Comparison of two snake genomes, spanning ~100 My of snake evolution, revealed extensive differences in their genomic repeat landscapes. Although both snakes contain diverse sets of repeat elements distributed across most major element types and superfamilies, the copperhead genome contains more of essentially all of these repeats (occupying 45% of the copperhead genome vs. 21% of the python genome), and many repeats have expanded recently. In comparison, the largest known difference in genomic repeat content between placental mammalian genomes occurs between the human and the mouse (46% vs. 38%, respectively), which are separated by ~75 My (Waterston et al. 2002
). Thus, for similar levels of temporal divergence, the difference in repeat content in these two snakes is exceptional. Furthermore, the greater repetitive content in the copperhead is not due to one or a few expanded repeat families but is distributed among a diversity of element families and subfamilies (243 collapsed TE families in the copperhead vs. 82 in the python).
TE-related transcripts appear to be expressed at much higher levels in copperhead tissue compared with python tissue, even when the greater genomic TE abundances in the copperhead genome are accounted for. Although we surveyed liver rather than gametic tissues for TE activity, the 23-fold greater overall levels of TE-related transcripts in the copperhead than in the python () suggests that TE transcription may be generally more active in the copperhead. This is true even in the case of CR1, for which there are 25 times more elements in the copperhead than in the python genome; there are 122 times more CR1-related transcripts in copperhead tissues than in python tissues ( and supplementary table S11
, Supplementary Material
online). If transcription levels have also increased in germ line tissues, they may have contributed to increased genomic TE insertion activity. One hypothesis, to explain the observations that TEs have higher transcription levels and have been more active in the copperhead versus python genomes, is that mechanisms known to control TE proliferation (e.g., CpG methylation and chromatin structural regulation; Yoder et al. 1997
; Lippman et al. 2004
; Feschotte 2008
) may be differentially effective in the two snakes. It is also possible that TEs may occur in greater proximity to transcriptional units in the copperhead genome, driving greater levels of read-through transcription. It is unclear, however, what mutational or selective force would have made TEs land and become fixed nearer transcriptional units in the copperhead than in the python. The increased transcription levels in the copperhead also suggest that TEs are more likely to influence flanking gene expression in the copperhead than in the python.
Prior to the present study, at least two plausible instances of horizontal transfer implicating snake TEs have been reported, involving Bov-B LINEs (Kordis and Gubensek 1998b
) and Sauria SINEs (Piskurek and Okada 2007
). This study provides novel evidence for additional horizontal transfer of TEs and by adding genomic data from two snake species in addition to the anole lizard, it provides the first large-scale comparative view into TE dynamics within squamate reptiles. Together, our sequence-based data and PCR-based confirmation of the absence of SPIN elements in the python (Feschotte C, unpublished data), in contrast to the abundance of recently inserted SPIN elements in the copperhead, provide compelling new evidence that, as with mammalian genomes (Pace et al. 2008
; Gilbert et al. 2010
), reptilian genomes have been differentially invaded by these elements. Although previous studies have already suggested horizontal transfer of Bov-B LINEs between squamate reptiles and mammals (Kordis and Gubensek 1997
), our analysis suggests the possibility of multiple transfer events into and/or out of squamate genomes (supplementary fig. S5
, Supplementary Material
online). The previous report of a poxvirus-mediated transfer of Squam1 SINE elements from viperid snakes to rodents demonstrates that viruses may sometimes mediate such horizontal transfer events (Piskurek and Okada 2007
). This transfer is thought to be dependent on the enzymatic machinery of a Bov-B LINE (Piskurek and Okada 2007
), and high transcript levels of Bov-B reverse transcriptase in snake tissues, such as those found in the copperhead, may thus increase the probability of horizontal transfer events.
