TP53 copy number expansion is associated with the evolution of increased body size and an enhanced DNA damage response in elephants

A major constraint on the evolution of large body sizes in animals is an increased risk of developing cancer. There is no correlation, however, between body size and cancer risk. This lack of correlation is often referred to as ‘Peto’s Paradox’. Here we show that the elephant genome encodes 20 copies of the tumor suppressor gene TP53 and that the increase in TP53 copy number occurred coincident with the evolution of large body sizes, the evolution of extreme sensitivity to genotoxic stress, and a hyperactive TP53 signaling pathway in the elephant (Proboscidean) lineage. Furthermore we show that several of the TP53 retrogenes (TP53RTGs) are transcribed and likely translated. While TP53RTGs do not appear to directly function as transcription factors, they do contribute to the enhanced sensitivity of elephant cells to DNA damage and the induction of apoptosis by regulating activity of the TP53 signaling pathway. These results suggest that an increase in the copy number of TP53 may have played a direct role in the evolution of very large body sizes and the resolution of Peto’s paradox in Proboscideans.


218
Consistent with transcription of the TP53RTG genes, we amplified PCR products at the 219 expected size for the TP53 and TP53RTG transcripts but did not amplify PCR products from

238
We attempted to identify the transcription start site of the TP53RTG12 gene using 239 several 5'-RACE methods, however, we were unsuccessful in generating PCR products from 240 either African Elephant fibroblast or placenta cDNA, or Asian elephant fibroblast cDNA.

241
Therefore, we designed a set of 34 PCR primers tiled across the region of scaffold_825 that 242 encodes the TP53RTG12 gene and used these primers to amplify PCR products from African 243 and Asian Elephant fibroblast cDNA generated from DNase treated RNA. We then 244 reconstructed the likely TP53RTG12 promoter, transcription start site, and exon-intron structure 245 from the pattern of positive PCR products. These data suggest that the major transcription

249
Next we tested the ability of the African and Asian elephant RTE1_LA elements and the 250 RTE1_LA consensus sequence (as a proxy for the ancestral RTE1_LA sequence) to function as 251 a promoter in transiently transfected African and Asian elephant fibroblasts when cloned into the 252 promoterless pGL4.10[luc2] luciferase reporter vector. We found that the African and Asian 253 elephant RTE1_LA elements increased luciferase expression 3.03-fold (t-test, P=2.41×10 -8 ) and 254 1.37-fold (t-test, P=2.60×10 -4 ), respectively, compared to empty vector controls ( Figure 5D).

255
However, luciferase expression from the pGL4.10[luc2] vector containing the RTE1_LA 256 sequence was not significantly different than the empty vector control in either Asian (0.96-fold; 257 t-test, P=0.61) or African elephant fibroblasts (0.95-fold; t-test, P=0.37; Figure 5D). These data 258 indicate that transcription of TP53RTG12 likely initiates within a RTE1_LA-derived promoter, but 259 9 that the ability of this RTE1_LA element to function as a promoter is not an ancestral feature of 260 RTE1_LA elements.

263
To determine if TP53RTG transcripts are translated, we treated African elephant, Asian 264 elephant, and hyrax cells with 50 J/m 2 UV-C (to stabilize TP53) and the proteasome inhibitor 265 MG-132 (to block protein degradation), and assayed for TP53/TP53RTG proteins by Western 266 blotting total cell protein with a polyclonal TP53 antibody (FL-393) that we demonstrated 267 recognizes Myc-tagged TP53RTG12. We identified bands in both African and Asian elephant 268 and hyrax total cell protein at the expected size for the full length p53, Δ133p53β/γ, and p53ψ-269 like isoforms of the TP53 protein (Khoury and Bourdon, 2010) as well as high molecular weight 270 bands corresponding to previously reported SDS denaturation resistant TP53 oligomers (Cohen 271 et al., 2008;Ottaggio et al., 2000) and (poly)ubiquitinated TP53 congugates (Sparks et al.,

