Many mammals hibernate to avoid food scarcity and harsh conditions during winter. Hibernation involves entering a state called torpor, which drastically reduces the amount of energy used by the body. During torpor, body temperature also decreases. This is particularly exemplified in ground squirrels, whose body temperature can hover at just above or even below the point of freezing. However, hibernating mammals cannot remain in this state continuously over the months of hibernation but instead cycle between bouts of torpor lasting for 1–3 weeks and brief periods of ‘arousal’ lasting between 12–24 hr, during which their body rapidly warms up.

The heat required to start warming up the hibernator is generated from a specialized form of fat called brown adipose tissue. Normally, the bursts of metabolic activity that are required to create this heat depend on certain proteins being produced. Making a protein involves ‘translating’ its sequence from template molecules called messenger RNA (mRNA), which are ‘transcribed’ from the gene that encodes the protein. During the low body temperatures experienced during torpor, both of these processes stop. So how is the hibernator able to quickly and efficiently heat itself up during the arousal periods of hibernation?

Grabek et al. investigated this by analyzing the relative levels of mRNA in the brown adipose tissue of hibernating 13-lined ground squirrels. Using a special technique to sample and sequence small fragments of mRNA taken from brown adipose tissue, Grabek et al. compiled a profile of the mRNA molecules present at different points in the torpor–arousal cycle and compared this with a similar profile taken from squirrels that were not hibernating.

From this analysis, Grabek et al. detected that a particular group of mRNA molecules that are required for producing heat increase in abundance during torpor, even though body temperature is low enough to stop gene transcription. This increased abundance does not occur because more of the mRNA molecules are made; instead, the mRNA molecules are modified to become more stable and long lasting. Once the animal warms up during arousal, gene transcription is reactivated and more new mRNA molecules are made.

Grabek et al. suggest that the key mRNAs required for brown adipose tissue function are selectively stabilized during torpor through a temperature-dependent protective mechanism. These mRNAs are then preferentially translated into proteins during arousal to rapidly and efficiently heat the hibernator. Most other mRNA molecules degrade throughout torpor, and so their numbers decline as replacements are not transcribed until body temperature briefly recovers during arousal. Whether this protective mechanism is also used in other tissues during torpor remains a question for future work.

Many mammals hibernate to conserve energy during extended periods of limited resource availability and harsh environmental conditions. As winter approaches in temperate climates, hibernators enter into a state of torpor. Torpor in ground squirrels involves active suppression of physiological processes to 2–5% of basal rates, which allows body temperature to lower to just above ambient, even as ambient temperatures fall to near freezing. This depressed state is not continuous throughout winter, however, instead it lasts for 1–3 weeks until it is punctuated by a spontaneous, rapid re-warming to 37°C; physiological rates during re-warming match or even exceed basal rates. The interbout arousal period is then sustained for 12–24 hr before torpor resumes. Cycles between torpor and arousal result in winter heterothermy or hibernation (Figure 1). Hibernation persists for 5–8 months before emergence in spring and maintenance of more typical mammalian homeostatic physiology throughout the summer period of growth and reproduction (Figure 1, see Carey et al., 2003; for review). Although of broad medical interest for their ability to tolerate these extraordinary physiological extremes (Carey et al., 2003; Andrews, 2007; Carey et al., 2012; Dave et al., 2012), many aspects of the hibernation phenotype remain poorly understood. Some of hibernation's most defining mysteries are the mechanisms that underlie the highly dynamic oscillations of the torpor–arousal cycle.

Figure 1

Download asset Open asset

Transcription and translation effectively cease at low body temperature during hibernation (van Breukelen and Martin, 2001; van Breukelen and Martin, 2002), yet organs maintain integrity and in some cases are quickly reactivated after 2 weeks of near inactivity in torpor. The need for immediate intense metabolic activation at low temperature is most pronounced in brown adipose tissue (BAT); early re-warming depends exclusively on non-shivering thermogenesis in this organ (Cannon and Nedergaard, 2004). Because of the constraints on gene expression during torpor, the rapid burst of metabolic activity that characterizes early re-warming may be particularly challenging for BAT.

To balance the decreased transcription, mRNA degradation during torpor also must be reduced to maintain cellular integrity and permit function in early arousal. Just as with transcription, low body temperature (i.e., Q10 effects) will slow rates of RNA degradation (Burka, 1969; Bremer and Moyes, 2014), but it is unclear how these two opposing activities will converge after two weeks to determine the steady-state abundance of specific RNAs at the end of a torpor bout. While the general consensus is that the transcriptome is largely stable during torpor (reviewed by Tessier and Storey (2014)), this view is based upon results (Frerichs et al., 1998; O'Hara et al., 1999; Knight et al., 2000; Williams et al., 2005) where ongoing degradation is not readily distinguishable from a stable transcriptome because of the sampling and normalization strategies employed. There are few clear examples of transcripts that diminish across a torpor bout (Epperson and Martin, 2002) and others that appear to increase (O'Hara et al., 1999), largely because few studies provide the necessary temporal resolution to quantify changes across a torpor bout.

In this study, we interrogate BAT mRNA dynamics in 13-lined ground squirrels across the torpor–arousal cycle and the circannual rhythm of hibernation (Figure 1B). We chose BAT because of its unique requirement to function quickly and maximally in the earliest moments of arousal, after spending two weeks at the transcriptionally prohibitive body temperatures of torpor. We used a transcriptional profiling approach developed for non-model organisms, EDGE (Hong et al., 2011), on five precisely timed sample groups to capture multiple phases of the torpor–arousal cycle (Figure 1B) and, for comparison, three groups from the non-hibernating, homeothermic portion of the year (Figure 1B).