The copperhead lineage also appears to have modified microsatellite evolutionary dynamics, including microsatellite seeding (Arcot et al. 1995
; Tay et al. 2010
) by a snake-specific CR1 LINE family (). Our analyses show that a high percentage of expanded microsatellite motifs were adjacent to readily identifiable snake1 CR1 LINEs (41.4% of all ATA and 22.7% of all AATAG SSR loci). These findings suggest that microsatellite seeding by these LINEs in the copperhead has occurred at a scale that is several orders or magnitude greater than any other example that we are aware of (Nadir et al. 1996
; Tay et al. 2010
). The similar SSR motif frequencies between the python and anole lizard are consistent with previous suggestions that SSR evolution and turnover rates in non-avian reptiles are generally lower than in mammals (Matsubara et al. 2006
; Shedlock et al. 2007
). In contrast, the increase in SSR content and radically different motif frequencies in copperhead indicate that SSR turnover rates in squamates can evolve even more rapidly than what is known from mammalian genomes.
Despite its substantial and recently expanded repeat content, the copperhead has a genome size that is among the smallest of snakes. This is surprising, as it is reasonable to expect that small genomes should have low repetitive content, as is the case in pythons and birds. We suggest that unidentified processes must be acting differentially in the copperhead to remove genomic sequence, potentially due to mechanistic differences in the biology of the two snake lineages and/or differences in selection on mutations that alter genome size. Further evidence of differential processes operating in these two snake lineages comes from our observation of differences in the relative abundance of 3′ truncated LINEs between species (). The excess of short LINEs in the copperhead is consistent with pressure to limit genome expansion, although 3′ LINE truncation has not been sufficient to balance the genome size equation (the total LINE element sequence is still considerably greater in the copperhead). The relative bias toward short elements in the copperhead could be caused by a greater tendency to generate shorter elements (at the time of insertion), a greater probability to fix shorter elements, and/or a greater probability to delete long elements at any time via ectopic recombination, which has been proposed to occur in the anole lizard genome (Novick et al. 2009
). Selection could be more effective in the copperhead than the python due to a larger effective population size rather than stronger selection against genome expansion, but this seems unlikely. It is expected that the Nearctic-distributed copperhead lineage suffered small effective population sizes due to bottlenecks during glacial cycles (Guiher and Burbrink 2008
), and the likelihood of large and stable effective populations in tropical lineages such as the python also contraindicates a primary role for population size differences as a sole explanation.
Selection to maintain a smaller genome size has been hypothesized numerous times in relation to extreme metabolic demands in flighted birds (Hughes and Hughes 1995
), although there is some controversy (Organ et al. 2007
). Previous studies have suggested that extreme metabolic demand in snakes (Secor and Diamond 1995
) has resulted in selection to decrease their mitochondrial genome size (Jiang et al. 2007
), extensive evolutionary redesign (Castoe et al. 2008
), and previously unprecedented molecular convergence in snake metabolic proteins (Castoe et al. 2009b
). It is therefore plausible that selection related to metabolic demands could have shaped snake nuclear genomes. Broader understanding of genomic repeat landscapes in snakes may shed greater light on this question. There are a range of alternative theories about the evolution of genome size and complexity (Lynch and Conery 2003
), and thus the role of selection in snake genome size and structure is a topic of considerable interest.
It is also an open question whether the biology of snake genomes may have contributed to the evolution of their extreme phenotypes and adaptations (Secor and Diamond 1995
; Cohn and Tickle 1999
; Fry et al. 2006
; Castoe et al. 2008
; Vonk et al. 2008
). Among the most conspicuous adaptations in snakes is that some lineages, including the ancestors of the copperhead, have evolved complex venom repertoires, largely by duplicating and repurposing existing genes to produce deadly toxins. Our evidence (supplementary table S10
, Supplementary Material
online), and that of others (Nobuhisa et al. 1998
; Fujimi et al. 2002
; Ikeda et al. 2010
), shows a tentative association between CR1 LINEs and venom genes. Our genomic sampling suggests that CR1 LINEs (as well as SSRs and other TEs) have expanded substantially in the copperhead lineage, and this expansion might be expected to lead to increased rates of recombination, unequal crossing over, and gene conversion (Witherspoon et al. 2009
; Stevison and Noor 2010
). These events could have, at least in part, facilitated the expansion and regulatory rewiring of venom gene families in venomous snakes.