Pathogenic Priming Confirmed: Dear Vaccine Manufacturers and FDA: Please Mandate the Exclusion of Unsafe Epitopes from All Future Vaccines

A new study found that the same immune proteins that I predicted would be targeted due to molecular mimicry are targeted in Long COVID in children. Future vaccines must exclude unsafe epitopes.

James Lyons-Weiler, Ph.D.

Dear Reader: Your paid subscription helps makes sure all of these articles stay free forever. Sharing is another way to help. This is a very, very important article: My message in April 2020 to screen out unsafe epitopes in COVID-19 vaccines was ignored. Here I show that we can know the causes of chronic illness from injection and infection and if regulations are updated we can prevent it. This also serves as a warning for the blithe use of plasmids that encode, inadvertently or intentionally, motifs that are bioactive in humans. Draft regulatory language included, F.O.C.

The immune proteins now shown to be dysregulated in Long COVID are the same ones I identified as high‑risk targets five years ago. Here I review the history, current findings and offer Proposed Policy Language for CFR Amendment to prevent chronic illness via mandated epitope screening.

In April 2020, I published a paper on “Pathogenic Priming” in the Journal of Translational Autoimmunity. It warned that certain epitopes in SARS-CoV-2 closely resembled human proteins, including immune-regulatory proteins. My concern was that vaccines would include epitopes that could induce chronic illness via autoimmunity. Remarkably, the results showed that nearly one-third of the proteins identified as potential targets of molecular mimicry were immune proteins themselves—the very regulators, receptors, and signaling components the body relies on to orchestrate a proper immune response.

In 2021, Aristo Vojdani and colleagues published a study in Frontiers in Immunology that provided laboratory validation for the hypothesis I put forward in April 2020 regarding pathogenic priming. They cited my work directly and extended it by demonstrating, through ELISA and inhibition assays, that antibodies raised against SARS-CoV-2 spike, nucleoprotein, envelope, and membrane proteins cross-react with a wide range of human tissue antigens. Out of 55 tested human proteins, 28 showed significant binding, confirming the reality of immune cross-reactivity driven by viral mimicry. This not only supports the principle of molecular mimicry but also validates the utility of epitope-level prediction in identifying unsafe sequences in vaccine design. Their findings provide hard empirical evidence that the immune system can be misdirected against self-structures due to unfiltered viral epitopes—exactly the risk I identified and mapped in early 2020.

My predictions were not a vague theoretical warning. They were precise, protein-level map of what could go wrong if unsafe epitopes were not excluded. And now, in 2025, a high-dimensional immunophenotyping study of pediatric Long COVID by Izquierdo-Pujol et al. has confirmed that the immune markers I flagged five years ago are now empirically shown to be disrupted in Long COVID. The overlap is not approximate, partial, or coincidental. It is a stunning match.

The new study, posted on bioRxiv in August 2025, analyzed peripheral immune cells in children with persistent symptoms after SARS-CoV-2 infection. They used 37-color spectral cytometry and developed a random forest classifier that distinguished Long COVID patients from controls with 79% accuracy. Among the top nine most predictive markers in that model, six were immune proteins I identified in my 2020 mimicry table as vulnerable to autoimmune targeting due to epitope homology with non-spike SARS-CoV-2 proteins.

The 2025 pediatric Long COVID study identified nine immune markers that most effectively distinguished affected children from controls using a random forest classifier. Of these nine, six—CCR6 (in four separate cell subsets), CD1c, and CD8—were among the exact proteins I predicted in April 2020 as vulnerable to molecular mimicry based on high local homology with SARS-CoV-2 peptides. These predictions were not generic. They were derived from epitope-level mapping and included only those human proteins with direct alignment to immunogenic viral epitopes. The match rate is striking: two-thirds of the most predictive markers in pediatric Long COVID were pre-identified in silico as autoimmune risk targets. This precise convergence between predictive modeling and empirical immunophenotyping offers compelling validation of the pathogenic priming hypothesis and confirms the utility of pre-screening viral epitopes for autoimmunity risk before incorporating them into vaccine constructs.

The strongest predictor? Suppressed CCR6 expression on monocytes and NK cells. Also prominent: dysregulation of CXCR5 on effector memory CD4+ T cells and dendritic cells, downregulation of CD1c on IgM-only and marginal zone–like B cells, suppressed CD8 on NK cells, and increased PD-1 expression.

That is not an immune system quietly standing down. That is the immune system redlining in neutral—stuck in a cycle of activation, checkpoint exhaustion, impaired migration, and ineffective antigen presentation. CCR6 and CCR7 are suppressed, which is exactly what one would expect from either chronic ligation or receptor-targeting autoimmunity. CXCR5 dysregulation undermines germinal center reactions and class-switching. CD1c loss impairs lipid antigen presentation. CD8 loss in NK cells disrupts cytotoxic function. Meanwhile, CD25, CD38, HLA-DR, PD-1, CD95, CD57, and CD127 are all elevated in T, B, and NK cells—a pattern that marks exhaustion, not quiescence.

