Fig. 7

Multiple alignment of amino acid sequences of Proteases potentially encoded by HERV-K(HML-2) loci. Because HML-2 Pro is translated via a ribosomal frameshift from the Gag ORF only HML-2 Pro sequences that also harbor a full-length Gag ORF are included. Note that other HML-2 loci may also encode protease in the case of translation bypassing Gag-Pro frameshifts. The HML-2 Pro ORF also encodes an upstream dUTPase. The C-terminal “last” dUTPase motif is included in the multiple alignment. Also indicated are a previously reported N-terminal auto-processing site for HML-2 Pro [44], and DTG, FLAP and GRDLL motifs conserved in retroviral aspartyl proteases. Note the early stop codons in two sequences that partially or entirely remove the GRDLL region. The HML-2 locus designations used here are a combination of two established naming systems; the first based on the location of HML-2 loci in chromosomal bands [58] and the second based on HUGO Gene Nomenclature Committee (HGNC)-approved designations of transcribed HML-2 loci [59]. HERV-K113 and the three bottom-most sequences are HML-2 sequences not present in the human reference genome [2, 60]. Also note that locus chr3q27.2_ERVK-11 harbors a fused Gag-Pro ORF that extends approximately 700 aa in the N-terminal direction. Locus 7p22.1_ERVK-6 represents the protease sequence used for in vitro and in vivo experiments in this study