delivery platform by Pooling Medicine, based on STV-C8 (Nature 2026) with nine-dimension optimization, patent design-around and independent IP.

How does Provec differ from LNP and lentivirus?

It uses membrane fusion rather than endosomal uptake (no endosome trap) and has no genomic integration risk, allowing repeat dosing and programmable targeting.

How can I collaborate on Provec?

Clinical, technical, and industry/capital partners are invited. Use the Inquiry float on any page or email wendy@100biotech.com with your organization, field and role.

The Full English Website

Every page on both Chinese websites has an English counterpart. The PoolingMed logo float at the bottom right of every page links to this complete English site index; the Inquiry float lets you send a structured cooperation request by email.

Last updated: 2026-11-22 · Source: https://www.poolingmed.com/provec-ai4s-case.html (Chinese original) · Maintained by Pooling Medicine · Contact: wendy@100biotech.com