2.1. Selecting a Proper His-Tag Position
To improve protein purification, a six-histidine tag (His-tag) was added to the RihC construct in two alternatives—to the N– or C–terminus.
Structural modeling was carried out to verify a proper His-tag position using a free version of the AlphaFold2 tool based on Google Collab (ColabFold v1.5.3) [
8,
9]. Models were built with the addition of a His-tag at the N– or C–terminus of the enzyme to explore whether the His-tag can affect the properties of the enzyme or not and whether it is available for binding to the metal chelate chromatography column.
To date, plenty of crystal structures are known for enzymes from the Rih family, incl. eight structures for RihC from various sources (
Table 1), and while these structures are known to be homodimers or homotetramers, there is no strict correlation between the degree of oligomerization and the source of the enzyme.
In this regard, the modeling of LreRihC was carried out with the addition of a His-tag at the N- or C-terminus (hereinafter LreRihC_HisN and LreRihC_HisC, respectively) in the form of a homodimer and homotetramer. The modeling results are presented in
Figure 1.
From this figure, it is clear that the tags at the N- and C-terminus of the enzyme are directed outward from the protein globule and should not seriously affect the intact enzyme structure and/or its catalytic properties. So, it was decided to obtain both variants of the LreRihC enzyme (with His-tag at the N- and C-terminus) and compare their properties.
2.4. Kinetic Parameters of LreRihC
Enzyme activity measurements were carried out according to the approach described in the corresponding section of the experimental part and more thoroughly discussed in [
12]. Four main ribonucleosides were chosen as substrates: uridine, cytidine, guanosine, and adenosine, as well as inosine, xanthosine, thymidine, 2′-deoxyribouridine, 5-methyluridine (also known as ribothymidine) and vidarabine. For each of the substrates, the enzymatic activity was measured; for some substrates, the dependence of the rate of the enzymatic reaction on the concentration of the substrate was obtained, and the kinetic parameters were calculated. Firstly, the enzyme’s kinetics with uridine (for both forms) were studied several times with a freshly purified enzyme batch to see if the kinetics converged within the margin of error. Both K
M and
kcat values for both forms were the same within the error. Secondly, for each of the substrates studied in terms of kinetics, we purified a new batch of both forms of the enzyme and repeated uridine kinetic studies each time to ascertain that the kinetic parameters obtained were correct while also repeating studies for that substrate once again to be sure that the kinetic values are correct. In total, for both forms of the enzyme for each substrate, the kinetics were studied twice (each substrate with a different batch of freshly purified enzymes), except for uridine, for which the kinetics were studied twice initially and then once each time parallel with the new substrate. All the kinetic parameters for each substrate were the same within the error margin.
The kinetic parameters are presented in
Table 2. This table also shows the kinetic parameters of other ribonucleoside hydrolases, C.
Supplementary Figure S4 shows the dependence of the enzymatic reaction rate on the substrate concentration in the concentration range of 0.5–5K
M, which was used to calculate the kinetic parameters for the LreRihC_HisN (
Figure S4A,C,E,G,I) and LreRihC_HisC (
Figure S4B,D,F,H,J) enzymes. For uridine and cytidine, it was possible to measure enzymatic activity in a concentration range of up to 100 mM for both forms of the enzyme; for adenosine, this was up to 30 mM (due to the poor solubility of this substrate when trying to obtain higher concentrations under conditions optimal for the enzyme to work); for inosine, this was up to 20 mM; and for xanthosine, this was up to 10 mM. Guanosine has low solubility under the reaction conditions; therefore, for this ribonucleoside, it was only possible to measure enzyme activity at a maximum substrate concentration of 2 mM. For vidarabine and ribothymidine, it was also decided to measure only the enzymatic activity with 2 mM of each substrate to compare to the activity with other substrates. The cleavage reactions of thymidine and 2′-deoxyuridine were not observed at any concentration of these substrates, from which it can be concluded that the LreRihC enzyme indeed catalyzes cleavage reactions of ribonucleosides. Moreover, since the enzyme showed activity with both purine and pyrimidine ribonucleosides, this enzyme truly belongs to the RihC class, i.e., nonspecific nucleoside hydrolases.
