Under extreme hypoxia, animals have to switch from oxidative phosphorylation to glycolysis. During glycolysis, glucose is processed via a series of intermediates to pyruvate with ATP yield. Pyruvate then has to be converted into various end products in order to restore NAD+. The most widespread of those are lactate, alanine, and ethanol, although many less common ones are known to exist9. Two pathways predominate in tetrapods: pyruvate is either directly converted to lactate, or reacts with glutamate in a transamination reaction yielding alanine and α-ketoglutarate. Glutamate concentration is then restored by the malate-aspartate shuttle reactions, while alanine is accumulated.
Both lactate and alanine are known to accumulate in various organs of the anoxia-tolerant turtles Chrysemys picta and Trachemys scripta10,11, as well as in the embryos of the annual killifish Austrofundulus limnaeus12. The naked mole-rat Heterocephalus glaber, a mammalian model of hypoxia tolerance, accumulates lactate at 5% oxygen concentrations13. The crucian carp Carassius carassius demonstrates elevated levels of lactate under anoxia14, despite the fact that ethanol is usually considered as its anaerobic end product15.
We found that both lactate and alanine accumulate in R. amurensis in response to extreme hypoxia. As shown in Fig. 3A, pyruvate conversion pathways were shifted towards lactate in heart and towards alanine, in liver.
We should note here that high alanine concentrations might also be the result of the amino acid degradation and the concomitant accumulation of the substances used to store and transport ammonium ions (glutamate, glutamine, and alanine, discussed below in the section “Amino acids”).
There was one more compound, 2,3-butanediol (2,3-BD), that was absent in normoxia but present in significant concentrations in extreme hypoxia (Fig. 3A). It is one of the end products of bacterial glycolysis, and is synthesized from pyruvate via acetoin by means of several alternative pathways16. In humans, 2,3-BD was detected as one of the products of ethanol degradation17,18,19, as well as in patients with certain forms of cancer20. Moreover, 2,3-BD is a marker of cardiac ischemia in humans, and is accumulated alongside lactate in ischemic pig hearts21,22. These facts suggest that 2,3-BD might be one of the end products of glycolysis alongside with lactate and alanine in R. amurensis, as well as in other vertebrates. One of its presumable advantages is that it is a neutral metabolite in contrast to lactate16. However, synthesis of 2,3-BD in bacteria has lower efficiency of NADH to NAD+ conversion compared to other glycolysis pathways, since only one molecule of 2,3-BD is formed from two molecules of pyruvate. Surprisingly, the pathways of 2,3-BD biosynthesis in animals are unknown; the findings that 2,3-BD in humans is formed after ethanol intake implies other reaction pathways.
Another putative end product of glycolysis is glycerol, also present in high amounts in extreme hypoxia but absent in normoxia (Fig. 5A). Glycerol is a well-known cryoprotectant in amphibians23,24,25,26. However, its presence in our sample could not be explained by adaptation to low temperatures, since control frogs were kept at the same temperature but did not accumulate any glycerol. The phenomenon of glycerol accumulation in response to hypoxia was never reported for animals, but is known to occur in yeast27,28 and trichomonads29. Under anoxia, Saccharomyces cerevisiae produces large amounts of glycerol in addition to the main product of glycolysis (ethanol), presumably as a means to restore NAD+ levels28. If this is the case for R. amurensis, glycerol may also be considered an end product of glycolysis. However, there is an alternative explanation: high glycerol levels may be the result of glycerophosphocholine hydrolysis. Concentrations of glycerophosphocholine were found to decrease both in heart and liver (Fig. 5A); however, no concomitant changes in the quantities of phosphocholine or choline were observed.
Increased concentrations of glucose in liver indicate activation of glycogenolysis. In heart, changes in glucose concentration were not significant, which probably reflects the balance of increased glucose delivery and consumption. We should note that glucose concentrations in normoxic frogs were also high, probably as a response to low temperatures. It is hypothesized that other sugars may play an important role in anoxia response. For example, fructose is an important energy source in the naked mole-rat under hypoxia13. However, we found only a relatively minor amounts of maltose in equal amount in extreme hypoxia and normoxia.
Rearrangement of energy pathways
The absence of oxygen leads to the termination of the electron transfer chain and to dramatic rearrangements of the Krebs cycle. A universal feature here is the accumulation of succinate, which was shown to be due to the reversal of the Krebs cycle30. Without oxygen, succinate dehydrogenase acts in reverse, forming succinate from fumarate. The malate-aspartate shuttle and the purine nucleotide pathway contribute to this process in ischemic mammalian tissues30,31,32. Succinate accumulation was also observed in various hypoxia-tolerant vertebrates10,11,12,13. We demonstrated that this phenomenon holds true for R. amurensis as well (Fig. 2C).
