Foldon trimerization domain proves immunogenic in vaccine constructs
The bacteriophage T4 Foldon domain has become a workhorse of recombinant vaccine design because it helps unstable viral proteins hold a trimeric shape. In this study, that convenience came with a clear immunologic cost: the domain itself elicited antibodies, and attempts to cloak it with engineered glycans only partly reduced recognition. The finding matters because Foldon now sits inside clinically relevant vaccine products, including the RSV subunit vaccines AREXVY and ABRYSVO®, and similar scaffolded antigens continue to move through development.
The central message is straightforward. A trimerization element added to preserve antigen structure can also become part of the immune target.
That tension sits at the heart of modern subunit vaccinology. Protein antigens often lose native conformation when produced outside the virus, and a trimerization tag can restore the arrangement needed for neutralizing epitopes to appear in the right geometry. Foldon has earned its place because it is compact, stable, and able to support expression of a broad range of viral glycoproteins, from influenza hemagglutinin to coronavirus spike and pneumovirus fusion proteins. Yet the same foreign structural motif may draw antibodies away from the intended viral target and toward the scaffold itself.
Engineering glycosylation into COBRA hemagglutinin Foldon constructs
Investigators used a computationally optimized broadly reactive antigen, or COBRA HA, appended at the C terminus with Foldon, an Avi-tag, and a hexahistidine tag. Guided by structural analysis of fibritin, three solvent-exposed positions in Foldon were selected for the introduction of N-linked glycosylation motifs. Those sites were labeled S1, S2, and S3. The design goal was not to rebuild the trimerization domain, but to place bulky glycans on exposed surfaces where antibodies might otherwise bind.
The modified proteins expressed in CHO cells and could be purified to homogeneity. Constructs bearing a single engineered glycosylation site behaved much like the unmodified wild-type protein on size-exclusion chromatography, supporting the idea that a limited amount of glycan engineering can be tolerated without collapsing the trimer. A construct carrying combined Site 1 and Site 3 modifications behaved differently and eluted later than wild-type protein, a pattern consistent with probable dissociation of HA protomers. In practical terms, the Foldon scaffold appears to have a narrow structural margin: enough substitutions can disturb the trimer even before immune effects are tested.
Partial glycan occupancy tempered antibody binding but did not erase immunogenicity
Mass spectrometry confirmed that the engineered glycosylation was only partially occupied at two of the three intended sites. That detail is important because glycan shielding depends on actual carbohydrate placement, not just the presence of a sequon in the sequence. If a glycosylation motif is not consistently decorated, exposed protein surfaces remain available to the immune system.
Antibody binding assays showed that two of the three COBRA HA Foldon mutant antigens bound anti-Foldon monoclonal and polyclonal antibodies less strongly than the wild-type Foldon construct. The reduction was real, but not complete. A monoclonal antibody, 3E11, did not cleanly separate the effect of amino acid substitution from the effect of added glycan, leaving open the possibility that sequence changes alone contributed to some of the altered recognition. In other words, glycosylation helped, but the scaffold still looked immunologically familiar enough to be noticed.
This is the crux of the paper. Shielding a protein surface can alter binding in vitro without eliminating the underlying immunogenicity in vivo.
Mouse immunization showed the scaffold remained visible to the immune system
When mice were immunized with the COBRA HA Foldon mutant antigens, the engineered glycans did not produce a significant reduction in anti-Foldon antibody elicitation. Sera from mice vaccinated with trimeric HA-Foldon, whether wild-type or mutant, reacted with Foldon. Two mutant constructs showed lower levels of reactivity, but those differences did not reach statistical significance.
For vaccine developers, the implication is uncomfortable but useful. A scaffold can be structurally masked to some degree and still remain immunogenic in a naïve host. The immune system does not need perfect access to a protein surface to learn it. Even partial exposure can be enough to generate a response, especially when the scaffold is presented repeatedly in the context of a particulate or highly ordered antigen.
The study also extends prior observations from other platforms. Foldon-specific antibodies have already been detected in mice, non-human primates, and humans after RSV F subunit vaccination. Similar anti-scaffold responses have been seen with other trimerization domains as well. The current work places that concern on firmer experimental ground by showing that Foldon itself is immunogenic and that glycoengineering offers only limited relief.
Clinical relevance for RSV, influenza, and other trimerized vaccines
The clinical relevance reaches beyond influenza HA. Foldon-containing antigens now appear in a widening array of recombinant vaccines, and the domain has already crossed the line from experimental scaffold to licensed product component. When a structural helper becomes standard engineering practice, its immunologic behavior stops being a technical footnote and starts influencing product design.
Two issues now stand out. First, if a trimerization domain generates its own antibody response, future booster doses may encounter a preexisting anti-scaffold environment. Second, if the scaffold contributes to the total antigenic load without improving protection, developers may need to decide whether to keep it, shield it further, remove it after expression, or redesign the antigen so it trimerizes on its own. Each route carries tradeoffs in stability, manufacturability, and conformational fidelity.
The data do not argue against Foldon use in every setting. They do argue against assuming that exogenous trimerization domains are immunologically invisible. In vaccine design, invisible scaffolds are rare, and this one clearly is not.
What the glycoengineering strategy did and did not solve
The study demonstrates feasibility, but only partial success. Glycosylation motifs can be introduced into Foldon without destroying expression in some constructs, and those modifications can reduce antibody binding in assay systems. Even so, the immune system still recognized the domain in vaccinated mice, and the partial occupancy of the engineered glycans likely limited the shielding effect. Structural interference from amino acid substitutions also complicated interpretation, since altered sequence alone appeared capable of changing monoclonal antibody recognition.
For vaccine engineers, the lesson is less about a failed experiment than about the limits of surface masking. A short trimerization domain is difficult to hide without disturbing the very geometry it is meant to preserve. The domain must remain compact enough to stabilize the antigen, yet foreign enough to be immunologically distinct. Those two requirements are in direct conflict.
The work leaves a narrow but practical question for the next round of design: can a trimerization scaffold be preserved, minimized, or removed without sacrificing the stability benefits that made it attractive in the first place? The current data show that partial glycan shielding is not enough to reliably silence Foldon in a naïve setting, and any vaccine platform that uses this domain should account for its own antigenicity when advancing into clinical testing.