Adaptive Evolution Coupled with Retrotransposon Exaptation Allowed for the Generation of a Human-Protein-Specific Coding Gene That Promotes Cancer Cell Proliferation and Metastasis in Both Haematological Malignancies and Solid Tumours: The Extraordinary Case of MYEOV Gene
The incidence of cancer in human is high as compared to chimpanzee. However previous analysis has documented that numerous human cancer-related genes are highly conserved in chimpanzee. Till date whether human genome includes species-specific cancer-related genes that could potentially contribute to a higher cancer susceptibility remains obscure. This study focuses on MYEOV, an oncogene encoding for two protein isoforms, reported as causally involved in promoting cancer cell proliferation and metastasis in both haematological malignancies and solid tumours. First we document, via stringent in silico analysis, that MYEOV arose de novo in Catarrhini. We show that MYEOV short-isoform start codon was evolutionarily acquired after Catarrhini/Platyrrhini divergence. Throughout the course of Catarrhini evolution MYEOV acquired a gradually elongated translatable open reading frame (ORF), a gradually shortened translation-regulatory upstream ORF, and alternatively spliced mRNA variants. A point mutation introduced in human allowed for the acquisition of MYEOV long-isoform start codon. Second, we demonstrate the precious impact of exonized transposable elements on the creation of MYEOV gene structure. Third, we highlight that the initial part of MYEOV long-isoform coding DNA sequence was under positive selection pressure during Catarrhini evolution. MYEOV represents a Primate Orphan Gene that acquired, via ORF expansion, a human-protein-specific coding potential.
Spyros I. Papamichos dedicates this writing to Thomas I. Karkalis.
It has formerly been suggested that at least some of human (Homo sapiens) major diseases could partially likely be related to genetic maladaptations during the recent evolutionary past . The incidence of cancer in human is rather high as compared to chimpanzee (Pan troglodytes) . However, previous analysis  has shown that numerous human cancer-related genes are highly conserved in chimpanzee, containing intact open reading frames (ORFs). Only minor differences were reported between species . Therefore whether human genome includes species-specific cancer-related genes or gene isoforms that could potentially contribute to a higher cancer susceptibility remains yet obscure.
Of note, MYEOV (also known as OCIM, National Center for Biotechnology Information [NCBI] Gene ID: 26579) was not included in the human cancer gene data set used in the analysis by Puente et al. . MYEOV is a noncensus cancer gene that during the last 15 years has been reported as causally involved in promoting cancer cell proliferation and metastasis in both haematological malignancies and solid tumours [4–11].
The gene has the potential to generate via alternative splicing six mRNA variants encoding for two protein isoforms [6, 9, 12], namely, for a 313-amino-acid (aa) peptide (MYEOV-313) as well as for a shorter 255-aa peptide (MYEOV-255). Solid Western-blot assays support the production of both the proteins [9, 12] while MYEOV-313 expression has been associated with poor prognosis in patients with multiple myeloma . Both MYEOV-313 and MYEOV-255 seem to be directed to the membrane  but are of, yet, unknown biological function.
MYEOV has been shown to be epigenetically regulated via a DNA-methylation mechanism  while at the protein synthesis level the gene seems to be tightly controlled by upstream ORFs (uORFs) . Importantly, MYEOV second pseudosignal generates a long, weak Kozak signal, uORF that could significantly impair translational efficiency [13–15].
In 2012, MYEOV was characterized as a “Class I” hominoid-specific de novo protein-coding gene , id est, deduced to encode for protein exclusively in human. Xie et al.  identified an ORF disrupting mutation that was present in MYEOV syntenic locus in multiple higher primates, ranging from chimpanzee to rhesus monkey (Macaca mulatta). Intriguingly, the “shared disabler”  located in MYEOV-313 start codon . Accordingly, the authors  excluded MYEOV origination via other molecular mechanisms known to generate novel genes  and annotated MYEOV as a human-specific de novo protein-coding gene, eventually concluding that it is likely de novo origination rather than ORF expansion that drove the origination of hominoid-specific genes , such as MYEOV.
Transposable elements (TEs) are known to exert a broad evolutionary impact, promoting genomic plasticity and eventually biological diversity [19, 20]. Exonized TEs have the potential to induce significant changes in functional noncoding regions of genes while, in extraordinary cases, some could even acquire a novel role as protein-coding modules [21–24], in a biological process called “exaptation” . Interestingly, it was relatively recently documented that TEs could preciously contribute in de novo gene-origination [26–29].