10
To determine the consequences of an enhanced TP53 response we treated primary 293 African and Asian elephant, hyrax, aardvark, and armadillo dermal fibroblasts with mitomycin C, 294 doxorubicin, nutlin-3a, or UV-C and measured cell viability (live-cell protease activity), 295 cytotoxicity (dead-cell protease activity), and the induction apoptosis (caspase-3/7 activation) 296 using an ApoTox-Glo Triplex assay. Consistent with the results from the luciferase assay, we 297 found that lower doses of mitomycin C or doxorubicin induced apoptosis in elephant cells than 298 the other species ( Figure 6B) and that the magnitude of the response was greater in elephant 299 than other species (Figure 6A). Similarly UV-C exposure generally induced more elephant cells 300 to undergo apoptosis than other species (Figure 6A). A striking exception to this trend was the 301 response of elephant cells to the MDM2 antagonist nutlin-3a, which elicited a strong TP53 302 transcriptional response ( Figure 6A) but did not induce apoptosis ( Figure 6B)

310
TP53RTG-specific siRNA or a scrambled siRNA control ( Figure 7A). Next we used a dual 311 luciferase reporter assay to measure activation of the TP53 pathway in response to treatment 312 mitomycin C, doxorubicin, UV-C, or nutlin-3a. As expected given our previous results, we found 313 that African elephant fibroblasts transfected with control siRNA induced TP53 signaling in 314 response to each treatment ( Figure 7A). In contrast, African elephant fibroblasts transfected 315 with TP53RTG-specific siRNA had significantly lower luciferase expression, and thus reduced 316 TP53 signaling, in response to either DNA-damaging agents (mitomycin C, doxorubicin, UV-C) 317 or MDM2 antagonism (nutlin-3a). TP53RTG knockdown also elevated baseline TP53 signaling 318 ( Figure 7B). These data suggest that TP53RTG proteins have at least two distinct functions, 319 inhibiting TP53 signaling in the absence of inductive signals and potentiation of TP53 signaling 320 after the induction of DNA damage.

348
TP53RTG proteins are unlikely to directly regulate TP53 target genes because they lack 349 critical residues required for nuclear localization, tetramerization, and DNA-binding ( Figure 9A).

350
Previous studies, for example, have shown the TP53 mutants lacking the tetramerization 351 domain and C-terminal tail are unable to bind DNA or transactivate luciferase expression from a 352 reporter vector containing TP53 response elements (Kim et al., 2012). Similarly the p53Ψ 353 isoform, which is truncated in the middle of the DNA binding domain and lacks the nuclear 354 localization signal and oligomerization domain, is unable to bind DNA and is transcriptionally 355 inactive (Senturk et al., 2014). Unexpectedly, we found that the GFP-tagged TP53RTG12   Figure 9A). These 368 data suggest at least two non-exclusive models of TP53RTG action: 1) TP53RTG proteins may 369 act as 'decoys' for the MDM2 complex allowing the canonical TP53 protein to escape negative 370 regulation ( Figure 10A); and 2) TP53RTG proteins may protect canonical TP53 from MDM2 371 mediated ubiqutination, which requires tetramerization (Kubbutat et al., 1998;Maki, 1999), by 372 dimerizing with canonical TP53 and thereby preventing the formation of tetramers ( Figure 10F).

374
The decoy model depends on the ability of TP53RTG proteins to physically interact with 375 MDM2. Previous crystallographic studies of the TP53/MDM2 interaction have shown that a trio 376 of residues in TP53 (F19, W23, and L26) insert deeply into a hydrophobic cleft in MDM2, which 377 stabilizes the interaction (Kussie et al., 1996). We identified a W23G substitution in all TP53RTG 378 proteins at a site that is invariant for tryptophan in TP53 proteins including African and Asian 379 elephant TP53 (Figure 10B), suggesting that TP53RTG proteins may be unable to physically 380 interact with MDM2. To infer the structural and functional consequences of the TP53RTG W23G 381 substitution we generated a homology model of the elephant TP53RTG12/MDM2 complex using 382 I-TASSER/ModRefiner (Roy et al., 2010;Xu and Zhang, 2011;Zhang, 2008) and the crystal 383 structure of the MDM2/TP53 dimer as a template (Kussie et al., 1996). We found that the 384 TP53RTG12 transactivation domain was inferred to be a short α-helix ( Figure 10C) and was 385 very similar to the template structure (RMSD: 1.756), however, the W23G substitution is 386 predicted to abolish crucial hydrophobic interactions between the amphipathic α-helix of TP53 387 and the hydrophobic cleft MDM2. Indeed, three methods (Pires et al., 2014) inferred that the 388 W23G substitution is destabilizing on the TP53RTG12/MDM2 interaction (mCSM ΔΔG=-2.42,

469
We also found that the increase in TP53RTG copy number occurred coincident with the    Rooney et al., 2002) and venom genes (Lynch, 2007).

