Hydrophobic interactions determine the optimum temperature of a housekeeping enzyme

TL;DR

This study shows that small hydrophobic interactions in a key enzyme's structure determine its optimal working temperature across different chordate species.

Contribution

The study identifies hydrophobic interactions in specific amino acid regions as the main determinant of AK1 optimum temperatures.

Findings

AK1 optimum temperatures correlate with the normal body temperature of 11 chordate species.

A predictive model using hydrophobic interactions in specific regions explains AK1 optimum temperatures with high accuracy.

Sequence similarity and phylogenetic relationships do not correlate with AK1 optimum temperatures.

Abstract

Chordates have adapted to diverse thermal environments, with poikilotherms adjusting to external temperatures and homeotherms maintaining stable body temperatures. While housekeeping enzymes conserve their activities, they function at different body temperatures among species. However, the determinants for the optimum temperatures of housekeeping enzymes largely remain unclear. In this study, we identified the determinants of the optimum temperatures of chordate adenylate kinase 1 (AK1), a key housekeeping enzyme. The optimum temperatures of AK1s were shown to be closely correlated with the normal body temperature of each chordate (11 species). A combination of enzymatic assays, computational analyses of numerous physicochemical interactions, and structural dynamics analyses of intact and mutant AK1s verified that the number of hydrophobic interactions among four amino acids in specific secondary structures (the 4th–17th regions) is a major determinant of the optimum temperatures of chordate AK1s. This allowed us to generate a predictive model for the optimum temperatures of native chordate AK1s: AK1 optimum temperature = 2.3587 × (the number of 4th–17th interactions) + 0.3663 × (the number of 13th–17th interactions) –5.9695, with a maximum error of ±1.70 °C. In contrast, sequence similarity and molecular phylogenetic relationships did not correlate with the optimum temperatures of chordate AK1s. Furthermore, these results suggest an evolutionary scenario for thermal adaptation of chordate AK1. Collectively, our study provides evidence that subtle hydrophobic interactions play a crucial role in determining the temperature preferences of a chordate housekeeping enzyme, offering new insights into the functional evolution and diversification of enzyme homologs.

Figures

Full text

Animals typically inhabit environments that align with their preferred temperatures. In homeotherms, body temperature is tightly regulated, while in poikilotherms, body temperature fluctuates in response to environmental conditions. Chordates, encompassing both poikilotherms (amphioxus, ascidians, cyclostomes, fish, amphibians, and reptiles) and homeotherms (birds and mammals), exhibit remarkable diversity in form and habitat, ranging from terrestrial to aquatic environments and spanning equatorial to polar regions. Poikilotherms are more susceptible to environmental temperature changes, which limit their habitats and active periods (1, 2, 3, 4, 5, 6, 7). In contrast, homeotherms maintain a constant body temperature through homeostatic mechanisms, enabling them to adapt to varying environmental temperatures (8, 9).

Given these differences, it is hypothesized that the proteins within these organisms have also evolved to function optimally at species-specific body temperatures. Temperature is thus a critical factor influencing protein function in both poikilotherms and homeotherms (10). The temperature preferences for an organism are implicated in the functionality of its proteins, including enzymes. Housekeeping enzymes, which are evolutionarily conserved across species, are believed to share core biochemical activities but their temperature preferences, including optimum temperatures, have yet to be fully investigated. Understanding the factors that determine the optimum temperatures of homologous enzymes could shed light on the evolutionary processes that drive diversity among organisms. Previous studies have used bioinformatics approaches, including sequence-activity databases such as BRENDA and machine learning, to provide insights into optimum temperatures for enzymes (11, 12). Moreover, structural information on the heat resistance of several enzymes in bacteria including thermophilic Bacillus and others has been reported (13, 14, 15, 16). Such approaches have revealed how particular amino acid residues and structural features contribute to enzyme thermostability, providing important insights into the evolutionary adaptations of thermophilic organisms.

In contrast, it should be noted that evolutionarily conserved housekeeping enzymes exhibit markedly different physicochemical properties between bacteria and vertebrates. For example, dihydrofolate reductases from Escherichia coli and mammals share an overall fold but differ greatly in distinct active-site dynamics (17, 18). Isocitrate dehydrogenase shows lineage-specific differences in allosteric regulation (19). Together, these examples illustrate that even highly conserved enzymes can evolve distinct kinetic and regulatory mechanisms depending on organismal physiology and thermal environment. In other words, these previous studies illustrate how the mechanisms identified in prokaryotic proteins may differ from those of their homologs in eukaryotes, including chordates. However, there is limited experimental evidence on temperature sensitivity and structural dynamics in these species.

Adenylate kinase 1 (AK1) is a housekeeping enzyme that catalyzes the conversion of one ATP and one AMP into two ADPs in both eukaryotes and prokaryotes. AK1 has been extensively studied as a model protein in structural chemistry and biology (20, 21, 22). Given its conserved enzymatic function across species, AK1 homologs are expected to sustain ADP production. While kinetic parameters such as turnover rates or substrate affinity may vary among chordate species adapted to different physiological backgrounds and thermoregulatory strategies, the optimum temperature of an enzyme is thought to be primarily constrained by universal physicochemical selection pressures acting on protein stability. Nevertheless, neither the optimum temperatures of nonmammalian AK1s nor their underlying structural mechanisms have been investigated, despite the expectation that AK1 activity aligns with host chordate body temperatures.