A total of 38 EDGE-tag libraries, representing 8 distinct sampling groups (Figure 1B), were sequenced, processed (Figure 2—figure supplement 1), and analyzed for changes associated with hibernation physiology. For each of the libraries, 90.1 ± 2.6% of the sequence reads aligned to ground squirrel genomic (Supplementary file 3A in Grabek et al., 2014) or mitochondrial DNA (Figure 2A). After normalization, filtering, and annotation (Figure 2B, Figure 2—figure supplement 1), 14,798 EDGE-tags representing 8,089 unique genes remained (Supplementary file 3B in Grabek et al., 2014). We first clustered the individual sample libraries by tag abundance using Random Forests (Breiman, 2001). Three main groups were evident (Figure 2C): (1)‘spring’, independent of ambient temperature, spring cold (SpC), and spring warm (SpW); (2) ‘winter warm’: interbout aroused (IBA), entrance (Ent), and summer active (SA); (3) ‘winter cold’: early torpor (ET), late torpor (LT), and early arousal (EAr). Notably, BAT samples harvested from winter animals at warm body temperature clustered separately from those at low body temperature. This separation indicates the transcriptome is dynamic across a torpor bout.

Figure 2 with 1 supplement see all

Download asset Open asset

The 14,798 tags were next tested for significant differential expression among the three main groups; changes were detected in 2,159 tags (14.6%; q < 0.05) representing 1,638 unique genes (Supplementary file 3C in Grabek et al., 2014). These correlated well with quantitative changes in the BAT transcriptome of this species reported previously (Hampton et al., 2013); 91% of overlapping differentially expressed transcripts exhibited changes in the same direction among comparable states (see ‘Materials and methods’). DIANA hierarchical clustering identified six expression patterns among the differentially expressed tags (Figure 2D and Supplementary file 3C in Grabek et al., 2014); those in Clusters 1 and 2 were generally increased in spring compared to winter, while those in Clusters 3–6 were increased in winter, particularly in late torpor and early arousal. Distinct from the spring-enriched tags, those increased in winter were overwhelmingly enriched (Huang da et al., 2009) for functions related to BAT activation, such as lipid metabolism, lipid droplet formation, lipid transport, mitochondria and the TCA cycle (Table 1 and Supplementary file 3D in Grabek et al., 2014).

Surprisingly, a preponderance of winter-increased tags (i.e., transcripts) reached their highest relative abundance during early torpor, late torpor, and/or early arousal (Figure 2D, Clusters 4–6) despite near cessation of transcription in hibernators at low body temperature (van Breukelen and Martin, 2002). One clue to resolve this apparent paradox was provided by RPPH1, the RNA subunit of RNaseP, whose relative abundance increased several 100-fold by early arousal (q < 10^−14, Figure 3A). Because RPPH1 is transcribed by Pol III, it is not typically polyadenylated (Baer et al., 1990) and should not have been recovered in these sequencing libraries. Nevertheless, RPPH1 acquired a long poly(A) tail at low body temperature (Figure 3B,C), explaining its presence in the libraries and increase in the cold.

Figure 3

Download asset Open asset

We considered three potential mechanisms that might explain increased transcript abundance at low body temperature: (1) elevated transcription; (2) relative stabilization; and (3) acquisition of a poly(A) tail. To probe these mechanisms, we quantified abundance and the effect of poly(A) tail length on the dynamics of RPPH1 and thirteen other transcripts, including three additional ncRNAs and ten mRNAs (Supplementary file 1A; note that there are two isoforms of LIPE), during the torpor–arousal cycle. The absolute abundance of these transcripts was measured by RT-qPCR in total RNA, and short and long poly(A) RNA fractions (Figure 4—figure supplement 1; Supplementary file 1A) from interbout aroused, late torpor, early arousal, and spring warm animals (n = 3). Two classes of RNA dynamics were apparent; transcripts were either decreased (labeled as Class I) or stabilized (labeled as Class II) during torpor but not newly transcribed.

Five Class I transcripts decreased during torpor, with poly(A) and total RNA mirroring the abundance of their EDGE tags (compare IBA to LT in Figure 4A; see also Figure 4—figure supplement 2A and Supplementary file 1B). Interestingly, during early arousal, when core body temperature was still low, some of these transcripts increased, likely because heat generated early in the arousal process has returned BAT to a temperature permissive for transcription (Osborne and Hashimoto, 2003). These transcripts were largely bearing long poly(A) tails, which also appeared to shorten during torpor (Figure 4—figure supplement 2B). Class I dynamics explain the DIANA Clusters 1–3, where RNA decreased during torpor but then increased at the elevated body temperature of interbout arousal (compare IBA to LT in Figure 2D), and likely the even larger collection of transcripts that were not differentially expressed (e.g., GAPDH, Figure 4—figure supplement 2A). Thus, it appears that most BAT transcripts slowly degrade over two weeks in torpor and are not replenished until body temperature recovers during the short euthermic period.

Figure 4 with 4 supplements see all

Download asset Open asset

Nine Class II transcripts were stabilized during torpor. While their EDGE-tags appeared to increase during torpor (Figure 4B,C, left), this increase was not mirrored in total RNA (Figure 4B,C, middle-left). We further sub-divided this class by differences in polyadenylation. Total RNA for five Class IIA transcripts remained stable among states (Figure 4B, middle-left; Figure 4—figure supplement 3). However, these transcripts increased in the short and long poly(A) fractions during late torpor and particularly early arousal (Figure 4B, right; Figure 4—figure supplement 3, Supplementary file 1C) with concurrent poly(A) tail lengthening (Figure 4—figure supplement 2C), correlating with their EDGE-tags (Figure 4B, left; Figure 4—figure supplement 3; Supplementary file 1B). Total RNA decreased in torpor and early arousal for four transcripts in Class IIB (Figure 4C, Figure 4—figure supplement 4), whereas their polyadenylated fraction remained stable (Figure 4C, middle-right; Figure 4—figure supplement 4), resulting in an apparent increase (Figure 4C, right; Figure 4—figure supplement 2D and 4; Supplementary file 1C) and consistent with the EDGE-tag pattern (Figure 4C, left; Figure 4—figure supplement 4; Supplementary file 1B). Thus, in contrast to Class I, Class II transcripts are stabilized throughout torpor with maintenance or acquisition of a poly(A) tail. The enhanced stability of this subset relative to all other transcripts apparently leads to their relative increases at low body temperature (Figure 2D, DIANA Clusters 4–6).