In this paper, we argue about MYEOV representing a human-specific de novo protein-coding gene. We show, via in silico analysis, that MYEOV arose de novo in Catarrhini and that it was adaptive evolution coupled with TE exaptation that eventually allowed for the generation of a human-specific primary ORF.
2. Materials and Methods
2.1. Inquiring into MYEOV Origin and Validating MYEOV Orthologous Sequences
BLASTN  search against the GenBank human genomic database and BLASTP  search against the GenBank nonredundant protein database were used to exclude MYEOV origination via gene duplication and to verify unambiguous MYEOV orthologs; the stringent filters described previously were applied [16, 35].
MYEOV orthologs annotated in the respective databases NCBI, Ensembl, and OrthoDB  were identified by searching with the gene name. Whole-genome shotgun (WGS) sequence contigs, including within the DNA segments used to curate the reference genomic sequences of MYEOV orthologs, were downloaded from the NCBI Gene database (http://www.ncbi.nlm.nih.gov/gene/). The WGS sequence contigs were used to build MYEOV locus alignment blocks between human and the species with annotated orthologs, via Clustal Omega . Each alignment block was manually scrutinized for validating the primary ORF (pORF) of the annotated orthologs.
To estimate the degree of evolutionary constraint on MYEOV coding DNA sequence during evolution in the Catarrhini genome, the average human-rhesus ratio was calculated ; MEGA6 software was used in the analysis . Ratios for MYEOV second and third exon coding DNA sequences were calculated separately.
Provided in the study by Finstermeier et al.  is a beautiful phylogenetic tree that clarifies the nomenclature and supports the mammalian classification and the estimated divergence ages used in the text.
2.2. Scanning MYEOV and MYEOV Evolutionary Antecedents for the Presence of Integrated TEs
MYEOV and MYEOV orthologous genomic sequences were scanned for the presence of TEs by RepeatMasker . The program was run in both default and sensitive mode, using a matrix optimal for MYEOV GC level to avoid false masking. Respective searches using PLOTREP , the TranspoGene database , and BLASTN [34, 44] were also performed.
MYEOV locus automated alignments downloaded from the UCSC Genome Browser Database [30–33] were used to extract syntenic DNA segments in numerous mammals that flank, 500 nucleotides (nts) in the 5′ and the 3′, MYEOV-313 start codon. BLASTN was used to align these DNA segments with corresponding WGS sequence contigs for extracting the original DNA sequence in each species. That is because the “Multiz Alignments of 100 Vertebrates” track of the UCSC Genome Browser Database excludes multiple inserts located between the alignment blocks of the synteny. Each WGS sequence contig was then realigned to human MYEOV. The DNA segments eventually extracted were scanned by RepeatMasker  for the presence of integrated TEs.
2.3. MYEOV Upstream ORF and Splice Site Computational Analysis
2.4. Searching for mRNA Expression Evidence in Nonhuman Primates
BLASTN searches with human MYEOV mRNA sequences against the GenBank expressed sequence tags (EST) database were used to identify matching expressed mRNA sequences in nonhuman primates.
2.5. MYEOV Common Single Nucleotide Polymorphisms and Confirmed Somatic Mutations Analysis
MYEOV confirmed somatic mutations annotated in the catalogue of somatic mutations in cancer (COSMIC)  were identified by searching with the gene name.
3. Results and Discussion
3.1. From a Eutherian-Mammal Noncoding Sequence to a Human-Protein-Specific Coding Gene
MYEOV BLASTN search against the GenBank human genomic database yields no significant similarity with any coding genomic sequences other than itself.
Syntenic alignments, extracted from the “Multiz Alignments of 100 Vertebrates” track of the UCSC Genome Browser Database [30–33], signify that the DNA segment where MYEOV locates emerged in eutherian mammals. The DNA segment is absent in all outgroups ranging from marsupials to lamprey. Of note a number of species, for example, Myomorpha, Sciuromorpha, and Lagomorpha, lack a large portion of the syntenic region (Supplementary Figure in Supplementary Material available online at http://dx.doi.org/10.1155/2015/984706). This finding could explicate the inability to detect a mouse myeov transcript via zoo blot analyses under low-stringency hybridization conditions, as reported previously .
MYEOV is reported in the Gene database from NCBI (http://www.ncbi.nlm.nih.gov/gene/) to have sixteen orthologous genes, exclusively present in higher primates (Anthropoidea). Fifteen protein-coding genes as well as a pseudo-ortholog in rhesus monkey are annotated.
MYEOV-313 BLASTP search against the GenBank nonredundant protein database  yields however an intriguing finding. The peptide predicted to be encoded from Bolivian squirrel monkey (Saimiri boliviensis) MYEOV ortholog presents higher amino acid coverage with MYEOV-313 protein compared to the respective peptide encoded from chimpanzee MYEOV ortholog, despite Bolivian squirrel monkey being a primate far more distantly related to human than chimpanzee is.