The pattern is complex but mechanistically coherent. These aren’t random perturbations; they are exactly the outcomes that would be predicted if the immune system were reacting to viral antigens that mimic its own regulatory molecules. Pathogenic priming is not just a model anymore—it’s a match.

And the affected children are not just recovering slowly. Their immune systems have been reprogrammed—trained, through mimicry, to treat parts of themselves as threats. The result is a smoldering, systemic dysregulation that persists long after the virus is gone. This is not just a post-viral syndrome. It is a biologically ingrained feedback loop, driven by epitopes that were never screened for similarity to human proteins.

To be clear, this dysregulation is visible in both natural infection and vaccine recipients—especially in those exposed to spike protein constructs that contain unfiltered epitopes. Other types of autoimmunity could also be at play beyond immune regulatory proteins—for example, the mimicry of Titin, a protein critical to cardiac muscle function, raises concern about post-infectious or post-vaccinal myocarditis, especially given its inclusion in the 2020 mimicry list.

The original vaccine designs did not screen for human mimicry. The regulators did not demand it. The manufacturers did not perform it. And now we are left to manage the consequences in the real world.

Based on the 2020 analysis, the matches to immune proteins—such as CCR6, CCR7, CD1c, PD-1, CD8, and CD25— were to non-spike proteins, including ORF1ab, NS3, NS7a, helicase, and other non-structural or accessory proteins. This distinction matters.

The spike-related epitopes identified in my 2020 analysis showed high local homology to structural and metabolic proteins. These included coiled-coil domain-containing proteins, aldehyde dehydrogenase family members, tetratricopeptide repeat scaffold proteins, titin (a very large heart protein) and keratin-associated proteins—elements involved in tissue integrity, enzymatic activity, and cytoskeletal maintenance.

While these spike-derived sequences did not match CCR6, CXCR5, CD8, CD1c, or PD-1—the proteins now empirically shown to be dysregulated in pediatric Long COVID—they still raise concern for pathogenic priming directed at non-immune tissues. For instance, cross-reactivity to keratin-associated proteins could contribute to dermatologic sequelae, while mimicry of coiled-coil proteins and cytoplasmic scaffolds may underlie persistent musculoskeletal symptoms. Although the spike protein did not house epitopes targeting immune control nodes, its inclusion in vaccines without screening for mimicry against tissue-structural proteins remains a design flaw that could explain part of the non-immunological symptomatology seen in long-haul COVID and post-vaccinal syndromes.

This must be accounted for in all future regulatory decisions.

The tools to do better exist. The analysis conducted in April 2020 is simple: identify the epitopes, and then study them for linear and structural homology to human proteins. If there is a match, a vaccine that contains antigens with those epitopes could be unsafe.

Epitope-scanning platforms can flag shared linear and conformational motifs between viral proteins and human proteins. Structure-guided reengineering can alter immunogenicity without sacrificing neutralization. Functional assays can pre-screen for autoreactivity. These are not speculative suggestions. They are practical, testable safeguards. See Vojdani et al. (2021) for proof.

This isn’t a call to abandon vaccination. It’s a demand that vaccine design enter 2026. Epitope-level safety must become standard. If a viral protein contains sequences that mimic essential human proteins, those sequences must be excluded. This is not negotiable. The history of immunology tells us what happens when the immune system sees self in non-self. Autoimmunity is not a side effect. It is a predictable consequence.

To the FDA: You must stop allowing vaccine candidates to proceed through the regulatory pipeline without epitope safety screening. Your current process treats the human proteome as if it does not matter. It does. You now have clear evidence that children are suffering long-term immune dysregulation due to unsafe viral epitope exposure. The burden of proof now lies with you: if you allow a vaccine to enter the market without homology analysis and pathogenic priming risk stratification, you are complicit in the harm that follows. A full regulatory revision is required—one that mandates automated homology scanning, includes conformational matching, and prohibits the inclusion of mimicry-prone sequences in any licensed vaccine. Silence and inaction are no longer options.

To manufacturers, medicine and public health: This is now proven. The precise proteins that are dysregulated in Long COVID are those I predicted five years ago. Ignoring this knowledge in the design of future vaccines would be a breach of scientific duty and void the social contract of “first, do no harm”.

To the public: You were told the vaccines were safe. But no one screened the antigens for homology to your immune system’s key regulators. That was not science. That was an omission. Your health deserves more than good intentions. It deserves rigorous design. Vaccines can be safe. But only if they are designed that way.