We next tested whether the observed transcript dynamics in torpor–arousal cycles could impact the corresponding protein by measuring PNPLA2. Three PNPLA2 protein isoforms, whose sizes were consistent with those predicted for mouse in UniProt (UniProt Consortium, 2014), were detected by Western blot (Figure 4D). All appeared to cycle, but only the 48-kD band changed significantly (Figure 4E,F), following the dynamics of the transcript with the long poly(A) tail despite no change in overall transcript abundance (Figure 4F). Hence, the dynamics of this PNPLA2 protein isoform appears to be explained by polyadenylation changes in its transcript.

To investigate how changes in rates of transcription and degradation could affect differential gene expression in torpor, we developed a mathematical model of transcript dynamics across the torpor–arousal cycle. We simulated a population of 50 ‘protected’ transcripts and a bulk population of 1,400 transcripts; these numbers are proportional to the 531 tags that were either increased or stabilized across a bout of torpor (DIANA Clusters 5 and 6, Figure 2D) compared to the 14,267 tags in the full dataset. For this simulated population, the abundance of each RNA transcript was governed by a differential equation describing temperature-dependent rates of RNA synthesis and degradation (Schwanhausser et al., 2011). To model RNA transcript dynamics across the torpor–arousal cycle (Figure 5—figure supplement 1), we introduced a representative 12-day body temperature profile, incorporating temperature-dependence into the rates of RNA synthesis and degradation based on Q10 effects (Burka, 1969; van Breukelen and Martin, 2002), as described in detail in Supplementary file 2.

In the 50-transcript subset, we implemented either fixed or temperature-dependent alterations to degradation and synthesis rates to determine the resulting protective effects on normalized transcript abundance following 10 days of torpor. We found that a temperature-dependent mechanism that protected a subset of transcripts relative to bulk RNA degradation (Figure 5A,B) was most consistent with the increased abundances observed experimentally. For a body temperature threshold of 10°C and degradation set to 3% of its rate in the warm animal, the relative abundance of the protected transcripts increased over twofold (Figure 5C,D), best reflecting the experimental data. This effect was dose-dependent with the level of protection and was relatively insensitive to thresholds above 10°C (Figure 5E). Although temperature-independent decreases in degradation rates also led to increases in the relative abundance of protected transcripts, this mechanism required implausible compensatory changes to either steady state RNA abundance or transcription rates in the warm animal (Figure 5—figure supplement 2). Due to the differential Q10 effects on transcription and degradation, increasing transcription rate did not produce relative abundance increases (Figure 5—figure supplement 3). Thus, in agreement with RT-qPCR data, mathematical modeling supports enhanced stabilization of a subset of transcripts via a temperature-dependent protective mechanism; this, rather than increased transcription, leads to the observed increase in their relative abundances at the low body temperature of torpor.

Figure 5 with 3 supplements see all

Download asset Open asset

Finally, to address the possible mechanism underlying protection of selected transcripts from degradation, we examined transcript 3′ untranslated region (UTR) sequences for shared motifs (Figure 6A); because ground squirrel 3′ UTRs are largely unannotated, we defined 3′ UTRs as the 500 nt region immediately downstream of the stop codon. This choice was validated by taking a random sampling of 3′ UTR sequences in all clusters, which returned enrichment for a motif resembling the polyadenylation signal (Colgan and Manley, 1997) when compared to a scrambled background set (Figure 6B; [Bailey et al., 2009]). To identify motifs unique to the protected RNA subset, the significantly changed transcripts were divided into two groups for comparison (Figure 6A): (1) the positive set of transcripts that appeared to be stabilized (e.g., Figure 2D, DIANA Cluster 6) or increased in torpor (e.g., Figure 2D, DIANA Clusters 4–5); and (2) the negative set of transcripts that appeared to decrease in torpor (e.g., Figure 2D, DIANA Clusters 1–3). When the positive set was compared to the negative set, EXTREME (Quang and Xie, 2014) identified two significantly enriched C-rich motifs (Motif 1 and 2, Figure 6C,D). Significantly, DIANA Cluster 5, comprised of the transcripts that most clearly increased in relative abundance from early to late in torpor, contained the greatest percentage of mRNAs with the two motifs (Motif 1: 59.1%; Motif 2: 66.7% of transcripts, Figure 6C,D). DIANA Cluster 6, comprised of transcripts that remained elevated and stable across a torpor bout, contained a higher proportion of transcripts with Motif 2 (50.8%, Figure 6D) relative to the other DIANA clusters (8.5–39.8%, Figure 6D) and the control, non-significant transcripts (21.1%, NS in Figure 6D). Additionally, the other winter-increased DIANA Clusters, 3 and 4, were relatively enriched for these motifs as compared to the spring-increased DIANA Clusters 1 and 2 (38.4–43% vs 8.5–23.3%, Figure 6C,D), suggesting that these motifs play a broader role in enhanced transcript stability and/or translation during winter heterothermy. Although the transcripts in DIANA Cluster 4 appeared elevated in early torpor (Figure 2D), their motif enrichment was similar to that of Cluster 3. In contrast to those in Clusters 5 and 6, these transcripts also appeared to largely degrade by late torpor (compare ET to LT, Figure 2D); hence their reduced motif enrichment is consistent with reduced transcript stability across a bout of torpor.

Figure 6

Download asset Open asset

To identify putative binding protein(s) for these enriched sequence motifs, we used TOMTOM (Gupta et al., 2007), searching against a database of RNA binding motifs (Ray et al., 2013; Ji et al., 2013). Our motifs significantly matched those reported by Ji et al. (2013) (Figure 6E), implying binding by a poly(C) binding protein. We detected expression of two poly(C) binding protein paralogs in our dataset: PCBP3 and PCBP4. While PCBP4 did not vary with hibernation physiology, PCBP3 expression increased significantly in the winter groups (q = 0.0067, Figure 6F). The enrichment of PCBP binding motifs in the subset of transcripts that increased at low body temperature, together with the increased PCBP3 abundance in winter, suggest a role for PCBP3 in protecting a subset of BAT transcripts via its binding to the 3′ UTR C-rich motifs during torpor.