Of note, OrthoDB  reports yet another MYEOV protein-coding ortholog in dolphin (Tursiops truncatus), raising the issue of the locus functionality in ancestral genomes to be subsequently lost in multiple lineages and later regained in higher primates [17, 49, 50].
The plausible interpretation of these intriguing findings is that some of the automated annotations of MYEOV orthologous ORFs provided by the NCBI Eukaryotic Genome Annotation Pipeline and the Ensembl Genebuild  contain inaccuracies . Elucidated below is the evolutionary path leading to MYEOV.
Interestingly, one of the first functional genetic modules that emerged during MYEOV evolution in the mammalian genome was the long uORF, reported previously to participate in the regulation of the gene’s translational efficiency  (Figure 1). The upstream ATG trinucleotide demarcating this long uORF seems fixed in Catarrhini but in depth phylogenetic analysis indicates that it could be of ancestral status (Figure 2). The trinucleotide appears transmuted in Platyrrhini probably due to either emergence of the ACG trinucleotide in a common Platyrrhini ancestor or parallel mutations in these species. With regard to the TGA trinucleotide originally delimiting the long uORF, it appears evolutionarily fixed in higher primates. Eventually the presence of these two trinucleotides in the genomic sequence allowed for the occurrence in Catarrhini of a primal 219nts long uORF (Figure 2).
Circa (ca.) 46 Ma (Catarrhini/Platyrrhini divergence time) , an important mutational event occurred in Catarrhini resulting in the de novo gain of a potentially functional ORF in a genomic region that is noncoding in other primates; that is the acquisition of MYEOV-255 start codon (Figure 2).
Ca. 15.2 Ma (Homininae/Ponginae divergence time)  another precious point mutation allowed in Hominines for the shortening of MYEOV long uORF (Figure 2), an event rather meaningful in the context of the gene’s translational regulation . Of note, an identical TGA trinucleotide occurring in Bolivian squirrel monkey syntenic region is likely due to a parallel mutation (Figure 2).
Finally, ca. 5.9 Ma (Homo/Pan separation time)  a momentous point mutation was introduced in human resulting in the acquisition of MYEOV-313 start codon (Figure 2). An identical trinucleotide appearing in white-tufted-ear marmoset (Callithrix jacchus) syntenic region is due to a parallel mutation and is likely nonfunctional.
Notably, the stop codon delimiting the relatively short pORFs in Old World monkeys (Cercopithecoidea) represents not an ancestral frame disrupting feature that the corresponding hominoid orthologs escaped from since this trinucleotide occurs exclusively in Old World monkeys (Supplementary Data Set 1). In addition it is off-frame with regard to MYEOV coding sequence (Figure 2, Supplementary Data Set 1). That is due to a 14-bp deletion, also fixed exclusively in Old World monkeys, locating short upstream from the stop codon (Figure 2). In the same sense, the stop codon demarcating the pORF in gibbon (Nomascus leucogenys) occurs exclusively in gibbon and orangutan (Pongo abelii). It would feel somewhat more plausible to speculate that parallel mutational events took place in both species rather than assuming the occurrence of a point mutation in a common hominoid ancestor, suffering refutation and evolutionary “refixation” to the antecedent state in succeeding lineages (Figure 2). Of note an orangutan-specific insertion, requiring decidedly additional validation due to shifting very early the respective orangutan MYEOV pORF, translocates out-of-frame this TGA trinucleotide (Figure 2, Supplementary Data Set 1).
All the above could signify that MYEOV third exon coding DNA sequence (Figure 2) was not under strong selective constraints during evolution in the Catarrhini genome [38, 52]. Reinforcing the above, this sequence yields an average human-rhesus ratio of 1.14. Accordingly, it was likely neutral drift-directed evolution  that drove the 3′ expansion of MYEOV pORF (Figure 2).
On the other hand, MYEOV second exon coding DNA sequence yields an average human-rhesus ratio of 3.12. Thus the 5′ expansion of MYEOV pORF during the gene’s evolution in the Catarrhini genome, including the acquisition of the human-specific start codon, was likely driven under positive selection pressure  (Figure 2).
3.2. Precious Exapted TEs Present in MYEOV Genomic Sequence
MYEOV scan by RepeatMasker reveals the presence of seven TEs in the genomic sequence (Supplementary Data Set 2). Importantly, nonredundant MYEOV modules seem to localize within three of these genetic elements.