In this study, we identified the determinants for the temperature preferences including the optimum temperatures of AK1 across chordate species. We demonstrated a prominent correlation between the optimum temperatures of AK1s and the body temperatures of chordates. Subsequently, we verified that the optimum temperatures of AK1s are determined by the number of hydrophobic interactions within several hydrophobic cores of AK1s in each species. Furthermore, we defined a simple theoretical equation linking the optimum temperature to the number of hydrophobic interactions in native AK1s, providing important insights into the molecular evolution and diversification of enzyme functions in chordates.

Results

Optimum temperature of chordate AK1s

The optimum temperatures of nonmammalian chordate AK1s have not been extensively examined. Thus, we measured AK1 activities from 11 chordate species at 15 to 65 °C (Figs. 1A, S1, and Table S1). The optimum temperatures were correlated with the normal body temperatures of each species, with poikilotherm values approximated from habitat temperatures (Fig. 1B). In all cases, AK1 optimum temperatures exceeded the body or environmental temperatures of the organisms and numerous similar cases have been reported, at least under in vitro conditions (15, 23, 24). The correlation between optimum temperatures and body/environmental temperatures was significant (r^2^ = 0.47). For example, Gallus gallus (body temperature: 41 °C) showed an optimum AK1 temperature of 53 °C, higher than those of mammals such as Homo sapiens (37 °C), Mus musculus (37.5 °C), and Heterocephalus glaber (∼32 °C) (8, 9), whose AK1 activities peaked at 50 °C. Similarly, Ornithorhynchus anatinus (body temperature: 28 °C) (25) exhibited a lower optimum temperature of 43 °C than other mammals. In poikilotherms, reptiles and tropical aquatic animals, such as Anolis carolinensis (20–40 °C) (1), Danio rerio (22–31 °C) (2), and Xenopus laevis (10–25 °C) (3), exhibited AK1 optima near 45 °C. Similarly, the lamprey Petromyzon marinus (17.8–21.8 °C) (4) also had an optimum of 45 °C, suggesting a common feature among poikilotherms. AK1s from marine invertebrate chordates, an urochordate Ciona intestinalis (10–23 °C) (5) and cephalochordate Branchiostoma floridae (15–23 °C) (6, 7) exhibited lower optimum temperatures at 36 °C and 25 °C, respectively. These findings suggest that AK1 optimum temperatures increased during the evolutionary transition from poikilotherms to homeotherms.Figure 1**Temperature-dependent activity of chordate AK1s, and the relationship between the optimum temperatures of chordate AK1s and their respective body temperatures, and relationship between optimum temperatures and sequence similarity of AK1.**A, temperature-dependent activities of chordate AK1s measured at 15 to 65 °C. Activity is expressed as relative activity (%), with the maximum measured activity for each AK1 defined as 100%. B, relationship between optimum temperatures of AK1 activity and body temperatures. Solid lines indicate the range of temperature in which each organism is normally active, and dashed lines indicate the range of temperature in which it can survive. C, molecular phylogenetic trees of AK1 sequences using MEGA6 (33), and amino acid sequence similarity to the Ciona intestinalis AK1 sequence plotted against the optimum temperature of enzymatic activity. Sequence similarity was calculated based on BLAST positives rather than sequence identity. The optimum temperatures were estimated based on temperature–activity profiles (panel A and Fig. S1), using bootstrap analysis. All values fall within ±3 °C at 95% confidence interval (p = 0.05). ∗ denotes homeothermic species, ∗∗ denotes semi-homeothermic species, and the remaining species are classified as poikilotherms. AK1, adenylate kinase 1.

Notably, AK1s with lower optimum temperatures (≤40 °C) retained over 40% activity even at temperatures 10 °C above their optimum temperatures, while those with higher optimum temperatures (≥50 °C) lost over 80% of activity under the same conditions. These results indicate distinct temperature preferences among chordate AK1s. Additionally, the optimum temperatures of AK1s were found to be close to the temperatures at which their structures became unstable due to thermal denaturation (Fig. 2). These results indicate that AK1s function at body temperatures lower than their optimum temperatures to maintain structural stability. Because the optimum temperatures of AK1s are governed mainly by structural stability rather than by catalytic efficiency, detailed kinetic analyses were not pursued in this study. In general, physicochemical features of an enzyme are correlated with its amino acid sequence. Therefore, we examined a relationship between the optimum temperature and sequence similarity of AK1s (Figs. 1C and S2 and S3). However, cladogenetic order and similarity of AK1 sequence are not implicated in the optimum temperature. For example, the AK1 sequence of G. gallus is more similar to that of C. intestinalis (83.1%) than that of D. rerio (74.5%), but the optimum temperature of AK1 of D. rerio (46 °C) is closer to that of C. intestinalis (36 °C) than that of G. gallus (53 °C). Taken together, these results corroborate that other factors besides sequence similarity serve as determinants for optimum temperatures of the highly conserved enzyme.Figure 2Thermal stability measurement of AK1 from H. sapiens, G. gallus, X. laevis, and C. intestinalis by CD spectrometry. The green curves represent the CD changes of AK1 during heating. The black curves are fits to the heating data, and the blue lines show native and denatured state of AK1s. The red lines and numbers indicate the midpoint temperature of denaturation (50% denatured). The light blue lines and numbers indicate the optimum temperature of AK1s, demonstrating that AK1s begin to deviate from their native state (orange triangles and numbers) at around these temperatures. AK1, adenylate kinase 1.

