Elevated Rate of Fixation of Endogenous Retroviral Elements in Haplorhini TRIM5 and TRIM22 Genomic Sequences: Impact on Transcriptional Regulation

All genes in the TRIM6/TRIM34/TRIM5/TRIM22 locus are type I interferon inducible, with TRIM5 and TRIM22 possessing antiviral properties. Evolutionary studies involving the TRIM6/34/5/22 locus have predominantly focused on the coding sequence of the genes, finding that TRIM5 and TRIM22 have undergone high rates of both non-synonymous nucleotide replacements and in-frame insertions and deletions. We sought to understand if divergent evolutionary pressures on TRIM6/34/5/22 coding regions have selected for modifications in the non-coding regions of these genes and explore whether such non-coding changes may influence the biological function of these genes. The transcribed genomic regions, including the introns, of TRIM6, TRIM34, TRIM5, and TRIM22 from ten Haplorhini primates and one prosimian species were analyzed for transposable element content. In Haplorhini species, TRIM5 displayed an exaggerated interspecies variability, predominantly resulting from changes in the composition of transposable elements in the large first and fourth introns. Multiple lineage-specific endogenous retroviral long terminal repeats (LTRs) were identified in the first intron of TRIM5 and TRIM22. In the prosimian genome, we identified a duplication of TRIM5 with a concomitant loss of TRIM22. The transposable element content of the prosimian TRIM5 genes appears to largely represent the shared Haplorhini/prosimian ancestral state for this gene. Furthermore, we demonstrated that one such differentially fixed LTR provides for species-specific transcriptional regulation of TRIM22 in response to p53 activation. Our results identify a previously unrecognized source of species-specific variation in the antiviral TRIM genes, which can lead to alterations in their transcriptional regulation. These observations suggest that there has existed long-term pressure for exaptation of retroviral LTRs in the non-coding regions of these genes. This likely resulted from serial viral challenges and provided a mechanism for rapid alteration of transcriptional regulation. To our knowledge, this represents the first report of persistent evolutionary pressure for the capture of retroviral LTR insertions.


Background
The human genome encodes in excess of 70 tripartite motifcontaining (TRIM) genes. This family of genes is characterized by the presence of a RING domain, one or two B-box domains, and a coiled-coil domain. The C-termini of these genes can consist of one or more of a select number of protein domains, with the majority of the TRIM genes in the human genome containing either a SPRY or PRY-SPRY (B30.2) domain at their C-terminus [1]. SPRY-containing TRIM genes have undergone rapid and extensive amplification via gene duplication and many of these genes display evidence of a history of positive selection [2,3,4,5,6,7,8]. Furthermore, a number of B30.2 domain containing TRIM genes have been implicated in host defense mechanisms owing to their transcriptional induction following interferon or viral stimulation [7,9,10] or to their inherent ability to inhibit viral replication [11,12,13]. Primate TRIM6, TRIM34, TRIM22 and TRIM5 are all interferon inducible [9], while TRIM5 and TRIM22 possess antiviral properties [11,12,13].
Studies involving TRIM5 have revealed antiretroviral activities for the alpha splice variant of this gene (TRIM5) from members of all branches of Haplorhini primates, which include hominoid, Cercopithecidae (Old World monkey), and Platyrrhini (New World monkey) species [5,14,15,16,17,18,19,20,21,22,23]. A more recent report has demonstrated that TRIM5a from Strepsirrhini (prosimian) lineages also possess antiviral function [24]. Antiretroviral activity appears to be a general feature of TRIM5 as cow and rabbit orthologs are capable of blocking retroviral infection [25,26,27]. However, antiviral activity of any specific TRIM5/ virus combination is governed by features in both TRIM5 and virus, being largely determined by the B30. 2  Several reports have demonstrated an anti-HIV-1 activity for the closely related gene TRIM22. The mechanism by which this anti-retroviral activity is exerted has yet to be fully elucidated, but appears to involve suppression of LTR transcription [13,41,42]. In addition to the effects on HIV-1 replication, TRIM22 has been shown to inhibit hepatitis B virus and encephalomyocarditis virus replication in tissue culture systems [43,44].
TRIM5 and TRIM22 along with TRIM6 and TRIM34 form a locus of four closely related TRIM genes within an olfactory receptor-rich region on chromosome 11 of the human genome. Genetic analysis of this locus indicates that these four genes have evolved by gene duplication from a common ancestral gene [1]. Host genes in direct evolutionary competition with pathogenic invaders are often subject to high levels of positive selective pressures and this is the case for both TRIM5 and TRIM22. Both genes have faced positive selective forces during primate evolution [4,5,45], but evolutionary constraints appear to have resulted in a dichotomous situation where along any given primate lineage only one of these genes (either TRIM5 or TRIM22) shows signs of positive selection [4]. Additionally, positive selective pressure on these genes has not been uniformly applied across the gene. The B30.2 region, consistent with its role in mediating viral recognition, has faced positive selective forces, while the tripartite motifencoding regions have been under purifying selection to maintain function of the RING, B-box and coiled-coil domains [5,12,45]. In contrast, over the same period of time TRIM6 and TRIM34 have been under constant purifying selection.
In addition to coding sequence changes, the structure and composition of the TRIM6/34/5/22 genomic locus has undergone lineage-specific changes during mammalian evolution. The genomic locus of all studied Haplorhini primates has the following gene order: TRIM6-TRIM34-TRIM5-TRIM22, where TRIM5 is present in opposite orientation to the rest of the genes ( Figure 1A). This genomic structure differs from that of the ancestral locus, which is maintained in the dog and other species [Diehl WE,unpublished data and 4], where TRIM5 is found in the same orientation as the other three genes. Rodent genomes have TRIM5 in the same orientation as primates, thus the inversion event most likely occurred prior to the rodent-primate divergence [6]. In contrast to what is found in primate genomes, TRIM5 has undergone gene duplication events in rodents that were accompanied by a resultant loss of the TRIM22 gene. Furthermore, in mice, but not rats, one duplication event resulted in two copies of TRIM34 being present in the genome [6]. In the cow genome, similar to in rodents, TRIM5 has undergone a number of gene duplication events in conjunction with a loss of the TRIM22 gene resulting in five TRIM5 orthologs as well as three pseudogenes. In contrast to the rodent genomes, the TRIM5 genes in cows are all present in the positive orientation [4].
A subtler means of genome modification involves insertion and removal of short nucleotide sequences, a phenomenon that often involves transposable elements. In the human genome, the most complete and fully characterized mammalian genome, approximately 43% of the DNA content is derived from retrotransposons, in the form of long interspersed nuclear elements (LINEs), short interspersed nuclear elements (SINEs) and endogenous retroviral (ERV) elements, while DNA transposons account for an additional 2% of the genome [46]. In certain contexts these elements have been shown to potentiate altered transcriptional regulation. LINE elements can modulate normal splicing [47,48], alter polyadenylation [49], interfere with transcriptional elongation [50], and provide novel promoter activity for neighboring genes [51]. SINE family members, including Alu elements, have been shown to have effects upon polyadenylation [52,53,54], alternate splicing patterns [54], and translation efficiency [reviewed in 55]. ERV sequences are generally found in the genome either as the mobilizationcompetent form of two LTRs flanking internal retroviral-derived The following species were examined in this study: human (Homo sapiens), chimpanzee (Pan troglodytes), white-cheeked gibbon (Nomascus leucogenys), olive baboon (Papio anubis), rhesus macaque (Macaca mulatta), guereza colobus (Colobus guereza), Peruvian red-necked owl monkey (Aotus nancymaae), common marmoset (Callithrix jacchus), Bolivian squirrel monkey (Saimiri boliviensis boliviensis), dusky titi (Callicebus moloch), and grey mouse lemur (Microcebus murinus). These species are highlighted using '*' as well as bold lettering. This phylogenetic tree was adapted from Bininda-Emonds et al. 2007, and uses the revised dates published with the corrigendum on the original article. doi:10.1371/journal.pone.0058532.g001 sequences or solo LTRs that are the product of homologous recombination between the two LTRs. The overwhelming majority of ERV sequences in the human genome are solo LTRs, which contain most of the viral transcriptional regulatory elements [46,56]. It has been shown that these LTRs can provide cellular genes with primary or alternate promoters, enhancer elements, protein-binding sites, and polyadenylation signals [57,58,59,60].
In order to determine what effects complex evolutionary pressures in the TRIM6/34/5/22 locus have had on the noncoding genic sequences, we retrieved primate sequences corresponding to the transcribed region of each of these TRIM genes, including introns, from publicly available sequence databases. These were assembled, aligned, and compared among 10 Haplorhini primates (three Hominoid, three Cercopithecidae, and four Platyrrhini) and one prosimian species. In Haplorhini primates, we observed a hierarchy of overall sequence conservation in these genes (TRIM6 . TRIM34 . TRIM22 . TRIM5) resulting from sequence insertions and deletions (indels) involving transposable elements. Differences in transposable element content are largely localized to the large first and fourth introns, with the first intron of TRIM5 and TRIM22 containing multiple lineagespecific ERV LTRs and dramatic gains and losses of LINE element content in the fourth intron. In contrast, the TRIM6/34/ 5/22 locus in prosimians has undergone duplication of TRIM5 and loss of TRIM22, with no accumulation of novel LTR insertions and limited LINE L1 changes in the fourth intron. We further tested the ability of one specific ERV LTR element to alter regulation of the gene it resides in. This element, LTR10D, is present in the first intron of TRIM22 in Hominoids and Cercopithecidae, but is absent in Platyrrhini. Using quantitative reverse transcriptase polymerase chain reaction (RT-PCR) on RNA from peripheral blood mononuclear cells (PBMCs), we found that TRIM22 was upregulated in response to p53 stimulation only in the LTR10D containing Old World primates. Based on these findings, we hypothesize that the antiviral TRIM genes of Haplorhini primates have evolved species-specific differences in transcriptional regulation, which are mediated by transposable element sequences in their non-coding regions.

