Abstract
Large bodied organisms have more cells that can potentially turn cancerous than smallbodied organisms with fewer cells, imposing an increased risk of developing cancer. This expectation predicts a positive correlation between body size and cancer risk, however, there is no correlation between body size and cancer risk across species (‘Peto’s Paradox’). Here we show that elephants and their extinct relatives (Proboscideans) may have resolved Peto’s Paradox in part through re-functionalizing a leukemia inhibitory factor pseudogene (LIF6) with pro-apoptotic functions. The LIF6 gene is transcriptionally up-regulated by TP53 in response to DNA damage, and translocates to the mitochondria where it induces apoptosis. Phylogenetic analyses of living and extinct Proboscidean LIF6 genes indicates its TP53 response element evolved coincident with the evolution of large body sizes in the Proboscidean stem-lineage. These results suggest that re-functionalizing of a pro-apoptotic LIF pseudogene may have been permissive (though not sufficient) for the evolution of large body sizes in Proboscideans.
Introduction
The risk of developing cancer places severe constraints on the evolution of large body sizes and long lifespans in animals. If all cells have a similar risk of malignant transformation and equivalent cancer suppression mechanisms, organism with many cells should have a higher risk of developing cancer than organisms with fewer cells. Similarly organisms with long lifespans have more time to accumulate cancer-causing mutations than organisms with shorter lifespans and therefore should also be at an increased risk of developing cancer, a risk that is compounded in large-bodied, long-lived organisms (Cairns, 1975; Caulin and Maley, 2011; Doll, 1971; Peto, 2015; Peto, 1975). Consistent with these expectations, there is a strong positive correlation between body size and cancer incidence within species. Larger dog breeds, for example, have higher rates of cancer than smaller breeds (Dobson, 2013) and human cancer incidence increases with increasing adult height for numerous cancer types (Green et al., 2011). In stark contrast, there are no correlations between body size or lifespan and cancer risk between species (Abegglen et al., 2015); this lack of correlation is often referred to as ‘Peto’s Paradox’ (Caulin and Maley, 2011; Leroi et al., 2003; Peto, 1975).
While the ultimate resolution to Peto’s paradox is that large bodied and/or long-lived species evolved enhanced cancer protection mechanisms, identifying and characterizing those mechanisms is essential for elucidating how enhanced cancer resistance and thus large bodies and long lifespans evolved. Numerous and diverse mechanisms have been proposed to resolve Peto’s paradox (Caulin and Maley, 2011; Dang, 2015; Katzourakis et al., 2014; Leroi et al., 2003; Maciak and Michalak, 2015; Nagy et al., 2007; Nunney, 1999; Takemoto et al., 2016), but discovering those mechanisms has been challenging because the ideal study system is one in which a large, long-lived species is deeply nested within a clade of smaller, short-lived species - all of which have sequenced genomes. Unfortunately, few lineages fit this pattern. Furthermore while comparative genomics can identify genetic changes that are phylogenetically associated the evolution of enhanced cancer protection, determining which of those genetic changes are causally related to cancer biology through traditional reverse and forward genetics approaches are not realistic for large species such as whales and elephants. Thus we must use other methods to demonstrate causality.
Among the most parsimonious mechanisms to resolve Peto’s paradox are a reduced number of oncogenes and/or an increased number of tumor suppressor genes (Caulin and Maley, 2011; Leroi et al., 2003; Nunney, 1999), but even these relatively simple scenarios are complicated by transcriptional complexity and context dependence. The multifunctional interleukin-6 class cytokine leukemia inhibitory factor (LIF), for example, can function as either a tumor suppressor or an oncogene depending on the context. Classically LIF functions as an extracellular cytokine by binding the LIF receptor (LIFR) complex, which activates downstream PI3K/AKT, JAK/STAT3, and TGFβ signaling pathways. The LIF gene encodes at least three transcripts, LIF-D, LIF-M, and LIF-T, which contain alternative first exons spliced to common second and third exons (Haines et al., 1999; Hisaka et al., 2004; Rathjen et al., 1990; Voyle et al., 1999). Remarkably while the LIF-D and LIF-M isoforms are secreted proteins that interact with the LIF receptor (Rathjen et al., 1990; Voyle et al., 1999), the LIF-T isoform lacks the propeptide sequence and is an exclusively intracellular protein (Haines et al., 1999; Voyle et al., 1999) that induces caspase-dependent apoptosis through an unknown mechanism (Haines et al., 2000).
Here we show that the genomes of Paenungulates (elephant, hyrax, and manatee) contain numerous duplicate LIF pseudogenes, at least one (LIF6) of which is expressed in elephant cells and is up-regulated by TP53 in response to DNA damage. LIF6 encodes a separation of function isoform structurally similar to LIF-T that induces apoptosis when overexpressed in multiple cell types and is required for the elephant-specific enhanced cell death in response to DNA-damage. These results suggest that the origin of a zombie LIF gene (a reanimated pseudogene that kills cells when expressed) may have contributed to the evolution of enhanced cancer resistance in the elephant lineage and thus the evolution large body sizes and long lifespans.
Results
Repeated segmental duplications increased LIF copy number in Paenungulates
We characterized LIF copy number in 53 mammalian genomes, including large, long-lived mammals such as the African elephant (Loxodonta africana), Bowhead (Balaena mysticetus) and Minke (Balaenoptera acutorostrata scammoni) whales, as well as small, long-lived mammals such bats and the naked mole rat. We found that most Mammalian genomes encoded a single LIF gene, however, the manatee (Trichechus manatus), rock hyrax (Procavia capensis), and African elephant genomes contained 7-11 additional copies of LIF (Figure 1). None of the duplicate LIF genes includes the 5’-UTR, coding exon 1, or a paired low complexity (CGAG)n/CT-rich repeat common to the canonical LIF genes in elephant, hyrax, manatee, tenrec, and armadillo (Figure 2A). Most of the duplicates include complex transposable element insertions composed of tandem tRNA-Asn-AAC/AFROSINE and AFROSINE3/tRNA-RTE/MIRc elements within introns one and two (Figure 2A). Fine mapping of the duplicate ends by reciprocal best BLAT indicates that there is no region of homology upstream of the tRNA-Asn-AAC/AFROSINE elements for duplicates that include exon 2, whereas duplicate LIF genes that lack exon 2 have ~150-300bp regions of homology just upstream of the paired AFROSINE3/tRNA-RTE/MIRc elements in intron 2. The LIF encoding loci in the hyrax and manatee genomes have not been assembled into large-scale scaffolds, but the African elephant LIF loci are located within a 3.5Mb block of chromosome 25 (loxAfr4).
