2.2.1. The Gammaretrovirus Genome
Gammaretroviruses have a simple genome (), that is, there are no known additional overlapping reading frames for nonstructural regulatory proteins such as those which occur in betaretroviruses, deltaretroviruses, and lentiviruses. Moloney MLV is the reference gammaretrovirus [28
]. Despite being simple, MLVs have some distinguishing features. The gag
gene may have alternative translational start sites, giving rise to both a myristoylated Gag polyprotein, which contains the inner structural proteins, and a glycosylated Gag membrane protein. Many of them have a phosphoprotein following the matrix protein p15, called p12. The major Gag protein is p30, the capsid protein. It is responsible for many of the antigenic cross-reactions which gave rise to the acronym Gag (“group-specific antigen”). Further, like many other gammaretroviruses, the MLLVs have one zinc finger in the NucleoCapsid, NC, portion of Gag (the p10 protein), instead of the customary two [29
]. The first finger is replaced with a highly charged sequence, binding to retroviral genomic RNA in a somewhat different way compared to two zinc finger retroviruses. The zinc finger status is here used as a taxonomic marker of a subset of gammaretroviruses [30
]. All MLLVs have one zinc finger. Occasional readthrough of a Gag stop codon creates Gag-Pol polyproteins.
Figure 2 Structure and genome of a gammaretrovirus. The nucleocapsid is built from a hexameric lattice . MA: matrix (p15), CA: capsid (p30), NC: nucleocapsid (p10), PR: protease, RT: reverse transcriptase (shown as a green dot), RH: RNAse H, IN: integrase, (more ...)
Another structural genomic aspect is that retroviruses with simple genomes like alpharetroviruses and gammaretroviruses occasionally may take up an oncogene in their genome, to form acutely transforming (“sarcoma”) viruses [28
]. They are often replication deficient. They then need a replication competent virus, a helper virus, to replicate. The so-called murine AIDS (MAIDS) virus variants are also defective, producing a new Gag protein (p60) which contains part of the p12 protein and a T cell neoepitope [31
]. Likewise, the feline leukemiavirus immunodeficiency-inducing defective virus (FeLV-T) has a mutated Env [34
]. Immunodeficiency associated with this variant is sometimes called “Feline AIDS,” FAIDS, although the feline immunodeficiency virus can cause another form of (FAIDS). Recombination between exo- and endogenous MLLV sequences is common in both mice and cats [35
MLVs can cause cancer in at least two ways, either through incorporation of an oncogene, or by integration near 5′ends of transcription units and associated CpG-rich portions [28
]. The propensity to integrate into or next to promoters is a gammaretroviral specialty [36
]. Random integration next to an oncogene is a frequent cause of leukemia in MLV-infected animals. Humans are not immune to this mechanism. MLV-based gene therapeutic vectors have the same target specificity [37
], see also [40
]. Thus, an MLLV infecting humans would be expected to cause leukemias or lymphomas.
Finally, the envelope proteins (Surface Unit; SU, gp70 and TransMembrane protein; TM, p15E) are central for tissue tropism, immunogenicity, and for immunosuppression. The latter contains the conserved so-called “immunosuppressive domain” (ISD) [41
] whose mode of action is still poorly known. Thus, despite their basic structural simplicity, MLLVs can display a complex pathobiology.
2.2.2. Occurrence among Vertebrates
A rich source of vertebrate information is the collection of ERV sequences in an early version of RetroBank [47
]. The program RetroTector (ReTe) [48
] was used to collect more than 40.000 proviruses from whole genome analyses of thirty vertebrate genomes. ReTe is based on a pattern recognition algorithm. It uses the order of and distances between conserved retroviral motifs to detect and characterize retroviral sequences from large genomic data sets. A score is calculated from the degree of fit to a collection of conserved motifs from all seven retroviral genera. The higher the score, the better the fit to a structural model which encompasses most orthoretroviral and also some retrovirus-like sequences. A provisional genus is designated by counting the best-fitting motifs from each genus.