TRIM5 Displays Increased Genetic Diversity Compared to Related TRIM Genes, as a Result of Turnover in Transposable Element Content in Intronic Sequences
To examine evolution of the primate TRIM6/34/5/22 locus at the genomic level, orthologous sequences were retrieved from NCBI nucleotide sequence databases (see Methods). In total, sequence information from ten Haplorhini and one prosimian species was retrieved from the online databases, including: human (Homo sapiens), chimpanzee (Pan troglodytes), white-cheeked gibbon (Nomascus leucogenys), olive baboon (Papio anubis), rhesus macaque (Macaca mulatta), guereza colobus (Colobus guereza), Peruvian rednecked owl monkey (Aotus nancymaae), common marmoset (Callithrix jacchus), Bolivian squirrel monkey (Saimiri boliviensis boliviensis), dusky titi (Callicebus moloch), and grey mouse lemur (Microcebus murinus). Thus, the species included in this study represent a broad sampling of Haplorhini lineages as well as a representative prosimian species to serve as an outgroup ( Figure 1B). Following identification and compilation of the genomic sequence information for these TRIM genes, multiple sequence alignments were generated for each TRIM gene's transcribed region (Files S1-S4). Due to differences in sequence composition and genomic architecture of the locus (see below) prosimian sequences were not included in the initial comparisons.
Next, percent nucleotide identity between species was calculated for each TRIM gene. It was observed that the degree of conservation for genomic sequences of genes in the TRIM6/34/ 5/22 locus have an inverse correlation with the previously reported dN/dS rate ratios for these genes ( Figure 2A) [4,61]. Thus, conservation of the genomic sequences for the genes in this locus is TRIM6 . TRIM34 . TRIM22 . TRIM5. With some species sharing as little as 44% genomic sequence identity. This simple analysis demonstrated that TRIM5 sequences are significantly less conserved, at the genomic level, than the other genes in this locus. As might be expected, due to the relative size difference and relaxed evolutionary constraints, most of the sequence divergence was isolated in the intronic regions (data not shown).
In order to determine the relative contribution that nucleotide substitutions versus indels have played in creating the divergent levels of sequence conservation between the TRIM genes, the different types of sequence changes were independently analyzed. It was observed that over the course of Haplorhini evolution the genomic sequences for all four TRIM genes have undergone similar nucleotide substitution rates ( Figure 2B). In contrast, very different rates of diversification resulting from indels were observed among the individual genes of the TRIM6/34/22/5 locus ( Figure 2C). The average rate of change due to indels in TRIM5 is significantly greater compared to all of the other genes, while the rate in TRIM22 was found to be significantly greater than either TRIM6 or TRIM34. While it is informative to compare the average rate of indel change in these genes, it should also be highlighted that the calculated rate between any two species can vary widely ( Figure 2C). This is especially true for TRIM5, where there is a large variance in the observed rate of indel-associated change. When the pairwise comparisons are separated by the evolutionary relationship of the species involved (Figure S1A-D) it becomes apparent that the rate of indel change in these genes is lineage specific. For example, in contrast to what is observed in other lineages, Old World primate TRIM5s display remarkably low rates of indel change, comparable to what is observed for the other TRIM genes. In spite of these lineage-specific differences, a strong inverse correlation between the rate of change due to indels and the percent nucleotide identity was observed for all TRIM genes in this analysis ( Figure 2D-G). In contrast, no correlation between the nucleotide substitution rate and the percent nucleotide identity was found ( Figure S2A-D). These observations strongly suggest that indels have been the main generator of sequence diversity in these genes during primate evolution.
The increased rates of sequence divergence resulting from indels observed for TRIM5 and TRIM22 could be explained by a proportionately greater accumulation of novel indels, the presence of one or more disproportionately large indels, or a combination of both. To assess the contribution of accumulation of novel indels, the rate of indel fixation was estimated in pairwise fashion by dividing the number of indels present between the sequences of two species by the estimated time since the last common ancestor ( Figure 3A). This analysis revealed a subtle but statistically significant elevation in the rate of indel accumulation for TRIM5 and TRIM22 compared to TRIM6 or TRIM34. To examine the contribution of indel size to the differential rate of nucleotide divergence, the average and median indel sizes for each gene were calculated (Figures 3B and data not shown). Widely divergent average indel sizes were found for these genes, with the average indel size in TRIM5 genes being significantly larger and the average indel size in TRIM6 significantly smaller than the other genes. In contrast, all TRIM genes examined were found to have a median indel size of approximately 2 nucleotides.
The observed difference between median and average indel size suggests an influence of proportionally few large indels in driving the sequence divergence rates of these genes. For a more detailed analysis of this, indels from each gene were separated into 100 nucleotide bins of increasing indel size. The number of binned indels for each gene are depicted on a log 10 scale in Figure 3C-F. This analysis highlights the fact that the overwhelming majority of indels are between 1 and 100 nucleotides in size for all TRIM genes examined, with between 81% and 87% of indels being under 10 nucleotides in size. Where these genes differ is in the number and size of indels greater than 300 nucleotides. TRIM6 has very few large indels (.300 nucleotides), TRIM22 and TRIM34 have intermediate numbers, while TRIM5 has, by far, the greatest number of large indels. From this data, we conclude that the disparate nucleotide conservation rates observed in these genes is the result of a combination of gene-specific variation in rates of indel accumulation and the differential fixation of large indels.
To ask whether the presence of variable numbers of large indels reflected accumulation of transposable elements, RepeatMasker software was used to identify transposable elements present in each of the TRIM genes. This analysis revealed that these genes contain members of all the major families of primate transposable elements, including LINEs, SINEs (MIRs and Alu elements), as well as DNA transposons and endogenous retroviral elements. Examples of all classes of LINEs, SINEs, and DNA transposons identified in this analysis are present in the genomes of both Haplorhini and Strepsirrhini primates, even if the specific insertions characterized here are not. In contrast, several of the endogenous retroviral LTRs identified in this locus are restricted to specific primate lineages.
Without regard to classification, the number of transposable elements identified in the different TRIM genes from each primate species was totaled ( Figure 4A). It was found that TRIM22 and TRIM5 have significantly more transposable elements than TRIM6 and TRIM34. However, this analysis ignores the nature or conservation of the transposable elements, and therefore the number of unique elements found in pairwise comparisons of sequences was also quantified ( Figure 4B). This analysis revealed that, in primates, TRIM6 has the fewest lineage-restricted transposable element insertions, TRIM34 and TRIM22 have an intermediate number, and TRIM5 has significantly more of these non-conserved transposable elements than all of the other genes. Furthermore, when these data were corrected for the evolutionary distances between these comparisons, similar results were observed with the same trend toward increasing rates of transposable element turnover going from TRIM6 to TRIM5 (data not shown). Comparing the number of unique transposable elements in each of the TRIM genes with the average indel size yields strong positive correlations ( Figure 4C-F). This correlation for TRIM6 was found to be significant in spite of the fact that colobus TRIM6 contains a novel LINE L1 element of approximately 6 kilobases in size, resulting in abnormally large average indel sizes in comparisons involving this species, as indicated with red dots ( Figure 4C). Taken together, these analyses point to a striking involvement of transposable elements in the evolution of genes in the TRIM6/34/ 5/22 locus, especially those genes whose coding sequences were under positive selective pressures in Haplorhini evolution. TRIM genes are variably conserved as a result of differential rates of indel turnover. Transcribed genomic sequences of the TRIM5, TRIM6, TRIM22, and TRIM34 genes were hand aligned. Using the formulas presented in the Materials and Methods, the nucleotide alignments were used to calculate the following statistics for all pairs of nucleotide sequences: the percent nucleotide identity (A), the nucleotide substitution rate (B), and the rate of indel change (C). The rate of change depicted in these figures is calculated as percent per million years, the black circles indicate separate pairwise sequence comparisons and the bars represent mean values. In panels A-C, statistical significance was calculated using the Friedman test, a one-way repeated measures ANOVA without assuming Gaussian distributions and using the Dunn's post-test to compare all genes against one another. A p-value of less than 0.05 is denoted by *, a p-value less than 0.01 is denoted by **, a p-value less than 0.001 is denoted by ***. For each gene, the correlation between nucleotide identity and the rate of indel change was examined using Spearman's rank correlation (D-G). The Spearman r and p-values resulting from this analysis are indicated in each panel. doi:10.1371/journal.pone.0058532.g002 Figure 3. TRIM5 and TRIM22 display elevated rates of indel fixation as well as fixation of larger indels. Using the formulas presented in the Materials and Methods, the nucleotide alignments were used to calculate the following statistics for all pairs of nucleotide sequences: the rate of indel creation (A) and the average indel size (B). Dots indicate separate pairwise sequence comparisons and the black bars represent mean values. Statistical significance was calculated using the Friedman test, a one-way repeated measures ANOVA without assuming Gaussian distributions and using the Dunn's post-test to compare all genes against one another. A p-value of less than 0.05 is denoted by *, a p-value less than 0.01 is denoted by **, a p-value less than 0.001 is denoted by ***. Indels present in pairwise comparisons of each gene were separated by size and placed into a corresponding 100 nucleotide 'bin'. The number of indels present in each bin is depicted for TRIM6 (C), TRIM34 (D), TRIM22 (E), and TRIM5 (F). doi:10.1371/journal.pone.0058532.g003 Figure 4. TRIM5 and TRIM22 contain more transposable elements than TRIM6 or TRIM34. The absolute number of transposable elements present in each TRIM gene was tallied for each species and the results are depicted in panel (A). Black dots represent the number of transposable elements found in a given primate species and the black bar represents the mean value. The quantitation shown in (A) was performed without regard for identity or conservation of the elements present, therefore the number of novel transposable elements was considered. In panel (B), the number of unique transposable elements present in pairwise comparisons of each TRIM gene is shown. The black dots indicate number of elements in an individual pairwise sequence comparison and the black bar represents the mean value. In panels (A) and (B), statistical significance was calculated using the Friedman test, a one-way repeated measures ANOVA without assuming Gaussian distributions and using the Dunn's post-test to compare all genes against one another. A p-value of less than 0.05 is denoted by *, a p-value less than 0.01 is denoted by **, a p-value less than 0.001 is denoted by ***. Correlations between average indel size and the number of unique transposable elements were examined TRIM6 (C), TRIM34 (D), TRIM22 (E), and TRIM5 (F). Statistical significance was assessed using Spearman's rank correlation the r and p-values resulting from this analysis are indicated in each panel. Comparisons involving colobus TRIM6, which contains a 6-kb LINE L1 element insertion, are indicated with red dots. doi:10.1371/journal.pone.0058532.g004

