**3. Results**

#### *3.1. Pregnancy-Induced Changes in Cerebellar Levels of Amino Acids and OGDHC Activity*

Table 1 compares the average levels of the OGDHC activity and 15 quantified amino acids in cerebella of the non-pregnan<sup>t</sup> and pregnan<sup>t</sup> rats.


**Table 1.** Influence of pregnancy on the levels of cerebellar amino acids and 2-oxoglutarate dehydrogenase complex (OGDHC) activity.

Amino acids are given in alphabetical order. Proteinogenic amino acids are indicated by their standard abbreviations, GABA-γ-aminobutyric acid. Levels of the amino acids and OGDHC activity are presented as mean ± SEM in μmol and μmol per min, correspondingly, per g of the tissue fresh weight. *p* values indicate significance of the differences between the two groups, estimated by the Mann-Whitney U-test. Significant (*p* ≤ 0.05) differences are indicated in bold italics. Seventeen non-pregnan<sup>t</sup> and eight pregnan<sup>t</sup> rats were used in the comparison.

The data presented in Table 1 show that most of the differences between the amino acid pools in cerebella of the non-pregnan<sup>t</sup> and pregnan<sup>t</sup> rats are within the quantification errors. However, pregnancy increases the levels of cerebellar glutamate and tryptophan by 22 and 57%, respectively.

*3.2. Hypoxia-Induced Changes in Cerebellar Pool of Amino Acids Depend on Physiological State*

Table 2 shows that hypoxia induces significant changes in the cerebellar pool of amino acids of non-pregnan<sup>t</sup> female rats. Half of the quantified amino acids (Arg, Glu. Lys, Met, Phe, Ser, Trp) undergo statistically significant increases.


**Table 2.** Influence of acute hypobaric hypoxia on the OGDHC activity and amino acid pool in cerebella of the non-pregnan<sup>t</sup> and pregnan<sup>t</sup> rats.

Amino acids are abbreviated as in Table 1. Levels of amino acids and OGDHC activity are presented as mean ± SEM in μmol and μmol per min, correspondingly, per g of the tissue fresh weight. *p* values indicate significance of the differences between the two groups, estimated by the Mann-Whitney U-test. Significant (*p* ≤ 0.05) differences are shown in bold italics. Number of animals in the groups: 17 control non-pregnant, 8 hypoxic non-pregnant, 8 control pregnan<sup>t</sup> and 10 hypoxic pregnant.

In contrast, under the same metabolic stress of pregnan<sup>t</sup> rats, their cerebellar levels of amino acids remain unaffected (Table 2). Remarkably, the tryptophan level, which is increased by pregnancy, exhibits a trend (0.08) to decrease (from 0.11 to 0.09 μmol per g FW) after exposure of the pregnan<sup>t</sup> rats to hypoxia, while in the non-pregnan<sup>t</sup> rats hypoxia induces an opposite change in the tryptophan level, which increases from 0.07 to 0.12 μmol per g FW (Table 2).

#### *3.3. Interdependence of Cerebellar Levels of OGDHC Activity and*/*or Amino Acids Varies under Di*ff*erent Physiological Settings*

Metabolic pathways of amino acids have not only common transporters for certain groups of amino acids, but also many common substrates. For instance, transaminase reactions link the levels of the corresponding pairs of the amino acids (aspartate and glutamate in the aspartate transaminase reaction), GABA is produced from glutamate etc. As noted in the introduction section, the amino acid metabolism in general is tightly coupled to the TCA cycle, whose metabolic flux is limited by the activity of OGDHC [4]. As a result, OGDHC inhibition strongly affects the amino acid pool in cerebellar granule neurons in culture [9] and amino acids levels in animal brain in vivo (reviewed in [6]). These metabolic features should cause interdependence of the tissue levels of amino acids. The interdependence is manifested in the correlated variations in the levels of amino acids and OGDHC activity in different animals, because the interindividual variability in the biochemical parameters is contributed by minor variations in the metabolic network organization. Indeed, such correlation analysis of the levels of the cerebellar OGDHC activity and 15 quantified amino acids in a sample of the non-pregnan<sup>t</sup> and pregnan<sup>t</sup> (Table 3) rats reveals a number of highly significant correlations between the studied biochemical parameters, pointing to a strongly interdependent content of di fferent amino acids in rat cerebellum, imposed by intercepts in their metabolism.

