The host interferon pathway upregulates intrinsic restriction factors in response to viral infection. Many of them block a diverse range of viruses, suggesting that their antiviral functions might have been shaped by multiple viral families during evolution. Virus-host conflicts have led to the rapid adaptation of viral and host proteins at their interaction hotspots. Hence, we can use evolutionary genetic analyses to elucidate antiviral mechanisms and domain functions of restriction factors. Zinc finger antiviral protein (ZAP) is a restriction factor against RNA viruses such as alphaviruses, in addition to other RNA, retro-, and DNA viruses, yet its precise antiviral mechanism is not fully characterized. Previously, an analysis of 13 primate ZAP identified 3 positively selected residues in the poly(ADP-ribose) polymerase-like domain. However, selective pressure from ancient alphaviruses and others likely drove ZAP adaptation in a wider representation of mammals. We performed positive selection analyses in 261 mammalian ZAP using more robust methods with complementary strengths and identified 7 positively selected sites in all domains of the protein. We generated ZAP inducible cell lines in which the positively selected residues of ZAP are mutated and tested their effects on alphavirus replication and known ZAP activities. Interestingly, the mutant in the second WWE domain of ZAP (N658A) is dramatically better than wild-type ZAP at blocking replication of Sindbis virus and other ZAP-sensitive alphaviruses due to enhanced viral translation inhibition. The N658A mutant inhabits the space surrounding the previously reported poly(ADP-ribose) (PAR) binding pocket, but surprisingly has reduced binding to PAR. In summary, the second WWE domain is critical for engineering a super restrictor ZAP and fluctuations in PAR binding modulate ZAP antiviral activity. Our study has the potential to unravel the role of ADP-ribosylation in the host innate immune defense and viral evolutionary strategies that antagonize this post-translational modification.Author summaryHost proteins and viral proteins that encounter one another are locked in a perpetual genetic arms race. In this evolutionary race, a mutation that confers a survival advantage will become more frequent in the population. By looking at the sequences of genes that are known to have antiviral roles in mammals, we can identify the exact sites where a host and viral protein have interacted and gain insight into how an antiviral protein works. Here, we identified these sites in zinc finger antiviral protein (ZAP), a host protein that blocks many different viruses. We found that changing one of the sites from the original amino acid to another dramatically improves ZAP’s antiviral activity against Sindbis virus, an alphavirus, due to improved inhibition of viral translation. Our mutation is also better at inhibiting other members in theAlphavirusgenus. We observed that our mutant ZAP has reduced ability to bind poly(ADP-ribose), a post-translational modification that is targeted by alphaviruses for productive infection. Our findings help us better understand how viruses have shaped the evolution of broad-spectrum host antiviral proteins, with great implications for the engineering of super restriction factors.