Lineage-specific Transposable Element Differences Localize to the First and Fourth Introns and the Difference in TE Composition in these Regions is Dominated by ERV LTR and LINE L1, Respectively
Transposable elements identified by RepeatMasker were then mapped back onto the multiple sequence alignments. The position and orientation of these transposable elements in relation to the exon/intron structure of the four TRIM genes are depicted in Figures 5 and 6. This analysis makes starkly clear the difference in magnitude of the genetic variability between the genes of the TRIM6/34/5/22 locus and highlights both the spatial location and involvement of specific transposable element families resulting in this variability.
TRIM6. In the species sampled, there are few differences in transposable element content between Haplorhini TRIM6 sequences ( Figure 5A). Alu elements account for the majority of lineage-specific transposable elements in TRIM6 where nine of these elements are found in a lineage-restricted manner. All nine of these Alu elements are present in multiple related species. Thus, all of these Alu insertions must have occurred prior to speciation and therefore can be inferred to be fixed in the species from which they are identified. We will refer to such elements shared by multiple species as being fixed throughout. In addition to the multiple lineage-specific Alu insertions, two other changes involving transposable elements were detected in TRIM6 sequences. In the fifth intron of Owl and Squirrel monkeys a deletion resulted in the loss of Alu-derived sequence. Also, as previously mentioned, the fourth intron of colobus contains a novel and nearly intact LINE L1 element. The colobus LINE L1 insertion is representative of many events in the TRIM6/34/5/22 locus in which a lineagespecific change was identified in a single species. Except for humans, each species analyzed is represented by the genome of a single individual, thus no inference can be made as to the prevalence of the observed insertions or deletions in the population. Such events could be fixed, present only in the individual sampled, or present at some frequency in between these extremes.
TRIM34. In Haplorhini primates, TRIM34 has more lineagerestricted transposable element content than TRIM6 ( Figure 5B). For instance, there are 25 independent Alu insertions in TRIM34, with only one fixed in all Haplorhini species sampled. The other 24 are lineage-restricted with 8 fixed in multiple species and 16 found only in a single species. Additionally, two deletion events were identified in the fourth intron of Haplorhini TRIM34. One removes approximately 3,400 nucleotides of Platyrrhini intronic sequence, while the other removes approximately 225 nucleotides from Old World (baboon, macaque and colobus) monkeys. In contrast to the TRIM6 sequences, which are devoid of ERV content, TRIM34 contains five ERV LTR elements. Three of these are found to be lineage-restricted, with one (ERV3-2_CJa-LTR) being a novel insertion found in the squirrel monkey genome. The other two ERV elements (PABL_B and MLT1K) appear to have been lost from Platyrrhini as a consequence of the large deletion event in the fourth intron. Finally, one novel LINE L1 insertion was identified in the fourth intron of baboon TRIM34.
TRIM22. The number of lineage-restricted transposable elements and complexity of intronic diversity of Haplorhini TRIM22 sequences is similar to that of TRIM34 ( Figure 5C). TRIM22 sequences contain 27 independent Alu insertions. Four of these are conserved in all species examined, while 23 Alu insertions are found to be lineage-restricted; 12 of these fixed in multiple species and 11 found only in a single species. Intron four of Platyrrhini TRIM22 contains a deletion of approximately 3,200 nucleotides, similar in size to the deletion in the fourth intron of Platyrrhini TRIM34. Three additional large deletions occurred in the first and fourth intron of TRIM22, where each is restricted to a single species. Similar to TRIM34, there are five lineage-restricted ERV LTR elements found in TRIM22, with one (MacERVK) being recently inserted into a single species, colobus in this instance, and one LTR (MLT1C) lost from Platyrrhini as a result of the large deletion in intron four. In contrast to TRIM34, three ERV LTRs inserted and became fixed during Haplorhini evolution. An LTR10D element became fixed in the first intron of TRIM22 early in Old World primate evolution and is present in all extant Cercopithecidae sampled. Later, an LTR2 element inserted and became fixed in the ancestral hominoid lineage and is found immediately downstream of the LTR10D element in all hominoid species sampled. Finally, an LTR element that appears to be a recombinant between LTR10A and LTR10B inserted and became fixed in intron two of the ancestral Platyrrhini TRIM22, where it is fixed in all species sampled.
TRIM5. In contrast to the three genes of this locus discussed above, Haplorhini TRIM5 sequences exhibit a significant amount of transposable element associated sequence divergence ( Figure 6). Thirty independent Alu insertions are found in the TRIM5 sequences examined. Of these, five are conserved amongst all Haplorhini sequences examined, while 25 are restricted to specific lineages; 11 of these are fixed in multiple species, while 14 are present in a single species. Neither the absolute number of Alu insertion events nor the lineage distribution in host genomes is dramatically different than either TRIM34 or TRIM22. A combination of at least 10 large deletion events in addition to numerous insertions cause the near complete replacement of both the first and fourth intronic sequences in Haplorhini primates. Only short regions of homology are conserved in all ten primate species. In the first intron, these regions consist of a 185-nucleotide region immediately following the first exon and a 730-nucleotide region immediately adjacent to the second exon, which includes sequence of LINE L2 origin. In the fourth intron, the conserved regions consist of 310 nucleotides immediately adjacent to the fourth intron, less than 50 nucleotides of LINE L1-derived sequence in the internal region of the intron, and approximately 2200 nucleotides at the 39 end of the intron. Interestingly, the majority of the transposable element content gained and lost from the first and fourth introns of TRIM5 is derived from a single class of transposable elements in both cases, these being ERV elements in the first intron and LINE L1 elements in the fourth intron.
Represented in the TRIM5 genes examined here are a total of ten ERV insertion events containing 11 LTR elements. None of these are conserved in all ten species examined as the result of lineage-specific deletion events excising one or more LTR elements. While all other ERV-derived sequences in all genes examined are solo LTRs (formed as the result of homologous recombination between the 59 and 39 LTRs of an ERV element), the HUERS-P3b sequence found in Platyrrhini primates is composed of internal ERV sequences linked to their 59 LTR9B element that is present in all species except human, chimpanzee and squirrel monkey. HUERS-P3b elements are non-traditional, replication-defective endogenous retroviral-like elements comprised of various fragments from what were at one time replication-competent ERVs. They contain a central LTR8A element, and are flanked by 59 and 39 LTR9B elements. The Platyrrhini TRIM5 HUERS-P3b sequence, in conjunction with its LTR9B 59 LTR, is found in the negative orientation with respect to TRIM5 transcription and is missing ,2500 nucleotides from its 39 terminus, including the 39 LTR. In squirrel monkeys, two separate deletion events have removed from this element the entire LTR9B 59 LTR as well as a significant portion of the internal LTR8A. In contrast, a solo LTR9B element is found in Old World primates consistent with homologous recombination between the 59 and 39 LTRs of the original intact HUERS-P3b retroelement and complete loss of the internal sequences.
The Prosimian TRIM6/34/5/22 Locus has an Altered Chromosomal Architecture and Contains Less Transposable Element Content than Haplorhini Primates We sought to determine if the differences in transposable element content and intronic turnover seen in Haplorhini species was a universal characteristic of these genes in primates. Thus, we examined this locus in the grey mouse lemur (Microcebus murinus). In this species, we discovered a TRIM6/34/5 locus with much different architecture than that of Platyrrhini or Catarrhini ( Figure 7A). In the grey mouse lemur a combination of a gene duplication event along with a deletion event has resulted in the presence of two TRIM34 and TRIM5 genes and the loss of TRIM22. Following this duplication, there appears to have been a loss of exons 5 through 8 of the TRIM34-1 gene, leaving a partially encoded pseudogene. Confirmation of this genomic structure was made through gene specific PCR amplification of intergenic sequences (data not shown). Such a genomic architecture, with multiple TRIM5 genes, has not previously been reported in primates, although expansions of TRIM5 copy number are found in mice, rats, and cows [4,6]. While the novel architecture of the grey mouse lemur's TRIM5 locus was unanticipated, the extant genes of this locus can still serve as an outgroup for the previously characterized Haplorhini TRIM genes. In this regard, transposable elements were identified and mapped back onto the mouse lemur TRIM genes ( Figures 7B-D). A sequence corresponding to the first exon of TRIM6 could not be identified in the grey mouse lemur genome sequence, therefore we included in our analysis sequences extending approximately 1500 nucleotides upstream of the promoter proximal MIR element, which is present in all TRIM6 genes examined ( Figures 5A and 7B). In comparison to the Haplorhini sequences, none of the Alu elements in the grey mouse lemur genome are found in any Haplorhini sequences (i.e., all 5 Alu elements present are unique to the lemur). Additionally, while the lemur sequence contains an MER33 element in the first intron, it lacks the second MER33 element present in the 39 UTR of Haplorhini species ( Figure 5A). With those exceptions, the transposable element content of the mouse lemur TRIM6 is nearly identical to what is fixed in all Haplorhini sequences examined ( Figures 7B, 5A).
Similar to TRIM6, all of the transposable elements present in the lemur TRIM34-2 gene are also found as conserved elements in Haplorhini TRIM34 sequences ( Figure 7C). Notably absent from TRIM34-2 is the LTR87 insertion fixed in the first intron of all Haplorhini TRIM34s examined and also identified in many TRIM5 sequences. This element appears to have been lost from grey mouse lemur TRIM34 genes prior to gene duplication as the result of a deletion event, as flanking sequences are also missing (data not shown).
Finally, the TRIM5 genes from the grey mouse lemur harbor fewer transposable elements compared to their Haplorhini counterparts; of those present, most (9 of 10) are also found in Haplorhini TRIM5 sequences (Figures 6 and 7D). The only grey mouse lemur specific transposable element is found in the first intron of TRIM5-1, a LINE L1 element, which was inserted after gene duplication in this lineage. Of the two ERV LTRs identified in the first intron of the TRIM5 genes, LTR87 is found in all Haplorhini species studied except human and chimpanzee, while the MER21B element is only found in Cercopithecidae and gibbon. These two LTR elements appear to be the minimal complement of LTRs present in the first intron of TRIM5 at the time of divergence of Haplorhini from Strepsirrhini. The fourth intron of both mouse lemur TRIM5s contain three LINE L1 elements. Fragments of all three can be found in all Haplorhini species. The Platyrrhini LINE L1 content in this intron most closely resembles that of the lemur, while the Cercopithecidae have largely lost these LINE L1 sequences in a series of deletions.  [4,5,45]. We observed that in Haplorhini species the length of the fourth intron correlates with the previously published degree of diversifying selective pressure placed on the genes [4]. The average length of the fourth intron in Haplorhini primates is as follows: 3387 nucleotides in TRIM6, 4809 nucleotides in TRIM34, 6078 nucleotides in TRIM22, and 10098 nucleotides in TRIM5 ( Figure 8A). Concomitant with the increase in the length of the fourth intron has been an increase in the amount of LINE L1derived content within this intron ( Figure 8B). Thus, the TRIM6 from most Haplorhini species lack any LINE L1 sequences, while Haplorhini TRIM34 averages 1950 nucleotides, TRIM22 averages 2875 nucleotides, and TRIM5 averages 5222 nucleotides of LINE L1-derived sequence.