Our results show that enhanced stabilization and polyadenylation of a crucial group of transcripts, with evidence of ongoing bulk RNA degradation, occur during torpor and are likely tied to rapid activation of BAT. Dynamic polyadenylation has been demonstrated to control temporal and spatial regulation of translation in many systems (Wu et al., 1998; Kojima et al., 2012), including maternal Xenopus oocyte maturation (reviewed in Vasudevan et al. [2006]). Generally, poly(A) tail elongation causes translational activation, whereas shortening leads to silencing and/or RNA degradation (Weill et al., 2012). Similar to oocyte maturation, in hibernation, the transcriptional machinery is silenced during the two week period of torpor; therefore, post-transcriptional mechanisms affecting mRNA stability and translation serve in the rapid switch between a hypo- and hyper-metabolic state in BAT. For instance, PNPLA2 catalyzes the first committed step in triacylglycerol hydrolysis, resulting in diacylglycerol and free fatty acid (Zimmermann et al., 2004). Its increased translation via poly(A) tail lengthening in early arousal would ensure immediate generation of free fatty acids for thermogenesis, while poly(A) shortening during interbout arousal offers a parsimonious means to silence protein translation and lower free fatty acids as metabolic activity declines. Additionally, dynamic polyadenylation likely controls translation of the other Class IIA transcripts. Although there is currently no evidence for this type of mechanism operating in human BAT, our results suggest a potential therapeutic strategy by which translation of key proteins can be prioritized for recovery from metabolic repression, and more specifically, for rapid activation of thermogenesis in BAT.

A role for post-transcriptional regulation in hibernation was posited previously, based on poor correlations between mRNA and protein levels (Shao et al., 2010). In this study, we provide evidence for stabilization of a specific, functionally relevant subset of mRNAs during torpor. Moreover, our findings are consistent with reports of mRNA degradation during torpor (Epperson and Martin, 2002) and with global maintenance of poly(A) tails and their increased length (Knight et al., 2000). These results also provide further explanation to histological observations, as RNP granules containing both ncRNAs and mRNAs are formed during torpor in BAT nucleoli (Malatesta et al., 2008; Malatesta et al., 2011). The Class II ncRNAs identified in our study are located in nucleoli, suggesting that sequestration into these RNP granules protects a subset of transcripts from the degradation affecting bulk BAT RNA, which is also consistent with the results of our mathematical modeling.

Our results propose that a mechanism underlying enhanced stabilization of a subset of RNAs involves their binding by one of the poly(C) binding proteins (PCBPs), as we detected their corresponding C-rich motifs in the 3′ regions of the torpor-stabilized mRNAs. Intriguingly, PCBPs are involved in many aspects of post-transcriptional control that are consistent with our observations, including enhanced stability of long-lived mRNAs (Makeyev and Liebhaber, 2002). Further roles include 3′ end-processing and alternative polyadenylation (Ji et al., 2011; Ji et al., 2013), and the addition, maintenance (Wang et al., 1999), and elongation of poly(A) tails (Vishnu et al., 2011). Finally, these proteins are involved in both translational silencing and enhancement (Makeyev and Liebhaber, 2002). Although these roles are established for the predominantly studied PCBP1 and PCBP2, these paralogs were not detected in our dataset. Rather, we detected increased expression of PCBP3 during winter heterothermy, a pattern that would be expected for a role involving enhanced mRNA stabilization and polyadenylation during torpor. Furthermore, PCBP3's pattern runs in contrast to most of the other RNA binding proteins detected in this dataset, which, if changed, were largely decreased during winter heterothermy (see Table 1, functional annotations for DIANA Clusters 1 and 2).

In addition to a PCBP, other 3′ UTR binding proteins may be involved in enhancing mRNA stability during torpor. The poly(A) binding protein PABP1 and the TIA-1/R RNA binding proteins were recently shown to localize to discrete sub-nuclear foci during torpor in the livers of 13-lined ground squirrels (Tessier et al., 2014). PABP1 specifically binds to the poly(A) tails of transcripts, influencing their length as well as overall transcript translation and stability (Burgess and Gray, 2010). Significantly, the PCBPs 1 and 2 functionally interact with PABP1 in order to prevent deadenylation and to maintain mRNA stability (Wang et al., 1999). While further research is needed to determine whether the protection identified here extends to other tissues during torpor and to thoroughly examine the role of PCBP3 or its homologs in this protection, our results of enhanced stabilization and polyadenylation for a subset of crucial transcripts suggest a mechanistic link to PABP1 observations in other organs (Knight et al., 2000; Tessier et al., 2014). Although there are reports of elevated RBM3, a cold-induced RNA binding protein, in several organs including BAT from ground squirrels and bears during hibernation (Williams et al., 2005; Yan et al., 2008; Fedorov et al., 2011), we found no evidence for enrichment of RBM3 recognition motifs (Liu et al., 2013; Ray et al., 2013) in our subset of stabilized transcripts.

More broadly, our study highlights the importance of how transcriptome data is interpreted. While it is generally assumed that changes in steady-state mRNA levels stem from changes in the rate of transcription, varying the rate of degradation also changes steady-state levels; this phenomenon has been observed at both mRNA and protein levels in the cold acclimation of fish (Sidell, 1977; Bremer and Moyes, 2014). Recently, the balance between mRNA synthesis and degradation rates was examined on a global scale in cultured mammalian cells, demonstrating complex gene-specific effects of transcription, processing, decay, and translation (Rabani et al., 2011; Schwanhausser et al., 2011). In our study, the modeling results shown in Figure 5 demonstrate that transcript-specific variability in the rate of degradation may cause a subset of transcripts to increase in relative abundance. These transcripts are degrading, but more slowly than most of the bulk transcripts and therefore exhibit a small increase in relative abundance across the torpor bout. Thus, many of the smaller fold changes observed in gene expression datasets from hibernators (Williams et al., 2005; Yan et al., 2008; Hampton et al., 2013; Schwartz et al., 2013) likely reflect intrinsic differences in the stability of specific mRNAs rather than specific mechanisms to regulate their transcription or decay. However, for changes greater than approximately twofold in a substantial number of transcripts, here ∼3.5% of the total, a specific regulatory mechanism appears to be required. Furthermore, particularly large fold changes, as observed here with RPPH1, likely reflect the addition or lengthening of poly(A) tails.