Determinants of optimum temperatures for AK1 activity

We thus focused on secondary structure interactions, since the number of interactions (hydrophobic interaction, hydrogen bonds, and charge interactions) between secondary structures was expected to be correlated with their optimum temperatures responsible for formation of their active protein configurations. Notably, interaction analysis (see Materials and Methods) revealed that the cumulative number of hydrophobic interactions showed a prominent correlation with the optimum temperatures, while hydrogen bonds and charge interactions had no significant effect (Fig. 3A). We subsequently examined the regions harboring hydrophobic interactions responsible for the determination of the optimum temperatures in AK1. The secondary structure of human AK1 comprises 28 regions, and the number of hydrophobic interactions between each region was estimated across all secondary structure pairs (Fig. 3B and Table S2). Notably, markedly abundant hydrophobic interactions were observed between region 13 and 17 (interaction 13–17), and between 6 and 27 (interaction 6–27), suggesting that these areas were involved in determination of the optimum temperatures. It is also noteworthy that no optimum temperature–related changes were observed in hydrogen bonds and charge interactions (Table S2).Figure 3**Hydrophobic interaction among secondary structures of AK1s.**A, the correlations were analyzed between optimum temperatures and each physicochemical interaction, such as hydrophobic interactions, hydrogen bonds, and charge interactions (electrostatic interaction), and total of them. B, twenty-eight divisions of AK1 and number of hydrophobic interactions between each pair of secondary structure regions (α-helix, β-strand, and others). Red rods, yellow arrows, and black bars indicate α-helixes, β-strands, turns, and others, respectively. Hydrophobic interactions of other organism AK1s shown in Table S2. C, correlation coefficients for the relationship between optimum temperatures and the number of hydrophobic interactions between each secondary structures (only region pairs with five or more hydrophobic interactions). D, all hydrophobic interactions between the fourth (light blue) and 17th (light green) secondary structures were governed by the positions and orientations of four amino acids: Ile12, Phe14, Phe107, and Ile111 (human AK1). AK1, adenylate kinase 1.

To examine the crucial determinants for the optimum temperature in greater detail, we calculated correlation coefficients for the region pairs with five or more interactions correlated with the optimum temperature. As shown in Figures 3C and S4, the correlation coefficient for the optimum temperature and the number of hydrophobic interactions 4 to 17 exhibited an exceptionally high value (correlation coefficient r^2^ = 0.96). These results confirmed the pivotal role of the interaction 4 to 17 as a determinant for the optimum temperature. Notably, interaction analysis (see Experimental procedures) demonstrated that all hydrophobic interactions between the 4th and 17th secondary structures were governed by the positions and orientations of four amino acids: Ile12 (in the fourth region), Phe14 (in the fourth region), Phe107 (in the 17th region), and Ile111 (in the 17th region) in human AK1 (Fig. 3D). In contrast, the abundantly present interaction between positions 6 and 27 exhibited a low correlation coefficient (r^2^ = 0.36). Several other regions also exhibit notable correlations between the optimum temperatures and their interaction counts (Fig. 3C). To comprehensively verify the influence of these interactions on the optimum temperature, we conducted linear multiple regression utilizing correlation coefficients of 0.65 or higher as criteria (Table S3). In general, multiple regression is considered significant, when the t value is 2 or higher and the p value is 0.05 or lower. Consequently, only "interaction 13 to 17" and "interaction 4 to 17" met these criteria among more than 400 interactions. Subsequently, we conducted another round of multiple regression analysis using only interaction 13 to 17 and 4 to 17 (Table S4). Collectively, these multiple regression analyses deduced the following formula to determine the optimum temperature:

$$[eqn]$$

This formula predicted optimum temperatures with high accuracy, with a maximum error of +1.70 °C for G. gallus (Table 1). In addition, the 4 to 17 interaction was 6.44-fold more influential than the 13 to 17 interaction. Collectively, interaction 4 to 17 and 13 to 17 constitute components of a shared hydrophobic core as crucial determinants for optimum temperatures of AK1 (Fig. S5).Table 1. Comparison of equation 1-deduced optimum temperatures with observed onesInteractionsObserved (°C)Predicted (°C)Difference4–1713–17Gallus gallus19345351.301.70Mus musculus19345051.30−1.30Homo sapiens19335050.93−0.93Heterocephalus glaber19305049.840.16Ornithorhynchus anatinus16334343.86−0.86Anolis carolinensis17344646.58−0.58Xenopus laevis17344646.58−0.58Danio rerio16354644.591.41Petromyzon marinus16344544.220.78Ciona intestinalis14223635.110.89Branchiostoma floridae10222525.68−0.68AK1, adenylate kinase 1.

