41 Structural Homology of HHV-6 Epitopes as Candidates for Molecular Mimicry Triggers of Type One Diabetes Mellitus Onset

Faculty Mentor: Julio Facelli (Biomedical Informatics, University of Utah)

Abstract

Molecular mimicry is a mechanism by which an infectious agent may trigger an autoimmune response through structural or sequence similarity to self-antigens. Identifying potential molecular mimics is critical to understanding autoimmune disorders, including type 1 diabetes mellitus (T1DM). Human herpesvirus 6B (HHV-6B), a common childhood virus implicated in multiple autoimmune conditions, has been epidemiologically linked to increased risk of T1DM. However, the connection between HHV-6B and T1DM remains unresolved, as HHV-6B epitopes listed in public databases show minimal sequence homology with known T1DM autoantigens. This study explores whether HHV-6B can act as a molecular mimic by assessing structural and electrostatic similarities between HHV-6B- derived and T1DM GAD65-derived epitopes. Epitope peptide structures and their interactions with HLA-DRB5*0101, a class II MHC allele linked to T1DM, were modeled using Boltz-1 and ChimeraX. HHV-6B epitopes studied here demonstrated high structural alignment with T1DM epitopes (RMSD < 1.0 Å) and shared electrostatic surface features. When modeling the epitope complexes with HLA-DRB5*0101, the HHV-6B and T1DM epitopes fit into the same binding region, suggesting a plausible mechanism for T cell cross-reactivity. These findings provide computational evidence that HHV-6B may act as a molecular mimic contributing to autoimmune responses in T1DM genetically susceptible individuals. This study demonstrates that structural modeling is a useful tool for identifying potential mimicry candidates that sequence-based methods may not find, underscoring the importance of integrating structure-based modeling into molecular mimicry prediction pipelines. Future research should explore additional HHV-6B epitopes and incorporate molecular docking to validate their pathogenic potential.

Introduction

An increasing body of evidence implicates molecular mimicry as a contributor to autoimmune diseases (ADs), including type 1 diabetes mellitus (T1DM) [1-4]. Molecular mimicry arises when foreign peptides share similarities with self-antigens, triggering a cross-reactive immune response. Although the etiology of T1DM is not fully understood, it likely involves an interplay of genetic, environmental, dietary, and viral triggers [5-6]. Studying molecular mimicry enhances understanding of T1DM onset and progression, guiding the development of targeted strategies for disease prevention and management.

Human herpesvirus 6 (HHV-6) exhibits a high global prevalence and has been frequently linked to autoimmune diseases, including multiple sclerosis [7-8], celiac disease [9], Hashimoto’s thyroiditis [10], and systemic lupus erythematosus [11-12]. Primary infection of HHV-6 affects more than 90% of individuals within the first two years of life, and adult seroprevalence exceeds 95% [13]. Following primary infection, HHV-6 establishes lifelong latency, and may reactivate and cause further complications to a wide range of cell types [14-15].

First discovered in 1985, HHV-6 was reclassified in 2012 into two distinct species: Human herpesvirus 6A (HHV-6A) and Human herpesvirus 6B (HHV-6B) [16]. The lack of clear distinction between these two variants in existing literature makes etiologic associations difficult to assess, so differentiation between the two, whenever possible, leads to more effective interventions. In the USA, UK and Japan, 97-100% of primary infections by these two viruses were caused by HHV-6B [17-20]. In a previous study, Mistry et al. [21] found that individuals in The Environmental Determinants of Diabetes in the Young (TEDDY) study cohort [22] who experienced a single episode of sixth disease (the clinical presentation of HHV-6B infection), at 12 months of age had a 4.49-fold increased risk of developing T1DM. Paradoxically, no HHV-6B epitopes currently listed in the Immune Epitope Database (IEDB) [23] appear to show meaningful homology with known autoimmune targets [24], including Glutamic Acid Decarboxylase (GAD65), a key autoantigen associated with T1DM onset [25-26].

Bach et al. [25] reported strong reactivity of GAD65-specific T cells toward several epitope candidates, including one from HHV-6. These contradictory findings may be explained because the results of the Bach study are not included in the IEDB. Moreover, the GAD-65 (GAD:248-259 and GAD:246-257) and HHV-6B U2 epitopes in that study exhibit low sequence homology [25]. Nevertheless, structural homology between GAD65 and infectious peptides (such as HHV-6B), even without amino acid similarities, can still produce a cross-reactive response [2,27-29]. The purpose of this study is to evaluate structural homology between the HHV-6B epitope identified in Bach’s study and the corresponding GAD65 epitopes, and to assess their binding potential to HLADRB5*0101, using Boltz-1 [30] and Chimera-X [31] modeling tools.