515
Under this model selection acts to maintain a minimal number of functional copies (functional 516 copy number) rather than total copy number, the increase in total copy number is driven by the 517 total number of loci and the rates of duplication, loss, and fixation. Thus the overall increase in 518 total TP53RTG copy number may be a selectively neutral processes, driven simply by higher 519 rates of duplication and/or fixation than loss. The 'decoy' model (Abegglen et al., 2015) has also been challenged because it would 551 allow for activation of the TP53 signaling pathway in the absence of DNA damage (Perez and 552 Komiya, 2016), which is generally lethal in animal models (Hoever et al., 1994;Lozano, 2010 tradeoffs including slower pre-and post-natal growth rates and reduced size (Maier et al., 2004), 563 a shortened lifespan (Maier et al., 2004), accelerated aging (Tyner et al., 2002), and reduced 564 fertility (Allemand et al., 1999;Maier et al., 2004), as well as developmental tradeoffs including  , 1996). Thus, increases in TP53 copy number protects against cancer but appears to 568 come with the cost of developmental delays, accelerated aging, and reduced fertility (Campisi, 569 2003;Donehower, 2002;Ferbeyre and Lowe, 2002;Rodier et al., 2007).

598
Collectively these data indicate that increased TP53 dosage comes with a cost, and 599 suggest that Proboscideans evolved a mechanism that reduced this cost, broke a major 600 developmental and evolutionary constraint on TP53 copy number, or perhaps evaded paying 601 the cost of increased TP53 copy number altogether. TP53RTG genes, for example, are 602 transcribed from a transposable element derived promoter that is evolutionarily younger than the 603 retrogenes. Thus, the initial TP53RTG genes were unlikely to be transcribed and incur a cost.

604
Similarly our phylogenetic analyses indicates that copy number expanded after the initial

773
Correlation between TP53/TP53RTG copy number and body size evolution

794
We also observed that the genomic region surrounding each TP53RTG gene contained

851
The TP53RTG genes are 80.0-82.7% identical to TP53 at the nucleotide level, with 204-852 231 total nucleotide differences compared to TP53. This level of divergence allows for many 853 reads to be uniquely mapped to each gene, there will also be significant read mapping 854 uncertainty in regions of these genes with few nucleotide differences. However, if read mapping 855 uncertainty was leading to false positive mappings of TP53 derived reads to TP53RTG genes 856 we would expect to observe expression of many TP53RTG genes, rather the robust expression 857 of a single (TP53RTG12) gene. We also counted the number of uniquely mapped reads to each 858 TP53RTG gene and TP53. We found that 0-8 reads were uniquely mapped to most TP53RTG 859 genes, except TP53RTG12 which had ~115 uniquely mapped reads across samples and TP53 860 which had ~3000 uniquely mapped reads across tissue samples. Thus we conclude that read 861 mapping uncertainty has not adversely affected our RNA-Seq analyses.

863
Gene expression data (PCR and Sanger sequencing)

864
We further confirmed expression of TP53RTG12 transcripts in elephant cells through

865
RT-PCR, taking advantage of differences between TP53RTG12 and TP53 sequences to design 866 two aligned primer sets, one TP53-specific, the other TP53RTG12 specific. TP53RTG12 867 primers were: 5´ ggg gaa act cct tcc tga ga 3´ (forward) and 5´ cca gac aga aac gat agg tg 3´ 868 (reverse). TP53 primers were: 5´ atg gga act cct tcc tga ga 3´ (forward) and 5´ cca gac gga aac 869 cat agg tg 3´ (reverse). The TP53 amplicon is expected to be 251 bps in length, while deletions 870 present in the TP53RTG12 sequence lead to a smaller projected amplicon size of 220 bps.

897
The RTE1_LA non-LTR retrotransposon has previously been described from the African 898 elephant genome, these elements are generally more than 90% identical to the consensus 899 RTE1_LA sequence but less than 70% identical to other mammalian RTEs 900

915
Clonetics/Lonza). Culture medium was removed from dishes just prior to UV treatment and 916 returned to cells shortly afterwards. Experimental cells were exposed to 50J/m 2 UV-C radiation 917 in a crosslinker (Stratalinker 2400, Stratagene), while control cells passed through media 918 changes but were not exposed to UV. A small volume (~3mL)