We applied this formula to estimate the optimum temperatures of AK1 in 57 chordate species (Table S5). As anticipated for homeothermic mammals (Mammalia) and birds (Aves), most species consistently exhibited 19 instances of the 4 to 17 interaction, resulting in minimal variation in predicted optimum temperatures. Interestingly, Tachyglossus aculeatus and Dasypus novemcinctus, which are phylogenetically closer to reptiles than other mammals and display weaker homeothermy, were predicted to have optimum temperatures approximately 7 °C lower than that of human AK1—consistent with their physiological characteristics. Body size in poikilothermic species also appeared to correlate with predicted optimum temperatures. For instance, Chondrichthyes, generally larger than Actinopterygii, showed higher predicted values. Similarly, Dermochelys coriacea, the largest among its close relatives (Caretta caretta and Trachemys scripta), exhibited a higher predicted optimum temperature than them. Collectively, these findings suggest that the optimum temperatures of AK1 in a broad range of chordates are governed by the interaction-based mechanism captured by Equation 1. In contrast, of particular interest is that, although Equation 1 predicts that AK1 homologs of prokaryotes such as E. coli (UniProtKB ID: B1XFR1) and Thermus thermophilus (NCBI ID: WP_011173701.1), based on their optimal growth temperatures, should exhibit more than 19 hydrophobic interactions in the 4 to 17 structure, they possess only 1 and 2 interactions, respectively (Table S6). Altogether, these data highlighted a fundamental mechanistic divergence between chordates and microorganisms in the determinants of AK1 thermal properties (13, 14, 15, 16).

Temperature-sensitive regions identified by H/D exchange analysis

The hydrophobic interaction-rich regions were expected to contribute to temperature sensitivity of AK1. To identify these regions, we employed hydrogen/deuterium (H/D) exchange using MALDI-TOF mass spectrometry. In this method, hydrogens in temperature-sensitive regions are replaced by deuteriums at higher temperatures, leading to the identification of temperature-sensitive regions responsible for the enzymatic stability (26, 27). AK1 from H. sapiens, G. gallus, X. laevis, and C. intestinalis was incubated in 90% deuterium water, followed by quenching at specific pH and temperature conditions and digestion with pepsin.

The AK1s were segmented into regions to quantify deuterium incorporation: C. intestinalis was partitioned into eleven regions, while the other species were analyzed in nine regions. For example, human AK1 was divided as follows: residues 1 to 14, 14 to 45 (derived by subtracting the H/D exchange data between 1–45 and 1–14), 46 to 75, 74 to 82, 83 to 106, 107 to 118, 119 to 165, 166 to 185, and 185 to 196 (similarly derived by subtracting data between 166–195 and 166–185). Of particular interest is that two common regions, residues 74 to 82 (including the 13th secondary structure) and residues 107 to 118 (including the 17th secondary structure), in H. sapiens and other species exhibited pronounced temperature sensitivity (Figs. 4A and S6). H/D exchange increased steadily up to the respective optimum temperatures and plateaued at approximately 90% at temperatures beyond the optimum temperatures. Region 185 to 196 also exhibited temperature-dependent deuterium incorporation; however, in contrast to regions 74 to 82 and 107 to 118, D_2_ incorporation in this region reached a plateau at temperatures below the optimum temperature. This behavior suggests that region 185 to 196 undergoes temperature-dependent structural changes that are not directly coupled to the determination of the optimum temperature. Moreover, interaction analysis (see Materials and Methods) demonstrated that the 13th and 17th secondary structures formed a single hydrophobic core (Fig. 4B). Collectively, these results indicated that the key regions for the hydrophobic interactions (13th and 17th secondary structures) are located in the temperature-sensitive regions, thereby demonstrating that the number of hydrophobic interactions is closely linked to temperature sensitivity in AK1. Furthermore, as the temperature approached the optimum, these regions incorporated more deuterium, suggesting increased structural flexibility that correlates with the temperature dependence of enzymatic activity (Fig. 4A). Together, the H/D exchange data confirm that the hydrophobic interactions within regions 4 to 17 and 13 to 17 are key determinants of the optimum temperatures of chordate AK1s.Figure 4**Identification of temperature-sensitive regions using H/D exchange analysis.**A, H/D exchange analysis of temperature-sensitive regions in H. sapiens, G. gallus, X. laevis, and C. intestinalis AK1. The upper panel provides an overview of the relationship between the 28 divided secondary structure segments and the primary sequence, with red squares on the sequence indicating pepsin digestion fragments used in the H/D exchange analysis. The purple and green boxes represent the time-course of deuterium uptake for residues 74 to 82 in H. sapiens and residues 107 to 118, respectively, tested at various temperatures. B, the temperature-sensitive regions in H. sapiens AK1. The ribbons, arrows, and wires represent α-helices, β-sheets, and others, respectively, while the structure depicted in sticks represents the substrate analog. The red dashed circle indicates the location of the hydrophobic core. The purple and green regions represent residues 74 to 82 and residues 107 to 118, respectively. AK1, adenylate kinase 1; H/D, hydrogen/deuterium.

