Implications of antibody-dependent enhancement of infection for SARS-CoV-2 countermeasures

To the Editor — For certain diseases, patients who have been previously infected by one strain of a virus and who are later infected by another strain can suffer outcomes that are worse than those infected only once. One explanation for this phenomenon is that differences between two viral serotypes can compromise the ability of antibodies induced by the first infection to neutralize the second one; instead, the antibodies elicited by the first infection ‘bridge’ the second viral strain to immunoglobulin G (IgG) antibody constant region (Fc) receptors on immune cells, such as macrophages. Because this bridging is believed to enable viral entry into immune cells, shifting the tropism of the virus1, the outcome manifests as an antibody-dependent enhancement (ADE) of infection and a potentially more serious recurrence of disease. This phenomenon is often observed when antibody concentrations decrease as a result of waning immunity; an antibody may neutralize potently at high concentrations but cause enhancement of infection at sub-neutralizing concentrations.

ADE has been observed with dengue virus2, Zika virus3, Ebola virus4 and, importantly in the context of COVID-19, coronaviruses (CoVs)5,6,7,8,9. Although no well-defined set of viral properties has been definitely established as causally linked to ADE, viruses with severe clinical manifestations of ADE show an ability to either replicate in macrophages or other immune cells or otherwise manipulate these cells’ immunological state10,11. We believe that it is important to consider ADE in the context of efforts to develop countermeasures against the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2). Indeed, data from previous CoV research strongly suggest that ADE may play a role in the virus’s pathology. If this is the case with SARS-CoV-2, then careful design and testing of vaccines or alternative approaches to prophylaxis will be needed to prevent ADE. Here, we highlight clinical and experimental observations from earlier CoV outbreaks and suggest strategies that may reduce ADE in treating the SARS-CoV-2 pandemic. Additionally, we outline how techniques in immunology and protein design can be leveraged to devise vaccines and biologics that avoid ADE.

Clinical samples from the 2003–2004 SARS-CoV outbreak indicate that the virus can infect immune cells, even though immune cells do not express the angiotensin-converting enzyme 2 (ACE-2) receptor, which mediates SARS-CoV entry into lung cells12,13. These results were corroborated by in vitro observations showing that some antibodies that bind viral spike (S) protein can facilitate uptake by human macrophages and B cells via their Fcγ receptors (FcγRs)14,15. In the case of Middle Eastern respiratory syndrome coronavirus (MERS-CoV), Fc-mediated targeting has been observed with neutralizing antibodies that bind directly to the receptor-binding domain of S protein8. For both viruses, this phenomenon is dependent on antibody concentration. Although ADE is believed to target primarily immune cell populations, it is worth noting that lung epithelial cells have also been shown to express functional neonatal Fc receptor, a non-classical FcγR16. RNA expression data assembled on the Human Protein Atlas17 show that FcγRs may have complex expression patterns that may warrant further exploration for resolving ADE mechanisms.

Although efficacy data on human CoV vaccines are lacking, results from preclinical models suggest that certain vaccine designs are more likely to induce ADE immune responses than others. For example, when macaques were vaccinated with inactivated SARS-CoV and challenged with live virus, some of the animals exhibited an ADE phenotype, with extensive macrophage and lymphocyte infiltration in the lungs and edema in the alveolar cavity18. In addition, macaques vaccinated with a modified vaccinia Ankara virus expressing SARS-CoV spike were more likely than controls to exhibit acute diffuse alveolar damage and severe acute lung injury upon viral challenge19. Mice and hamsters vaccinated with trimeric spike protein vaccines were protected from SARS-CoV infection but showed evidence of developing antisera that facilitated ACE2-independent pseudovirus entry20,21. When four different SARS-CoV vaccines developed for human use were tested in mice (two different whole virus vaccines, a recombinant spike protein, and a virus-like particle), they all triggered pulmonary immunopathology upon viral challenge22. Similar effects were also seen in mice vaccinated with virus replicon particles expressing SARS-CoV nucleocapsid protein23 and inactivated MERS-CoV24.

Circumstantial clinical evidence for ADE in SARS-CoV patients includes the observation that macrophages treated with the virus and sera from patients who succumbed to the infection showed a cytokine profile that was similar to that observed in macaques that experienced fatal acute lung injury following vaccination and viral challenge19. This effect could be diminished by blocking FcγR, suggesting that, as in the macaque model, this effect is mediated by antibodies. More indirectly, antisera from SARS-CoV patients were protective in viral infectivity assays when used at a high concentration but enhanced infection when highly diluted25.

