A LEGO-style approach to building virus-like particles in insect cells

By Bio-IT World Staff

November 13, 2026 | Maren Schubert’s lab at the Helmholtz Center for Infection Research has spent a decade refining an alternative to the field’s default route for making virus-like particles (VLPs) in insect cells. The conventional path — the baculovirus expression vector system, run in HighFive cells derived from the cabbage looper moth — reliably delivers high yields and underpins products, including an early bivalent HPV vaccine still prescribed outside the U.S. But it takes about a month per construct, which was untenable when Schubert, then a PhD student, needed 40 proteins for a high-throughput expression screen.

Her solution: skip the virus. Plasmids carrying the gene of interest are delivered into the insect cells with polyethylenimine, and the expressed proteins self-assemble and bud out as particles—a workflow she compares to the HEK cell system. Yields are comparable or better, and the output is faster, purer, and more modular, which is where the LEGO analogy comes in: swap components, get the VLP you want.

The technical argument is as much about quality as speed. Baculovirus kills its host cells, degrading protein quality, and co-expressed baculoviral particles contaminate the product. For VLPs mimicking enveloped viruses such as SARS or hantavirus, that contamination sits in the membrane and survives purification. Two quirks of the plasmid method: it rewards very high cell passage numbers—Schubert’s cells have been subcultured more than 600 times, growing larger, faster, and more productive—and it demands far more input DNA, since there’s no self-amplifying virus.

For drug discovery, the payoff is exploratory speed. Schubert’s group has built harmless replicas of RSV, Ebola, and other pathogens to probe viral mechanism and test vaccine and drug approaches, and is now supplying smallpox proteins and VLPs for antibody analysis in an EU orthopoxvirus project, useful because live virus stocks are restricted. The limiting step is translational: purification is hard for any complex protein, and plasmid systems lack the stability needed for GMP manufacturing. Clinical trials, Schubert says, require GMP readiness first.

Next week Schubert sits on a PEGS Europe panel on the data problem behind AI-driven protein design, the same method can give different results in different labs, and she has joined EXPERT, a 70-plus-institution effort to standardize how expression and purification data are reported.

Schubert presents at PEGS Europe in Lisbon, November 16–19. Read Christina Nunez’s story at PEGS 365 for her full account of the method, the tradeoffs, and what comes next.