Gammaretrovirus-like sequences were detected in all of the 30 genomes (those reported in [48
] plus the turkey genome). Those scoring above 1000 by RetroTector, and with only one zinc finger (n
= 2534, from marmoset 32, dog 41, guinea pig 211, horse 4, duckbill 16, lemur 43, orangutan 82, rhesus 204, pig 79, tree shrew 11, lizard 162, cow 37, human 143, opossum 393, mouse 515, chimpanzee 192, rat 361, and zebra finch 8), were selected from RetroBank. The mouse genome assembly employed was mm8, from a C57 black mouse. Some were from the MLLV subset, as defined in . A study of their taxonomy was initiated by clustering and consensus sequence calculation at the 85% similarity level. It resulted in 75 interhost Pol consensus sequences which together with reference Pols were used to build the tree shown in the simplified form in . This is part of JBs ongoing work with retroviral taxonomy and will be reported in a more complete form in future papers. Especially many seemingly intact, potentially infectious MLLV proviruses were found in the mouse. In this review, we will concentrate on MLLV of mice and mention other rodents, pigs, felines, primates, and some marsupials. Of 7646 retroviral sequences detected in the mm8 assembly, 1461 were gammaretrovirus-like [47
]. Some of the latter (300 proviruses) scored higher than 2000 by ReTe (Figures and ). This is a high score, achieved only by complete or virtually complete proviruses. Indeed, they all turned out to be complete proviruses with very few stop or shift (indel) mutations which could incapacitate the virus. As mentioned above, the 300 proviruses included 35 which had no such mutations. They were structurally “intact” by bioinformatic criteria. The 35 had less than 0.5% LTR divergence. They are marked with green arrows in the Pol tree presented in . Thus, the 35 proviruses have hallmarks of being infectious and also belong to the most recently integrated murine ERVs.
Figure 3 Neighbor-joining (NJ) tree based on Pol amino acid sequences of 300 high ReTe scoring MLLVs, the same as in . The three high-scoring murine gamma groups (G1–G3) segregate in a similar way as in a gag nt-based tree (). Bootstrap (more ...)
gag sequences of 300 high scoring mouse gammaretroviral sequences were aligned together with reference sequences. MLV sequences with ascribed tropism, from GenBank, were also added. The tree was rooted with a divergent rabbit gammaretrovirus-like sequence (more ...)
Three major groups of high scoring murine gammaretroviral proviruses, named gamma 1–3 (G1–G3), were observed.
Group G1 (188 members, 10 with open reading frame (ORF) in gag, pro, pol,
) members encompassed the “Mus musculus
endogenous retrovirus” (MmERV; GenBank Id AC005743 [13
], as interpreted by RetroTector online [49
]). Mus dunni
ERV (AF053745) [14
] is highly related. The most similar nonmouse viruses were from rat chromosomes 7 and 17 (nr4 assembly), and more distantly, gibbon ape leukemia virus (GaLV, PCGGPE) and koala retrovirus (KoRV, AF151794) sequences.
Group G2 (59 members, 3 with ORF in gag, pro, pol,
) contained the GLN retroviruses described by Ribet et al. [15
]. It was most related to rat sequences at chromosomes 7 and 9.
A group of porcine ERVs (PERVs) located at chromosomes 9, 10, 12, and 4 (susScr10 assembly) were 74% similar to the consensus of Group G2, and 70% to the consensus of Group G1. MuRRS [18
] and MuERVC [19
] sequences were ancestral to groups G1 and G2 at the level of 64% similarity to their consensuses.
Group G3 (53 members, 22 with ORF in gag, pro, pol,
) encompassed the ampho-, eco-, xeno-, poly-, and modified polytropic MLVs [50
]. Most of the MLVs which have been prominent in retrovirological research for half a century are ecotropic [51
]. Amphotropic MLVs are primarily exogenous, while the others are mainly endogenous. The recombinant endogenous Mus spretus
] emerged between modified polytropic and xenotropic proviruses. The HEMV provirus was at the root of the G3 branch [17
The three major groups were discernible in trees made with several techniques resulting from alignment of nucleotides and protein sequences of the three genes gag, pol, and env. They represent three evolutionarily recent bursts of gammaretroviral proliferation in the mouse and its immediate progenitors. The third group, which includes the retroviruses reported in the human diseases, prostate cancer and ME/CFS, contains the highest proportion of structurally intact proviruses. It may thus have the greatest zoonotic potential.