The ADE mechanisms driving viral replication or T-helper type 2 cell–mediated pathology of SARS-CoV and MERS-CoV remain unresolved. In the case of SARS-CoV, infected macrophages exhibit little or no induction of interferon-β, leading to the hypothesis that viral suppression of the immune response results in unchecked viral replication in respiratory epithelial cells26. This may result in high viral loads that lead to further tissue damage and drive pathologic adaptive immune responses. Antiserum against the S protein can enhance viral load in macrophages27, suggesting a model in which antibody-mediated CoV–FcγR binding increases virus uptake by macrophages, functionally inactivating them by virally mediated immunosuppression.

Whether SARS-CoV-2 can cause ADE effects remains an open question. Epidemiological studies investigating ADE in individuals with multiple SARS-CoV-2 infections or cross-reactivity to common-cold-causing CoVs will likely take several years. However, given that ADE has been observed with the closely related SARS-CoV, we believe that the question of ADE effects in SARS-CoV-2 should be urgently resolved using experimental immunology. For example, measuring cross-reactivity of antibodies of unexposed individuals to SARS-CoV-2 could help identify related viruses capable of generating ADE-causing antibodies. Similar characterization of cellular responses, some of which have already been published28, would help reveal cross-reactive T cell memory responses that may be involved in amplifying common cold B cell responses. Such cross-reactive T cell responses have been linked to plasma leakage syndrome in dengue fever29.

Titration experiments using pseudotyped viruses carrying S protein should be used to analyze sera from recovered patients. If antibodies against SARS-CoV-2 with ADE potential are detected, vaccine development efforts can leverage the full suite of modern technologies around epitope mapping, protein design, adjuvant design and delivery to maximize safety. Specifically, patient sera can be used to build probabilistic maps of ADE-associated epitopes using high-throughput peptide-based scanning methods30,31,32 and hydrogen–deuterium exchange mass spectrometry33,34. Sequencing of patient antibody repertoires35,36 and computational approaches37,38 can then extend epitope mapping data to structural predictions of antibody–epitope interactions. As a cautionary note, no effort has yet been able to identify epitopes that can fully avoid the problem of ADE for any single viral pathogen associated with the phenomenon. However, studies on mouse antibodies raised against immunogenic SARS-CoV epitopes suggest that regions of the spike protein do vary in their propensity to cause ADE18. If it turns out that no SARS-CoV-2 antigens fully avoid ADE, then vaccine development efforts should aim at eliciting a T cell response against non-surface proteins, perhaps by borrowing approaches from neoantigen vaccines39.

Assuming that antibodies with a low risk for ADE can be identified, the candidate molecules can then be further filtered for their ability to elicit T cell immunity, an important aspect of vaccine design. To predict epitopes that are recognized by T helper cells, state-of-the-art major histocompatibility complex (MHC) prediction methods can rank candidates likely to be presented on MHC-II molecules40,41,42. These predictions could be complemented with screens that use DNA-barcoded tetramer libraries to identify viral peptides that bind to patient-derived T cells43,44,45, or using the T-Scan technology46. The chosen epitopes known to have a low risk of eliciting ADE antibodies will need to fold properly, mimicking the native conformation in the virus, to elicit productive B cell responses in vaccinated individuals. Computational tools for de novo protein folding can help identify minimal fragment sizes containing properly folded epitopes47,48 or can help stabilize the fragments by introducing disulfide bonds or scaffold domains49,50. For example, prolines have been introduced in the fusion domain to stabilize the pre-fusion conformation of S protein51,52.

Because there is no guarantee that epitopes that exhibit low ADE risk are also immunogenic, vaccine programs may have to boost titers by using adjuvants and/or maximizing the intrinsic immunogenicity of the various vaccine platforms. Gene-based vaccination platforms, such as plasmid DNA or synthetic mRNA, should allow a greater immune response to weakly immunogenic antigens by delivering them with or fusing them to protein-based adjuvants53,54. Furthermore, one such adjuvant, CD40L, has been shown to reduce the pulmonary pathology otherwise associated with an adenoviral MERS-CoV vaccine55. Although the mechanism for this observation has yet to be elucidated, the data underscore the need for testing specific adjuvants that may reduce the risk of ADE — perhaps by shifting either the isotype or the breadth of the resulting immune response.

