AbstractThe distribution of fitness effects (DFE) of mutation plays a central role in constraining protein evolution. The underlying mechanisms by which mutations lead to fitness effects are typically attributed to changes in protein specific activity or abundance. Here, we reveal the importance of a mutation’s collateral fitness effects, which we define as effects that do not derive from changes in the protein’s ability to perform its physiological function. We comprehensively measured the collateral fitness effects of missense mutations in the E. coli TEM-1 β-lactamase antibiotic resistance gene using growth competition experiments in the absence of antibiotic. At least 42% of missense mutations in TEM-1 were deleterious, indicating that for some proteins, collateral fitness effects occur as frequently as effects on protein activity and abundance. Deleterious mutations caused improper post-translational processing, incorrect disulfide-bond formation, protein aggregation, changes in gene expression, and pleiotropic effects on cell phenotype. Deleterious collateral fitness effects occurred more frequently in TEM-1 than deleterious effects on antibiotic resistance in environments with low concentrations of the antibiotic. The surprising prevalence of deleterious collateral fitness effects suggests they may play a role in constraining protein evolution, particularly for highly-expressed proteins, for proteins under intermittent selection for their physiological function, and for proteins whose contribution to fitness is buffered against mutations with deleterious effects on protein activity and protein abundance.Significance StatementMutations provide the source of genetic variability upon which evolution acts. Deleterious protein mutations are commonly thought of in terms of how they compromise the protein’s ability to perform its physiological function. However, mutations might also be deleterious if they cause negative effects on one of the countless other cellular processes. The frequency and magnitude of such collateral fitness effects is unknown. Our systematic study of mutations in a bacterial protein finds widespread collateral fitness effects that were associated with protein aggregation, improper protein processing, incomplete protein transport across membranes, incorrect disulfide-bond formation, induction of stress-response pathways, and unexpected changes in cell properties. Our results suggest that deleterious collateral fitness effects may be an important constraint on protein evolution.

The distribution of fitness effects of mutation plays a central role in constraining protein evolution. The underlying mechanisms by which mutations lead to fitness effects are typically attributed to changes in protein specific activity or abundance. Here, we reveal the importance of a mutation’s collateral fitness effects, which we define as effects that do not derive from changes in the protein’s ability to perform its physiological function. We comprehensively measured the collateral fitness effects of missense mutations in theEscherichia coli TEM-1β-lactamase antibiotic resistance gene using growth competition experiments in the absence of antibiotic. At least 42% of missense mutations inTEM-1were deleterious, indicating that for some proteins collateral fitness effects occur as frequently as effects on protein activity and abundance. Deleterious mutations caused improper posttranslational processing, incorrect disulfide-bond formation, protein aggregation, changes in gene expression, and pleiotropic effects on cell phenotype. Deleterious collateral fitness effects occurred more frequently inTEM-1than deleterious effects on antibiotic resistance in environments with low concentrations of the antibiotic. The surprising prevalence of deleterious collateral fitness effects suggests they may play a role in constraining protein evolution, particularly for highly expressed proteins, for proteins under intermittent selection for their physiological function, and for proteins whose contribution to fitness is buffered against deleterious effects on protein activity and protein abundance.

Mammalian gene expression is a complex process regulated in part by CpG methylation. The ability to target methylation for de novo gene regulation could have therapeutic and research applications. We have previously developed a dCas9-MC/MN protein for targeting CpG methylation. dCas9-MC/MN is composed of an artificially split M.SssI methyltransferase (MC/MN), with the MC fragment fused to a nuclease-null CRISPR/Cas9 (dCas9). Guide RNAs directed dCas9-MC/MN to methylate target sites in E. coli and human cells but also caused some low-level off-target methylation. Here, in E. coli, we show that shortening the dCas9-MC linker increases methylation of CpG sites located at select distances from the dCas9 binding site. Although a shortened linker decreased methylation of other CpGs proximal to the target site, it did not reduce off-target methylation of more distant CpG sites. Instead, targeted mutagenesis of the methyltransferase’s DNA binding domain, designed to reduce DNA affinity, significantly and preferentially reduced methylation of such sites.

Biomolecular condensates composed of proteins and RNA, known as ribonucleoprotein granules, are one approach by which cells regulate post-transcriptional gene expression. The formation of ribonucleoprotein granules typically involves the liquid-liquid phase separation of intrinsically disordered proteins with a target mRNA, sequestering the mRNA into the condensate. This sequestration regulates gene expression by inhibiting translation or facilitating RNA processing, such as splicing. Here, we designed a recombinant fusion of the human Pumilio2 homology domain (Pum2) RNA-binding protein and a synthetic intrinsically disordered protein that exhibits liquid-liquid phase separation to create synthetic ribonucleoprotein granules that extrinsically regulate the expression of a target gene. We show that this fusion protein selectively binds an RNA transcript of interest that contains a Pum2-binding RNA sequence at its 3’-end and sequesters the mRNA within a biomolecular condensate. Sequestration of a target mRNA within the condensate largely reduces the translation of the target RNA in protocells and in E. coli compared to cells that do not sequester the target mRNA in a synthetic condensate. We demonstrate that the viability of E. coli that harbor a cytotoxic protein can be rescued by sequestering the mRNA encoding the cytotoxic protein within a synthetic condensate. Finally, we use RNA-seq to determine that the target RNA of interest is preferentially sequestered in synthetic ribonucleoprotein granules. This approach enables modulation of cell function via the formation of a synthetic biomolecular condensate that spatiotemporally regulates the expression of a target protein.