Our data suggest a model (Figure 7) of RNA dynamics in hibernator BAT wherein key RNAs for BAT function are selectively stabilized during torpor while bulk transcripts decline through degradation in the absence of new transcription. Stabilization likely occurs by a temperature-dependent protective mechanism that is in place before body temperature reaches 5°C, such as PCPB3 binding to the 3′ UTRs of protected transcripts, which then leads to their relative increase as torpor progresses. At the end of torpor and onset of arousal, the stabilized mRNA subset with the longest poly(A) tails is translated immediately as BAT temperature becomes permissive. As BAT temperature rises, further polyadenylation of the remaining stabilized RNAs facilitates their translation, transcription resumes (Osborne et al., 2004), and, during interbout arousal, transcripts that were previously degraded during torpor are replenished to their baseline levels. This dynamic cycle of transcription, degradation, stabilization, and polyadenylation in BAT leads to translation of the correct transcripts at the correct time with minimal energy expenditure. Specifically: (1) energy intensive translation during early arousal is directed to proteins needed for BAT activation; (2) the cell is not dependent on de novo transcription at the onset of the short bursts of metabolic activity, which could delay thermogenesis and induce stress; (3) inhibition of translation via shortening of poly(A) tails while body temperature is high or begins to decline conserves energy compared to mRNA degradation and subsequent re-synthesis. Thus, given a general suppression of transcription by low body temperature during two-week torpor periods, stabilization and dynamic polyadenylation provide an alternative mechanism to prioritize transcripts for immediate translation when BAT metabolic activity rapidly resumes.

Figure 7

Download asset Open asset

1. 1. Grabek KR
   2. Diniz Behn C
   3. Barsh GS
   4. Hesselberth JR
   5. Martin SL

   (2014) Data from: Enhanced stability and polyadenylation of select mRNAs support rapid thermogenesis in the brown fat of a hibernator

   Available at Dryad Digital Repository under a CC0 Public Domain Dedication. Contains: Supplementary file 3. (A) EDGE-tag raw counts. (B) EDGE-tag normalized counts. (C) EDGE-tags significant for changed expression. (D) List of All DAVID Functional Annotation Clusters.

   http://dx.doi.org/10.5061/dryad.5hh54

1. 1. Anders S
   2. Huber W

   (2010) Differential expression analysis for sequence count data

   Genome Biology 11:R106.

   https://doi.org/10.1186/gb-2010-11-10-r106

   • Google Scholar
2. 1. Andrews MT

   (2007) Advances in molecular biology of hibernation in mammals

   Bioessays 29:431–440.

   https://doi.org/10.1002/bies.20560

   • Google Scholar
3. 1. Baer M
   2. Nilsen TW
   3. Costigan C
   4. Altman S

   (1990) Structure and transcription of a human gene for H1 RNA, the RNA component of human RNase P

   Nucleic Acids Research 18:97–103.

   https://doi.org/10.1093/nar/18.1.97

   • Google Scholar
4. 1. Bailey TL
   2. Boden M
   3. Buske FA
   4. Frith M
   5. Grant CE
   6. Clementi L
   7. Ren J
   8. Li WW
   9. Noble WS

   (2009) MEME SUITE: tools for motif discovery and searching

   Nucleic Acids Research 37:W202–W208.

   https://doi.org/10.1093/nar/gkp335

   • Google Scholar
5. 1. Bailey TL
   2. Elkan C

   (1994)

   Fitting a mixture model by expectation maximization to discover motifs in biopolymers

   Proceedings / International Conference on Intelligent Systems for Molecular Biology 2:28–36.

   • Google Scholar
6. 1. Bailey TL
   2. Gribskov M

   (1998) Combining evidence using p-values: application to sequence homology searches

   Bioinformatics 14:48–54.

   https://doi.org/10.1093/bioinformatics/14.1.48

   • Google Scholar
7. 1. Breiman L

   (2001) Random forests

   Machine Learning 45:5–32.

   https://doi.org/10.1023/A:1010933404324

   • Google Scholar
8. 1. Bremer K
   2. Moyes CD

   (2014) mRNA degradation: an underestimated factor in steady-state transcript levels of cytochrome c oxidase subunits?

   The Journal of Experimental Biology 217:2212–2220.

   https://doi.org/10.1242/jeb.100214

   • Google Scholar
9. 1. Burgess HM
   2. Gray NK

   (2010) mRNA-specific regulation of translation by poly(A)-binding proteins

   Biochemical Society Transactions 38:1517–1522.

   https://doi.org/10.1042/BST0381517

   • Google Scholar
10. 1. Burka ER

    (1969) Characteristics of RNA degradation in the erythroid cell

    The Journal of Clinical Investigation 48:1266–1272.

    https://doi.org/10.1172/JCI106092

    • Google Scholar
11. 1. Cannon B
    2. Nedergaard J

    (2004) Brown adipose tissue: function and physiological significance

    Physiological Reviews 84:277–359.

    https://doi.org/10.1152/physrev.00015.2003

    • Google Scholar
12. 1. Carey HV
    2. Andrews MT
    3. Martin SL

    (2003) Mammalian hibernation: cellular and molecular responses to depressed metabolism and low temperature

    Physiological Reviews 83:1153–1181.

    https://doi.org/10.1152/physrev.00008.2003

    • Google Scholar
13. 1. Carey HV
    2. Martin SL
    3. Horwitz BA
    4. Yan L
    5. Bailey SM
    6. Podrabsky J
    7. Storz JF
    8. Ortiz RM
    9. Wong RP
    10. Lathrop DA

    (2012) Elucidating nature's solutions to heart, lung, and blood diseases and sleep disorders

    Circulation Research 110:915–921.

    https://doi.org/10.1161/CIRCRESAHA.111.255398

    • Google Scholar
14. 1. Colgan DF
    2. Manley JL

    (1997) Mechanism and regulation of mRNA polyadenylation

    Genes & Development 11:2755–2766.

    https://doi.org/10.1101/gad.11.21.2755

    • Google Scholar
15. 1. Dave KR
    2. Christian SL
    3. Perez-Pinzon MA
    4. Drew KL

    (2012) Neuroprotection: lessons from hibernators

    Comparative Biochemistry and Physiology Part B, Biochemistry and Molecular Biology 162:1–9.