In detail, the constitutively used donor of MYEOV first splicing junction locates in a region reported from RepeatMasker to match an antisense orientation L2 repeat of the long interspersed nuclear element (LINE) class (Figure 1). Of note, this TE provided also the stop codon delimiting MYEOV primal 219nts long uORF, discussed above (Figure 2).
The stop codon that is common to both MYEOV-313 and MYEOV-255 protein isoforms locates in a region reported from RepeatMasker to match an antisense orientation mammalian-wide interspersed repeat (MIR) of the short interspersed nuclear element (SINE) class (Figure 1).
Most importantly the three alternative acceptors of MYEOV first splicing junction and the initial region of MYEOV-313 coding sequence, including the start codon, are located in a region reported to match a sense orientation LINE L2a repeat (Figure 1). This finding represents a rare exaptation event [22, 55] but is somewhat equivocal in the context of the RepeatMasker match appearing highly degenerated and yielding a score relatively close to the Smith-Waterman cutoffs; that is 180 for the ancient MIR, L2, and MER5 sequences . At this point it should be noted however that the accurate annotation of a L2 family-repeat content and the valid estimation of its original extent are issues very hard to meet since it is not unusual for these ancient, inactive elements to be degenerated beyond recognition [56, 57].
A subsequent MYEOV scan using PLOTREP  further reinforces that a segment of MYEOV-313 initial coding sequence was likely provided by a TE of the L2 family (Supplementary Data Set 3). The relatively short sequence extracted by PLOTREP, while corresponding to a minor segment of the RepeatMasker match, matches identically to the respective data extracted from the TranspoGene database and to MYEOV DNA segment masked by BLASTN, when a DUST-driven  filter for human-specific repeats is applied. Of note, RepeatMasker’s results not corresponding well with the respective data extracted from PLOTREP , the TranspoGene database , and BLASTN represent not an atypical finding; this inconsistency is due to other programs using much smaller databases than RepeatMasker. RepeatMasker’s mammalian libraries represent heavily manipulated and expanded versions of the respective Repbase libraries .
The above data led us to subsequently inquire into the status of the L2a repeat in MYEOV evolutionary antecedents. Results validated the presence of L2 family-repeat-relics in the corresponding syntenic region in, at least, seven eutherian mammals (Figure 3, Supplementary Data Set 4). Importantly, it is well known that L2s and MIRs underwent active retroposition prior to the placental mammalian radiation [22, 57]. It is also known that cases of TEs inserted independently in nearby syntenic genomic regions in species identical-by-descent represent extremely rare events with exceedingly low probability to occur . In line with these data, it is highly likely that RepeatMasker not only defining divergently the repeat’s size and exact boundaries in each species but also providing different subfamily names for the repeat is due to the high degree of degeneration of the repeat during its evolution in the mammalian genome. Overall, it is particularly plausible that the molecular L2 repeat-fossils, present in the nearby syntenic regions of the seven mammals (Figure 3), were derived from the same L2 transposable element that was introduced in the common eutherian ancestor.
Another line of evidence to support the above, likely the most conclusive one, would be the probabilistic reconstruction of MYEOV locus ancestral sequence . Indeed, it has been suggested that using RepeatMasker to scan the inferred boreoeutherian ancestor sequence of a given genomic locus would provide more accurate information on the original extent of the ancient TEs included within than running the software in the corresponding human sequence , because the ancient repeats would appear much less degenerated in the ancestral sequence. Of note, a series of prerequisites should be met for accurately reconstructing an ancestral sequence . Most importantly, problematic sampling of major lineages and outgroups seems to significantly decrease the accuracy of ancestral sequence reconstruction [61, 62]. Consequently, Myomorpha, Sciuromorpha, and Lagomorpha lacking a large segment of MYEOV syntenic region likely represent a severe drawback in optimally performing the sampling procedure.
3.3. MYEOV Full Splicing Potential Was Acquired Relatively Late during the Gene’s Evolution in the Catarrhini Genome
As shown in Figure 4, only one MYEOV canonical splicing signal, id est, allowing for RNA processing by the standard U2 type spliceosome , predated the radiation of Haplorhini. Three canonical splicing signals arose in higher primates while two precious standard acceptors appeared later in time.
3.4. Transcription of MYEOV Locus in Human and Nonhuman Primates
In human, transcription of MYEOV genomic locus is driven by a strongly active cryptic promoter sequence locating in MYEOV 5′ untranslated region (5′ UTR) , between the second and third pseudosignal of the gene (Figure 1). Should this cryptic promoter precede or follow the emergence of the translatable ORF in Catarrhini and be active in other higher primates as well remains to be clarified via subsequent studies, especially because the expressed sequence tags- (EST-) coverage of the genomic sequence in nonhuman primates is low.