Functional correlation between hydrophobic interactions and optimum temperatures using region-Exchanged mutants

To further explore the correlation between hydrophobic interactions and optimum temperature, we generated mutants by swapping temperature-sensitive regions between human AK1 (optimum temperature: 50 °C) and Ciona AK1 (36 °C). These regions, identified through H/D exchange analysis, included residues 74 to 82 (13th secondary structure) and residues 107 to 118 (17th secondary structure).

We replaced the temperature-sensitive regions of Ciona AK1 with those of Human AK1, generating Hu_13th_Ciona AK1 and Hu_17th_Ciona AK1 mutants. It is also noteworthy that these mutants were shown to possess a greater number of hydrophobic interactions within the regions 4 to 17 than native Ciona AK1 by interaction calculation (Table 2). The optimum temperatures of Hu_13th_Ciona AK1 and Hu_17th_Ciona AK1 were predicted to increase based on Equation 1, and, consistent with this prediction, these mutants exhibited higher optimum temperatures compared to native Ciona AK1, as shown in Fig. S7A and Table 2.Table 2. Optimum temperatures of swap mutants4–17 Interactions13–17 InteractionsOptimal temperatureRelative activityelative activity (WT = 100%) %Predicted °CObserved °CCiona AK1 Wild142235.11**↑36↑100 Region 1 → human152036.74↑40↑32 Region 2 → human142435.844234Human AK1 Wild193350.9350100 Region 1 → ciona193451.30↑55↑93 Region 2 → ciona192949.47↓43↓**59Hydrophobic interactions and optimum temperature of the temperature-sensitive region-swapped AK1 mutants (Human and Ciona). The predicted optimum temperatures were calculated with Equation 1. Arrows indicate the increase and decrease of optimum temperatures in mutants, compared with that of the WT AK1. Relative activity is expressed as a percentage of the maximal activity of the corresponding WT AK1 (WT = 100%); absolute activities of the WT enzymes are provided in Table S1.AK1, adenylate kinase 1.

Subsequently, we generated Ciona_13th_Human AK1 and Ciona_17th_Human AK1 mutants, in which the temperature-sensitive regions of human AK1 were replaced with the corresponding regions from Ciona AK1. Notably, the number of hydrophobic interactions within the 4th–17th secondary structures of these mutants was predicted to be more and fewer than those of native Human AK1 (Table 2). As estimated by Equation 1, Ciona_13th_Human AK1 and Ciona_17th_Human AK1 also showed higher and lower optimum temperatures (Fig. S7A and Table 2). Taken together, these results further highlighted the crucial role of hydrophobic interactions in determining the optimum temperatures.

To further examine the contribution of individual amino acids within the residues 74 to 82 to temperature sensitivity, we created single amino acid-reverted mutants (Ci_13th_89re, 90re, 91re, 92re, and 94re Human AK1). Enzymatic assays revealed that none of these single mutations restored the optimum temperature to 55 °C (Fig. S7B), indicating that optimum temperature is not determined by a single amino acid but by the collective effects of multiple hydrophobic interactions affected by the temperature-sensitive regions. Collectively, these results indicate that the temperature-dependent dynamics of the 13th and 17th regions regulate the number of hydrophobic interactions, which is responsible for the elucidation of the optimum temperatures of chordate AK1s.

Discussion

Animals maintain body temperatures suitable for their survival, but the optimum temperatures of a non-mammalian chordate housekeeping enzyme, AK1, have not been systematically measured. Furthermore, the molecular and structural determinants of the optimum temperatures remain unknown. As shown in Figure 1C, neither sequence similarity nor molecular phylogenetic relationships of AK1, a highly conserved housekeeping enzyme present in all organisms, were correlated with its optimum temperatures or temperature-dependent activities. Current structural analyses, including X-ray crystallography, cryo-EM, and computational predictions including the AlphaFold systems (28), have not provided insights into the molecular determinants underlying temperature sensitivity.