To the FDA: Your responsibility is not merely to evaluate toxicity or short‑term reactogenicity. Your responsibility is to ensure biological safety at the molecular level. Yet the vaccines authorized under emergency pathways were permitted to proceed without any epitope‑level homology screening—despite the availability of computational tools capable of identifying mimicry between viral epitopes and human immune proteins. That omission is no longer defensible. The 2025 immunophenotyping results show clear dysregulation of CCR6, CD1c, CD8, and other immune proteins I flagged in 2020 as high‑risk targets of molecular mimicry from non‑spike SARS‑CoV‑2 proteins. You now have empirical evidence that these predicted risks have materialized in children. Failure to incorporate systematic mimicry screening into the regulatory framework moving forward would constitute a breach of regulatory duty. The FDA must mandate full-proteome epitope analysis, exclude mimicry-prone regions from all vaccine constructs, require independent verification of epitope safety, and demand longitudinal surveillance for autoimmune markers. Anything less invites a repeat of the same avoidable harm. Regulatory science must evolve—now—not after the next wave of injury.

In 2020, I mapped the risk. In 2025, pediatric Long COVID immunology confirmed it. Long COVID medical therapies should include immunologists.

We cannot unsee what has now been made visible. And the cost of pretending otherwise is measured not just in trust—but in lives.

We have had our warning. The US Code of Federal Regulations should be updated to mandate epitope safety in biologics to reduce the risk of vaccine induced autoimmunity via molecular mimicry.

Regulatory Blind Spot: Current Law Requires Safety Testing—But Not at the Epitope Level

It is a matter of federal law that all constituent materials used in vaccines must be demonstrated to be safe. Under 21 CFR § 610.15, the FDA requires that “no constituent material shall be introduced into a product unless it has been shown to be safe and suitable for the intended use.” This includes all proteinaceous materials—whether full proteins, peptides, or recombinant fragments. However, nowhere in current law or regulation is there a requirement that these proteins be screened for epitope-level molecular mimicry to human proteins. The regulations are protein-aware, but not epitope-aware. This is a critical omission in the context of pathogenic priming.

My 2020 analysis showed that SARS-CoV-2 contains dozens of immunogenic peptides with high local sequence homology to human proteins—particularly those involved in immune regulation. The autoimmune risk is not a property of the entire viral protein, but of localized mimetic epitopes. The failure to evaluate candidate vaccine antigens at this level has now resulted in real-world harm: the immune proteins I flagged are now dysregulated in children with Long COVID. The empirical confirmation of this mechanism demands a regulatory response.

If the FDA continues to allow vaccine developers to include viral proteins or peptides without screening them for autoimmune potential, it is failing to meet the standard of “safe and suitable for the intended use.” In fact, it is allowing the inclusion of material that is, by design, structurally similar to the targets of the human immune system. The time has come to close this regulatory blind spot.

Proposed Policy Language for CFR Amendment

Proposed addition to 21 CFR § 610.15 – Constituent Materials

(c) For any biological product containing proteinaceous constituents intended to induce an immune response—including but not limited to full-length proteins, subunits, peptides, or epitopes—the manufacturer must demonstrate, through validated in silico and/or in vitro methods, that no immunogenic region exhibits high local sequence or structural homology to human proteins known to participate in immune regulation, tissue-specific integrity, or essential organ function.

(d) Epitope-level analysis must include screening for molecular mimicry using alignment algorithms capable of detecting high-affinity T-cell or B-cell cross-reactivity. Where mimicry is identified, such sequences must be excluded from inclusion in vaccine constructs unless robust justification can be made that such inclusion does not pose a risk of autoimmune targeting or immune dysregulation.

(e) The applicant shall provide evidence that all antigenic components included in the final product have been evaluated for epitope-level safety, including cross-reactivity profiling against a representative panel of human tissue antigens.

(f) The Secretary may publish guidance, updated periodically, identifying required databases, minimum alignment thresholds, and validation standards for epitope safety analysis.

Redesign with intent. The time to act was yesterday. The next best time is now.

Citations

Lyons-Weiler J. Pathogenic priming likely contributes to serious and critical illness and mortality in COVID-19 via autoimmunity. Journal of Translational Autoimmunity. 2020 Apr 6;3:100051. doi: 10.1016/j.jtauto.2020.100051

Izquierdo-Pujol J, Pedreño-López N, Pidkova T, Nevot M, Urrea V, Laguía F, Muñoz-López F, Dalmau J, Gonzalez-Aumatell A, Carreras-Abad C, Mendez M, Rodrigo C, Massanella M, Blanco J, Carrillo J, Trinité B, Martinez-Picado J, Morón-López S. Pediatric Long COVID Is Characterized by Myeloid CCR6 Suppression and Immune Dysregulation. bioRxiv. 2025 Aug 22:2025.08.22.671713. doi: 10.1101/2025.08.22.671713

Vojdani A, Vojdani E, and others. 2021. Reaction of Anti–SARS‑CoV-2 Spike Protein, Nucleoprotein, Envelope Protein, and Membrane Protein Antibodies With Different Tissue Antigens. Frontiers in Immunology. 2021.