For all studied substrates for which the kinetic parameters of the enzymatic reaction were obtained, classical Michaelis–Menten dependence is observed for both forms of the LreRihC enzyme at low substrate concentrations, as seen in
Figure S4. Moreover, this should not create problems with the use of this enzyme since it is known that the physiological concentration of nucleosides and nitrogenous bases in humans in blood plasma and other extracellular fluids is 0.4–6 μM [
13]. Intracellular concentrations are usually slightly higher. Unfortunately, there are no exact data on the physiological concentrations of nucleosides in bacteria, but we assume that they should not greatly exceed the concentrations of these substances in humans.
Table 2.
Kinetic parameters of RihC enzymes with different substrates. Parameters for the enzymes obtained in this work are in bold.
Table 2.
Kinetic parameters of RihC enzymes with different substrates. Parameters for the enzymes obtained in this work are in bold.
Enzyme | LreRihC_HisN | LreRihC_HisC | EcoRihC b | CfaRihC | SenRihC | LmaRihC |
---|
kcaturidine, s−1 | 167 ± 6 | 134 ± 6 a | 10.85 ± 0.23 | 143 | 46 ± 3 c | 32 ± 6 |
KMuridine, µM | 320 ± 40 | 320 ± 40 a | 408 ± 184 | 1220 ± 40 | 1060 ± 100 c | 234 ± 112 |
kcatcytidine, s−1 | 112 ± 4 | 58 ± 6 | 1.12 ± 0.53 | 20 | 7.8 ± 1.3 c | 0.36 ± 0.05 |
KMcytidine, µM | 680 ± 80 | 620 ± 60 | 682 ± 298 | 4700 ± 500 | 9200 ± 1200 c | 422 ± 175 |
kcatinosine, s−1 | 30 ± 5 | 18 ± 6 | 4.31 ± 0.22 | 32 | 9.0 ± 0.15 c 8.1 ± 0.14 d | 119 ± 34 |
KMinosine, µM | 2500 ± 600 | 2600 ± 600 | 422 ± 225 | 380 ± 30 | 650 ± 60 c 1280 ± 130 d | 445 ± 209 |
kcatxanthosine, s−1 | 57 ± 8 | 40 ± 6 | 6.30 ± 0.05 | ND | 62 ± 8 c 26.3 ± 0.3 d | ND |
KMxanthosine, µM | 1200 ± 200 | 1300 ± 200 | 454 ± 165 | ND | 5900 ± 1000 c 790 ± 50 d | ND |
kcatadenosine, s−1 | 118 ± 4 | 65 ± 7 | 1.15 ± 0.47 | 4.3 | 2.06 ± 0.07 c | 0.57 ± 0.04 |
KMadenosine, µM | 420 ± 50 | 480 ± 90 | 416 ± 249 | 460 ± 30 | 160 ± 20 c | 185 ± 46 |
kcatguanosine, s−1 | ND | ND | ND | 2 | ND | 0.59 ± 0.03 |
KMguanosine, µM | ND | ND | ND | 420 ± 10 | ND | 140 ± 23 |
Source | This work | This work | [7] | [10,14] | [15] | [14] |
From the table, it is clear that the position of the His-tag does not affect the Michaelis constants of the enzyme LreRihC with all substrates, but it does affect the value of the reaction rate constant kcat, and for each of the substrates, it is observed that kcat of LreRihC_HisN is higher than kcat of LreRihC_HisC. The reaction of uridine cleavage for both forms of this enzyme has the lowest KM and the highest kcat, which indicates the greatest efficiency of the enzyme with this substrate. It was not possible to measure the kinetic parameters for the reaction with guanosine due to the low solubility of this substrate under the operating conditions of the enzyme. When comparing the kinetic parameters of the LreRihC enzymes obtained in this work with enzymes from other sources, it can be noted that the Michaelis constants are comparable to those of other enzymes, although for LreRihC, the least preferrable is inosine; however, the catalytic constants, even for the least preferred LreRihC substrate, are quite high compared to all other RihC enzymes (comparable to kcat for more preferred substrates for other RihC enzymes). Overall, our enzyme seems to have high turnover rates for all the substrates, which may possibly play a part in the enzyme’s role when it is produced by L. reuteri in response to the presence of Klebsiella.
To understand the relative efficiency of guanosine cleavage by this enzyme, a comparison of the activity of the enzyme with different substrates at the same concentrations was made (a concentration of 2 mM was chosen as it was the maximum achievable concentration of guanosine in the working conditions). In addition, the activity of the enzyme was studied in relation to other nucleosides: two 2′–deoxyribonucleosides (thymidine and 2′–deoxyuridine), one ribonucleoside (5-methyluridine) and one arabinoside (vidarabine, an analog of adenosine). The results are presented in
Figure 3A.
