A detailed understanding of the molecular mechanisms capable of driving regenerative growth in vertebrates may provide important insights into the treatment of diverse human diseases. Because traditional vertebrate model systems offer limited insight into natural organ regenerative processes, non-traditional model systems, including snakes in general and Burmese pythons in particular, hold great potential for providing unique insights into vertebrate regenerative organ growth processes. In this study we have found that multiple integrated growth pathways, in addition to multiple stress-response pathways, appear to underlie the coordinated organ regenerative process in Burmese pythons upon feeding. Despite distinct patterns of gene expression associated with growth for each organ, pathway and upstream regulatory molecule analyses reveal substantial similarities in pathways associated with post-feeding, extreme-growth responses across multiple organs. Specifically, we found evidence for a consistent interactive role of three major types of pathways underlying growth responses in python organs following feeding, including the related growth pathways mTOR and PI3K/AKT, lipid-signaling pathways such as PPAR and LXR/RXR, and stress-response/cell-protective pathways including NRF2.
mTOR and other growth pathways underlying organ growth
Across the four organs examined, we found evidence for the involvement of the mTOR signaling pathway as a key integrator of growth signals underlying post-feeding regenerative organ growth. This pathway integrates processes for the use of energy and nutrients to regulate growth and homeostasis [30]. mTOR interacts with multiple other pathways, including PI3K/AKT, several lipid metabolism and signaling pathways [30, 31], and the NRF2-mediated oxidative stress response [32, 33] – all of which are also active in multiple organs during growth (Figs. 3–5). mTOR complex 1 (mTORC1) is the most well-characterized of the two mTOR complexes and integrates signaling from growth factors, energy status, oxygen, and amino acids to promote cell growth when activated [31]. The TSC1/2 complex transmits upstream signals from growth factor and insulin signaling to modulate the activity of mTORC1 and its interaction with other pathways including PI3K/AKT [30, 31, 34]. The effector kinases of these external pathways inactivate the TSC complex through phosphorylation, thus, indirectly activating mTORC1 [30, 31]. AKT can also directly activate mTORC1 through phosphorylation of an mTORC1 inhibitor. In a low energy state, AMPK inhibits mTORC1 by phosphorylating regulatory associated protein of mTORC1 (RAPTOR) [30, 31]. mTORC2 signaling is less well-understood, but is known to respond to growth factors through PI3K signaling [30].
CPA of gene expression in the first 24 h after feeding indicate that involvement of the mTOR signaling pathway is significant in the small intestine (predicted activation), but insignificant in both the heart (predicted activation) and kidney (activation state undetermined). The liver lacked evidence of involvement of the mTOR signaling pathway from CPA (Figs. 3–4). In URM analysis, the mTOR molecule itself was predicted to be downregulated in the heart, liver, and intestine with no presence in the kidney, which contrasts our CPA results (Figs. 3–4). However, URMA-predicted activation of the mTORC1 complex is supported in both the kidney and small intestine with undefined involvement in the heart, and the liver shows no signal for mTORC1 (Fig. 4). Interestingly, CPA indicate mTORC1 is downregulated in the small intestine at 0–1DPF (Fig. 6), yet this downregulated state of mTORC1 is based only on the downregulation of a single gene, G protein subunit beta 1 like (GNB1L), which IPA identifies as a subunit of the mTORC1 complex. In contrast, AMPK signaling is predicted to be downregulated in the kidney and small intestine, indicative of elevated ATP levels and active mTORC1 [30, 31] (Fig. 3). It is notable that nearly all genes in the mTOR pathway were associated with python orthologs that were observed as expressed across our dataset (see Additional file 1: Table S2), which suggests that our inferences of non-responsive genes within the mTOR pathway are biologically meaningful (e.g., true negatives), rather than representative of a lack of data. Thus, mTOR signaling in python tissues during regenerative organ growth may include non-canonical features compared to typical models of mTOR signaling that account for the partial responsiveness of genes and targets inferred from our CPA.
Our results identify mTOR as a central regulator and integrator of a number of diverse growth signals that drive post-feeding regenerative organ growth in Burmese pythons. Insulin signaling represents a key-regulating factor of the mTOR pathway [31], and we found multiple lines of evidence indicating roles of insulin signaling in post-feeding growth responses. Specifically, 0–1DPF URMA inferred the activation of INSR and insulin, and the inhibition of INSIG1 and INSIG2, in the kidney, small intestine, and liver, and the inverse of these activation patterns in the heart. INSIG1 and INSIG2 are negative regulators of SCAP [35, 36], which in turn regulates SREBP activity. Consistent with inferences of inhibition of INSIG1–2, URMA predicted the upregulation of SREBF1 and SREBF2, which provide evidence of an increase in sterol-regulatory element activity coincident with organ growth [36, 37] (Fig. 4). In addition to the interaction of insulin signaling and mTOR activity, we also found multiple lines of evidence for PI3K/AKT signaling that would interact with mTOR. Our URMA indicates significant downregulation of PTEN, an upstream regulator of the PI3K/AKT pathway, across all four organs, and CPA predicts activation of the PI3K/AKT signaling pathway in the small intestine and liver.