To address these issues, we integrated biochemical assays, computational estimates of hydrophobic interactions within enzyme cores, and H/D exchange analyses for an evolutionarily conserved housekeeping enzyme, AK1. These approaches verified a prominent correlation between the optimum temperatures of chordates and the optimum temperatures of AK1 (Fig. 1B), and uncovered a previously unrecognized correlation between the optimum temperature and the number of hydrophobic interactions among specific secondary structures (4th–17th and 13th–17th secondary structures) (Fig. 4) modulated by temperature-sensitive regions in AK1: residues 74 to 82 (13th secondary structure) and residues 107 to 118 (17th secondary structure) (Fig. 4B). These results indicated that the temperature-dependent conformational dynamics of these secondary structure regions regulate the number of hydrophobic interactions between the 4th and 17th regions in a temperature-dependent manner. Furthermore, predicted optimum temperatures of native AK1s using Equation 1 were consistent with the observed optimum temperatures with a maximum deviation of 1.70 °C (Table 1). Because Equation 1 was derived from experimentally characterized AK1s, this predictive accuracy reflects the internal consistency of the empirical relationship between hydrophobic interactions and optimum temperature, and independent validation will be required to assess its broader generality. These results also indicate that chordates utilize a mechanism fundamentally distinct from that of bacteria, such as Thermus and Escherichia, which possess extremely few 4 to 17 interactions (Table S6). Consequently, the optimum temperatures of these bacterial AK1s are expected to be determined through alternative mechanisms, such as extensive hydrogen-bonding networks or other non-hydrophobic interactions, as previously reported in bacterial enzyme studies (13, 14, 15, 16). Interestingly, by contrast, comparative analyses of AK1 from a psychrophilic bacterium (Bacillus globisporus) and a mesophilic bacterium (Bacillus subtilis) also demonstrated that optimization of hydrophobic packing in the CORE domain is a primary determinant of structural stabilization (16), suggesting that hydrophobic interactions are partially involved in determining the optimum temperatures of bacterial AK1. Thus, the hydrophobic interaction-regulatory mechanisms in the determination of optimum temperatures might have diverged via distinct evolutionary processes of AK1 homologs (Tables S2 and S6). Although hydrophobic interactions between the 13th and 17th secondary structure elements were also correlated with optimum temperature, their contribution appears to be modulatory rather than determinant. At present, we do not observe a clear mechanistic relationship between the 13 to 17 interaction and optimum temperature comparable to that of the 4 to 17 interaction, suggesting that the 13th region primarily fine-tunes the hydrophobic environment surrounding the 4 to 17 interaction network.

AK universally adopts a conserved three-domain architecture consisting of the CORE, LID, and NMP domains (Fig. S8). Previous studies on non-chordate adenylate kinases have collectively shown that these domains can fold and unfold semi-independently, that functional conformational transitions are accommodated primarily by localized unfolding in the LID and NMP domains, and that the CORE domain is thermodynamically more stable due to optimized hydrophobic packing, thereby playing a central role in determining the overall thermal stability of the enzyme (16, 29, 30).

In chordate adenylate kinases, including human AK1, our analysis further indicates that the largest intramolecular interaction network is localized within the CORE region (Fig. S8). Importantly, the hydrophobic interactions among the fourth, 13th, and 17th secondary structure elements form a densely connected network within the CORE domain, with the 4 to 17 interaction constituting the central stabilizing component of this network. These interactions support the hydrophobic core that underpins the structural stability of AK. Together, these observations support the view that the CORE domain serves as the principal stability-determining region of AK, and that this CORE-centered stabilization mechanism is conserved in chordate AK1. Therefore, in chordate AK1, where optimal enzymatic activity is closely linked to protein stability, it is reasonable that a hydrophobic cluster including the fourth, 13th, and 17th secondary structure elements plays a key role in determining the optimum temperature.

Of particular interest is that hydrophobic interactions between the 4th and 17th secondary structures were primarily determined by the positions and orientations of four amino acids: Ile12, Phe14, Phe107, and Ile111 (in human AK1) (Fig. 3D). These residues were conserved among chordate AK1s, with the exception of B. floridae AK1, which harbors Val12 instead (Fig. 5). Moreover, it is noteworthy that the number of hydrophobic interactions between these amino acids was positively correlated with optimum temperatures. For example, B. floridae AK1 (optimum temperature: 25 °C) exhibited only 10 interactions, while Ciona intestinalis AK1 (36 °C) formed 14 interactions involving the same residues (Fig. 6). Organisms with higher optimum temperatures exhibited 16 to 19 interactions among the same residues, reflecting the strong link between hydrophobic interactions and temperature sensitivity (Fig. 6). Interestingly, these key residues were distant from the active site across all chordate species, highlighting the indirect role of structural environment in regulating optimum temperatures (Figs. 4B, and 6).Figure 5**Hydrophobic interactions between 4th and 17th secondary structures.Light blue and green are colored fourth of secondary structure and 17th on tertiary structure of human AK1 and sequence alignment of chordate AK1, respectively. Structures drawn with sticks are Ile12, Phe14, Phe107, Ile111, and their amino acids are surrounded by open squares on the alignment. Dotted lines show hydrophobic interactions. AK1, adenylate kinase 1.Figure 6The relationship between hydrophobic interactions of 4 to 17 and optimum temperatures of each chordate and the evolutionary process of the number of hydrophobic interactions.**Light blue and green are colored fourth of secondary structure and 17th on tertiary structure of AK1 from B. floridae, D. rerio, and *H.*sapiens, respectively. Structures drawn with sticks are Ile12 (Val12 in B. floridae AK1 structure), Phe14, Phe107, Ile111, and dotted lines show hydrophobic interactions. Hydrophobic interactions originally acquired in each animal species are indicated as follows: hydrophobic interaction indicated by red dot lines in B. floridae AK1, brown ones in C. intestinalis AK1, blue ones in D. rerio, O. anatinus, and P. marinus AK1s, pink ones in X. laevis and A. carolinensis AK1s, and black ones H. sapience, H. glaber, M. musculus, and G. gallus AK1s. AK1, adenylate kinase 1.