From the presented data, it is clear that with all basic ribonucleosides, both forms of the enzyme exhibit catalytic activity, with the greatest activity observed for uridine and the least for guanosine. The reaction does not occur with 2′–deoxyribonucleosides, which confirms the important role of the 2′–OH group of the ribonucleoside in catalysis since this group is coordinated by the calcium ion in the active site of the enzyme, thus correctly orienting the substrate for catalysis. Moreover, with vidarabine (an analog of adenosine with arabinose instead of ribose), the 2′–OH group of which is located differently in space than that of adenosine, enzymatic activity is still observed, although it is significantly lower than the activity with adenosine. This means that the 2′–OH position of the sugar is also important for catalysis. From the same figure, it can be seen that the activity of LreRihC_HisC is lower than the activity of LreRihC_HisN for each substrate studied, which is directly related to the kinetic properties of these two forms and confirms the negative effect of C-terminus His-tag on catalytic properties of LreRihC.
To confirm the role of calcium ion in catalysis by this enzyme, the enzymatic activity of uridine cleavage was measured in the presence of 10 mM EDTA, as well as 5, 10 or 50 mM CaCl
2. These substances were added to enzyme samples, which were then incubated for 24 h at 4 °C, and an enzymatic reaction was performed with 20 mM of uridine. The obtained data are presented in
Figure 3B.
The experiment showed that the activity of both enzymes in the presence of EDTA was almost 20 times lower. The addition of calcium ions almost did not change the enzymatic activity for samples with 5 mM and 10 mM CaCl2, and for the sample with 50 mM CaCl2, the activity increased by 20%. Upon the subsequent addition of calcium ions to the enzyme sample containing EDTA (concentration of CaCl2 was 50 mM in the final solution) and incubation, the activity returned to almost the previous level. From this, we can conclude that calcium is indeed important for the catalysis of RihC. EDTA coordinates and pulls calcium ions away from the active site of the enzyme, causing the substrate to not be properly oriented for catalysis and activity to drop significantly. In this case, activity is restored when calcium ions are added to such an enzyme solution. The addition of additional calcium ions has virtually no effect on activity, and the slight increase can be explained by the fact that during protein synthesis in the cells, there may have been such an amount of available calcium ions that a small number of protein globules folded without it. The addition of a calcium solution also allows these certain protein molecules to catalyze reactions, thereby increasing activity.
Since LreRihC_HisN had the best kinetic properties and better expression results, further studies were carried out specifically for this enzyme.
2.5. Temperature Stability of LreRihC_HisN
Differential scanning calorimetry (DSC) was used to study the temperature stability of LreRihC. During the experiment, the temperature of the sample cell and the comparison cell was linearly increased, and the change in heat capacity was monitored. Since protein denaturation is a phase transition, the experimental curve shows an increase in the change in heat capacity of the sample with a peak at the temperature at which the maximum rate of denaturation occurs and which is used as a characteristic of the protein’s thermal stability (the so-called phase transition temperature, T
m). DSC data for LreRihC_HisN are presented in
Figure 4A.
The figure shows that Tm for this enzyme is 59.5 °C. This DSC curve is fitted by a single calorimetric domain. The protein is irreversibly inactivated.
The kinetics of the thermal inactivation of LreRihC_HisN were also studied at different temperatures. The temperatures chosen were 45, 50, 55 and 60 °C based on DSC data. The dependences of residual enzyme activity on incubation time are presented in the semi-logarithmic coordinates in
Figure 4B. All values of residual activity are averaged over three experiments for each point on the graph. To study the activity, 30 mM uridine was used. Enzyme samples were incubated in 0.1 M NaPB pH 7.0.
At 60 °C, enzyme activity drops by two times within 5 min of incubation. In general, all observed dependencies remained linear at different enzyme concentrations, which indicates that the thermal inactivation process is monomolecular. The rate constants of this process for each temperature are presented in the figure. From the data obtained, using the equations of the transition state theory (TST), the activation parameters ΔH≠ and ΔS≠, as well as the T20 parameter (the temperature at which the enzyme activity drops by two times in 20 min), were found. ΔH≠ was calculated to be 260 ± 30 kJ mol−1, ΔS≠ was 490 ± 90 J mol−1 K−1 and T20 was 53.7 °C.