Over Representation of Newly Inserted LTR Elements in TRIM22 and TRIM5 Compared to TRIM34
One of the most striking differences in terms of transposable element content between the four TRIM genes studied is the abundance and variability of ERV elements in TRIM5, and to a lesser extent TRIM22, compared to the other genes. Nine unique ERV elements were found to have been fixed in TRIM5 genes during primate evolution. The HUERS-P3b element in Platyrrhini species contains an internal LTR8A, so these nine ERV elements result in ten novel LTR elements. During the same evolutionary time period, eight unique LTR sequences were fixed in TRIM22 genes, three unique LTRs were fixed into TRIM34, and no LTRs were fixed in TRIM6 (Figures 5 and 6; Table 1). Of the five LTR elements present in primate TRIM34 (Figure 5B), the LTR87 and MLT1K fragments could be identified in the same relative location in the TRIM5 loci from other mammalian species, including cats (data not shown). This argues that fixation of these two LTRs predates both the gene duplication event(s) that gave rise to TRIM5 as well as the speciation events that separated carnivores from primates, thus they were not counted in this analysis. No other LTRs from any of these genes were found to be conserved in both primate and non-primate TRIM gene sequences (data not shown).
Previous studies report a strong bias against LTRs within transcribed genic regions being in the positive orientation relative to the harboring gene, with approximately one in five LTR elements in the positive orientation. In vitro experiments suggest that this orientation bias occurs by negative evolutionary selection post-insertion and not as a result of bias during integration [62,63,64]. We therefore examined the orientation of primatespecific LTR elements in TRIM34, TRIM22 and TRIM5 (Table 1). Due to an inversion event resulting in a tandem but oppositely oriented repeat, the MLT1C element present in the fourth intron of TRIM22 was counted as two elements, one in the positive orientation and one in the negative orientation. It is clear from Table 1 that the orientation of LTR elements in the TRIM genes is non-uniform across genes. None of the three primate-specific LTR insertions into TRIM34 are in the positive orientation. Two out of eight LTRs in TRIM22 are in the positive orientation, although one of those is part of the previously mentioned MLT1C element, consistent with the previously reported ,20% orientation bias for genic LTR elements. In contrast to both TRIM34 and TRIM22, greater than half of the LTR elements present in the TRIM5 gene are found in the positive orientation. This is similar to the 50% probability of positive orientation based exclusively on integration. No orientation bias was observed for Alu elements in genic regions of Haplorhini TRIM5 (Table 1). Figure 8. LINE L1 associated evolution in the fourth intron. The number of nucleotides of LINE L1 origin (A), and the size of the fourth intron (B) were calculated for each of the TRIM genes examined. In both panels, the red dots represent the fourth intron of New World primates, while the blue dots represent the fourth intron of Old World primates. The black bars represent average values. Statistical significance was calculated using the Friedman test, a one-way repeated measures ANOVA without assuming Gaussian distributions and using the Dunn's post-test to compare all genes against one another. A p-value of less than 0.05 is denoted by *, a p-value less than 0.01 is denoted by **, a p-value less than 0.001 is denoted by ***. doi:10.1371/journal.pone.0058532.g008 Differential Transcriptional Regulation of TRIM22