    https://doi.org/10.1016/j.cbpb.2012.01.008

    • Google Scholar
16. 1. Diaz-Uriarte R

    (2007) GeneSrF and varSelRF: a web-based tool and R package for gene selection and classification using random forest

    BMC Bioinformatics 8:328.

    https://doi.org/10.1186/1471-2105-8-328

    • Google Scholar
17. 1. Epperson LE
    2. Martin SL

    (2002) Quantitative assessment of ground squirrel mRNA levels in multiple stages of hibernation

    Physiological Genomics 10:93–102.

    https://doi.org/10.1152/physiolgenomics.00004.2002

    • Google Scholar
18. 1. Fedorov VB
    2. Goropashnaya AV
    3. Toien O
    4. Stewart NC
    5. Chang C
    6. Wang H
    7. Yan J
    8. Showe LC
    9. Showe MK
    10. Barnes BM

    (2011) Modulation of gene expression in heart and liver of hibernating black bears (Ursus americanus)

    BMC Genomics 12:171.

    https://doi.org/10.1186/1471-2164-12-171

    • Google Scholar
19. 1. Frerichs KU
    2. Smith CB
    3. Brenner M
    4. DeGracia DJ
    5. Krause GS
    6. Marrone L
    7. Dever TE
    8. Hallenbeck JM

    (1998) Suppression of protein synthesis in brain during hibernation involves inhibition of protein initiation and elongation

    Proceedings of the National Academy of Sciences of USA 95:14511–14516.

    https://doi.org/10.1073/pnas.95.24.14511

    • Google Scholar
20. 1. Grabek KR
    2. Diniz Behn C
    3. Barsh GS
    4. Hesselberth JR
    5. Martin SL

    (2014) Data from: Enhanced stability and polyadenylation of select mRNAs support rapid thermogenesis in the brown fat of a hibernator

    Dryad Digital Repository.

    https://doi.org/10.5061/dryad.5hh54

    • Google Scholar
21. 1. Gupta S
    2. Stamatoyannopoulos JA
    3. Bailey TL
    4. Noble WS

    (2007) Quantifying similarity between motifs

    Genome Biology 8:R24.

    https://doi.org/10.1186/gb-2007-8-2-r24

    • Google Scholar
22. 1. Hampton M
    2. Melvin RG
    3. Andrews MT

    (2013) Transcriptomic analysis of brown adipose tissue across the physiological extremes of natural hibernation

    PLOS ONE 8:e85157.

    https://doi.org/10.1371/journal.pone.0085157

    • Google Scholar
23. 1. Hampton M
    2. Melvin RG
    3. Kendall AH
    4. Kirkpatrick BR
    5. Peterson N
    6. Andrews MT

    (2011) Deep sequencing the transcriptome reveals seasonal adaptive mechanisms in a hibernating mammal

    PLOS ONE 6:e27021.

    https://doi.org/10.1371/journal.pone.0027021

    • Google Scholar
24. 1. Hindle AG
    2. Martin SL

    (2014) Intrinsic circannual regulation of brown adipose tissue form and function in tune with hibernation

    American Journal of Physiology Endocrinology and Metabolism 306:E284–E299.

    https://doi.org/10.1152/ajpendo.00431.2013

    • Google Scholar
25. 1. Hong LZ
    2. Li J
    3. Schmidt-Kuntzel A
    4. Warren WC
    5. Barsh GS

    (2011) Digital gene expression for non-model organisms

    Genome Research 21:1905–1915.

    https://doi.org/10.1101/gr.122135.111

    • Google Scholar
26. 1. Huang da W
    2. Sherman BT
    3. Lempicki RA

    (2009) Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources

    Nature Protocols 4:44–57.

    https://doi.org/10.1038/nprot.2008.211

    • Google Scholar
27. 1. Janicke A
    2. Vancuylenberg J
    3. Boag PR
    4. Traven A
    5. Beilharz TH

    (2012) ePAT: a simple method to tag adenylated RNA to measure poly(A)-tail length and other 3' RACE applications

    RNA 18:1289–1295.

    https://doi.org/10.1261/rna.031898.111

    • Google Scholar
28. 1. Ji X
    2. Kong J
    3. Liebhaber SA

    (2011) An RNA-protein complex links enhanced nuclear 3' processing with cytoplasmic mRNA stabilization

    The EMBO Journal 30:2622–2633.

    https://doi.org/10.1038/emboj.2011.171

    • Google Scholar
29. 1. Ji X
    2. Wan J
    3. Vishnu M
    4. Xing Y
    5. Liebhaber SA

    (2013) alphaCP Poly(C) binding proteins act as global regulators of alternative polyadenylation

    Molecular and Cellular Biology 33:2560–2573.

    https://doi.org/10.1128/MCB.01380-12

    • Google Scholar
30. 1. Knight JE
    2. Narus EN
    3. Martin SL
    4. Jacobson A
    5. Barnes BM
    6. Boyer BB

    (2000) mRNA stability and polysome loss in hibernating Arctic ground squirrels (Spermophilus parryii)

    Molecular and Cellular Biology 20:6374–6379.

    https://doi.org/10.1128/MCB.20.17.6374-6379.2000

    • Google Scholar
31. 1. Kojima S
    2. Sher-Chen EL
    3. Green CB

    (2012) Circadian control of mRNA polyadenylation dynamics regulates rhythmic protein expression

    Genes & Development 26:2724–2736.

    https://doi.org/10.1101/gad.208306.112

    • Google Scholar
32. 1. Liu Y
    2. Hu W
    3. Murakawa Y
    4. Yin J
    5. Wang G
    6. Landthaler M
    7. Yan J

    (2013) Cold-induced RNA-binding proteins regulate circadian gene expression by controlling alternative polyadenylation

    Scientific Reports 3:2054.

    https://doi.org/10.1038/srep02054

    • Google Scholar
33. 1. Makeyev AV
    2. Liebhaber SA

    (2002) The poly(C)-binding proteins: a multiplicity of functions and a search for mechanisms

    RNA 8:265–278.