Animal models often serve as invaluable tools for determining the safety and efficacy of vaccines. In fact, preclinical studies employing various animals, including mice, hamsters, ferrets and macaques, provided evidence that SARS-CoV vaccines are capable of causing an ADE response18,19,20,21,22,23,56,57. However, the immune system of each of these animals differs compositionally and functionally from the human immune system. Specifically, Fc receptors differ in sequence, structure and function between animal models and humans58,59. Such differences make it difficult to study human antibodies in animal models and to translate the effects of a vaccine in a model to human use. However, the use of transgenic models, such as human FcγR-expressing mice60, may more faithfully model human immune responses to candidate vaccines and antibody therapeutics. Ultimately, non-human primates represent the best option for building models of human ADE responses, as evidenced by the ability of SARS-CoV Chinese macaque vaccination/challenge model to recapitulate certain aspects of sera from patients with SARS19. However, failures of some HIV vaccine candidates have highlighted the need for caution even with non-human primate models59.

While vaccine testing is ongoing, monoclonal antibodies will be a compelling option for protecting immunocompromised populations, including the elderly, or healthcare workers who may be exposed to high viral loads. Monoclonal antibodies and associated approaches, such as convalescent sera, should be carefully tested for ADE effects. One powerful potential safeguard could involve mutating the Fc-binding domain of the monoclonal antibody to retain its neutralizing potential while preventing uptake in immune cells61,62. There are known mutations that abrogate the binding of antibodies to Fcγ receptors, including LALA (L234A L235A), LALA-PG (L234A L235A P329G), and elimination of the glycosylation site at N29763,64,65. Notably, introduction of LALA-PG and elimination of the glycosylation binding site have been demonstrated to completely eliminate the effector functions of the Fc region, whereas introduction of the LALA mutation leaves minimal, but sometimes detectable, activity. Additionally, recent efforts have suggested that alteration of the Fc portion can be performed not just to eliminate binding to FcγRs, but conversely, to allow enhanced immune responses to viruses. For example, the F241A mutation in the Fc region results in an antibody with a more robust endogenous immune response caused by more efficient CD23 uptake and greater immunogen formation in situ66. Whether such immune-enhancing mutations alter the risk of ADE is unknown.

Although the development of vaccines and therapeutics for SARS-CoV-2 remains urgent, we must proceed with caution, using the full armory of vaccine and protein design tools at our disposal to rationally minimize the risk of ADE.

References

  1. 1.

    Taylor, A. et al. Immunol. Rev. 268, 340–364 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  2. 2.

    Katzelnick, L. C. et al. Science 358, 929–932 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  3. 3.

    Bardina, S. V. et al. Science 356, 175–180 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  4. 4.

    Kuzmina, N. A. et al. Cell Rep. 24, 1802–1815.e5 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

  5. 5.

    Petersen, N. C. & Boyle, J. F. Am. J. Vet. Res. 41, 868–876 (1980).

    CAS  PubMed  Google Scholar 

  6. 6.

    Vennema, H. et al. J. Virol. 64, 1407–1409 (1990).

    CAS  PubMed  PubMed Central  Google Scholar 

  7. 7.

    Yip, M. S. et al. BMC Proc. 5 (Suppl. 1), 80 (2011).

  8. 8.

    Wan, Y. et al. J. Virol. 94, e02015–e02019 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  9. 9.

    Tetro, J. A. Microbes Infect. 22, 72–73 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  10. 10.

    Chen, Y. C. & Wang, S. Y. J. Virol. 76, 9877–9887 (2002).

    CAS  PubMed  PubMed Central  Google Scholar 

  11. 11.

    Bosio, C. M. et al. J. Infect. Dis. 188, 1630–1638 (2003).

    CAS  PubMed  Google Scholar 

  12. 12.

    Li, L. et al. J. Clin. Virol. 28, 239–244 (2003).

    CAS  PubMed  PubMed Central  Google Scholar 

  13. 13.

    Gu, J. et al. J. Exp. Med. 202, 415–424 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  14. 14.

    Jaume, M. et al. J. Virol. 85, 10582–10597 (2011).

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15.

    Yip, M. S. et al. Virol. J. 11, 82 (2014).