Indeed, BLASTN searches with human MYEOV mRNA sequences against the GenBank EST database yield only one, however precious, unambiguous hit (NCBI accession: DC527428). This EST validates transcription of the genomic locus in chimpanzee.
3.5. MYEOV-313 Start Codon Includes Not Common Polymorphic Sites in Human
According to the single nucleotide polymorphism database from NCBI (http://www.ncbi.nlm.nih.gov/snp/), located in human MYEOV gene are 17 common polymorphic sites. None locates in MYEOV-313 start codon.
In the same context, out of the 7 common polymorphic sites located in MYEOV long-ORF, all representing single nucleotide substitutions, none induces a nonsense codon. Out of the 6 common polymorphisms included in either MYEOV 5′ UTR or MYEOV intragenic regions, none locates within core splicing signals.
In this context, it is highly plausible that all individuals could be able to encode for the 313-aa peptide.
3.6. MYEOV Peptide(s) May Be Involved Per Se in Promoting the Malignant State
In line with data extracted from COSMIC , only 60 MYEOV confirmed somatic mutations were identified in 72 out of 23780 unique samples tested.
The above could signify that the peptide(s) produced by MYEOV may be involved per se, rather than mutated, in promoting the malignant state.
(i) MYEOV is a Primate Orphan Gene that acquired, via ORF expansion, a human-protein-specific coding potential. Thus it represents the first cancer-related gene identified to present significantly differing protein-coding potential between human and chimpanzee.
(ii) Should MYEOV role in promoting cancer cell proliferation and metastasis be attributed to MYEOV-313 is a tantalizing question that warrants further investigation, especially in the context that all individuals could be able to encode for this peptide and also because MYEOV-313 may be related per se in promoting tumour propagation and thus could be specifically targeted for. After all, targeting a human-specific peptide would theoretically cause less adverse effects than targeting components of evolutionary conserved signaling cascades.
(iii) It was adaptive evolution coupled with TE exaptation that allowed for the expansion of MYEOV pORF. It is tempting to speculate that it could have all started with the relatively strong core splicing signals provided by the LINE L2a repeat (Figure 4), creating an evolutionary “hotspot” within MYEOV sequence that was subsequently positively selected for throughout the course of Catarrhini evolution .
The sequence data from this study are archived at the Gene database from NCBI (http://www.ncbi.nlm.nih.gov/gene/), the Ensembl Genome Browser Database (http://www.ensembl.org/index.html), and the UCSC Genome Browser Database (http://genome.ucsc.edu/).
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Spyros I. Papamichos gratefully acknowledges Dr. Vassiliki Kotoula for providing fruitful mentoring.
Supplementary Material includes one Figure and four Data Sets. Depicted in Supplementary Figure 1 is that the DNA segment where MYEOV locates emerged in Eutherians. Supplementary Data Set 1 includes sequence data from selected segments in MYEOV syntenic region in numerous Vertebrates, extracted from the "Multiz Alignments of 100 Vertebrates" track of the UCSC Genome Browser Database (http://genome.ucsc.edu/), reinforcing the results reported in the text. Supplementary Data Set 2 includes the results of MYEOV genomic sequence scan by Repeat Masker; run in both default and sensitive mode. Supplementary Data Set 3 contains the short MYEOV coding sequence extracted identically from PLOTREP, TranspoGene and BLASTN as corresponding to a L2 repeat segment. Supplementary Data Set 4 includes RepeatMasker results verifying the presence of L2 family repeat-relics in MYEOV syntenic region in six eutherian mammals
J. W. G. Janssen, J.-W. Vaandrager, T. Heuser et al., “Concurrent activation of a novel putative transforming gene, myeov, and cyclin D1 in a subset of multiple myeloma cell lines with t(11;14)(q13;q32),” Blood, vol. 95, no. 8, pp. 2691–2698, 2000.View at: Google Scholar
S. E. Calvo, D. J. Pagliarini, and V. K. Mootha, “Upstream open reading frames cause widespread reduction of protein expression and are polymorphic among humans,” Proceedings of the National Academy of Sciences of the United States of America, vol. 106, no. 18, pp. 7507–7512, 2009.View at: Publisher Site | Google Scholar
Y. Xing and C. Lee, “Evidence of functional selection pressure for alternative splicing events that accelerate evolution of protein subsequences,” Proceedings of the National Academy of Sciences of the United States of America, vol. 102, no. 38, pp. 13526–13531, 2005.View at: Publisher Site | Google Scholar