Single-residue mutation experiments further confirmed that no single amino acid determined the optimum temperature (Fig. S7B). Instead, the collective impact of multiple residues within the hydrophobic core formed by two specific temperature-sensitive regions, residues 74 to 82 (including 13th secondary structure) and residues 107 to 118 (including the 17th secondary structure), was essential for determining temperature sensitivity of AK1 (Fig. 4). Moreover, the four amino acids, Ile12, Phe14, Phe107, and Ile111, are almost completely conserved in chordate AK1s (Figs. 5, and 6), whereas each chordate species AK1 exhibited distinct optimum temperatures and the number of hydrophobic interactions that are correlated with their body temperatures (Figs. 1 and 2 and Equation 1). Moreover, the temperature sensitivity of the two aforementioned regions was species-specific (Fig. 4). In combination, these results verified that the variation in optimum temperatures arises not from the “presence or absence” of the four conserved amino acids but from the broader structural environment surrounding these amino acids which is formed by the temperature-sensitive regions.

Given that ancestral chordates emerged over 530 million years ago, their fundamental biochemical systems, including ADP production by AK1, have remained largely conserved throughout their evolutionary history, despite diversification in lifestyle, habitat, and internal environment (31). The present study suggests a plausible evolutionary scenario (Fig. 7).

• 1.Amino acid mutations altered the number of hydrophobic interactions in AK1.
• 2.These interactions regulated the enzyme’s optimum temperature.
• 3.The resulting thermal properties of multiple proteins, including AK1 and other metabolic enzymes, might have contributed to smoother adaptation to environmental temperatures, thereby supporting ecological adaptability and biodiversity. Figure 7The evolutionary scenario for determination of the optimum temperatures of AK1. The evolutionary scenario for the diversification and accommodation of chordates, driven by the optimum temperature of housekeeping enzyme homologs defined by the number of hydrophobic interactions. AK1, adenylate kinase 1.

Collectively, the present study suggests that the number of hydrophobic interactions in housekeeping enzymes may affect the life cycles, lifestyles, and habitats of extant chordates. Further studies are needed to explore whether such a correlation between hydrophobic interactions and optimum temperature in specific temperature-sensitive hydrophobic regions holds for nonchordate AK1s or other housekeeping enzyme homologs.

In this study, we elucidated the relationship between body temperature, the temperature preferences of an evolutionarily conserved enzyme of chordates, and the number of hydrophobic interactions within specific regions. These findings provide unprecedented insights into the molecular mechanisms underlying functional evolution and diversification of a housekeeping enzyme, which could not be uncovered through molecular phylogenetics or structural analyses alone. Our results are expected to open new avenues for exploring the molecular and functional evolution of housekeeping enzyme homologs and the organisms that harbor them.

Experimental procedures

Cloning of AK1

Total RNAs (1 μg) of H. sapiens, G. gallus, X. laevis, and C. intestinalis were purified from HEK293MSR cell, bird muscle, egg, and ovary, respectively, and were reverse-transcribed to template complementary DNA at 55 °C for 60 min using the oligo(dT) anchor primer and SuperScriptIII RNase H-Reverse Transcriptase (Invitrogen). Each species AK1 complementary DNA was amplified by polymerase chain reaction using the primers listed in Table S7. Genes coding AK1s from D. rerio, B. floridae, H. glaber, M. musculus, A. carolinensis, O. anatinus, and P. marinus were designed for high expression in E. coli and synthesized by GenScript. The ORF of AK1 from G. gallus was subcloned in-frame into pGex 6p-1 (GE Healthcare) at EcoRI/XhoI. The ORF of AK1 from others was subcloned at EcoRI/SalI. The cutting sequence for PreScission Protease was inserted into the 5′ terminal of the AK1 ORF by PCR using the primers listed in Table S8.

Subcloned inserts were sequenced on an ABI PRISM 310 Genetic Analyzer (Applied Biosystems) using a Big-Dye sequencing kit (Applied Biosystems) and pGEX sequencing primers.

Mutation of AK1

Swap mutants of their temperature-sensitive regions from H/D exchange analysis were made by PCR using pGex 6p-1 vector inserted Human and Ciona AK1 and the primers listed in Table S9. AK1s expressed in E. coli were purified by the same method as WT. Single-residue mutants reverted to the human form in residues 74 to 82 of Ciona13thHuman AK1 were made by same methods.