    https://doi.org/10.1017/S1355838202024627

    • Google Scholar
34. 1. Malatesta M
    2. Biggiogera M
    3. Baldelli B
    4. Barabino SM
    5. Martin TE
    6. Zancanaro C

    (2008) Hibernation as a far-reaching program for the modulation of RNA transcription

    Microscopy Research and Technique 71:564–572.

    https://doi.org/10.1002/jemt.20587

    • Google Scholar
35. 1. Malatesta M
    2. Zancanaro C
    3. Biggiogera M

    (2011) Immunoelectron microscopic characterization of nucleolus-associated domains during hibernation

    Microscopy Research and Technique 74:47–53.

    https://doi.org/10.1002/jemt.20872

    • Google Scholar
36. 1. Meijer HA
    2. Bushell M
    3. Hill K
    4. Gant TW
    5. Willis AE
    6. Jones P
    7. de Moor CH

    (2007) A novel method for poly(A) fractionation reveals a large population of mRNAs with a short poly(A) tail in mammalian cells

    Nucleic Acids Research 35:e132.

    https://doi.org/10.1093/nar/gkm830

    • Google Scholar
37. 1. O'Hara BF
    2. Watson FL
    3. Srere HK
    4. Kumar H
    5. Wiler SW
    6. Welch SK
    7. Bitting L
    8. Heller HC
    9. Kilduff TS

    (1999)

    Gene expression in the brain across the hibernation cycle

    The Journal of Neuroscience 19:3781–3790.

    • Google Scholar
38. 1. Osborne PG
    2. Gao B
    3. Hashimoto M

    (2004) Determination in vivo of newly synthesized gene expression in hamsters during phases of the hibernation cycle

    The Japanese Journal of Physiology 54:295–305.

    https://doi.org/10.2170/jjphysiol.54.295

    • Google Scholar
39. 1. Osborne PG
    2. Hashimoto M

    (2003) State-dependent regulation of cortical blood flow and respiration in hamsters: response to hypercapnia during arousal from hibernation

    The Journal of Physiology 547:963–970.

    https://doi.org/10.1113/jphysiol.2002.033571

    • Google Scholar
40. 1. Peddigari S
    2. Li PW
    3. Rabe JL
    4. Martin SL

    (2013) hnRNPL and nucleolin bind LINE-1 RNA and function as host factors to modulate retrotransposition

    Nucleic Acids Research 41:575–585.

    https://doi.org/10.1093/nar/gks1075

    • Google Scholar
41. 1. Quang D
    2. Xie X

    (2014) EXTREME: an online EM algorithm for motif discovery

    Bioinformatics 30:1667–1673.

    https://doi.org/10.1093/bioinformatics/btu093

    • Google Scholar
42. 1. Rabani M
    2. Levin JZ
    3. Fan L
    4. Adiconis X
    5. Raychowdhury R
    6. Garber M
    7. Gnirke A
    8. Nusbaum C
    9. Hacohen N
    10. Friedman N
    11. Amit I
    12. Regev A

    (2011) Metabolic labeling of RNA uncovers principles of RNA production and degradation dynamics in mammalian cells

    Nature biotechnology 29:436–442.

    https://doi.org/10.1038/nbt.1861

    • Google Scholar
43. 1. Ray D
    2. Kazan H
    3. Cook KB
    4. Weirauch MT
    5. Najafabadi HS
    6. Li X
    7. Gueroussov S
    8. Albu M
    9. Zheng H
    10. Yang A
    11. Na H
    12. Irimia M
    13. Matzat LH
    14. Dale RK
    15. Smith SA
    16. Yarosh CA
    17. Kelly SM
    18. Nabet B
    19. Mecenas D
    20. Li W
    21. Laishram RS
    22. Qiao M
    23. Lipshitz HD
    24. Piano F
    25. Corbett AH
    26. Carstens RP
    27. Frey BJ
    28. Anderson RA
    29. Lynch KW
    30. Penalva LO
    31. Lei EP
    32. Fraser AG
    33. Blencowe BJ
    34. Morris QD
    35. Hughes TR

    (2013) A compendium of RNA-binding motifs for decoding gene regulation

    Nature 499:172–177.

    https://doi.org/10.1038/nature12311

    • Google Scholar
44. 1. Risso D
    2. Schwartz K
    3. Sherlock G
    4. Dudoit S

    (2011) GC-content normalization for RNA-Seq data

    BMC Bioinformatics 12:480.

    https://doi.org/10.1186/1471-2105-12-480

    • Google Scholar
45. 1. Schwanhäusser B
    2. Busse D
    3. Li N
    4. Dittmar G
    5. Schuchhardt J
    6. Wolf J
    7. Chen W
    8. Selbach M

    (2011) Global quantification of mammalian gene expression control

    Nature 473:337–342.

    https://doi.org/10.1038/nature10098

    • Google Scholar
46. 1. Schwartz C
    2. Hampton M
    3. Andrews MT

    (2013) Seasonal and regional differences in gene expression in the brain of a hibernating mammal

    PLOS ONE 8:e58427.

    https://doi.org/10.1371/journal.pone.0058427

    • Google Scholar
47. 1. Shao C
    2. Liu Y
    3. Ruan H
    4. Li Y
    5. Wang H
    6. Kohl F
    7. Goropashnaya AV
    8. Fedorov VB
    9. Zeng R
    10. Barnes BM
    11. Yan J

    (2010) Shotgun proteomics analysis of hibernating arctic ground squirrels

    Molecular & Cellular Proteomics 9:313–326.

    https://doi.org/10.1074/mcp.M900260-MCP200

    • Google Scholar
48. 1. Sidell BD

    (1977) Turnover of cytochrome C in skeletal muscle of green sunfish (Lepomis cyanellus, R.) during thermal acclimation

    Journal of Experimental Zoology 199:233–250.

    https://doi.org/10.1002/jez.1401990208

    • Google Scholar
49. 1. Svetnik V
    2. Liaw A
    3. Tong C
    4. Culberson JC
    5. Sheridan RP
    6. Feuston BP

    (2003) Random forest: a classification and regression tool for compound classification and QSAR modeling

    Journal of Chemical Information and Computer Sciences 43:1947–1958.