Methods

The study by Bach et al. [25] identified candidate mimicry peptides by screening protein databases using epitope motifs derived from GAD65 sequences and performing the corresponding immunological studies. Of interest here, this study focuses on three key epitopes with little apparent sequence homology: GAD:248-259 (epitope 3), GAD:246-257 (epitope 4), and HHV-6 U2 (epitope 7), using the numbering from Table 2 in [25].

Boltz-1 [30] is an open-source deep learning model that predicts biomolecular structures and binding sites with accuracy comparable to AlphaFold3 [27]. Boltz-1 uses a diffusion-based generative approach to predict protein structures and potential binding sites based on amino acid sequences provided in FASTA format [32-33].

In this study, Boltz-1 (v.0.4.1) was executed on NVIDIA H100 NVL GPUs (94 GB HBM3 memory) with 64 GB of system RAM per node, using the Granite GPU cluster at the Utah Center for High Performance Computing (https://www.chpc.utah.edu/).

FASTA files were created for the three epitopes considered here (Table 1). The sequence for the HLA- DRB5*0101 molecule, known to bind GAD65-derived epitopes with high affinity [25,34], was obtained from UniProt (https://www.uniprot.org/) and converted into FASTA format. Boltz-1 calculations were performed for all three epitopes, the HLA-DRB5*0101 molecule, and the HLA- DRB5*0101 molecule with each individual epitope.

Boltz-1 generates four structural models per input, but all results presented here are based on the first model, which had the highest accuracy. Visualization, root mean square deviation (RMSD), and electrostatic surface potential calculations were performed with UCSF ChimeraX (v.1.10rc202505210000) [31]. RMSD values were computed using the “matchmaker” tool in ChimeraX, which superimposes protein or nucleic acid structures through pairwise sequence alignment [35]. Electrostatic surface potentials were calculated using ChimeraX’s Coulombic Surface Coloring tool.

Results

In all cases selected for analysis, model 0 from Boltz-1 consistently exhibited the highest confidence scores, and these models were used in subsequent analyses.

Structural Superposition of Isolated Epitopes

Superposition of the structures of the two GAD65 protein epitopes and the HHV-6 epitope revealed a structural match with RMSD values of 0.788 Å and 0.616 Å, respectively, shown in Table 2. These isolated structures each share a coil secondary structure, as depicted in Figure 1.

 

Electrostatic Surface Potentials

Electrostatic potential mapping revealed that the epitopes shared similar minimum, mean, and maximum Coulombic values, exhibited in Table 3. These comparable charge distributions indicate that the epitopes will bind to HLA-DRB5*0101 in a related fashion.

Table 3: Coulombic Values of the GAD65 protein epitopes and the HHV-6 epitope in kcal/(mol·e) at 298 K.

 

Epitope-HLA Complex Structures

The structural models generated by Boltz-1, for each epitope-MHC (HLA-DRB5*0101) complex are presented in Fig. 2. The conformation of the HLA-DRB5*0101 molecules (grey) does not change significantly upon binding to the epitopes, but all the epitopes show structural reorganization, losing the coil conformation observed in their isolated predicted structures. The binding location of the three epitopes is consistent with existing evidence of binding sites for other well-characterized HLA-epitope complexes [36].

Discussion

This study builds on the findings of Bach et al. [25] by demonstrating that peptides with low sequence similarity can nonetheless exhibit significant structural and electrostatic homology, resulting in good candidates for molecular mimicry. Using Boltz-1 and ChimeraX, we show that GAD65-derived and HHV-6B epitopes share similar electrostatic surface potentials, demonstrating that these epitopes may bind with comparable affinity to HLA-DRB5*0101, a class II MHC allele previously linked to T1DM onset [25,34]. This electrostatic resemblance supports the potential for a mechanistic basis for T cell cross-reactivity.

These findings support clinical results derived from the TEDDY [22] cohort, which reported a 4.49- fold increased risk for T1DM following HHV-6B infection in infancy [21]. These findings provide computational evidence supporting HHV-6B as a plausible molecular mimic contributor to T1DM pathogenesis.