Evidence from previous studies also support the role of mTOR, PI3K/AKT, and AMPK signaling mechanisms in python post-feeding growth, at least in the heart. Western blots of python cardiac tissue post-feeding support the inference of early activation of mTOR and PI3K/AKT pathways by demonstrating that phosphorylated AKT and MTOR proteins increase significantly in abundance between 12 and 24 h post-feeding [11]. These western blots also demonstrated phosphorylated AMPK protein was upregulated within 24 h post-feeding, but lagging temporally behind the peak in phosphorylated MTOR and AKT [11], consistent with the antagonistic relationship between AMPK and MTOR/AKT [30]. These independent lines of evidence for the roles of mTOR, PI3K/AKT, and AMPK signaling in python post-feeding organ growth confirm our inferences of the central roles of these pathways, and support the power of pathway and URM inferences for inferring signaling mechanisms.
MAPK and related pathways also appear to be prominently involved in organ growth responses post-feeding, which is sensible given their known interactions with multiple growth pathways, including PI3K/AKT signaling and mTOR [38,39,40]. Our data reveal the involvement of MAPK signaling most clearly in the heart, with significant enrichment and predicted inhibition of p38 MAPK signaling and significant activation of ERK5 signaling (Fig. 3). ERK5 is a member of the Mitogen-activated protein kinases (MAPKs) that is crucial to cell proliferation and activated in response to growth factors and oxidative stress [41, 42]. MYC is a downstream transcription factor regulated by the MAPK pathway and ERK5 specifically [43, 44], and an essential regulator of development and cell proliferation [45,46,47]. Our URMA predict significant activation of MYC in all four organs, indicating a broad role of active MAPK signaling in post-feeding organ growth in the python.
NRF2 – protective function and interaction with growth pathways
One of the strongest and most consistent signals in the canonical pathway and upstream regulatory molecule analyses was the involvement of the NRF2-mediated oxidative stress response pathway. Commonly associated with anti-aging and longevity [48,49,50], injury repair, and mitigation of inflammation [51], evidence for the central involvement of the NRF2-mediated oxidative stress response pathway in the small intestine, liver, and kidney begs the question of whether there is an important yet largely unappreciated role for stress-response signaling pathways in growth responses, and regenerative organ growth in particular.
The NRF2 pathway was significantly upregulated in small intestine, kidney, and liver within the first day following feeding (Fig. 3), and the NRF2 transcription factor (NFE2L2) was one of the most significant and highest in magnitude URMs predicted in these three organs (p-values < 1.55e-10, z-scores > 3.0) (Fig. 4). The 24 h period following feeding in Burmese pythons involves unparalleled rates and magnitudes of organ growth, and also includes massive upregulation of metabolism – up to 44-fold increases in aerobic metabolism, which is among the highest fluctuation known for any vertebrate [3]. It is, therefore, sensible that activation of NRF2 is related to these major shifts in oxidative metabolism, and associated generation of reactive oxygen species [1, 2, 6, 9]. An open question, however, is what broader role the activation of NRF2 may play in facilitating the extraordinary growth responses associated with feeding in pythons. For example, post-fed Burmese python blood plasma has been shown to convey resistance to apoptosis to mammalian cells, even with exposure to high fatty acid concentrations that would otherwise cause cell death [11, 17]; such cell-protective qualities may be related to signals that activate NRF2 and/or other stress-response pathways. Interestingly, in addition to cell-protective roles of NRF2, this pathway also contains multiple points of integration with various growth pathways, including those activated in python organ regenerative growth.