LIF duplicates may result from independent duplication events in the elephant, hyrax, and manatee lineages, ancestral duplications that occurred in the Paenungulate stem-lineage followed by lineage-specific duplication and loss events, or some combination of these processes. We used Bayesian phylogenetic methods to reconstruct the LIF gene tree and gene tree reconciliation to reconstruct the pattern of LIF duplication and loss events in Paenungulates. Consistent with a combination of ancestral and lineage-specific duplications, our phylogenetic analyses of Paenungulate LIF genes identified well-supported clades containing loci from multiple species as well as clades containing loci from only a single species (Figure 2B). The reconciled tree identified 17 duplication and 14 loss events (Figure 2C). These data indicate that the additional LIF genes result from repeated rounds of segmental duplication, perhaps mediated by recombination between repeat elements.
Duplicate LIF genes are structurally similar to the LIF-T
Barring transcription initiation from cryptic upstream sites encoding in frame start codons, all duplicate LIF genes encode N-terminally truncated variants that are missing exon 1, lack the propeptide sequence, and are similar in primary structures to LIF-T (Figure 3A). While some duplicates lack the N-terminal LIFR interaction site (Figure 3A), all include the leucine/isoleucine repeat required for inducing apoptosis (Figure 3A) (Haines et al., 2000). Crucial residues that mediate the interaction between LIF and LIFR (Figure 3B) (Hudson et al., 1996; Huyton et al., 2007) are relatively well conserved in duplicate LIF proteins, as are specific leucine/isoleucine residues that are required for the pro-apoptotic functions of LIF-T (Figure 3C)(Haines et al., 2000). Haines et al. (2000) suggested that the leucine/isoleucine residues of LIF-T are located on a single face of helix B, and may form an amphipathic α-helix. Similar to LIF-T, leucine/isoleucine residues of duplicate LIF proteins are located on a single face of helix B (Figure 3D). These data suggest that at least some of the structural features that mediate LIF functions, in particular the pro-apoptotic function(s) of LIF-T, are conserved in duplicate LIFs.
Elephant LIF6 is up-regulated by TP53 in response to DNA damage
If expansion of the LIF gene repertoire plays a role in the evolution of enhanced cancer resistance, then one or more of the LIF genes should be transcribed. To determine if duplicate LIF genes were transcribed, we assembled and quantified elephant LIF transcripts with HISAT2 (Kim et al., 2015) and StringTie (Pertea et al., 2015) using deep 100bp paired-end RNA-Seq data (>138 million reads) we previously generated from Asian elephant dermal fibroblasts (Sulak et al., 2016), as well as more shallow (~30 million reads) singe-end sequencing from Asian elephant peripheral blood mononuclear cells (PBMCs) (Reddy et al., 2015), African elephant dermal fibroblasts (Cortez et al., 2014) and placenta (Sulak et al., 2016). We identified transcripts corresponding to the LIF-D, LIF-M, and LIF-T isoforms of the canonical LIF1 gene, and one transcript of a duplicate LIF gene (LIF6) in Asian elephant dermal fibroblasts (Figure 4A). The LIF6 transcript initiates just downstream of canonical exon 2 and expression was extremely low (0.33 transcripts per million), as might be expected for a pro-apoptotic gene. No other RNA-Seq dataset identified duplicate LIF transcripts.
Previous studies have shown that TP53 regulates basal and inducible transcription of LIF in response to DNA damage through a binding site located in LIF intron 1 (Baxter and Milner, 2010; Hu et al., 2007), suggesting that duplicate LIF genes may be regulated by TP53. Therefore we computationally predicted TP53 binding sites within a 3kb window around Atlantogenatan LIF genes and identified binding site motifs in the first intron of African elephant, hyrax, manatee, tenrec, and armadillo LIF1 genes whereas the only duplicate LIF gene with a putative TP53 binding site was elephant LIF6; note that the putative TP53 binding sites around LIF1 and LIF6 are not homologous (Figure S1). Next we treated African elephant primary dermal fibroblasts with the DNA damaging agent doxorubicin (DOX) or the MDM2 antagonist nutlin-3a and quantified the transcription of canonical LIF1, duplicate LIF genes, and the TP53 target gene BAX by qRT-PCR. DOX treatment induced LIF6 expression 8.18-fold (Wilcox test, P=1.54×10−6) and nutlin-3a induced LIF6 expression 16.06-fold (Wilcox test, P=1.00×10−4), which was almost completely attenuated by siRNA mediated TP53 knockdown (Figure S2 and Figure 4B). Treatment with DOX (Wilcox test, P=1.55×10−4) or nutlin-3a (Wilcox test, P=1.55×10−4) also up-regulated the TP53 target gene BAX (Figure 4B), which again was almost blocked by knockdown of TP53 (Figure 4B). In contrast neither treatment up-regulated LIF1 (Figure 4B) and we observed no expression of the other duplicate LIF genes in African elephant fibroblasts or any LIF duplicate in hyrax fibroblasts treated with DOX or nutlin-3a. These data suggest that while LIF6 encodes a transcribed gene in elephants, transcription of the other LIF duplicates is either induced by different signals or they are pseudogenes.
To test if the putative TP53 binding site upstream of elephant LIF6 was a functional TP53 response element, we cloned the −1100bp to +30bp region of the African elephant LIF6 gene into the pGL3-Basic[minP] luciferase reporter vector and tested its regulatory ability in dual luciferase reporter assays. We found that the African elephant LIF6 upstream region had no effect on basal luciferase expression in transiently transfected African elephant fibroblasts (Wilcox test, P=0.53). In contrast, both DOX (Wilcox test, P=1.37×10−8) and nutlin-3a (Wilcox test, P=1.37×10−8) strongly increased luciferase expression (Figure 4C), which was almost completely abrogated by deletion of the putative TP53 binding-site in DOX (Wilcox test, P=1.37×10−8) and N3a (Wilcox test, P=1.37×10−9) treated cells (Figure 4C). Next we performed ChIP-qPCR to determine if the TP53 binding-site upstream of LIF6 is bound by TP53 in African elephant fibroblasts treated with DOX or nutlin-3a using a rabbit polyclonal TP53 antibody (FL-393) that we previously demonstrated recognizes elephant TP53 (Sulak et al., 2016). DOX treatment increased TP53 binding 14.26-fold (unequal variance t-test, P=0.039) and nutlin-3a increased TP53 binding 10.75-fold (unequal variance t-test, P=0.058) relative to ChIP-qPCR with normal mouse IgG control antibody. This increased binding was almost completely attenuated by siRNA mediated TP53 knockdown (Figure 4D).
