In a 2017 Nature publication (Takata et al.), a team of virologists led by Professor Paul D. Bieniasz (The Rockefeller University) found that zinc-finger antiviral protein (ZAP) discriminates self from nonself (viral) RNAs based on their CG dinucleotide content. Now, together with crystallographers form the University of Michigan (Meagher et al. 2019), the group describe how ZAP selectively targets CG-rich viral sequences.

HIV has many tricks to escape host innate immunity. And when HIV encounters innate immune factors that it can’t overcome, the virus can simply… evolve.  The zinc-finger antiviral protein (ZAP) provides a protective effect against a variety of RNA viruses, yet HIV is largely resistant to inhibition by endogenous ZAP. How does HIV manage to avoid ZAP while other RNA viruses meet their demise? In work led by Professor Paul D. Bieniasz (The Rockefeller University), Takata et al. (2017) found that ZAP discriminates self from nonself (viral) RNAs based on their CG dinucleotide content, and that HIV smartly avoids ZAP by mimicking the CG-poor dinucleotide content of endogenous human mRNAs.

Why do human mRNAs have a low GC content?
Vertebrates show a very low level of CG dinucleotides in their mRNAs, which is probably due to structural constraints related to high dinucleotide stacking energy, supercoiling, and chromatin packing. It’s also likely that CG-specific DNA methyl transferases and spontaneous methyl-cytosine deamination have driven C-to-T mutations, and over hundreds of millions of years, these mutations have accumulated and contributed to this low CG content.

If ZAP uses CGs to identify nonself… wouldn’t that select for even lower CG content?
Absolutely. It could also be that ZAP per se contributes to the evolution of low CG content in the DNA of host cells. When CG-rich DNA sequences are transcribed, the resulting RNAs would be targeted by ZAP (mistakenly identified as non-self), blocking their function and resulting in a pressure for the genome to remove these CGs. Anyway, how ever we got there, human mRNAs have a low CG content and ZAP uses this characteristic to hunt out suspicious CG-rich RNA viruses. In response, most RNA viruses,  including HIV, appear to mimic this low CG content of their hosts to evade ZAP. In fact, ZAP antiviral activity can be predicted by the CG content of the virus.

How ZAP targets CG-rich viral sequences
Now, together with crystallographers form the University of Michigan, the team are back to provide atomic detail on how ZAP is able to selectively recognize CG-rich viral RNAs as foreign (Meagher et al. 2019). By solving the x-ray of the ZAP domain that recognizes RNA in complex with a CG-rich RNA oligomer, the team found that ZAP’s four zinc fingers create a basic patch on the protein surface and identified a pocket on the highly basic second zinc finger (ZnF2) that can accommodate a CG dinucleotide, but no other dinucleotide combination. Any dinucleotide combination other than CG would produce steric clashes.

This mechanism has now been described at the atomic detail and certainly seems that, despite all our efforts, the viruses are always a step ahead.