    PubMed  PubMed Central  Google Scholar 

  16. 16.

    Spiekermann, G. M. et al. J. Exp. Med. 196, 303–310 (2002).

    CAS  PubMed  PubMed Central  Google Scholar 

  17. 17.

    Thul, P. J. et al. Science 356, eaal3321 (2017).

    PubMed  PubMed Central  Google Scholar 

  18. 18.

    Wang, Q. et al. ACS Infect. Dis. 2, 361–376 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  19. 19.

    Liu, L. et al. JCI Insight 4, e123158 (2019).

    Google Scholar 

  20. 20.

    Kam, Y. W. et al. Vaccine 25, 729–740 (2007).

    CAS  PubMed  Google Scholar 

  21. 21.

    Jaume, M. et al. Hong. Kong Med. J. 18 (Suppl. 2), 31–36 (2012).

  22. 22.

    Tseng, C. T. et al. PLoS One 7, e35421 (2012).

    CAS  PubMed  PubMed Central  Google Scholar 

  23. 23.

    Deming, D. et al. PLoS Med. 3, e525 (2006).

    PubMed  PubMed Central  Google Scholar 

  24. 24.

    Agrawal, A. S. et al. Hum. Vaccin. Immunother. 12, 2351–2356 (2016).

    PubMed  PubMed Central  Google Scholar 

  25. 25.

    Wang, S. F. et al. Biochem. Biophys. Res. Commun. 451, 208–214 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  26. 26.

    Cheung, C. Y. et al. J. Virol. 79, 7819–7826 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  27. 27.

    Yip, M. S. et al. Hong. Kong Med. J. 22 (Suppl. 4), 25–31 (2016).

  28. 28.

    Braun, J. et al. Preprint at medRxiv https://doi.org/10.1101/2020.04.17.20061440 (2020).

  29. 29.

    Rothman, A. L. Nat. Rev. Immunol. 11, 532–543 (2011).

    CAS  PubMed  Google Scholar 

  30. 30.

    Larman, H. B. et al. Nat. Biotechnol. 29, 535–541 (2011).

    CAS  PubMed  PubMed Central  Google Scholar 

  31. 31.

    Price, J. V. et al. Nat. Med. 18, 1434–1440 (2012).

    CAS  PubMed  PubMed Central  Google Scholar 

  32. 32.

    Guo, N. et al. EBioMedicine 2, 859–867 (2015).

    PubMed  PubMed Central  Google Scholar 

  33. 33.

    Baerga-Ortiz, A., Hughes, C. A., Mandell, J. G. & Komives, E. A. Protein Sci. 11, 1300–1308 (2002).

    CAS  PubMed  PubMed Central  Google Scholar 

  34. 34.

    Puchades, C. et al. Sci. Rep. 9, 4735 (2019).

    PubMed  PubMed Central  Google Scholar 

  35. 35.

    Lee, J. et al. Nat. Med. 22, 1456–1464 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  36. 36.

    Briney, B., Inderbitzin, A., Joyce, C. & Burton, D. R. Nature 566, 393–397 (2019).

    PubMed  PubMed Central  Google Scholar 

  37. 37.

    Jespersen, M. C., Peters, B., Nielsen, M. & Marcatili, P. Nucleic Acids Res. 45 W1 W24–W29 (2017).

  38. 38.

    Weitzner, B. D. et al. Nat. Protoc. 12, 401–416 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  39. 39.

    Sahin, U. et al. Nature 547, 222–226 (2017).

    CAS  PubMed  Google Scholar 

  40. 40.

    Chen, B. et al. Nat. Biotechnol. 37, 1332–1343 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  41. 41.

    Racle, J. et al. Nat. Biotechnol. 37, 1283–1286 (2019).

    CAS  PubMed  Google Scholar 

  42. 42.

    Sarkizova, S. et al. Nat. Biotechnol. 38, 199–209 (2020).

    CAS  PubMed  Google Scholar 

  43. 43.

    Dash, P. et al. Nature 547, 89–93 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  44. 44.

    Glanville, J. et al. Nature 547, 94–98 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  45. 45.

    Zhang, S. Q. et al. Nat. Biotechnol. 36, 1156–1159 (2018).

    CAS  Google Scholar 

  46. 46.

    Kula, T. et al. Cell 178, 1016–1028.e13 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  47. 47.