Finally, we transiently transfected elephant fibroblasts with either a negative control siRNA or siRNAs targeting TP53 and a LIF6 expression vector and assayed cell viability, cytotoxicity, and apoptosis using an ApoTox-Glo assay 18 hours after treatment with DOX or control media. We found that LIF6 expression with negative control siRNAs augmented the induction of apoptosis by DOX (Wilcox test, P=0.033; Figure 4E and Figure S3). Knockdown of TP53 did not inhibit the induction of apoptosis (Wilcox test, P=0.033; Figure 4E and Figure S3), suggesting TP53 knockdown was insufficient to alter the induction of apoptosis; note that while siRNA mediated knockdown significantly reduced TP53 transcript levels (Figure S2), we were unable to validate knockdown of the TP53 protein because the FL-393 antibody that recognizes elephant TP53 is no longer available. Interestingly, however, LIF6 transfection induced apoptosis in elephant fibroblasts treated with control media and negative control siRNAs (Wilcox test, P=0.008), suggesting that LIF6 can induce apoptosis in the absence of DNA damage similar to LIF-T (Figure 4E and Figure S3). Thus, we conclude that elephant LIF6 is transcriptionally up-regulated by TP53 in response to DNA damage and may have pro-apoptotic functions.
Elephant LIF6 contributes to the augmented DNA-damage response in elephants
We have previously shown that elephant cells evolved to be extremely sensitive to genotoxic stress and induce apoptosis at lower levels of DNA damage than their closest living relatives, including the African Rock hyrax (Procavia capensis capensis), East African aardvark (Orycteropus afer lademanni), and Southern Three-banded armadillo (Tolypeutes matacus) (Sulak et al., 2016). To test the contribution of LIF6 to this derived sensitivity, we designed a set of three siRNAs that specifically target LIF6 and reduce LIF6 transcript abundance ~88% (Figure S2). Next, we treated African elephant dermal fibroblasts with DOX or nutlin-3a and either LIF6 targeting siRNAs or a control siRNA and assayed cell viability, cytotoxicity, and apoptosis using an ApoTox-Glo assay 24 hours after treatment. Both DOX (Wilcox test, P=3.33×10−9) and nutlin-3a (Wilcox test, P=3.33×10−9) reduced cell viability ~85%, which was attenuated 5-15% by LIF6 knockdown in DOX (Wilcox test, P=1.33×10−8) or nutlin-3a (Wilcox test, P=3.33×10−9) treated cells (Figure 5A). While neither DOX nor nutlin-3a induced cytotoxicity (Figure 5A), both DOX (4.05-fold, Wilcox test, P=3.33×10−9) and nutlin-3a (2.64-fold, Wilcox test, P=3.33×10−9) induced apoptosis (Figure 5A).
To determine if LIF6 expression was sufficient to induce apoptosis, we transiently transfected a LIF6 expression vector in to African elephant dermal fibroblasts and assayed cell viability, cytotoxicity, and apoptosis using the ApoTox-Glo assay 24 hours after transfection. We again found that LIF6 overexpression induced apoptosis in the absence of either DNA damage by DOX or TP53 activation by nutlin-3a treatment (Wilcox test, P=3.11×10−4), and augmented apoptosis induced with DOX (Wilcox test, P=0.02). Induction of apoptosis by LIF6 was almost completely blocked by co-treatment with the irreversible broad-spectrum caspase inhibitor Z-VAD-FMK (Wilcox test, P=1.55×10−4) but not cyclosporine A (Wilcox test, P=0.23), which inhibits opening of the opening of the mitochondrial permeability transition pore (Figure 5B and Figure S4). These data suggest that LIF6 contributes to the enhanced apoptotic response that evolved in the elephant lineage, likely through a mechanism that induces caspase-dependent apoptosis.
Elephant LIF6 induces mitochondrial dysfunction and caspase-dependent apoptosis
To infer the mechanism(s) by which LIF6 contributes to the induction of apoptosis, we first determined the sub-cellular localization of a LIF6-eGFP fusion protein in African elephant dermal fibroblasts. Unlike LIF-T, which has diffuse cytoplasmic and nuclear localization (Haines et al., 2000), LIF6-eGFP was located in discrete foci that co-localized with MitoTracker Red CM-H2XRos stained mitochondria (Figure 6A). Mitochondria are critical mediators of cell death, with distinct pathways and molecular effectors underlying death through either apoptosis (Karch et al., 2013; Tait and Green, 2010) or necrosis (Tait and Green, 2010; Vaseva et al., 2012). During apoptosis, for example, the Bcl-2 family members Bax/Bak form large pores in the outer mitochondrial membrane that allow cytochrome c to be released into the cytosol thereby activating the caspase cascade (Karch et al., 2013; Tait and Green, 2010). In contrast, during necrosis, Bax/Bak in the outer membrane interact with the cyclophilin D (CypD) and the inner membrane complex leading to the opening of the mitochondrial permeability transition pore (MPTP), swelling, and eventual rupture (Tait and Green, 2010; Vaseva et al., 2012).