Mediated by Fixation of an Intronic LTR10D Element
Transcriptional regulation of the TRIM genes involved in this study has not been the subject of detailed investigation. Of these genes, the regulation of TRIM22 transcription has been the best characterized. It has been reported that expression of human TRIM22 is upregulated upon p53 induction, that a p53-binding site is located in the first intron of this gene, and that mutations abolishing p53 binding to this site result in the loss of p53 responsiveness [65]. Following this, the p53-binding site was found to be located in an LTR10D element, and that a network of p53regulated genes exists as the result of fixation events involving either the LTR10 or MER61 families of ERV elements [57]. While these elements can be found in both Old and New World primate genomes, an LTR10D is only present in TRIM22 of Old World primate lineages ( Figure 5C). Based on the previous reports and our results demonstrating the presence of this element in Old World primates and absence of this element in New World primates, we hypothesized that TRIM22 of Old World primates would respond to p53 induction by upregulating transcriptional levels, while TRIM22 in New World monkeys would be refractory to p53 activation. In order to test this hypothesis, TRIM22 expression changes were assessed following induction of DNA damage, which is known to induce p53 [57,65], in peripheral blood mononuclear cells (PBMC) from healthy humans, rhesus macaques, and squirrel monkeys.
Pan-primate quantitative one-step RT-PCR was used to detect changes in TRIM22 mRNA expression ( Figure 9A). In accordance with previous work [57,65,66], TRIM22 expression levels in human and rhesus PBMCs displayed a significant increase in expression of 4.09-and 2.64-fold, respectively. In contrast, no significant change in TRIM22 RNA levels was observed in squirrel monkey PBMCs, although treatment with 5-fluorouracil tended to reduce TRIM22 expression. Thus, the general pattern of TRIM22 expression level changes following p53 induction was consistent with the presence or absence of the p53-binding site containing LTR10D element in a given species.
In order to verify that the changes to TRIM22 expression were specific for this gene, RNA levels were measured for the control genes murine double minute 2 (MDM2), TRIM5, and b-actin. MDM2 is a well-studied inhibitor of p53 activity whose expression is directly controlled by p53 activation [67,68,69], while transcription of TRIM5 and b-actin have been shown to be largely unaffected by DNA damage [66]. Thus, MDM2 expression serves as a positive control for p53 induction, while TRIM5 and b-actin serve as indicators of global transcriptional changes. When MDM2 expression levels were assessed, it was observed that this gene was upregulated in the PBMCs of all three species (Figure 9B), demonstrating that the differential transcriptional regulation seen for TRIM22 was not due to a failure to induce p53 using the DNAdamaging agents. In contrast, TRIM5 and b-actin expression levels remained largely unchanged in PBMCs following DNA damage ( Figure 9C-D). These results clearly demonstrate that TRIM22 is specifically upregulated in human and rhesus PBMCs following p53 induction, and that the p53-binding site in the first intron does not mediate transcriptional regulation of neighboring genes.