    Shaw, D. E. et al. Proc. Intl Conf. High Performance Computing, Networking, Storage and Analysis (SC14) 41–53 (IEEE, 2014).

  48. 48.

    Alford, R. F. et al. J. Chem. Theory Comput. 13, 3031–3048 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  49. 49.

    Correia, B. E. et al. Nature 507, 201–206 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  50. 50.

    Bhardwaj, G. et al. Nature 538, 329–335 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  51. 51.

    Pallesen, J. et al. Proc. Natl Acad. Sci. USA 114, E7348–E7357 (2017).

    CAS  PubMed  Google Scholar 

  52. 52.

    Wrapp, D. et al. Science 367, 1260–1263 (2020).

    CAS  PubMed  PubMed Central  Google Scholar 

  53. 53.

    Hu, H. et al. Clin. Vaccin. Immunol. 14, 894–901 (2007).

    CAS  Google Scholar 

  54. 54.

    Hewitt, S. L. et al. Sci. Transl. Med. 11, eaat9143 (2019).

    CAS  PubMed  Google Scholar 

  55. 55.

    Hashem, A. M. et al. J. Infect. Dis. 220, 1558–1567 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  56. 56.

    Weingartl, H. et al. J. Virol. 78, 12672–12676 (2004).

    CAS  PubMed  PubMed Central  Google Scholar 

  57. 57.

    Yang, Z. Y. et al. Proc. Natl Acad. Sci. USA 102, 797–801 (2005).

    CAS  PubMed  Google Scholar 

  58. 58.

    Mestas, J. & Hughes, C. C. W. J. Immunol. 172, 2731–2738 (2004).

    CAS  PubMed  Google Scholar 

  59. 59.

    Crowley, A. R. & Ackerman, M. E. Front. Immunol. 10, 697 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  60. 60.

    Smith, P., DiLillo, D. J., Bournazos, S., Li, F. & Ravetch, J. V. Proc. Natl Acad. Sci. USA 109, 6181–6186 (2012).

    CAS  PubMed  Google Scholar 

  61. 61.

    Hezareh, M., Hessell, A. J., Jensen, R. C., van de Winkel, J. G. & Parren, P. W. J. Virol. 75, 12161–12168 (2001).

    CAS  PubMed  PubMed Central  Google Scholar 

  62. 62.

    Kang, T. H. & Jung, S. T. Exp. Mol. Med. 51, 1–9 (2019).

    PubMed  Google Scholar 

  63. 63.

    Tao, M. H. & Morrison, S. L. J. Immunol. 143, 2595–2601 (1989).

    CAS  PubMed  Google Scholar 

  64. 64.

    Chappel, M. S. et al. Proc. Natl Acad. Sci. USA 88, 9036–9040 (1991).

    CAS  PubMed  Google Scholar 

  65. 65.

    Lo, M. et al. J. Biol. Chem. 292, 3900–3908 (2017).

    CAS  PubMed  PubMed Central  Google Scholar 

  66. 66.

    Maamary, J., Wang, T. T., Tan, G. S., Palese, P. & Ravetch, J. V. Proc. Natl Acad. Sci. USA 114, 10172–10177 (2017).

    CAS  PubMed  Google Scholar 

Download references

Acknowledgements

We thank N. Dhar, J. Quinn, E. Webster, Z. Shriver, A. Rovner and C. T. Wu for critically reviewing the manuscript. N.E. and H.R. were supported by Schmidt Futures.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Hannu Rajaniemi.

Ethics declarations

Competing interests

N.E, T.G, M.K.K, G.M.C., J.M.T. and H.R. are shareholders in and/or employees of Helix Nanotechnologies Inc., which is developing gene-encoded vaccines for SARS-CoV-2. A full list of G.M.C.’s technology transfer, advisory roles and funding sources can be found at http://arep.med.harvard.edu/gmc/tech.html. J.M.T. is a shareholder and employee of SmartPharm Therapeutics, which is developing gene-encoded antibody vaccines for SARS-CoV-2.

Additional information

Editorial note: This article has been peer reviewed.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Eroshenko, N., Gill, T., Keaveney, M.K. et al. Implications of antibody-dependent enhancement of infection for SARS-CoV-2 countermeasures. Nat Biotechnol 38, 789–791 (2020). https://doi.org/10.1038/s41587-020-0577-1

Download citation

Further reading