To test if LIF6 induced apoptosis was specific to elephant cells and independent of LIF receptor (LIFR) mediated signaling, we transiently transfected Chinese hamster (Cricetulus griseus) ovary (CHO) cells, which do not express LIFR (Orellana et al., 2018), with the LIF6 expression vector and assayed the induction of apoptosis with the ApoTox-Glo assay. Overexpression of LIF6 induced apoptosis 5.38-fold (Wilcox test, P=3.33×10−9) 24 hours after transfection, consistent with a pro-apoptotic function independent of LIFR (Figure 6B). Induction of apoptosis by LIF6, however, was almost completely blocked by co-treatment with Z-VAD-FMK (Figure 6B) but not cyclosporine A (CsA) (Figure 6B). To test if LIF6 induced apoptosis is dependent upon Bax and Bak, we overexpressed LIF6 in Bax/Bak knockout mouse embryonic fibroblasts (MEFs) but did not observe an induction of apoptosis (Wilcox test, P=0.14; Figure 6C and Figure S5). In contrast LIF6 overexpression induced apoptosis in wild-type MEFs (Wilcox test, P=0. 3.10×10−4; Figure 6C and Figure S5). During apoptosis, collapse of the mitochondrial membrane potential (MMP) coincides with the opening of the mitochondrial transition pores, leading to the release of proapoptotic factors into the cytosol. Consistent with this mechanism, we found that LIF6 overexpression, treatment with DOX, or with nutlin-3a induced loss of MMP in CHO cells 48 hours after transfection (Wilcox test, P=7.40×10−7; Figure 6D). Thus LIF6 is sufficient to induce mitochondrial dysfunction and apoptosis mediated through Bax/Bak and independent of MPTP opening.
Elephant LIF6 is a refunctionalized pseudogene
We reasoned that most duplicate LIF genes are (likely) pseudogenes because elephant LIF6 is deeply nested within the duplicate LIF clade, is the only expressed duplicate, and is the only duplicate with a TP53 response element, suggesting elephant LIF6 re-evolved into a functional gene from a pseudogene ancestor. To test this hypothesis and reconstruct the evolutionary history of the LIF6 gene in the Proboscideans with greater phylogenetic resolution, we annotated the LIF6 locus in the genomes of Elephantids including the African Savannah elephant (Loxodonta africana), African Forest elephant (Loxodonta cyclotis), Asian elephant (Elephas maximus), woolly mammoth (Mammuthus primigenius), Columbian mammoth (Mammuthus columbi), and straight-tusked elephant (Palaeoloxodon antiquus), as well as the American Mastodon (Mammut americanum), an extinct Mammutid. We found that the genomes of each extinct Proboscidean contained a LIF6 gene with coding potential similar to the African and Asian elephant LIF6 genes as well as the TP53 binding-site, indicating that LIF6 evolved to be a TP53 target gene in the stem-lineage of Proboscideans.
While functional genes evolve under selective constraints that reduce their dN/dS (ω) ratio to below one, pseudogenes are generally free of such constraints and experience a relaxation in the intensity of purifying selection and an elevation in their dN/dS ratio. Therefore, we used a random effects branch-site model (RELAX) to test for relaxed selection on duplicate LIF genes compared to canonical LIF genes. The RELAX method fits a codon model with three ω rate classes to the phylogeny (null model), then tests for relaxed/intensified selection along lineages by incorporating a selection intensity parameter (K) to the inferred ω values; relaxed selection (both positive and negative) intensity is inferred when K<1 and increased selection intensity is inferred when K>1. As expected for pseudogenes, LIF duplicates (other than Proboscidean LIF6 genes) had significant evidence for a relaxation in the intensity of selection (K=0.36, LRT=42.19, P=8.26×10−11) as did the Proboscidean LIF6 stem-lineage (K=0.00, LRT=3.84, P=0.05). In contrast, Proboscidean LIF6 genes had significant evidence for selection intensification (K=50, LRT=4.46, P=0.03). We also found that the branch-site unrestricted statistical test for episodic diversification (BUSTED), which can detect gene-wide (not site-specific) positive selection on at least one site and on at least one branch, inferred a class of strongly constrained sites in (ω=0.00, 23.74%), a class of moderately constrained sites (ω=0.64, 75.85%), and a few sites that may have experienced positive selection in Proboscidean LIF6 genes (ω=10000.00, 0.41%; LRT=48.81, P≤0.001). These data are consistent with the reacquisition of constraints after refunctionalization.
Finally we inferred a Bayesian time-calibrated phylogeny of Atlantogenatan LIF genes, including LIF6 from African and Asian elephant, woolly and Columbian mammoth, straight-tusked elephant, and American Mastodon, to place upper and lower bounds on when the Proboscidean LIF6 gene may have refunctionalized (Figure 7A). We found that estimated divergence date of the Proboscideans LIF6 lineage was ~59 MYA (95% HPD: 61-57 MYA) whereas the divergence of Proboscideans was ~26 MYA (95% HPD: 23.28 MYA). These data indicate that the Proboscidean LIF6 gene refunctionalized during the evolutionary origin of large body sizes in this lineage, although precisely when within this time interval is unclear (Figure 7B). Thus LIF6 was reanimated sometime before the demands of maintaining a larger body existed in the Proboscidean lineage, suggesting LIF6 is permissive for the origin of large bodies but is not sufficient.
Discussion
Identifying the molecular and genetic mechanisms that underlie the evolution of enhanced cancer resistance in large and long-lived species is important for understanding how animals resolved Peto’s paradox and may provide a better understanding of cancer biology, but has been hampered by a lack of genomes for species that differ dramatically in body size yet are relatively recently diverged. Fortunately sequenced genomes are available for several living and extinct elephant species including the African Savannah elephant, African Forest elephant, Asian elephant, woolly and Columbian mammoths, straight-tusked elephant, the American Mastodon, an extinct Mammutid that diverged from the elephant lineage ~25 million years ago, as well as manatee and rock hyrax which are the closest living relatives of Proboscideans. This allowed us to use comparative genomic methods to identify genetic changes that occurred during the evolution of large bodies in the Proboscidean lineage, such as gene duplication and loss events.
A comprehensive analyses of genetic changes associated with the resolution of Peto’s paradox in the elephant lineage has yet to be performed, but candidate gene studies have identified functional duplicates of the master tumor suppressor TP53 as well as putative duplicates of other tumor suppressor genes (Abegglen et al., 2015; Caulin et al., 2015; Sulak et al., 2016). Caulin et al, for example, characterized the copy number of 830 tumor-suppressor genes (Higgins et al., 2007) across 36 mammals and identified 382 putative duplicates, including five copies of LIF in African elephants, seven in hyrax, and three in tenrec. Here we show that an incomplete duplication of the LIF gene in the Paenungulate stem-lineage generated a duplicate missing the proximal promoter and exon 1, generating a gene with similar structure to the LIF-T isoform (Haines et al., 1999), which functions as an intra-cellular pro-apoptotic protein independently from the LIFR-mediated signaling. Additional duplications of this original duplicate increased LIF copy number in Paenungulates, however, most LIF duplicates lack regulatory elements, are not expressed in elephant or hyrax fibroblasts (manatee cells or tissues are unavailable), and, with the exception of elephant LIF6, are likely pseudogenes.
