Abstract
All retroviruses encode a Gag polyprotein containing an N-terminal matrix domain (MA) that anchors Gag to the plasma membrane and recruits envelope glycoproteins to virus assembly sites. Membrane binding by the Gag protein of HIV-1 and most other lentiviruses is dependent on N-terminal myristoylation of MA by host N-myristoyltransferase enzymes (NMTs), which recognize a six-residue "myristoylation signal" with consensus sequence: M1GXXX[ST]. For unknown reasons, the feline immunodeficiency virus (FIV), which infects both domestic and wild cats, encodes a non-consensus myristoylation sequence not utilized by its host or by other mammals (most commonly: M1GNGQG). To explore the evolutionary basis for this sequence, we compared the structure, dynamics, and myristoylation properties of native FIV MA with a mutant protein containing a consensus feline myristoylation motif (MANOS) and examined the impact of MA mutations on virus assembly and ability to support spreading infection. Unexpectedly, myristoylation efficiency of MANOS in Escherichia coli by co-expressed mammalian NMT was reduced by ~70% compared to the wild-type protein. NMR studies revealed that residues of the N-terminal myristoylation signal are fully exposed and mobile in the native protein but partially sequestered in the MANOS chimera, suggesting that the unusual FIV sequence is conserved to promote exposure and efficient myristoylation of the MA N terminus. In contrast, virus assembly studies indicate that the MANOS mutation does not affect virus assembly, but does prevent virus spread, in feline kidney cells. Our findings indicate that residues of the FIV myristoylation sequence play roles in replication beyond NMT recognition and Gag-membrane binding.