The in-silico pipeline developed here can be adapted to identify other molecular mimics based on structural and electrostatic similarity, even without sequence homology. This approach may have wide applicability, given that molecular mimicry has been implicated in the pathogenesis of multiple sclerosis, systemic lupus erythematosus, rheumatoid arthritis, Sjogren’s disease, systemic sclerosis, and autoimmune hepatitis [2]. HLA-DRB5 molecules have also been linked to multiple sclerosis [37] and immune thrombocytopenia [38], so continuing to investigate HLA-epitope complexes may identify biomarkers and suppressors for autoimmune disease.

Limitations

While Boltz-1 provided reasonable structural predictions for all peptide and complex models, with confidence values ranging from 0.549 to 0.721, it is difficult to evaluate the tool’s exact accuracy as there is very little known about Boltz-1 modeling short peptide interactions. Further exploration using established methods for molecular binding and docking is necessary to confirm the biological relevance of Boltz-1 in predicting molecular mimicry.

Conclusion

Using novel bioinformatics tools, this study demonstrates that peptides lacking sequence homology can still exhibit significant structural and electrostatic similarity. These features may enable molecular mimicry, contributing to the onset of autoimmune diseases such as T1DM. Moreover, the results align with previous work that demonstrated that molecular mimicry could occur between structurally similar peptides without significant sequence homology [25].

While previous studies have shown that investigating sequence homology is an efficient and less computationally expensive approach for identifying potential mimics [24,39], and that many candidates showing sequence homology also show structural homology [40], this work highlights that structural modeling is essential for uncovering mimicry candidates that sequence-based methods may not find. These findings emphasize the importance of integrating structure-based modeling into molecular mimicry pipelines, especially in the context of autoimmune disease research.

These results support the hypothesis that HHV-6B infection may contribute to the onset or exacerbation of autoimmune diseases through molecular mimicry. Future studies should investigate additional HHV-6B-derived epitopes using molecular docking to evaluate the generalizability and potential immunogenicity of these mimicry interactions.

Acknowledgements

This work was partially supported (MJ) by the SPUR program from the Office of Undergraduate Research at the University of Utah and by a CTSA award to the Utah Clinical and Translational Science Institute UM1TR004409 (JCF). Computational resources were provided by the Utah Center for High Performance Computing. Molecular graphics and analyses were performed with UCSF ChimeraX, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from National Institutes of Health R01- GM129325 and the Office of Cyber Infrastructure and Computational Biology, National Institute of Allergy and Infectious Diseases.

References

[1] Cusick MF, Libbey JE, Fujinami RS. Molecular mimicry as a mechanism of autoimmune disease. Clin Rev Allergy Immunol. 2012;42(1):102-111. doi:10.1007/s12016-011-8294-7

[2] Rojas M, Restrepo-Jiménez P, Monsalve DM, et al. Molecular mimicry and autoimmunity. J Autoimmunity. 2018;95:100-123. doi:10.1016/j.jaut.2018.10.012.

[3] Blank M, Barzilai O, Shoenfeld Y. Molecular mimicry and auto-immunity. Clinic Rev Allerg Immunol. 2007;32(1):111-118. doi:10.1007/BF02686087

[4] Roep BO, Hiemstra HS, Schloot NC, et al. Molecular mimicry in type 1 diabetes: immune cross-reactivity between islet autoantigen and human cytomegalovirus but not Coxsackie virus. Ann N Y Acad Sci. 2002;958:163-165.

[5] Ogrotis I, Koufakis T, Kotsa K. Changes in the Global Epidemiology of Type 1 Diabetes in an Evolving Landscape of Environmental Factors: Causes, Challenges, and Opportunities. Medicina (Kaunas). 2023;59(4):668. Published 2023 Mar 28. doi:10.3390/medicina59040668

[6] Selgrade MK, Cooper GS, Germolec DR, Heindel JJ. Linking environmental agents and autoimmune disease: an agenda for future research. Environ Health Perspect. 1999;107 Suppl 5(Suppl 5):811-813. doi:10.1289/ehp.99107s5811

[7] Cheng W, Ma Y, Gong F, et al. Cross-reactivity of autoreactive T cells with MBP and viral antigens in patients with MS. Front Biosci (Landmark Ed). 2012;17(5):1648-1658. Published 2012 Jan 1. doi:10.2741/4010

[8] Friedman JE, Lyons MJ, Cu G, et al. The association of the human herpesvirus-6 and MS. Multiple Sclerosis. 1999;5(5):355-362. doi:10.1177/135245859900500509.