While we are unable to do the kinds of reverse and forward genetic experiments that traditionally establish causal associations between genotypes and phenotypes, we were able to use primary African elephant and hyrax dermal fibroblasts to functionally characterize LIF duplicates. We found, for example, that the elephant LIF6 gene is transcribed at very low levels under basal conditions, but is up-regulated by TP53 in response to DNA damage. One of the constraints on the refunctionalization of pseudogenes is that they must evolve new cis-regulatory elements to direct their expression, but random DNA sequences can evolve into promoters with only a few substitutions suggesting de novo origination of regulatory elements may be common (Yona et al., 2017). There should be strong selection against the origin of constitutively active enhancers/promoters for pro-apoptotic pseudogenes, however, because their expression will be toxic. These results imply refunctionalizing LIF pseudogenes may impose a potential evolutionary cost. One of the ways to avoid that cost is through the gain of inducible regulatory elements that appropriately respond to specific stimuli, such as a TP53 signaling. Indeed our phylogenetic analysis indicates that a TP53 response element up-stream of LIF6 evolved before the divergence of mastodons and the modern elephant lineage, suggesting that LIF6 refunctionalized in the stem-lineage of Proboscideans coincident with the origin of large body sizes and thus may have been permissive for the large bodies.
The precise mechanisms by which mitochondrial dysfunction leads to apoptosis are uncertain, however, during early stages of apoptosis the pro-death Bcl-2 family members Bax and Bak hetero- and homo-oligomerize within the mitochondrial outer membrane leading to permeabilization (MOMP) and the release of pro-apoptotic protein such as cytochrome c (Karch et al., 2013; 2015). In contrast, during necrosis the collapse of the MMP and the opening of the mitochondrial permeability transition pore (MPTP) leads to mitochondrial swelling, rupture, and cell death (Ly et al., 2003). Our observations that cyclosporine A (CsA) did not inhibit LIF6 induced apoptosis, and that LIF6 overexpression did not induce apoptosis in Bax/Bak null MEFs suggests that LIF6 functions in a manner analogous to the pro-apoptotic Bcl-2 family members by inducing the opening of the outer mitochondrial membrane pore. Furthermore our observation that LIF6 overexpression incudes apoptosis in elephant dermal fibroblasts, Chinese hamster ovary cells, and mouse embryonic fibroblasts indicates the LIF6 mechanism of action is neither of cell-type nor species specific. The molecular mechanisms by which LIF6 induces apoptosis, however, are unclear and the focus of continued studies.
Experimental Procedures
Identification of LIF genes in Mammalian genomes
We used BLAT to search for LIF genes in 53 Sarcopterygian genomes using the human LIF protein sequences as an initial query. After identifying the canonical LIF gene from each species, we used the nucleotide sequences corresponding to this LIF CDS as the query sequence for additional BLAT searches within that species genome. To further confirm the orthology of each LIF gene we used a reciprocal best BLAT approach, sequentially using the putative CDS of each LIF gene as a query against the human genome; in each case the query gene was identified as LIF. Finally we used the putative amino acid sequence of the LIF protein as a query sequence in a BLAT search.
We thus used BLAT to characterize the LIF copy number in Human (Homo sapiens; GRCh37/hg19), Chimp (Pan troglodytes; CSAC 2.1.4/panTro4), Gorilla (Gorilla gorilla gorilla; gorGor3.1/gorGor3), Orangutan (Pongo pygmaeus abelii; WUGSC 2.0.2/ponAbe2), Gibbon (Nomascus leucogenys; GGSC Nleu3.0/nomLeu3), Rhesus (Macaca mulatta; BGI CR_1.0/rheMac3), Baboon (Papio hamadryas; Baylor Pham_1.0/papHam1), Marmoset (Callithrix jacchus; WUGSC 3.2/calJac3), Squirrel monkey (Saimiri boliviensis; Broad/saiBol1), Tarsier (Tarsius syrichta; Tarsius_syrichta2.0.1/tarSyr2), Bushbaby (Otolemur garnettii; Broad/otoGar3), Mouse lemur (Microcebus murinus; Broad/micMur1), Chinese tree shrew (Tupaia chinensis; TupChi_1.0/tupChi1), Squirrel (Spermophilus tridecemlineatus; Broad/speTri2), Mouse (Mus musculus; GRCm38/mm10), Rat (Rattus norvegicus; RGSC 5.0/rn5), Naked mole-rat (Heterocephalus glaber; Broad HetGla_female_1.0/hetGla2), Guinea pig (Cavia porcellus; Broad/cavPor3), Rabbit (Oryctolagus cuniculus; Broad/oryCun2), Pika (Ochotona princeps; OchPri3.0/ochPri3), Kangaroo rat (Dipodomys ordii; Broad/dipOrd1), Chinese hamster (Cricetulus griseus; C_griseus_v1.0/criGri1), Pig (Sus scrofa; SGSC Sscrofa10.2/susScr3), Alpaca (Vicugna pacos; Vicugna_pacos-2.0.1/vicPac2), Dolphin (Tursiops truncatus; Baylor Ttru_1.4/turTru2), Cow (Bos taurus; Baylor Btau_4.6.1/bosTau7), Sheep (Ovis aries; ISGC Oar_v3.1/oviAri3), Horse (Equus caballus; Broad/equCab2), White rhinoceros (Ceratotherium simum; CerSimSim1.0/cerSim1), Cat (Felis catus; ICGSC Felis_catus 6.2/felCat5), Dog (Canis lupus familiaris; Broad CanFam3.1/canFam3), Ferret (Mustela putorius furo; MusPutFur1.0/musFur1), Panda (Ailuropoda melanoleuca; BGI-Shenzhen 1.0/ailMel1), Megabat (Pteropus vampyrus; Broad/pteVam1), Microbat (Myotis lucifugus; Broad Institute Myoluc2.0/myoLuc2), Hedgehog (Erinaceus europaeus; EriEur2.0/eriEur2), Shrew (Sorex araneus; Broad/sorAra2), Minke whale (Balaenoptera acutorostrata scammoni; balAcu1), Bowhead Whale (Balaena mysticetus; v1.0), Rock hyrax (Procavia capensis; Broad/proCap1), Sloth (Choloepus hoffmanni; Broad/choHof1), Elephant (Loxodonta africana; Broad/loxAfr3), Cape elephant shrew (Elephantulus edwardii; EleEdw1.0/eleEdw1), Manatee (Trichechus manatus latirostris; Broad v1.0/triMan1), Tenrec (Echinops telfairi; Broad/echTel2), Aardvark (Orycteropus afer afer; OryAfe1.0/oryAfe1), Armadillo (Dasypus novemcinctus; Baylor/dasNov3), Opossum (Monodelphis domestica; Broad/monDom5), Tasmanian devil (Sarcophilus harrisii; WTSI Devil_ref v7.0/sarHar1), Wallaby (Macropus eugenii; TWGS Meug_1.1/macEug2), and Platypus (Ornithorhynchus anatinus; WUGSC 5.0.1/ornAna1).