The NRF2-mediated oxidative stress response pathway interacts with multiple pathways predicted in our canonical pathway analysis [52,53,54,55,56,57] (Figs. 3–4). The PI3K/AKT signaling pathway, predicted to be upregulated upon feeding in both the liver and small intestine, is essential for regulating the antioxidant functions of NRF2, and studies have shown that inhibition of this signaling pathway leads to attenuation of NRF2 activities [58, 59]. This interaction is evident when examining the role of NRF2 in the proliferation of cancer cells. Studies have shown that NRF2 is able to redirect glucose and glutamine into anabolic pathways through activation of PI3K/AKT signaling [60]. The activated PI3K/AKT pathway leads to greater accumulation of NRF2 in the nucleus, which allows NRF2 to enhance metabolic activities as well as promote cell proliferation and cytoprotection [60]. The PI3K/AKT signaling pathway activates mTOR activity in response to growth factors, and this and previous studies [11] have shown that PI3K/AKT and mTOR signaling are key growth pathways underlying organ regenerative growth in the Burmese python. Therefore, there appears to be strong and coordinated links between growth signaling (via PI3k/AKT and mTOR) and stress response signaling via NRF2 underlying organ growth in pythons following feeding. Like mTOR, a large majority of genes in the NRF2 pathway were associated with python orthologs and were observed as expressed across our dataset (see Additional file 1: Table S2), which indicates that our inferences of non-responsive genes within the NRF2 pathway are likely true negatives, rather than artifacts due to a lack of ortholog identification in the python. Accordingly, predicted but unobserved expression responses in the NRF2 pathway in pythons suggest that the absence of expected responses may represent novel or non-canonical aspects of python biology or of the organ regeneration response in pythons.
In addition to NRF2-mediated oxidative stress response, evidence for the involvement of other stress response signaling mechanisms in python post-feeding organ growth was also observed. EIF2 signaling, important in translational control and responsiveness to conditions of environmental stress [61, 62], is strongly downregulated in the intestine, yet, absent in the other three organs (Fig. 3). Acute phase response signaling, which is involved in restoring homeostasis following inflammation or injury [63], is predicted to be strongly downregulated in the liver and moderately upregulated (but non-significant in the heart; Fig. 3). The precise roles of these additional stress response mechanisms in regenerative organ growth in the python remains an open question, although there is strong and consistent signal for the involvement of multiple stress response pathways overall in python post-feeding organ growth.
Role of lipid signaling in driving growth
Previous studies have shown evidence that molecules responsible for triggering python post-feeding organ growth circulate in the blood of the Burmese python [11, 64]. Riquelme et al. demonstrated that post-feeding python plasma was capable of inducing cardiomyocyte growth in pythons and mice, and that fasted python plasma supplemented with three particular fatty acids successfully stimulated cardiomyocyte growth in mice [11]. Because these fatty acids only facilitated a growth response in the presence of fasted Burmese python serum, it is likely that python plasma contains additional factors required for successful post-feeding regenerative growth and that fatty acids are only partially responsible for stimulating growth responses. In the heart, we found significant enrichment and predicted activation for the LXR/RXR activation pathway as well as predicted activation of this pathway (although insignificant enrichment with P > 0.01) in the small intestine (Fig. 3). LXR is a potent activator of the SREBP-1c gene [65], and our data predict clear and significant activation of both SREBF1 and SREBF2 upon feeding in the kidney, liver, and small intestine with significant downregulation and undefined direction for SREBF1 and SREBF2 in the heart, respectively (Fig. 4). When activated, these proteins directly enhance genes important for the uptake and synthesis of various lipids. SCAP, important for the activation of these SREB molecules, is also predicted to be strongly activated in the kidney, liver, and small intestine (Fig. 4) [35, 36, 66].
We also examined PPAR signaling as a potential pathway for lipid signaling during this regenerative growth, given the central role of PPAR in mediating fatty acid signaling as well as its effects on gene expression [67]. PPAR has also been identified as an important regulator of cell survival during wound repair and regeneration [68]. Although CPA did not detect significant PPAR signaling activation, URMA significantly predicted PPARA, PPARG, PPARGC1A, and PPARGC1b involvement across organs, typically inhibited in the heart and activated in the other three organs in 0–1DPF comparisons (Fig. 4). Given the variations in pathway and URM inferences between the heart and the other three organs, the question of whether fatty acids also play a similar stimulatory role in regenerative growth in the small intestine, liver, and kidney as they do in the heart remains. Our results do, however, argue for a poorly understood yet central role of lipid-signaling in these growth responses, and suggest that the unusually strong bioactivity of fatty acids may elicit growth through conserved canonical pathway signaling mechanisms.