Discussion
Pathogenic invaders can impose major selective pressures on genes involved in the antimicrobial response. Variability in TRIM5, and the genomic locus in which it resides, reflects the consequences of such selective pressures. This is embodied by: the variation in copy number of TRIM5, TRIM22, and TRIM34 between species [5,6,26], elevated rates of non-synonymous change in the coding sequence of TRIM5 and TRIM22 [4,5,45], maintenance of balanced polymorphisms as well as cross-species sharing of polymorphisms in TRIM5 [70,71], and convergent evolution of exon replacement via cDNA retrotransposition [36,72,73,74,75,76,77].
For the four TRIM genes studied here, we found that the degree of nucleotide divergence between species for each gene parallels the dN/dS rate ratio previously reported for the coding regions of that gene [4,61]. For example, TRIM6, the gene in this locus with the lowest dN/dS rate ratio, was observed to have on average 91% nucleotide identity between Haplorhini species. In contrast, TRIM5 has been shown to have the highest dN/dS rate ratio, and genomic sequences had dramatically lower conservation (with only 65% average nucleotide identity between the species included in this study). This high level of diversity correlated with elevated rates of transposable element turnover, much of which was due to alterations in either ERV LTR content in the first intron or LINE L1 content in the fourth intron. In between these extremes, TRIM22 and TRIM34 were both found to have between 83% and 85% nucleotide identity in these species. Both of these genes feature a large deletion removing much of the LINE L1 content in the fourth intron. However, in contrast to TRIM34, TRIM22 contains lineage-specific LTR insertions in the first intron.
It is possible that positive selective forces on the coding regions of TRIM5 and TRIM22 have indirectly lead to the differences in fixation of transposable elements observed. This is possibly the case for the evolutionary forces acting on the fourth intron of TRIM22 and TRIM5, where increasing LINE L1 content and Figure 9. Transcription of TRIM22 from different primate species is differentially regulated following p53 induction. Following a 3-day stimulation with PHA, PBMCs from humans, rhesus macaques, or squirrel monkeys were treated with doxorubicin or DMSO control for 24 hours. Total RNA was harvested and levels of (A) TRIM22, (B) MDM2, (C) TRIM5, and (D) b-actin mRNA were assessed using SYBR green-based qPCR with 50ng per reaction input RNA. Shown are the fold changes measured following drug treatment compared to DMSO controls, as calculated using the DDC(t) method. Error bars represent 6 SEM from three independent experiments with n = 7 human subjects, n = 6 rhesus subjects, and n = 4 squirrel monkeys subjects. doi:10.1371/journal.pone.0058532.g009 associated intronic length (3387 nucleotides in TRIM6, 4809 nucleotides in TRIM34, 6078 nucleotides in TRIM22, and 10098 nucleotides in TRIM5) are correlated with increasing positive selection of the B30.2 coding region. Taken together, we hypothesize that the fourth intron may function as a ''flexible spacer'' between the evolutionarily conserved exons encoding the RING, B-box, and coiled-coil domains and the rapidly evolving B30.2 domain.
Differential fixation of LTR elements in the first intron of the TRIM5 and TRIM22 genes is unlikely to be solely driven by positive selection in the coding domains. Three main arguments can be made against this. First, the entire coding regions of these genes have not been under uniform selective forces during primate evolution. Rather, only the B30.2 domains of these genes have been under strong positive selection [5,45], and in the case of TRIM22 this was restricted to Catarrhini species. Moreover, the more than 8 kb that separates the first intron of TRIM5 and TRIM22 genes from the B30.2-domain encodes the tripartite motif, which has been under purifying selective pressure during primate evolution. Second, the differential fixation of LTR elements is restricted to TRIM22 and TRIM5 from Haplorhini species and the majority of the LTRs fixed in these species predominantly reside in the promoter-proximal first intron. If the fixation of these elements was driven purely through linkage with positively selected polymorphisms in the B30.2 domain encoding exons one would expect them to be more evenly distributed throughout the genes. Third, the capture of ERV elements is unique to primates; other mammalian species whose TRIM5 orthologs have been under positive selection have not diversified in this way (data not shown). Thus we believe it is unlikely that the insertion of LTR elements was solely the result of linkage disequilibrium with the positively selected B30.2 domain.
Intriguing correlative evidence points to the evolutionary forces that have resulted in the fixation of LTR elements as being unique to Haplorhini. In these species, genes of the TRIM6/34/5/22 locus lie adjacent to one another in the following order: TRIM6, TRIM34, TRIM5, TRIM22. Three of these genes, TRIM6, TRIM34, and TRIM22, are oriented such that transcription proceeds in the same direction, while transcription of TRIM5 occurs in the opposite direction ( Figure 1A). As a result, in Haplorhini primates TRIM5 and TRIM22 are situated in such a manner that only ,5 kb separates the transcriptional start sites of these genes. While extensive promoter mapping studies for these genes have not been conducted, whole genome mapping has localized H3K27Ac, DNase hypersensitivity, and transcription factors to the sequences surrounding the first exons of both TRIM5 and TRIM22, localizing the core promoter domains to these regions [78,79]. In Haplorhini primates, both TRIM5 and TRIM22 possess intracellular antiviral functions and over the course of primate evolution likely faced multiple rounds of challenge from viruses with divergent tissue tropisms, that required flexibility in matching TRIM gene expression to the target tissue of the challenge virus. The relatively short intergenic region providing core promoter function for both of these genes would serve to limit the evolutionary flexibility of the core promoter, but fixation of LTR elements proximally to this region in TRIM5 and TRIM22 could have served as a novel adaptive evolutionary mechanism to modulate transcriptional requirements placed upon these apposed genes.
The capture of LTR sequences into the TRIM locus is not a significant feature of non-primate mammalian species [Diehl, W.E. unpublished data] and is even very limited in prosimians. However, in these non-primate species, there is no evidence for TRIM22 possessing antiviral properties. Indeed, independent deletion events in multiple non-primate species, including all those known to have undergone TRIM5 gene duplication (mouse, rat, cow, and grey mouse lemur), have resulted in the loss of TRIM22 [4,6,26]. Similarly, recent analyses of the porcine genome show that it encodes only TRIM6, TRIM34, and a single copy of TRIM5 [Diehl, W.E. unpublished data]. Thus a functional role favoring maintenance and adaptive evolution of TRIM22 may be restricted to Haplorhini species. Combined, these observations suggest that a gain of antiviral function for TRIM22 in primates in combination with the restricted flexibility of the genomic organization of this locus favored serial fixation of ERV LTRs in the promoter proximal regions of these neighboring genes.
There is a growing body of literature demonstrating the profound influences, both at the cellular and organismal level, mediated by alterations in non-coding DNA content, in general, and ERV LTRs, in particular [80,81,82,83,84,85,86,87,88,89,90]. In humans, well known examples of such regulation include: the human salivary amylase gene whose expression in the salivary glands is mediated by an HERV-E 59 LTR, which acts as an enhancer element for a cellular promoter [88] and the human fetal gamma globin gene where an ERV9 LTR element ,40 kb upstream acts to stimulate expression during fetal development and suppress expression after birth [87]. A recent report has shown that a similar ERV LTR capture event has occurred in an antiviral gene in certain lineages of mice, including C57BL. In these strains, a xenotropic murine leukemia virus insertion into the second intron of apolipoprotein B mRNA-editing, enzymecatalytic, polypeptide-like 3 (APOBEC3), resulting in a greater than 4-fold increase in expression in the spleens of mice, and in conjunction with other changes, resulted in an enhanced antiviral phenotype in these mice [58]. Such findings clearly demonstrate that ERV elements and other non-coding DNA, frequently described as ''junk DNA'', may indeed contribute to evolution of biological function(s). The recent report of trans-species balancing selection on the LTR12D and PABL_A elements in the first intron of TRIM5 in both humans and chimpanzees [70] indicating an evolutionary benefit from both the ERV elements as well as the conserved polymorphisms within these elements. This observation thus provides additional evidence for a positive role of LTRs in modulating function within the TRIM locus.
In exploring the cross-species differences in transcriptional control the ERV elements provide to the neighboring genes we focused on TRIM22. This gene provides a more tractable model due to its limited transposable element diversity as well as the existence of studies examining its transcriptional regulation. These previous studies demonstrated that a p53-binding site, found in the LTR10D element located in the first intron, is necessary and sufficient for providing p53 responsiveness to this gene in human cells [57,65,66]. Because this LTR10D element is present in Catarrhini TRIM22 but is absent from that of Platyrrhini, we tested the hypothesis that the presence or absence of this element might alter p53 responsiveness at a species level. Indeed, we found that p53-mediated transcriptional control of TRIM22 expression in response to DNA damage in PBMCs was restricted to those species harboring the LTR10D element. Thus, it does not appear that New World monkeys have acquired p53 responsiveness for TRIM22 via some other mechanism. In contrast TRIM5 expression was unaffected by p53 upregulation in all species tested, and while these TRIM5 genes contain multiple different ERV insertions, all lack an LTR10D element. Together, these findings clearly demonstrate a lineage-specific, differential regulation of transcription resulting from fixation of an ERV LTR element fixed within an antiviral TRIM gene. By extension, we would argue that all of the LTR elements fixed in TRIM5 and TRIM22 have the potential to alter transcriptional regulation of these genes in response to specific stimuli, e.g. DNA damage, or in a tissue-or lineage-specific manner as demonstrated for murine APOBEC3 [58]. The regulatory control exerted by intronic LTR elements present in TRIM5 and TRIM22 provide the opportunity for a complex interaction between the core promoter and various LTR elements present in a specific gene. The potential for their involvement in transcriptional regulation represents a previously unrecognized consequence of the evolutionary pressures placed on these TRIM genes. A better understanding of the genetic basis for transcriptional control of these genes will assist in defining how TRIM5 and TRIM22 exert their antiviral effects in vivo.