Preparation of AK1

AK1 expression vectors were transformed into BL21 (DE3) and the bacteria were incubated in LB broth with ampicillin at 37 °C to an A_600_ of 0.8. Protein expression was induced by the addition of 1 mmol/L IPTG for 5 h at 37 °C. Cell pellets were resuspended in a buffer containing 20 mM Tris–HCl (pH 7.5), 100 mM NaCl, and 0.1% Tween 20, sonicated twice for 5 min on ice, then centrifuged at 20,000×g for 30 min. The supernatant was incubated with Glutathione Sepharose 4B (GE Healthcare) for 1 h at 4 °C. The resin was then washed five times with the same buffer, and the glutathione S-transferase tag was cleared by the addition of Precision Protease (GE Healthcare) and further incubation for 16 h at 4 °C. Glycine and proline were added to the N-terminal of AK1 to allow digestion of the glutathione S-transferase tag by Precision Protease. The AK1 protein was stored at −80 °C as an AK1 solution in PBS with 10% glycerol and 9.3 mM magnesium acetate.

Activity of AK1

The reactions of AK1s were started by mixing 2 μl of AK1 solution with 18 μl aliquots of 0.5 mg/ml ATP and AMP in PBS with 10% glycerol and 9.3 mM magnesium acetate. After 3 min, the reactions were stopped by adding 2 μl of 20% acetic acid. These reacted solutions were deproteinized with ziptip C-18 (Millipore). An LC/MS-8030 triple-quadrupole mass spectrometer (Shimadzu Corporation) was used for the LC–MS quantification of AK1-product ADP. The mass spectrometer was coupled with a Nexera high-performance liquid chromatograph (Shimadzu Corporation), and all data analyses were performed using the LabSolutions software (https://www.shimadzu.com/) (version 5.53, Shimadzu Corporation). ADP was separated on a reverse-phased column Scherzo SM-C18 250 × 4.6 mm (Imtakt Corporation) using a linear gradient from 20 to 150 mM ammonium acetate in 50% acetonitrile.

Interaction analysis of chordate AK1 structures

Structural predictions of chordate AK1s were carried out using Modeller 9.21 (32) employing template structures that were bound for bis(adenosine)-5′-pentaphosphate. A total of one hundred structures were generated for each AK1. Subsequently, all interatomic distances were computed based on their coordinates, and atom pairs involved in three distinct types of interactions (hydrophobic interactions, hydrogen bonds, and charge interactions) were defined based on these distances. First, pairs associated with hydrophobic interactions were defined as carbon-carbon or carbon-sulfur (in the case of methionine) pairs that were not covalently linked to nitrogen or oxygen, with distances in the range of 3.4 to 4.6 Å. Second, pairs related to hydrogen bonds were defined as a pair between nitrogen, oxygen, and sulfur atoms in cysteine in which the distance is shorter than 3.5 Å, and as an exception, the distance between oxygen atoms was shorter than 3.2 Å. Third, pairs related to charge interactions were defined as a nitrogen-oxygen pair between side chains in positive and negative amino acids, and these distances are the same as those used for hydrogen bonds.

Regression coefficients were determined by least-squares fitting. Correlation analyses by single and multiple regression were calculated using ORIGINPRO 8J (OriginLab, www.originlab.com). Multiple regression analysis was performed to derive an empirical relationship between the optimum temperature of AK1 and the number of hydrophobic interactions among specific secondary structure elements. The regression model was constructed using experimentally determined optimum temperatures and corresponding interaction counts obtained from the same set of AK1 homologs. These data were treated as a training dataset to identify the combination of interaction terms that best explained the observed variation in optimum temperature.

H/D exchange reaction

Deuterium incorporation was initiated by mixing 9 μl of deuterium oxide with 1 μl aliquots of approximately 50 μM-AK1 solution for each species in 0.2 ml tubes at 5 °C intervals in the range of 15 °C to 60 °C. The H:D atomic ratio was 1:9. The reaction quench and pepsin digestion were performed as described previously (27).

MALDI mass spectrometry

Quenched samples were measured with Ultraflex III, MALDI TOF/TOF instrument (Bruker Daltonics Inc) as described previously (27). The percent deuterium content of fragments were calculated by Scipas DX (27) and kinetics of H/D exchange analysis were calculated from average masses using ORIGINPRO 8J (OriginLab). In this study, reactions at all timepoints were independently prepared in different tubes (196 reactions). The data error was estimated using a 107 to 118 fragment of hAK1 reacted at 45 °C for 1 min in the reaction transition period, which is thought to have a large data error, and was found to be ±2.45% (n = 3).

Thermostability of AK1

The thermal stability of AK1 was monitored using a J-725 spectrodichrometer (JASCO Corporation). The protein concentration was maintained at 0.1 to 0.5 mg/ml for CD measurements at 222 nm. The sample solutions were placed in a quartz cell with a 1 mm light path, and the temperature was increased from 15 °C to 85 °C at a rate of 1 °C/min.

Structure visualizations

Structure visualizations were created in PyMol v.2.4.0 (https://www.pymol.org/).

Data availability

All data in this study are presented in the main article and the supporting information.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.