In general, the kinetics of thermal inactivation are in good agreement with the DSC results. At 45 °C, the protein is still mostly stable and has a small thermal inactivation rate constant, which corresponds to the very beginning of the DSC peak; at 50 °C, the inactivation constant increases, but not by a significant amount, and DSC shows that this temperature still corresponds to the beginning of the peak, and at 55 °C and 60 °C, the inactivation constants increase sharply, which corresponds to the maximum of the DSC peak.
2.6. Structural Studies of LreRihC
Since structural studies were carried out only for LreRihC_HisN, further in this section, we use the abbreviated name for this enzyme—LreRihC. The crystal structure of LreRihC was determined at a 1.9 Å resolution. There were four almost identical protein subunits (RMSD between subunits do not exceed 0.13 Å) in the asymmetric unit of the crystal, which belongs to the P2
1 space group. Contact analysis revealed that the protein is a tetramer in a crystal (
Figure 5A), which is in accordance with solution studies. A structural comparison of the LreRihC with known structures of RihC from different organisms (
Table 1,
Figure 5) showed that the assembly is similar to protozoan LmaRihC, which is also tetramer (
Figure 5B), while bacterial GvaRihC and plant ZmaRihC are dimers (
Figure 5C,D). Despite that, the LreRihC tetramer is organized differently compared to that from LmaRihC, which means different intersubunit interfaces (
Figure 5A,B).
Table 3 presents all shortened names that are used hereinafter in text for RihCs from different organisms.
The subunit comparison revealed that all the structures have similar folds. However, LreRihC is mostly similar to the protozoan LmaRihC, LbrRihC, and CfaRihC with the least similarity to bacterial GvaRihC and BanRihC (
Table 1), and there are three polypeptide regions, which result in high RMSD. Region 1 (residues 272–283 in LreRihC) in all structures is a flexible loop with different conformations (
Figure 5E,F). Region 2 (residues 226–234) is unstructured in LreRihC and BanRihC, in contrast to other RihCs (
Table 1), where this region is an α-helix. Region 3 (residues 77–99) in LreRihC and LmaRihC has unstructured conformation (
Figure 5E), while in GvaRihC and ZmaRihC, this region contains one α-helix (
Figure 5F,G).
The LreRihC subunit maintains the typical α/β fold consisting of 11–stranded β–sheet surrounded by α–helices. The active site of LreRihC is formed by the residues D16, D20, D21, N45, T128, N162, E168, N170, H237 and D238. Each active site contains one Ca
2+ ion, which is coordinated by carboxylates of three conserved aspartates (D16, D21, and D238), T128 backbone oxygen and three water molecules (
Figure 6A). The subunit superposition of the LreRihC and apo forms of known RihCs (
Table 1) demonstrated similar calcium coordination (
Figure 6). The lack of solvent molecules coordinating calcium ions in some structures (GvaRihC, ZmaRihC, and LmaRihC) (
Figure 6B–D) seems to be a result of relatively low resolution. It is also worth noting that in ZmaRihC, instead of threonine (128) residue, there is a leucine (123) residue coordinating calcium via its main chain oxygen, which, however, does not alter the coordination (
Figure 6C).
To achieve insights into possible mechanisms of substrate binding by LreRihC, we further compared its active site with those from the holo form of bacterial BanRihC in complex with alpha-D-ribofuranose (
Table 1,
Figure 7A,B). The analysis showed the similar environment of the cavity for ribose moiety binding formed by D21 (D14 in BanRihC), T128 (T125), N162 (N160), E168 (E171), N170 (N173) and D238 (D247). It is noteworthy that the side chains of D20 and N45 in LreRihC are oriented away from the substrate-binding cavity (
Figure 7A), in contrast to holo BanRihC, where the side chains of corresponding D13 and D38 might participate in substrate coordination (
Figure 7B, orange dotted lines). Further structural comparison with protozoan CfaRihC in holo form (
Table 1) showed that, in this case, corresponding D14 and N39 (D20 and N45 in LreRihC) also participate in ribose moiety binding (
Figure 7C). However, in the CfaRihC apo form (
Figure 7D), D14 is salt bridged with H241, and N39 is oriented away from the active site, which is similar to the apo form of LreRihC.