[9] Kaisa M. Kemppainen, Kristian F. Lynch, Edwin Liu, et al. Factors That Increase Risk of Celiac Disease Autoimmunity After a Gastrointestinal Infection in Early Life – ClinicalKey. Clinical Gastroenterology and Hepatology. 2017;15(5):694-702.

[10] Caselli E, Zatelli MC, Rizzo R, et al. Virologic and Immunologic Evidence Supporting an Association between HHV-6 and Hashimoto’s Thyroiditis. PLOS Pathogens. 2012;8(10):e1002951. doi:10.1371/journal.ppat.1002951.

[11] Broccolo F, Drago F, Paolino S, et al. Reactivation of human herpesvirus 6 (HHV-6) infection in patients with connective tissue diseases. Journal of Clinical Virology. 2009;46(1):43-46. doi:10.1016/j.jcv.2009.05.010

[12] Sokolovska L, Cistjakovs M, Matroze A, Murovska M, Sultanova A. From Viral Infection to Autoimmune Reaction: Exploring the Link between Human Herpesvirus 6 and Autoimmune Diseases. Microorganisms. 2024;12(2):362. doi:10.3390/microorganisms12020362

[13] King O, Syed HA, King O, Syed HA, Khalili YA. Human Herpesvirus 6. In: StatPearls. StatPearls Publishing; 2025. Published May 18, 2025. http://www.ncbi.nlm.nih.gov/books/NBK540998/

[14] Dunn N, Kharlamova N, Fogdell-Hahn A. The role of herpesvirus 6A and 6B in multiple sclerosis and epilepsy. Scandinavian Journal of Immunology. 2020;92(6):e12984. doi:10.1111/sji.12984

[15] Leibovitch EC, Caruso B, Ha SK, et al. Herpesvirus trigger accelerates neuroinflammation in a nonhuman primate model of multiple sclerosis. Proceedings of the National Academy of Sciences of the United States of America. 2018;115(44):11292-11297.

[16] Ablashi D, Agut H, Alvarez-lafuente R, et al. Classification of HHV-6A and HHV-6B as distinct viruses. Arch Virol. 2014;159(5):863-70. doi:https://doi.org/10.1007/s00705-013-1902-5

[17] Dewhurst S, McIntyre K, Schnabel K, Hall CB. Human herpesvirus 6 (HHV-6) variant B accounts for the majority of symptomatic primary HHV-6 infections in a population of U.S. infants. J Clin Microbiol. 1993;31(2):416-418. doi:10.1128/jcm.31.2.416-418.1993

[18] Yamamoto T, Mukai T, Kondo K, Yamanishi K. Variation of DNA sequence in immediate-early gene of human herpesvirus 6 and variant identification by PCR. J Clin Microbiol. 1994;32(2):473-476. doi:10.1128/jcm.32.2.473-476.1994

[19] Danielle M. Zerr, Amalia S. Meier, Stacy S. Selke, et al. A Population-Based Study of Primary Human Herpesvirus 6 Infection – ProQuest. The New England Journal of Medicine. 2005;352(8):768-776.

[20] Thäder-Voigt A, Jacobs E, Lehmann W, Bandt D. Development of a microwell adapted immunoblot system with recombinant antigens for distinguishing human herpesvirus (HHV)6A and HHV6B and detection of human cytomegalovirus. Clinical Chemistry and Laboratory Medicine. 2011;49(11):1891-1898.

[21] Mistry S, Gouripeddi R, Raman V, Facelli JC. Sequential data mining of infection patterns as predictors for onset of type 1 diabetes in genetically at-risk individuals. Journal of Biomedical Informatics. 2023;142:104385. doi:10.1016/j.jbi.2023.104385.