Visual inspection indicates that the correlation patterns change along with physiological settings (Table 3, the lower left triangles in A and B). The sets of strongly interacting parameters di ffer between the non-pregnan<sup>t</sup> and pregnan<sup>t</sup> rats. For instance, pregnancy drastically decreases the interdependences of Glu, Phe, Met with other amino acids and OGDHC, compared to those manifested in the cerebellum of non-pregnan<sup>t</sup> rats. On the other hand, some features of the metabolic network are preserved in both the non-pregnan<sup>t</sup> and pregnan<sup>t</sup> rats. That is, lysine and tryptophan show a low number of interactions in both physiological states (Table 3). Independent of pregnancy, none of the cerebellar amino acids show statistically significant correlation with OGDHC activity, although cerebellar tryptophan tends to positively correlate with OGDHC in both non-pregnan<sup>t</sup> (R = 0.79, 0.061) and pregnan<sup>t</sup> (R = 0.85, 0.07) rats (Table 3).

In our previous work [23], an approach to quantify subtle physiological shifts in metabolic networks, based on the data of correlation analysis, has been developed. It employs reduction of the correlation matrix dimension, resulting in a small number of easily comparable parameters presented in Table 4. The parameters are: the summarized and averaged correlation coe fficients for each element of the matrix, which characterize an overall degree of interdependence of the elements in a specific (patho) physiological state, and the total number of significant (*p* < 0.05) correlations (positive or negative) in the matrix, which indicates how many strong and very strong interdependences of the studied elements are inherent in a metabolic network. According to the data presented in Table 4, the summarized and averaged correlation coe fficients are significantly lower in the pregnan<sup>t</sup> rats, compared to the non-pregnan<sup>t</sup> ones, pointing to generally lower interdependences in the amino acid levels in the former than the latter. The lower interdependence is consistent with the lower number of statistically significant correlations in the pregnan<sup>t</sup> vs. non-pregnan<sup>t</sup> rats (38 vs. 62 for the positive correlations and 0 vs. 4 for the negative correlations, Table 4), although this di fference does not reach statistical significance.


**Table 3.** Correlation matrices characterizing interdependence between the amino acid levels and OGDHC activity in the cerebella of the non-pregnant and pregnant

#### *Cells* **2020**, *9*, 139


**Table 3.** *Cont.*



For the OGDHC activity (OGDHC) and each of the amino acids, the sum of its correlation coefficients (absolute values) to other amino acids (Σ), average correlation coefficient (X), and total number of statistically significant positive (+) and negative (−) correlations are shown. At the bottom, the sum (Σs, positive and negative correlations) or average (Xs) of all the values in the row and *p* values of the differences between the parameters in the control and hypoxia groups, estimated by the Wilcoxon signed rank test, are shown, with the statistically significant differences between the control and hypoxia groups in bold. #-Significant (*p* ≤ 0.05) differences between the parameters of the corresponding non-pregnant and pregnant groups.

#### *Cells* **2020**, *9*, 139

#### *3.4. Concerted Hypoxia-Induced Shift to the Negative Correlations between the Levels of Amino Acids and OGDHC Activity Is not Observed in Pregnancy*

Exposure of rats to an environmental challenge, such as acute hypobaric hypoxia, strongly changes the correlation matrices (Table 3, the upper right triangles) and their overall parameters (Table 4). Moreover, the hypoxia-induced changes in the interdependences are well-detectable in both the non-pregnan<sup>t</sup> and pregnan<sup>t</sup> rats (Tables 3 and 4), in contrast to the changes in the average levels of amino acids and OGDHC activity, which are detectable in the non-pregnan<sup>t</sup> rats only (Table 2). That said, the correlation analysis reveals a very di fferent response to hypoxia of the interdependences between OGDHC activity and/or amino acids in the non-pregnan<sup>t</sup> and pregnan<sup>t</sup> rats. Thus, not only the control correlation matrices, but also those after exposure of animals to hypoxia are very di fferent in non-pregnan<sup>t</sup> and pregnan<sup>t</sup> rats (Table 3, the upper right triangles). As shown in Table 3 hypoxia induces a high number of negative correlations of amino acids with OGDHC in non-pregnan<sup>t</sup> rats. Induction by hypoxia of the negative correlations between the amino acid levels and OGDHC activity exclusively in non-pregnan<sup>t</sup> rats (Table 3, the upper right triangle) thus coincides with the hypoxic reactivity of the average levels of amino acids in non-pregnan<sup>t</sup> rats only (Table 2). In contrast, in pregnan<sup>t</sup> rats hypoxia induces one positive correlation between OGDHC and methionine and one negative correlation between the amino acids tryptophan and phenylalanine (Table 3, the upper right triangle).