In fact, the ancestors of humans were not spared infections with retroviruses related to MLLV. The so-called HERV-T is highly similar to MLLV (). It has around 30 representatives in the human genome [52
]. More distantly related are ERV-E and ERV-9W. None of the three are structurally intact in the human genome (J.B., unpublished). They are different enough from the murine G3 MLLVs to not interfere with the nucleic acid based methods for XMRV/HMRV detection (J.B., unpublished). Judging from the degree of mutational damage, HERV-T sequences integrated in a human primate progenitor genome around 30 million years ago [52
]. MLLVs include the pig endogenous gammaretroviruses, PERV A, B, and C, several of which are infectious and are a problem for xenotransplantation of porcine organs to humans. The murine MLLV groups G1-3 also contain structurally intact proviruses. Very little attention has been paid to groups G1 and G2, while group G3 (“MLVs”) has been thoroughly investigated. The number of references regarding the G3 group would be staggering, and out of scope for this review. The receptors and host range of groups G1 and G2 are largely unknown. However, the GLN retroviruses seem to have a tropism similar to ecotropic MLVs [15
]. Group G3 is known to contain MLLVs with several envelope-determined tropisms, see, for example, [53
2.2.3. Known and Probable Instances of Transspecies Transfer of Gammaretroviruses
As seen in , MLLVs can infect a wide range of hosts. For example, the XPR1 receptor, which is used by xeno- and polytropic MLVs, is common among vertebrates [57
]. In some cases, prey-predator relations probably contributed to the transmission [59
]. In other cases, there are no such known relations. The wide range of hosts is reflected in the panorama of their receptors [55
]. They are known to spread via several modes: often via saliva (e.g., into wounds of fighting animals) and sexual contact [26
]. Moreover, chimpanzees seem to have been infected with MLLVs from baboon and other primates [59
], while baboons and cats also have common MLLVs [73
]. Thus, MLLVs and similar gammaretroviruses have a tendency to spread between vertebrates.
Figure 5 MLLVs have spread among vertebrates in recent evolutionary time. The approximate time estimates to the last common MLLV progenitor are based on references given in the text, and on the phylogenetic analysis of . Some gibbon apes in captivity have (more ...)
However, a barrier against spread to humans may be the strong anti-α
-galactosyl antibodies in humans, which can neutralize viruses coming from species with different glycosylation patterns, like the mouse [74
]. Once the virus has entered the body, its sugars will follow the human glycosylation pattern, and the virus will not any longer be neutralized. Therefore, this barrier is not absolute.
A variety of other restrictions, like the APOBEC cytidine deaminases [50
], tetherin [53
], and TRIMs [77
] also affect retroviral spread between species. However, restrictions may be as important within a natural host as between hosts [80
]. The high XMRV replication in the cell line 22Rv1 [82
] and the ready growth of XMRV in LNCap cells [83
], both RNAse L-deficient human prostate cancer cell lines, plus the ability to grow in human PBMCs [84
], indicate the ability of XMRV to grow in human cells [84
] and the importance of an intact interferon system for the defence against it. These barriers to spread could probably be overcome, and humans be infected by XMRV, although the infectivity in vivo
is hard to predict. It was recently reported that XMRV can grow in human PBMCs [84
Judging from the wide spread of MLLVs, a zoonotic spread of XMRV/HMRV from mouse to human, directly or indirectly via another vertebrate, is not impossible. Humans are occasionally exposed to animals which harbor MLLVs. For example, microbes known to spread to humans from pets are viruses (arena-, hanta-, pox-, orthomyxo-, and rhabdoviruses), bacteria (chlamydiae, salmonella, tularemia, and leptospira), protozoa (toxoplasmosis), and helminths (worms). Rabbits, mice, rats, and guinea pigs are frequent as pets. The frequency of animal contact should therefore be recorded in epidemiological investigations regarding MLLVs, like XMRV/HMRV, in humans.
Like other MLLVs, a human MLLV would be expected to spread via kissing, sex, intravenous drug use, blood donation, and possibly via breast feeding. Enough systematic tests for MLLVs in the corresponding body fluids have not been performed. There should also be an overrepresentation of XMRV/HMRV in intravenous drug users and in patients infected with other sexually transmitted microbes, like HIV [85
]. This needs more study. Credible transmission chains between ME/CFS patients (with the exception of outbreaks), between PC patients, from ME/CFS to PC, and from PC to ME/CFS have not been reported (cf. ).
How well does XMRV/HMRV fulfill the expected properties of an MLV spread to humans + means an argument for, and—an argument against expectation?