    https://doi.org/10.1021/ci034160g

    • Google Scholar
50. 1. Tessier SN
    2. Audas TE
    3. Wu CW
    4. Lee S
    5. Storey KB

    (2014) The involvement of mRNA processing factors TIA-1, TIAR, and PABP-1 during mammalian hibernation

    Cell Stress and Chaperones 19:813–825.

    https://doi.org/10.1007/s12192-014-0505-8

    • Google Scholar
51. 1. Tessier S
    2. Storey K

    (2014) To be or not to be: the regulation of mRNA fate as a survival strategy during mammalian hibernation

    Cell Stress and Chaperones 19:763–776.

    https://doi.org/10.1007/s12192-014-0512-9

    • Google Scholar
52. 1. UniProt Consortium

    (2014) Activities at the universal protein resource (UniProt)

    Nucleic Acids Research 42:D191–D198.

    https://doi.org/10.1093/nar/gkt1140

    • Google Scholar
53. 1. van Breukelen F
    2. Martin SL

    (2001)

    Translational initiation is uncoupled from elongation at 18 degrees C during mammalian hibernation

    American Journal of Physiology Regulatory, Integrative and Comparative Physiology 281:R1374–R1379.

    • Google Scholar
54. 1. van Breukelen F
    2. Martin SL

    (2002) Reversible depression of transcription during hibernation

    Journal of Comparative Physiology B, Biochemical, Systemic, and Environmental Physiology 172:355–361.

    https://doi.org/10.1007/s00360-002-0256-1

    • Google Scholar
55. 1. Vasudevan S
    2. Seli E
    3. Steitz JA

    (2006) Metazoan oocyte and early embryo development program: a progression through translation regulatory cascades

    Genes & Development 20:138–146.

    https://doi.org/10.1101/gad.1398906

    • Google Scholar
56. 1. Vishnu MR
    2. Sumaroka M
    3. Klein PS
    4. Liebhaber SA

    (2011) The poly(rC)-binding protein alphaCP2 is a noncanonical factor in X. laevis cytoplasmic polyadenylation

    RNA 17:944–956.

    https://doi.org/10.1261/rna.2587411

    • Google Scholar
57. 1. Wang Z
    2. Day N
    3. Trifillis P
    4. Kiledjian M

    (1999)

    An mRNA stability complex functions with poly(A)-binding protein to stabilize mRNA in vitro

    Molecular and Cellular Biology 19:4552–4560.

    • Google Scholar
58. 1. Weill L
    2. Belloc E
    3. Bava FA
    4. Mendez R

    (2012) Translational control by changes in poly(A) tail length: recycling mRNAs

    Nature Structural & Molecular Biology 19:577–585.

    https://doi.org/10.1038/nsmb.2311

    • Google Scholar
59. 1. Williams DR
    2. Epperson LE
    3. Li W
    4. Hughes MA
    5. Taylor R
    6. Rogers J
    7. Martin SL
    8. Cossins AR
    9. Gracey AY

    (2005) Seasonally hibernating phenotype assessed through transcript screening

    Physiol Genomics 24:13–22.

    https://doi.org/10.1152/physiolgenomics.00301.2004

    • Google Scholar
60. 1. Wu L
    2. Wells D
    3. Tay J
    4. Mendis D
    5. Abbott MA
    6. Barnitt A
    7. Quinlan E
    8. Heynen A
    9. Fallon JR
    10. Richter JD

    (1998) CPEB-mediated cytoplasmic polyadenylation and the regulation of experience-dependent translation of alpha-CaMKII mRNA at synapses

    Neuron 21:1129–1139.

    https://doi.org/10.1016/S0896-6273(00)80630-3

    • Google Scholar
61. 1. Yan J
    2. Barnes BM
    3. Kohl F
    4. Marr TG

    (2008) Modulation of gene expression in hibernating arctic ground squirrels

    Physiol Genomics 32:170–181.

    https://doi.org/10.1152/physiolgenomics.00075.2007

    • Google Scholar
62. 1. Zimmermann R
    2. Strauss JG
    3. Haemmerle G
    4. Schoiswohl G
    5. Birner-Gruenberger R
    6. Riederer M
    7. Lass A
    8. Neuberger G
    9. Eisenhaber F
    10. Hermetter A
    11. Zechner R

    (2004) Fat mobilization in adipose tissue is promoted by adipose triglyceride lipase

    Science 306:1383–1386.

    https://doi.org/10.1126/science.1100747

    • Google Scholar

Author details

1. Katharine R Grabek

   1. Department of Cell and Developmental Biology, University of Colorado School of Medicine, Aurora, United States
   2. Human Medical Genetics and Genomics Program, University of Colorado School of Medicine, Aurora, United States

   Present address

   Department of Genetics, Stanford University School of Medicine, Stanford, United States

   Contribution

   KRG, conceived the project, generated and analyzed EDGE, RT-qPCR and western blot and motif data and drafted the manuscript, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

2. Cecilia Diniz Behn

   Department of Applied Math and Statistics, Colorado School of Mines, Golden, United States

   Contribution

   CDB, developed and implemented the mathematical modeling, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

3. Gregory S Barsh

   Department of Research, HudsonAlpha Institute for Biotechnology, Huntsville, United States

   Contribution

   GSB, assisted with EDGE-tag study design, EDGE data generation, analysis and interpretation, and assisted with revising of manuscript, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents

   Competing interests

   The authors declare that no competing interests exist.

4. Jay R Hesselberth

   Department of Cell and Developmental Biology, University of Colorado School of Medicine, Aurora, United States

   Contribution

   JRH, assisted with processing and annotation of EDGE-tag libraries, interpretation of data and writing of manuscript, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

5. Sandra L Martin

   1. Department of Cell and Developmental Biology, University of Colorado School of Medicine, Aurora, United States
   2. Human Medical Genetics and Genomics Program, University of Colorado School of Medicine, Aurora, United States

   Contribution

   SLM, directed the project, interpreted data and drafted manuscript with KRG, Conception and design, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents.

   For correspondence

   sandy.martin@ucdenver.edu

   Competing interests

   The authors declare that no competing interests exist.

• 2,336

  views

• 246

  downloads

• 26

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.