[22] TEDDY Study Group. The Environmental Determinants of Diabetes in the Young (TEDDY) Study. Ann N Y Acad Sci. 2008;1150:1-13. doi:10.1196/annals.1447.062

[23] Vita R, Mahajan S, Overton JA, et al. The Immune Epitope Database (IEDB): 2018 update. Nucleic Acids Research. 2019;47(D1):D339-D343. doi:10.1093/nar/gky1006

[24] Mistry S, Gouripeddi R, Facelli JC. Prioritization of infectious epitopes for translational investigation in type 1 diabetes etiology. Journal of Autoimmunity. 2023;140:103115. doi:10.1016/j.jaut.2023.103115

[25] Bach JM, Otto H, Jung G, et al. Identification of mimicry peptides based on sequential motifs of epitopes derived from 65-kDa glutamic acid decarboxylase. Eur J Immunol. 1998;28(6):1902-1910. doi:10.1002/(SICI)1521-4141(199806)28:06<1902::AID-IMMU1902>3.0.CO;2-J

[26] Towns R, Pietropaolo M. GAD65 autoantibodies and its role as biomarker of Type 1 diabetes and Latent Autoimmune Diabetes in Adults (LADA). Drugs Future. 2011;36(11):847. doi:10.1358/dof.2011.036.11.1710754

[27] Kohm AP, Fuller KG, Miller SD. Mimicking the way to autoimmunity: an evolving theory of sequence and structural homology. Trends in Microbiology. 2003;11(3):101-105. doi:10.1016/S0966-842X(03)00006-4

[28] Trier NH, Houen G. Antibody Cross-Reactivity in Auto-Immune Diseases. Int J Mol Sci. 2023;24(17):13609. doi:10.3390/ijms241713609

[29] Ständer SHD, Reboul CF, Le SN, et al. Structure and dynamics of GAD65 in complex with an autoimmune polyendocrine syndrome type 2-associated autoantibody. Nat Commun. 2025;16:2275. doi:10.1038/s41467-025-57492-4

[30] Wohlwend J, Corso G, Passaro S, et al. Boltz-1 Democratizing Biomolecular Interaction Modeling. Published online May 6, 2025:2024.11.19.624167. doi:10.1101/2024.11.19.624167

[31] Meng EC, Goddard TD, Pettersen EF, et al. UCSF ChimeraX: Tools for structure building and analysis. Protein Science. 2023;32(11):e4792. doi:10.1002/pro.4792

[32] Modeling an RNA–Ligand Complex with Boltz-1. DiPhyx. April 23, 2025. Accessed July 21, 2025. https://diphyx.com/stories/rna-ligand-complex/

[33] Alphafold – Center for High Performance Computing – The University of Utah. Accessed July 21, 2025. https://chpc.utah.edu/documentation/software/alphafold.php#boltz

[34] Zhao LP, Alshiekh S, Zhao M, et al. Next-Generation Sequencing Reveals That HLA-DRB3, -DRB4, and -DRB5 May Be Associated With Islet Autoantibodies and Risk for Childhood Type 1 Diabetes. Diabetes. 2016;65(3):710-718. doi:10.2337/db15-1115

[35] Meng EC, Pettersen EF, Couch GS, Huang CC, Ferrin TE. Tools for integrated sequence-structure analysis with UCSF Chimera. BMC Bioinformatics. 2006;7(1):339. doi:10.1186/1471-2105-7-339

[36] Sidney J, Peters B, Sette A. Epitope prediction and identification- adaptive T cell responses in humans. Seminars in Immunology. 2020;50:101418. doi:10.1016/j.smim.2020.101418

[37] Caillier SJ, Briggs F, Cree BAC, et al. Uncoupling the Roles of HLA-DRB1 and HLA-DRB5 Genes in Multiple Sclerosis. The Journal of Immunology. 2008;181(8):5473-5480. doi:10.4049/jimmunol.181.8.5473

[38] Ye Q, Ying Q, Chen Y, Liao C, Li A. HLA-DRB5 promotes immune thrombocytopenia via activating CD8+ T cells. Open Med (Wars). 2024;19(1):20240955. doi:10.1515/med-2024-0955

[39] Maldonado-Catala P, Gouripeddi R, Schlesinger N, Facelli JC. Molecular mimicry impact of the COVID-19 pandemic: Sequence homology between SARS-CoV-2 and autoimmune diseases epitopes. ImmunoInformatics. 2025;18:100050. doi:10.1016/j.immuno.2025.100050

[40] Gardner R, Wilkins J, Mistry S, Gouripeddi R, Facelli JC. Structural Homology of Epitope Binding Mimicry in the Onset of Type 1 Diabetes Mellitus. In: 2023 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). IEEE; 2023:3708-3713. doi:10.1109/BIBM58861.2023.10385653