Evolution of a globally unique SARS-CoV-2 Spike E484T monoclonal antibody escape mutation in a persistently infected, immunocompromised individual

Abstract

Prolonged infections in immunocompromised individuals may be a source for novel SARS-CoV-2 variants, particularly when both the immune system and antiviral therapy fail to clear the infection, thereby promoting adaptation. Here we describe an approximately 16-month case of SARS-CoV-2 infection in an immunocompromised individual. Following monotherapy with the monoclonal antibody Bamlanivimab, the individual's virus was resistant to this antibody via a globally unique Spike amino acid variant (E484T) that evolved from E484A earlier in infection. With the emergence and spread of the Omicron Variant of Concern, which also contains Spike E484A, E484T may arise again as an antibody-resistant derivative of E484A.

Predicting the antigenic evolution of SARS-COV-2 with deep learning

“…A reverse mutation, R493Q, for example, was found in a persistently infected, immunocompromised individual 44 . Other mutations found by our model, like E340K 45 , E484T 33 , G485R 46 , and F490L/E484G 47 , are also found in immunocompromised patients treated with monoclonal antibodies. Moreover, the unique mutations found in the emerging variants, BA.4/5, the L452R, F486V, and the reverse mutation R493Q, are also captured by our model.…”

“…Next, with our model as the scoring function, we used the genetic algorithm 31,32 to generate synthetic RBD variants with high ACE2 binding and antibody escape potential. Interestingly, the in silico directed evolution shares similar mutations with the adaptive evolution in immunocompromised COVID-19 patients [33][34][35] and newly emerging variants like BA.4/5. Further computational inference and docking experiments of top generated sequences reveal synthetic sequences with comparable or even better escaping potential than the Omicron variant, which may become the high-risk variants in the future.…”

“…Inspired by Darwin's theory of natural evolution, the genetic algorithm mimics the evolutionary process in the genome, where mutations, crossover and selection happen, letting candidate solutions of a population with higher fitness scores have a higher probability of surviving and producing the next generation of offspring. For SARS-CoV-2, it has been proven that similar progress happens in immunocompromised infected patients who got treated with the monocle antibodies 33 . Hence, we used the genetic algorithm to model the antigenic evolution process and search for the potential risky variants that might appear in the future.…”

Wastewater surveillance using ddPCR accurately tracked Omicron emergence due to altered N1 probe binding efficiency

“…The residual clinical sample standards were received as viral RNA from the Wisconsin State Laboratory of Hygiene. They consisted of viral RNA that was isolated from nasopharyngeal swab specimens using the total nucleic acid kit for the Maxwell RSC instrument (Promega, Madison, WI) according to Halfmann et al 10 Raw droplet amplification data for each variant and total N1 and N2 marker concentration were extracted from the Bio-Rad QuantaSoft software and processed using R package twoddpcr (version 1.11.0) according to Feng et al 8 To validate the observed diminished fluorescence corresponding to a mismatch at the 5′ end of the N1 probe, we mimicked the same type of substitution in an artificial N2 template. We synthesized a 350 bp DNA template with an A:C change in the third nucleotide position of the N2 probe binding site (Twist Bioscience, South San Francisco, California, USA) (Fig.…”

“…The residual clinical sample standards were received as viral RNA from the Wisconsin State Laboratory of Hygiene. They consisted of viral RNA that was isolated from nasopharyngeal swab specimens using the total nucleic acid kit for the Maxwell RSC instrument (Promega, Madison, WI) according to Halfmann et al 10 Raw droplet amplification data for each variant and total N1 and N2 marker concentration were extracted from the Bio-Rad QuantaSoft software and processed using R package twoddpcr (version 1.11.0) according to Feng et al 8…”