Phylogenetic analyses and gene tree reconciliation of Paenungulate LIF genes
The phylogeny of LIF genes was estimated using an alignment of the LIF loci from the African elephant, hyrax, manatee, tenrec, and armadillo genomes and BEAST (v1.8.3) (Rohland et al., 2010). We used the HKY85 substitution, which was chosen as the best model using HyPhy, empirical nucleotide frequencies (+F), a proportion of invariable sites estimated from the data (+I), four gamma distributed rate categories (+G), an uncorrelated random local clock to model substitution rate variation across lineages, a Yule speciation tree prior, uniform priors for the GTR substitution parameters, gamma shape parameter, proportion of invariant sites parameter, and nucleotide frequency parameter. We used an Unweighted Pair Group Arithmetic Mean (UPGMA) starting tree. The analysis was run for 10 million generations and sampled every 1000 generations with a burn-in of 1000 sampled trees; convergence was assessed using Tracer, which indicated convergence was reached rapidly (within 100,000 generations). We used Notung v2.6 (Chen et al., 2000) to reconcile the gene and species trees.
Gene expression data (Analyses of RNA-Seq data and RT-PCR)
To determine if duplicate LIF genes were basally transcribed, we assembled and quantified elephant LIF transcripts with HISAT2 (Kim et al., 2015) and StringTie (Pertea et al., 2015) using deep 100bp paired-end RNA-Seq data (>138 million reads) we previously generated from Asian elephant dermal fibroblasts (Sulak et al., 2016), as well as more shallow (~30 million reads) singe-end sequencing from African elephant dermal fibroblasts (Cortez et al., 2014) and placenta (Sulak et al., 2016), and Asian elephant peripheral blood mononuclear cells (PBMCs) (Reddy et al., 2015). HISAT2 and StringTie were run on the Galaxy web-based platform (https://usegalaxy.org) (Afgan et al., 2016) using default settings, and without a guide GTF/GFF file.
We determined if LIF transcription was induced by DNA damage and p53 activation in African elephant Primary fibroblasts (San Diego Frozen Zoo) using RT-PCR and primers designed to amplify elephant duplicate LIF genes, including LIF1-F: 5’-GCACAGAGAAGGACAAGCTG-3’, LIF1-R: 5’-CACGTGGTACTTGTTGCACA-3’, LIF6-F: 5’-CAGCTAGACTTCGTGGCAAC-3’, LIF6-R: 5’-AGCTCAGTGATGACCTGCTT-3’, LIF3-R: 5’-TCTTTGGCTGAGGTGTAGGG-3’, LIF4-F: 5’-GGCACGGAAAAGGACAAGTT-3’, LIF4-R: 5’-GCCGTGCGTACTTTATCAGG-3’, LIF5-F: 5’-CTCCACAGCAAGCTCAAGTC-3’, LIF5-R: 5’-GGGGATGAGCTGTGTGTACT-3’. We also used primers to elephant BAX to determine if it was up-regulated by TP53: BAX-F: 5’-CATCCAGGATCGAGCAAAGC-3’, BAX-R: 5’-CCACAGCTGCAATCATCCTC-3’. African elephant Primary fibroblasts were grown to 80% confluency in T-75 culture flasks at 37°C/5% CO2 in a culture medium consisting of FGM/EMEM (1:1) supplemented with insulin, FGF, 6% FBS and Gentamicin/Amphotericin B (FGM-2, singlequots, Clonetics/Lonza). At 80% confluency, cells were harvested and seeded into 6-well culture plates at ~10,000 cells/well. Once cells recovered to 80% confluency they were treated with either vehicle control, 50um Doxorubicin, or 50um Nutlin-3a.
Total RNA was extracted using the RNAeasy Plus Mini kit (Qiagen), then DNase treated (Turbo DNA-free kit, Ambion) and reverse-transcribed using an olgio-dT primer for cDNA synthesis (Maxima H Minus First Strand cDNA Synthesis kit, Thermo Scientific). Control RT reactions were otherwise processed identically, except for the omission of reverse transcriptase from the reaction mixture. RT products were PCR-amplified for 45 cycles of 94°/20 seconds, 56°/30 seconds, 72°/30 seconds using a BioRad CFX96 Real Time qPCR detection system and SYBR Green master mix (QuantiTect, Qiagen). PCR products were electrophoresed on 3% agarose gels for 1 hour at 100 volts, stained with SYBR safe, and imaged in a digital gel box (ChemiDoc MP, BioRad) to visualize relative amplicon sizes.
Statistical methods
We used a Wilcox or T-test test for all statistical comparisons, with at least four biological replicates. The specific statistical test used and number replicates for each experiment are indicated in figure legends.
Luciferase assay and cell culture
We used the JASPAR database of transcription factor binding site (TFBS) motifs (Mathelier et al., 2015) to computationally predict putative TFBSs within a 3kb window around Atlantogenatan LIF genes and identified matches for the TP53 motif (MA0106.3), including a match (sequence: CACATGTCCTGGCAACCT, score: 8.22, relative score: 0.82) ~1kb upstream of the African elephant LIF6 start codon. To test if the putative p53 binding site upstream of elephant LIF6 was a functional p53 response element, we synthesized (GeneScript) and cloned the −1100bp to +30bp region of the African elephant LIF6 gene (loxAfr3_dna range=scaffold_68:4294134-4295330 strand=+ repeatMasking=none) and a mutant lacking the CACATGTCCTGGCAACCT sequence into the pGL3-Basic[minP] luciferase reporter vector.