The obtained data (Fig. 2A–F) indicate that extreme hypoxia dramatically reduces available energy reserves. This is demonstrated by extreme changes in ATP/ADP and PCr/Cr ratios. However, the absence of xanthine and hypoxanthine, the products of purine degradation that are observed in high concentrations during ischemia in mammals11 suggests that this stress is reversible.
What metabolic features are associated with hypoxia tolerance? Brungaard11 suggested that two major patterns are observed in red-eared slider turtles, i.e., (1) high ATP/ADP ratio (relative to intolerant species) and no signs of their degradation to AMP, xanthine, and hypoxanthine; (2) low succinate/fumarate ratio. For the Siberian wood frog, rule no. 2 holds true (Fig. 3B). However, ATP/ADP ratio, as well as the absolute quantities of ATP in extreme hypoxia and even in normoxia are low, similar to those in mammalian heart under ischemia (Fig. 1A). Moreover, degradation to AMP (but not xanthine or hypoxanthine) is observed. We can state, however, that such changes are far from lethal: the samples were taken from frogs exposed to 17-day extreme hypoxia, while we know that they could survive this state for further three to 5 months4. Low ATP levels compared to turtles may be accounted for by different systematic position of these model organisms. We can thus conclude that low ATP content does not impede hypoxia survival.
Two amino acids are believed to play an important role in hypoxia: alanine, the end product of glycolysis, and aspartate that is depleted by the malate-aspartate shuttle. As seen in Fig. 4 and Table S1, alanine was among the most abundant and the most affected by extreme hypoxia both in liver and heart (changes in phenylalanine concentrations were higher, but the background content was low). Aspartate was the only amino acid that decreased in response to extreme hypoxia (Fig. 4). One can also see that changes in aspartate levels were much higher in heart compared to liver, in agreement with higher alanine concentrations in that tissue.
Increased concentrations of all amino acids except for aspartate can be explained by inhibition of ribosomal synthesis in order to save energy. Translation arrest is observed in red-eared slider turtles in response to anoxia33,34. However, the increase of free amino acids is not a universal response in anoxia-tolerant animals. Increased concentrations of all amino acids are observed only in embryos of the annual killifish Austrofundulus limnaeus12. Red-eared slider turtles demonstrate a mixed response whereby five amino acids including alanine are upregulated, and the rest, downregulated11. A similarly ambiguous response is also characteristic for Carassius carassius14.
However, the obtained data imply that glutamine and glutamate must have a more direct role in hypoxia response (Fig. 4). Moreover, increases in concentrations of these amino acids are much more pronounced in liver compared to heart. One possible option is that these amino acids accumulate as the result of protein and amino acid catabolism. Deamination of amino acids yields glutamate that is used to hold excess ammonium35. It is normally converted back to α-ketoglutarate, but this process should be impeded under the paucity of NAD+ during hypoxia. In some organs such as muscle, glutamate is then converted to glutamine or alanine that are transported to the liver, where they are catabolized35. Under extreme hypoxia, their processing may be impeded, so this may be the reason we observe high quantities of these amino acids in liver but not in heart.
A number of compounds changed their concentrations in response to extreme hypoxia. Sarcosine (upregulated in both tissues) and taurine (downregulated in liver) are usually regarded as osmolytes9, but may also have multiple other functions36,37,38. Changes in concentrations of choline, phosphocholine, glycerophosphocholine, and serine-phosphoethanolamine are probably associated with changes in phospholipid metabolism39. Moreover, glycerphosphocholine is an osmolytes9, while choline is the precursor of acetylcholine, an important neuromediator. Taurine is also considered a neuroregulatory substance; increased taurine concentrations were also observed in anoxic carp muscle14. Multiple functions of the aforementioned substances make it hard to state which role they play in hypoxia response.
Reoxygenation stress is an important factor in hypoxia-related damage. Adaptation to this stress is thus a vital part of hypoxia response. In accordance with this, concentrations of glutathione increased 2.5-fold in liver under extreme hypoxia (Fig. 5C). Concentrations of glutathione in the heart were high both in hypoxic and control frogs. Heart is one of the most important and sensitive organs, so high intrinsic levels might reflect preadaptation to reoxygenation stress. Similarly high concentrations of glutathione are observed in red-eared slider turtles40.