Interstrain recombinants of human cytomegalovirus uncouple glycoprotein display, virion infectivity, and spread characteristics

A prevailing model holds that human cytomegalovirus (HCMV) spreads cell-to-cell spread upon initial culturing from clinical specimens and that de novo mutations in the UL128-131 genes reduce expression of glycoprotein complex gH/gL/UL128-131 (pentamer) relative to gH/gL/gO (trimer), which in turn enhances virion infectivity and favors cell-free spread. The clone Merlin-BAC (ME) expresses far more pentamer than trimer and is highly cell-associated, whereas clone TB40-BAC4 (TB) expresses mostly trimer and spreads predominantly cell-free. A single G>T polymorphism in TB relative to ME was shown to impair UL128 mRNA splicing, reducing pentamer expression and enhancing infectivity and cell-free spread. However, enhanced cell-free spread of ME due to pentamer suppression did not come at the expense of efficient cell-to-cell spread, and cell-to-cell spread of TB was especially poor despite highly infectious intracellular virus. Most of the nucleotide diversity in the HCMV genome is due to 17 genes that have up to 14 alleles each, and TB and ME match at only two of the 17. Here, we report a set of recombinant HCMV generated by coinfecting cells with TB and ME. Trimer:pentamer ratio and virion infectivity largely aligned with the TB or ME UL128, with "TB-like" recombinants displaying more trimer and higher infectivity and "ME-like" recombinants displaying more pentamer and lower infectivity. Strikingly, some recombinants had shifts in preference for cell-free vs cell-to-cell spread without predicted changes to trimer:pentamer ratio or virion infectivity, demonstrating uncoupling of these phenotypes.IMPORTANCEThe emerging picture of HCMV genetic diversity in vivo prompts a reevaluation of how in vitro-characterized phenotypes, such as the abundance of different viral envelope glycoproteins, virion infectivity, and tendency toward cell-free or direct cell-to-cell spread, reflect viral characteristics in vivo. Laboratory examination of HCMV phenotypes has included a limited sampling of the apparent in vivo genetic diversity. A widely held model that directly links cell-free and cell-to-cell spread characteristics to glycoprotein display and virion infectivity also presumes that HCMV is predominantly cell-associated in vivo. These have implications for intervention strategies, such as calling into question the therapeutic benefit of neutralizing antibodies. Our finding that spread characteristics can be uncoupled from glycoprotein display and virion infectivity suggests a model that includes both cell-free and cell-associated spread as bona fide wild-type phenotypes for different allelic haplotypes. This model would allow a broader examination of neutralizing antibodies as correlates of protection.