Samples from Human and Non-human Primates
Human blood samples were obtained from volunteers enrolled in the Emory University Institutional Review Board approved ''Emory Vaccine Center's Healthy Adults Study'' (Protocol #555-2000). Volunteers in this study were healthy adults, aged from 23 to 62, who had signed a written informed consent form.
Rhesus macaque and squirrel monkey blood samples were obtained in accordance with NIH guidelines from animals housed at the Yerkes National Primate Research Center. These animals were maintained in accordance with standards of the Public Health Service Policy on Humane Care and Use of Laboratory Animals and the Emory University Institutional Animal Care and Use Committee. Emory University's Animal Welfare Assurance Number is A3180-01. In addition the Yerkes National Primate Research Center is fully accredited by the International Association for Assessment and Accreditation of Laboratory Animal Care. All Yerkes nonhuman primates are fed a primate specific diet (Purina) supplemented daily with fresh fruit. The animals are monitored daily for clinical well-being and psychosocial enrichment. The collection of blood for this study was reviewed and approved by the Emory University Institutional Animal Care and Use Committee (protocol #028-2009Y) and the collection procedures conformed to the principles described in the Guide for the Care and Use of Laboratory Animals. All macaques and squirrel monkeys included in this study were unrelated adult males with no obvious chronic illnesses. All of the rhesus macaques included in the study were of Indian origin and SIV negative.
Grey mouse lemur liver necropsy samples were harvested at the Duke Lemur Center in accordance with NIH guidelines from animals who died of natural causes. These samples were provided to us via an MTA agreement with the Duke Lemur Center.

Gene Identification
Genes of the TRIM5/6/22/34 cluster from human, chimpanzee, and rhesus macaque were retrieved from the respective genomic sequence databases at the National Center for Biotechnology Information (NCBI). Genome assemblies used for these analyses are as follows: Homo sapiens, NCBI Build 36. Bacterial artificial chromosome (BAC)-derived sequences of the TRIM5 locus were also obtained from white-cheeked gibbon (Nomascus leucogenys), olive baboon (Papio anubis), guereza colobus (Colobus guereza), Peruvian red-necked owl monkey (Aotus nancymaae), common marmoset (Callithrix jacchus), Bolivian squirrel monkey (Saimiri boliviensis boliviensis), dusky titi (Callicebus moloch), and grey mouse lemur (Microcebus murinus). The TRIM gene sequences from these primate species were identified by using NCBI's trace Basic Local Alignment Search Tool (BLAST), with human TRIM genes as query sequences while searching all available primate sequencing databases. In all cases, sequence data obtained in this manner was generated by the NIH Intramural Sequencing Center's (www. nisc.nih.gov) Comparative Vertebrate Sequencing Initiative. When available, sequence information from multiple overlapping BAC clones were compiled in order to provide the most complete coverage for all four TRIM genes. Genbank accession numbers for the BAC clones used in this study are as follows: AC198823, AC193710, and AC191889 (gibbon); AC147862 and AC148062 (baboon); AC174629 (colobus); AC183999 and AC174391 (owl); AC148555 and AC148636 (marmoset); AC192681 (squirrel); AC173941 and AC172720 (titi); AC197314, AC172705, and AC183331 (grey mouse lemur).
The genomic sequences used in this analysis include a previously described Owl monkey TRIM5-cyclophilin A (TRIM-Cyp) fusion gene that arose due to the retrotransposition of a processed cyclophilin A mRNA into the seventh intron of TRIM5 [36,75]. A similar, independent cyclophilin A retrotransposition event immediately 39 of TRIM5 in macaques has resulted in a TRIM-cyp allele in rhesus macaques [72,73,74,76,77]. However, as this allele is not represented in the macaque genome sequence, it was not included in the analysis.