Early phases of organ regression following digestion
Physiological studies have shown that python post-feeding organ growth peaks between 1DPF and 3DPF [1, 2, 5, 9] and that phenotypes begin to decline by 4DPF [2, 3, 7, 9]. Thus, as post-feeding growth phenotypes reverse from 1DPF to 4DPF, we expected to observe shifts towards the fasted state, such as the reversal or inhibition of growth-associated pathways. Relative to comparisons between fasting and 1DPF, comparisons between 1DPF and 4DPF yielded nearly an order of magnitude fewer significantly differentially expressed genes (Table 1). Accordingly, expression heatmaps (Fig. 2) and expression profile summaries (see Additional file 1: Figure S1) show that expression profiles of many genes at 4DPF tend to remain elevated (i.e., similar to levels at 1DPF), or exist at intermediate levels (between fasted and 1DPF levels of expression). We did not observe any particularly informative trends in canonical pathways and upstream regulator molecule predictions (see Additional file 1: Figure S7) associated with shifts in gene expression from 1DPF to 4DPF, and this result is not surprising given the relatively small number of genes that significantly change between these time points. Among the predicted pathways were several that are related to stress response and biosynthesis (see Additional file 1: Figure S7), although a lack of predicted direction of activation prevents detailed interpretation of the involvement of nearly all pathways predicted between 1DPF and 4DPF. The only pathway predicted as significant and with a direction of activation between 1DPF and 4DPF was the mitotic roles of polo-like kinase pathway, which was activated in the small intestine (see Additional file 1: Figure S7). It therefore remains an open question whether atrophy and other processes involved in reverting to the fasting state are controlled actively (via a new signal that stimulates the apoptotic and atrophy processes), passively (the signal (s) that stimulates the initial cascade of responses fades or stops), or some combination of the two mechanisms. Collectively, our results suggest that comparisons between the 1DPF to 4DPF time points may not be sufficient to predict the physiological mechanisms involved in phenotypic regression with adequate power. Further experiments, possibly with multiple later-stage time point sampling, may be required to address outstanding questions about how these growth phenotypes are reversed.
Comparison of python organ regeneration to other regenerative model systems
Organ regeneration in snakes represents an extreme and unique phenotype among vertebrates. However, other examples of regenerative growth do exist among vertebrates, such as limb regeneration in salamanders [69], fin regeneration in fish [70], and regenerative heart growth in zebrafish [71, 72] and prenatal mammals [73]. This begs the question of whether or not these regenerative responses share common mechanisms, and as we continue to better understand the mechanisms driving regenerative growth in snakes, such key comparisons can begin to be made. While none of these other vertebrate regenerative growth systems directly parallel regenerative organ growth in snakes, regeneration of heart tissue in zebrafish is the most analogous comparison, as it occurs in adult organisms and represents regenerative growth of organ tissue specifically. Following injury or amputation of cardiac tissue, zebrafish hearts grow primarily by dedifferentiation and subsequent proliferation of cardiomyocytes [72]. Conversely, python hearts grow only by hypertrophy [3, 11, 74], and therefore may be driven by largely different regenerative mechanisms. The python small intestine, liver, and kidney, however, do grow via by hypertrophy and hyperplasia [3, 5, 11, 12]; while they represent different organ systems than the zebrafish heart, they may be driven by similar pathways that regulate cell proliferation in general. Indeed, there are parallels between zebrafish and python responses in the shared involvement of p38 MAPK signaling, a negative regulator of cardiomyocyte proliferation in zebrafish [71] that we infer to be inhibited in the Burmese python heart between fasting and 1DPF (Fig. 3). Additionally the mitotic roles of polo-like kinase pathway, which was the only pathway we predicted as significant and with a direction of activation between 1DPF and 4DPF (activated in the small intestine; see Additional file 1: Figure S7) is also involved in zebrafish regenerative heart growth. Cell-cycle regulation by polo-like kinase 1 is an important component of cardiomyocyte proliferation in zebrafish [72], and therefore may be playing a similar role in the python small intestine, although it is notable that it was not predicted as significant between fasting and 1DPF, when growth is presumably greatest in this organ [3, 5]. Other pathways involved in zebrafish regenerative growth, such as IGF signaling, FGF signaling, HIPPO signaling, and TGF-Beta signaling [71], were not inferred as significant based on canonical pathway analyses of either post-feeding time interval in our study of the Burmese python. TGFB1 and IGF1 growth factors were, however, inferred in our URMA analysis of the fasting to 1DPF interval (see Additional file 1: Figure S2), suggesting that there may still be some involvement of these growth factors in the regulation of regenerative growth in the Burmese python. A key conclusion based on our study is that, to our knowledge, mTOR signaling and NRF2-mediated oxidative stress response pathways have not been implicated in zebrafish regenerative growth. Thus, regenerative organ growth in the Burmese python appears to remain quite unique among vertebrates, both in the nature of the phenotype, and now in the molecular mechanisms underlying growth.