The overall interdependence parameters (Table 4) point to statistically significant di fferences in the hypoxic responses of cerebellum in both the non-pregnan<sup>t</sup> and pregnan<sup>t</sup> rats. This finding indicates that analysis of the correlation matrices (Table 4) detects the responses to hypoxic stress with a much higher sensitivity, compared to the average levels of amino acids, which did not reveal any changes in the cerebellum of pregnan<sup>t</sup> rats after hypoxia (Table 2). Yet both markers of hypoxic stress point to di fferent responses to hypoxia in the di fferent physiological states. The strong negative correlations between the levels of amino acids and OGDHC activity after hypoxic exposure of non-pregnan<sup>t</sup> rats (Table 3) coincide with the hypoxia-increased levels of the amino acids in these rats (Table 2) and with the overall increase in the interdependence between the OGDHC activity and/or amino acid levels, indicted by the summarized and averaged correlation coe fficients and number of statistically significant correlations (Table 4). Contrary, in pregnan<sup>t</sup> rats, demonstrating no changes in the average levels of amino acids after hypoxia (Table 2), no negative correlations of the amino acid levels with OGDHC activity are induced by hypoxia (Table 3), and the overall interdependence is diminished, based on the summarized and averaged correlation coe fficients between the studied parameters (Table 4). Remarkably, the number of statistically significant positive correlations between the parameters is increased by hypoxia in pregnan<sup>t</sup> rats too (from 38 to 63, Table 4). However, in accordance with a generally decreased interdependence, evident from the decreased correlation coe fficients (summarized and average, Table 4), the increased number of statistically significant positive correlations is expressed less than in non-pregnan<sup>t</sup> rats (Table 4). The most significant contribution to increases in the positive correlations is provided by di fferent amino acids; in non-pregnan<sup>t</sup> rats they are histidine, lysine, and tyrosine, whereas in pregnan<sup>t</sup> rats they are methionine, serine, and phenylalanine (Tables 3 and 4). Thus, hypoxia a ffects di fferent metabolic pathways in the cerebella of pregnan<sup>t</sup> and non-pregnan<sup>t</sup> rats.

#### *3.5. Physiological Consequences of the Hypoxia-Induced Changes in the Interdependent Levels of OGDHC Activity and*/*or Amino Acids in Cerebellum*

In hypoxic experiments, rats are exposed to acute hypobaric hypoxia until they collapse, which is registered as apnea. As described in the methods section, the time between the established hypoxic condition (5% O2) and apnea, i.e., the life time (LT), characterizes individual and/or group di fferences in the resistance to hypoxia according to the arbitrary intervals of the LT, indicated in the legend to Figure 1. Figure 1 shows that distribution of a sample of the rats into the animals with varied resistance to hypoxia di ffers dependent on the physiological state. Pregnant rats are characterized by decreased fraction of the animals with high resistance to hypoxia, with the corresponding increase

in the fraction of the low resistant rats. The difference is also manifested in a higher average value of LT for non-pregnan<sup>t</sup> rats (369 ± 45.1 s), compared to LT of the pregnan<sup>t</sup> rats (277 ± 37 s) (n = 41; 0.048, according to Mann-Whitney test). Thus, the reactivity to hypoxia is higher (Figure 1) when no significant changes in the OGDHC activity or amino acid levels, neither any negative correlations between these parameters occur in the cerebellum, as observed in the pregnan<sup>t</sup> rats (Tables 2–4).

**Figure 1.** Distribution of the non-pregnan<sup>t</sup> (*n* = 41) and pregnan<sup>t</sup> (*n* =31) rats according to their relative resistance to acute hypobaric hypoxia. HR—highly resistant rats (dark-grey), spent under hypoxia ≥10 min; MR—medium resistant rats (light-grey), spent under hypoxia between 5 and 10 min; LR—low resistant rats (middle-grey), spent under hypoxia ≤5 min. % of each sub-group to total number of animals in the group is shown in shades of grey. The indicated *p* value is estimated using Fisher exact test.

Accordingly, the hypoxia-induced changes in the average levels (Table 2) and/or interdependence (Tables 3 and 4) of cerebellar OGDHC activity and amino acids, which are observed in the non-pregnan<sup>t</sup> rats, are of compensatory significance. That is, the changes in the amino acid pool (Table 2) and its dependence on OGDHC activity (Table 3), observed in the non-pregnan<sup>t</sup> rats and absent in the pregnan<sup>t</sup> ones, obviously enable the former to resist hypoxia better than the latter (Figure 1).