African elephant Primary fibroblasts (San Diego Frozen Zoo) were grown to 80% confluency in T-75 culture flasks at 37°C/5% CO2 in a culture medium consisting of FGM/EMEM (1:1) supplemented with insulin, FGF, 6% FBS and Gentamicin/Amphotericin B (FGM-2, singlequots, Clonetics/Lonza). At 80% confluency, 104 cells were harvested and seeded into 96-well white culture plates. 24 hours later cells were transfected using Lipofectamine LTX and either 100g of the pGL3-Basic[minP], pGL3-Basic[minP] −1100bp to +30bp, pGL3-Basic[minP] - 1100bp-+30bp Δp53TFBS luciferase reporter vectors and 1ng of the pGL4.74[hRluc/TK] Renilla control reporter vector according the standard protocol with 0.5 ul/well of Lipofectamine LTX Reagent and 0.1ul/well of PLUS Reagent. 24 hours after transfection cells were treated with either vehicle control, 50um Doxorubicin, or 50um Nutlin-3a. Luciferase expression was assayed 48 hours after drug treatment, using the Dual-Luciferase Reporter Assay System (Promega) in a GloMax-Multi+ Reader (Promega). For all experiments luciferase expression was standardized to Renilla expression to control for differences transfection efficiency across samples; Luc./Renilla data is standardized to (Luc./Renilla) expression in untreated control cells. Each luciferase experiment was replicated three independent times, with 8-16 biological replicates per treatment and control group.
ChIP-qPCR and cell culture
African elephant Primary fibroblasts were grown to 80% confluency in T-75 culture flasks at 37°C/5% CO2 in a culture medium consisting of FGM/EMEM (1:1) supplemented with insulin, FGF, 6% FBS and Gentamicin/Amphotericin B (FGM-2, singlequots, Clonetics/Lonza). 104 cells were seeded into each well of 6-well plate and grown to ~80% confluency. Cells were then treated with either a negative control siRNA or equimolar amounts of a combination of three siRNAs that specifically target the canonical TP53 transcript using Lipofectamine LTX according to the suggested standard protocol. The next day, cells were treated with either water, DMSO, 50uM Doxorubicin, or 50uM Nutlin-3a in three biological replicates for each condition. After 18 hrs of incubation with each drug, wells were washed three times with ice cold PBS and PBS replaced with fresh media, and chromatin cross linked with 1% fresh formaldehyde for 10 minutes. We used The MAGnify Chromatin Immunoprecipitation System (ThermoFischer #492024) to perform chromatin immunoprecipitation according to the suggested protocol. However rather than shearing chromatin by sonication, we used the ChIP-It Express Enzymatic Shearing Kit (Active Motif # 53009) according to the suggested protocol. Specific modifications to the MAGnify Chromatin Immunoprecipitation System included using 3ug of the polyclonal TP53 antibody (FL-393, lot #DO215, Santa Cruz Biotechnology).
We used qPCR to assay for enrichment of TP53 binding from the ChIP-Seq using the forward primer 5’-TGGTTTCCAGGAGTCTTGCT-3’ and the reverse primer 5’-CATCCCCTCCTTCCTCTGTC-3’. 100ng of ChIP DNA was used per PCR reaction, which was amplified for 45 cycles of 94°/20 s, 56°/30 s, 72°/30 s using a BioRad CFX96 Real Time qPCR detection system and SYBR Green master mix (QuantiTect, Qiagen). Data are shown as fold increase in TP53 ChIP signal relative to the background rabbit IgG ChIP signal and standardized to the control water for DOX or DMSO for nutlin-3a treatments.
ApoTox-Glo Viability/Cytotoxicity/Apoptosis experiments
T75 culture flasks were seeded with 200,000 African Elephant primary fibroblasts, and grown to 80% confluency at 37°C/5% CO2 in a culture medium consisting of FGM/EMEM (1:1) supplemented with insulin, FGF, 6% FBS and Gentamicin/Amphotericin B (FGM-2, singlequots, Clonetics/Lonza). 5000 cells were seeded into each well of two opaque-bottomed 96-well plates. In each plate, half of the columns in the plate were transfected with pcDNA3.1/LIF6/eGFP (GenScript) using Lipofectamine LTX (Thermo Scientific 15338100); the other half were mock transfected with the same protocol without any DNA. In the plate designated for the 18hr timepoint, each column was treated with either: 50uM (-)-Nutlin-3 (Cayman 18585); 20uM Z-VAD-FMK (Cayman 14463); 2uM Cyclosporin A (Cayman 12088); 50uM Doxorubicin (Fisher BP251610); DMSO (Fisher BP231100); or DPBS (Gibco 14190136). For the 24hr timepoint, the same schema for treatment was used, but with half-doses. Each treatment contained eight biological replicates for each condition. After 18 hrs of incubation with each drug, cell viability, cytotoxicity, and Caspase-3/7 activity were measured using the ApoTox-Glo Triplex Assay (Promega) in a GloMax-Multi+ Reader (Promega). Z-VAD-FMK readings were normalized to the PBS-treated, mock-transfected cells; all others were normalized to the DMSO-treated, mock-transfected cells.
T75 culture flasks were seeded with 250,000 wild-type (ATCC CRL-2907) and Bak/Bax double knockout (ATCC CRL-2913) mouse embryonic fibroblasts (MEFs), or Chinese hamster ovary cells (CHO-K1, Thermo R75807) and allowed to grow to 80% confluency at 37°C/5% CO2 in a culture medium consisting of high-glucose DMEM (Gibco) supplemented with GlutaMax (Gibco), Sodium pyruvate (Gibco), 10% FBS (Gibco), and Penicillin-Streptomycin (Gibco). 3000 cells were seeded into each well of an opaque, bottomed 96-well plate. Half of the columns in the plate were transfected with pcDNA3.1/LIF6/eGFP (GenScript) using Lipofectamine LTX (ThermoFisher Scientific 15338100); the other half were mock transfected with the same protocol without any DNA. 6 hours post-transfection, the transfection reagents and media from each well was replaced: for the 24-hour timepoint, drug-supplemented media was placed within the wells; for the 48-hour timepoint, untreated media was placed in the wells, and then replaced with treatment media 24-hours later. Each column was treated with either: 50uM (-)-Nutlin-3 (Cayman 18585); 20uM Z-VAD-FMK (Cayman 14463); 2uM Cyclosporin A (Cayman 12088); 50uM Doxorubicin (Fisher BP251610); DMSO (Fisher BP231100); or DPBS (Gibco 14190136). Each treatment contained eight biological replicates for each condition. After 18 hrs of incubation with each drug, cell viability, cytotoxicity, and Caspase-3/7 activity were measured using the ApoTox-Glo Triplex Assay (Promega) in a GloMax-Multi+ Reader (Promega). Z-VAD-FMK readings were normalized to the PBS-treated, mock-transfected cells; all others were normalized to the DMSO-treated, mock-transfected cells.