Sequence Analysis
Transcribed sequences in the TRIM6/34/5/22 cluster corresponding to the human alpha splice variant of TRIM5, isoform 2 of TRIM6, isoform 4 of TRIM34, and isoform 1 of TRIM22 were compiled for further examination. With the exception of TRIM6, these sequences correspond to the longest of the described transcripts. The Genbank mRNA accession numbers corresponding to these transcripts are NM_033034. RepeatMasker options used were 'slow' speed/sensitivity and selecting 'other mammal' as DNA source, except in the case of human sequences where 'human' was selected as the DNA source. Grey mouse lemur sequences were further analyzed using the CENSOR algorithm [91] to identify lemur-specific repetitive elements.

Quantifying the Rate of Indel Turnover
In order to quantify indel differences, the multiple sequence alignments for each TRIM gene were broken up into 45 pairwise alignments and extraneous gaps were removed using the Gapstreeze program (www.hiv.lanl.gov/content/sequence/ GAPSTREEZE/gap.html). Unsequenced regions in the source sequence were dealt in the most conservative manner possible: by assuming that the missing region exactly matched that of the second species of the comparison.

Calculating Genetic Diversity
In order to assess the amount and type of genetic diversity observed in the various TRIM genes, we utilized the following calculations. The percent nucleotide identity was calculated using the following formula: 2 Ã # of conserved nucleotides total # of nucleotides taxa 1ztaxa 2 ½ Ã 100 The nucleotide substitution rate was calculated using the following formula:

Isolation and Culture of Blood Cells
Blood samples were collected in sodium heparin Vacutainer tubes (BD Biosciences; San Jose, CA). PBMCs were isolated following centrifugation of whole blood over a Ficoll cushion. For all experiments, freshly isolated PBMCs were plated at approximately 2610 6 cells per well in 6-well plates and cultured for 3 days in RPMI medium containing 15% fetal bovine serum, 10 U/ ml penicillin, 10 mg/ml streptomycin, and 3 mg/ml phytohemagglutinin (PHA). Following this stimulation, induction of p53 was accomplished by addition of 0.6 mg/ml doxorubicin to the culture medium. Culture of PBMCs in the presence of 0.1% DMSO served as a carrier control.

Transcriptional Profiling
At various times following induction of p53, cells were harvested and RNA was isolated using TRIzol (Invitrogen; Carlsbad, CA) according to the manufacturer's instructions. A one-step, SYBR green-based quantitative RT-PCR was then performed on 50 ng of input RNA. Primer sets targeting TRIM22, TRIM5, MDM2, and b-actin were designed and validated to be gene specific, while also targeting regions conserved between all primate species included in this study. The sequences of the primers are as follows: 'TRIM22 qPCR F' All qPCR reactions were performed with 125 nM forward primer and 250 nM reverse primer concentrations with the following conditions: 55uC for 10 minutes (RT step) and 95uC for 5 minutes followed by 45 cycles of 95uC for 30 seconds, genespecific annealing temperature for 30 seconds, 72uC for 30 seconds followed by melting curve analysis. The annealing temperatures were as follows: 64uC for TRIM22, 56.5uC for TRIM5, 61uC for MDM2, 57uC for b-actin. Relative fold change in gene transcription was calculated via the DDC(t) method using DMSO treatment values as the baseline.
For the genotyping of the intronic LTR10D, unstimulated human and rhesus PBMCs were used as a source of genomic DNA, which was isolated using the DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer's protocols. This was used as source material for PCR amplification of the LTR element using 'TRIM22 LTR10D F' (59-GACCATT-CATTTCTTCAATCTAGGTAC-39) and 'TRIM22 LTR10D R' primer (59-ATCAAAATGACAGAATAGGAATGTGGG-39). PCR was performed with the Phusion enzyme with the following thermocycler conditions: 95uC for 2 minutes followed by 35 cycles of 95uC for 15 seconds, 56uC for 15 seconds, and 72uC for 1 minute 30 seconds and a final 7-minute extension at 72uC. Direct sequencing of the PCR product was performed using the PCR primers.
Grey mouse lemur DNA was extracted from frozen liver samples using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI), according to the manufacturer's protocol. This was used as source material for PCR verification of the TRIM6/34/5 genomic architecture observed in BAC clone sequences found in GenBank. PCR primers were designed to specifically bind to the extreme 59 and 39 regions of each gene present in the BAC clones, such that PCR amplifies the intergenic region separating the genes. Presence of PCR product as well as definitive sizing via agarose gel electrophoresis were used as positive indicators of genomic architecture found in GenBank BAC clones. More detailed primer, strategy, and protocol information for these PCR reactions is available upon request.

Sequence Information
Four unique rhesus macaque LTR10D alleles were discovered ( Figure S3 and Table S1) and have been deposited to GenBank with the following accession numbers: allele 1, HM104186; allele 2, HM104187; allele 3, HM104188; allele 4, HM104189. These sequences differ from one another at several nucleotide positions. However, none of these polymorphic sites are located within, or immediately adjacent to, the p53 binding site.
Human and rhesus macaque p53 nucleotide sequences were in silico translated and aligned ( Figure S4). Amino acid sequences generated in this study match previously described human p53 72P and 72R alleles and the rhesus macaque protein sequence (GenBank accession numbers NP_000537, AAD28628, and NP_001040616, respectively). While only one predicted amino acid sequence was observed, three rhesus macaque p53 alleles were identified. The allele 1 sequence corresponds to that of the rhesus genome sequence (GenBank accession #NM_001047151), and the other two alleles only differ by synonymous polymorphisms. The sequences corresponding to alleles 2 and 3 have been deposited to GenBank with the accession numbers HM104190 and HM104191, respectively. Figure S1 Reduced rate of indel change in TRIM5 from Old World monkeys. Nucleotide sequence alignments were used to calculate the rate of indel change between all pairs of species in TRIM6 (A), TRIM34 (B), TRIM22 (C), and TRIM5 (D). These analyses have been broken down for comparisons within and between the following evolutionarily distinct groups of primates: hominids, cercopithecidae (Old World monkeys), and platyrrhini (New World monkeys). Dots indicate separate pairwise sequence comparisons and the black bars represent mean values. (EPS) Figure S2 No correlation between percent nucleotide identity and the nucleotide substitution rate. The correlation between percent nucleotide identity and the nucleotide substitution rate was examined using linear regression analysis for TRIM5 (A),  Figure S4 Nucleic acid alignment of the LTR10D element found within the first intron of Old World primate TRIM22. Nucleotide sequences comprising and flanking the Old Worldspecific LTR10D element were trimmed from the greater TRIM22 nucleotide alignment. The sequence corresponding to the LTR10D element is numbered from 1 to 516; sequences upstream and downstream are numbered in reference to this element using -and + symbols, respectively. The red highlighted nucleotides correspond to the target site duplication that occurred during ERV integration. The nucleotides highlighted in green correspond to the p53-binding region. Nucleotides highlighted in blue correspond to the nucleotides found to be polymorphic in rhesus macaques.

(EPS)
Table S1 Allelic nucleotide polymorphisms found in the LTR10D elements within rhesus macaque TRIM22. a Nucleotide positions correspond to those shown in Figure S4.