Sigh, fine. Want to get REALLY specific:
Hominoid-Specific De Novo Protein-Coding Genes Originating from Long Non-Coding RNAs
Here we identified 24 hominoid-specific
de novo protein-coding genes with precise origination timing in vertebrate phylogeny. Strand-specific RNA–Seq analyses were performed in five rhesus macaque tissues (liver, prefrontal cortex, skeletal muscle, adipose, and testis), which were then integrated with public transcriptome data from human, chimpanzee, and rhesus macaque.
On the basis of comparing the RNA expression profiles in the three species, we found that most of the hominoid-specific de novo protein-coding genes encoded polyadenylated non-coding RNAs in rhesus macaque or chimpanzee with a similar transcript structure and correlated tissue expression profile.
....
We found that
most of the hominoid-specific de novo protein-coding genes encoded long non-coding RNAs in rhesus macaque or chimpanzee, with similar transcript structure and correlated tissue expression profile, but the protein-coding genes often had higher transcriptional abundance. According to the rule of parsimony, we conclude that
at least a portion of long non-coding RNAs, especially those with active and regulated transcription,
may serve as a birth pool for protein-coding genes that are then further optimized at the transcriptional level, a pattern insensitive to the parameters used in the identification and analysis of
de novo genes.
...
We analyzed the characteristics of these
de novo protein-coding genes. Consistent with previous reports
[4], we found that the gene products were smaller, with a median length of 150.5 amino-acids, compared with 416 amino-acids in the human genome, suggesting the difficulty in
de novo origination of long ORFs [reference omitted]. The transcripts were generally expressed at lower levels compared with the human genome background [reference omitted]. Nineteen of the 24
de novo genes (79.2%) showed evidence to co-opt the transcriptional context, including seventeen (70.8%) that overlapped with another gene (nine anti-sense and nine at the same strand) and five (20.8%) located downstream of bi-directional promoters [table reference omitted]. Consistent with the case of
FLJ33706 or
C20orf203 [8],
Alu elements contributed to eight exons in eight genes (33.3%), and to three splicing junction sites in two genes [table reference omitted], further supporting the previous observation that repeat elements might be involved in the origination of some
de novo genes
[11]
...
Brain and testis may be favorable locations for newly created de novo genes
The out-of-testis hypothesis states that the testis is a hotbed of new gene origination, given its permissive chromatin state allowing widespread transcription
[13],
[29]. We analyzed the expression profiles of the
de novo protein-coding genes we identified and tested the out-of-testis pattern by following the idea described in
[29]. First, we investigated the relative abundance of each
de novo gene across testis, skeletal muscle, liver, heart, adipose, prefrontal cortex and cerebellum in terms of RNA-Seq read counts.
As shown in [table reference omitted], the de novo genes were most often transcribed in cerebellum. Testis ranked number two or three, depending on the species. Second, we tested whether class I (human-specific)
de novo genes are more often biased to testis compared to class II (human/chimpanzee shared)
de novo genes. As discussed in
[29], Class I should be more often expressed in testis if the out-of-testis hypothesis stands. As shown in [table reference omitted], different from this prediction, class II
de novo genes had a much higher proportion of reads expressed in testis. Thus,
de novo genes may not generally conform to the out-of-testis hypothesis. Testis-biased expression of
de novo genes had been reported so far in the fruit fly
[9],
[12]. By contrast, none of the previously reported human-specific
de novo genes appeared to have such a pattern:
i) the three
de novo genes identified in
[7] are broadly transcribed;
ii) higher expression levels of
FLJ33706 [8] were observed in several brain regions such as cerebellum or cerebral cortex compared to testis (relative expression levels, 2.4–4.9
versus 1.5);
iii) sixty
de novo genes identified in
[5] are most often transcribed in cerebral cortex with testis following as the second top tissue. In summary, testis may still be the location of some
de novo genes,
but brain appears to be a favorable location for the origination of many human de novo genes.
....