For knockdown experiments T75 culture flasks were seeded with 200,000 African Elephant primary fibroblasts, and grown to 80% confluency at 37°C/5% CO2 in a culture medium consisting of FGM/EMEM (1:1) supplemented with insulin, FGF, 6% FBS and Gentamicin/Amphotericin B (FGM-2, singlequots, Clonetics/Lonza). 5000 cells were seeded into each well of two opaque-bottomed 96-well plates. In each plate, pairs of rows were transfected with either Silencer™ Select Negative Control No. 1 siRNA (Thermo 4390843), P53 siRNA (Dharmacon) (Sulak et al. 2016), and either with or without pcDNA3.1/LIF6/eGFP (GenScript) using Lipofectamine LTX (Thermo Scientific 15338100). In the plate designated for the 18hr timepoint, each column was treated with either: 50uM Doxorubicin (Fisher BP251610); or an equivalent dilution of Ethanol (Fisher BP2818100. For the 24hr timepoint, the same schema for treatment was used, but with half-doses. Each treatment contained eight biological replicates for each condition. After 18 hrs of incubation with each drug, cell viability, cytotoxicity, and Caspase-3/7 activity were measured using the ApoTox-Glo Triplex Assay (Promega) in a GloMax-Multi+ Reader (Promega). All data were normalized to the ethanol-treated scrambled siRNA control samples. siRNAs were designed to specifically-target the elephant LIF6 gene. Sequences of the three LIF6-specific siRNAs used are as follows: 1) 5’-GAAUAUACCUGGAGGAAUGUU-3’, 2) 5’-GGAAGGAGGCCAUGAUGAAUU-3’, 3) 5’-CACAAUAAGACUAGGAUAUUU-3’ (Dharmacon). We also validated efficiency of the knockdown via qRT-PCR using the primer sets described earlier, which specifically the LIF6 gene, and confirmed the combination of all three LIF6 siRNAs was ~88%.
To determine if LIF6 was sufficient to induce apoptosis we synthesized and cloned (GeneScript) the African elephant LIF6 gene into the pcDNA3.1+C-DYK expression vector, which adds at DYK epitope tag immediately C-terminal to the LIF6 protein. We transiently transfected Chinese hamster ovary (CHO) cells or MEFs with LIF6_pcDNA3.1+C-DYK expression vector using Lipofectamine LTX according to manufacturer protocol and as described above, and assayed cell viability, cytotoxicity, and the induction of apoptosis using an ApoTox-Glo triplex assay. Mitochondrion membrane potential was assayed in CHO cells using the fluorometric Mitochondrion Membrane Potential Kit (Sigma MAK147) 48 hours after transfection.
Evolutionary analyses of LIF genes
We used a Bayesian approach to date LIF duplication events implemented in BEAST (v1.8.3) (Rohland et al., 2010), including all identified African elephant, hyrax, and manatee LIF duplicates, as well as cannonical LIF genes from armadillo, sloth, aardvark, golden mole, and LIF6 genes from Asian elephant, woolly and Columbian mammoth, straight-tusked elephant, and American Mastodon (Palkopoulou et al., 2018). We used the GTR substitution, which was chosen as the best model using HyPhy, empirical nucleotide frequencies (+F), a proportion of invariable sites estimated from the data (+I), four gamma distributed rate categories (+G) with the shape parameter estimated from the data, an uncorrelated random local clock to model substitution rate variation across lineages, a Yule speciation tree prior, uniform priors for the GTR substitution parameters, gamma shape parameter, proportion of invariant sites parameter, and nucleotide frequency parameter. We used an Unweighted Pair Group Arithmetic Mean (UPGMA) starting tree. The analysis was run for 10 million generations and sampled every 1000 generations with a burn-in of 1000 sampled trees; convergence was assessed using Tracer, which indicated convergence was reached rapidly (within 100,000 generations).
To constrain nodes we used normal priors with estimated confidence intervals, the root node was constrained to be 105 MYA, the root of Xenarthra was constrained to be 66 MYA, the root of Afrosoricida was constrained to be 70 MYA, the root of Afrosoricida-Macroselidea divergence constrained to be 75 MYA, the Elephantidea root was constrained to be 7.5 MYA, the Afrotheria root was constrained to be 83 MYA, the Paeungulata root was constrained to be 68 MYA, and the Proboscidea root was constrained to be 16 MYA. Divergence dates were obtained from www.timetree.org using the ‘Expert Result’ divergence dates.
We used the RELAX method to (Wertheim et al., 2015) test if duplicate LIF genes experienced a relaxation of the intensity of selection using the DataMonkey web server (Delport et al., 2010). The alignment included all duplicate LIF genes identified in the African elephant, hyrax, and manatee genomes, as well as cannonical LIF genes from armadillo, sloth, aardvark, golden mole, and LIF6 genes from Asian elephant, woolly and Columbian mammoth, straight-tusked elephant, and American Mastodon. Alignment confidence was assessed using GUIDANCE2 (Sela et al., 2015) with the MAFFT (Katoh et al., 2005) algorithm and 100 bootstrap replicates.
Author contributions
Writing - Original Draft: V.J.L and J.M.V; Writing - Review & Editing: V.J.L and J.M.V; Conceptualization: V.J.L and J.M.V; Investigation: J.M.M, M.S., S.C., and V.J.L.; Methodology: J.M.M, M.S.; Formal Analysis: V.J.L and J.M.V; Supervision and Project Administration: V.J.L.
Declaration of interests
The authors declare no competing interests.
Acknowledgements
The authors would like to thank E. Palkopoulou and D. Reich (Harvard Medical School), the Broad Institute, M. Hofreiter (University of Potsdam) and H. Poinar (McMaster University) for providing access to the Proboscidean genomes, and U.M. Moll for providing the Bax/Bak KO MEF cell line.