In CfaRihC residues H82, Y229 and H241 are the only residues that coordinate the nitrogenous base of the substrate [
5,
10]. In the case of LreRihC, H237 (H241 in CfaRihC) has a similar orientation (
Figure 7A,C); however, H86 and Y226 (H82 and Y229 in CfaRihC) are located on flexible regions 3 and 2, respectively (
Figure 5E–G), and thus are far away from the active site, similar to the apo CfaRihC structure. Based on structural comparison, we could speculate that these loops as well as residues D20 and N45 in LreRihC, might change their conformation upon substrate binding, as seen in known holo structures.
We further compared the structures of CfaRihC in apo (1MAS) and holo (2MAS) forms to see if the conformation of the enzyme changes significantly upon inhibitor binding, which in the case of 2MAS closely resembles the transition state (
Figure 8A).
From
Figure 8A, it can be seen that there are some conformational changes in the RihC structure when the substrate is bound (since the pAPIR ligand represents the transition state), but overall, the structures align well. The regions highlighted in ovals were the regions that were unresolved in our LreRihC crystal structure. Because of this, we resolved these regions and obtained a model (
Figure 8B).
Figure 8B shows that the unresolved regions apparently do not contain any secondary structure elements. This appears to be quite significant because when we aligned our LreRihC model and crystal structures of CfaRihC (1MAS and 2MAS) and LmaRihC (1EZR), we discovered that near to the active site, all three crystal structures have a lengthier α-spiral (
Figure 8C) than LreRihC. Both CfaRihC and LmaRihC were chosen as those with kinetic parameters studied. There is also a bit of difference in Region 3 (lower oval in
Figure 8C), but out of all the structures, only in 2MAS is it significantly closer to the active site. While this may be interesting in regards of conformation changes in RihC for catalysis, it may not explain the difference in catalytic properties (back in
Table 2), while the differences in Region 2 (upper oval in
Figure 8C) may do so.
Since only protozoan CfaRihC and LmaRihC both had their structural and kinetic parameters defined and LreRihC is a bacterial enzyme, we decided to make an amino acid alignment of all RihCs that had their kinetic parameters studied. This includes the aforementioned protozoan CfaRihC and LmaRihC as well as bacterial EcoRihC, SenRihC (both of which do not have crystal structures) and our enzyme LreRihC. The alignment is shown in
Supplementary Figure S5. The region that we think may explain the differences in terms of the kinetic parameters of these enzymes is located near 220–235 residues in the alignment (numbered for SenRihC), where 226 is the position wherein the LreRihC α-spiral becomes a loop region but in CfaRihC and LmaRihC the α-spiral continues making this element of the structure more rigid. It can be noticed in the alignment too since CfaRihC and LmaRihC both have more amino acid residues in that region that constitute the rest of said α-spiral. It is worth noting that both EcoRihC and SenRihC are similar to LreRihC and thus should also have a short α-spiral with a loop following after in that region. This is an interesting fact because in terms of
kcat values, LreRihC is quite similar to protozoan CfaRihC, but in terms of K
M values, it is more like bacterial EcoRihC (the exceptions being inosine and xanthosine, which are preferred by EcoRihC but much less preferred by LreRihC). Thus, we decided to model both EcoRihC and SenRihC using AlphaFold2 and compare them to our model structure (
Figure 8D). The two regions of interest are marked with yellow ovals. While it is true that residues 220–240 (upper oval) make a short α-spiral and the loop region afterwards, just like in our structure, it also can be seen that in both EcoRihC and SenRihC, this region is closer to the active site (and to be more precise, to the nucleobase-binding part of it) than in LreRihC. It is also clear that the 75–85 residues region (lower oval) of EcoRihC and SenRihC while being almost identical in secondary structure composition to that of LreRihC, is also closer to the active site cavity. The possible explanation may be that, in our enzyme, the structure may be more flexible than in EcoRihC and SenRihC (Region 3 being a bit further from the active site in LreRihC), thus the active site cavity may more easily adapt during the different parts of catalysis for the transition state forming and nucleobase leaving, and this may explain higher
kcat values even for less preferred LreRihC substrates compared to other RihCs. As for inosine and xanthosine being the least preferred for LreRihC compared to other RihCs while we cannot be absolutely certain, we assume that the keto-group in position 6 of the nucleoside (hence inosine, guanosine, and xanthosine are also called 6-oxopurines) may be stabilized more poorly than the amino group in the same position for adenosine, making inosine and xanthosine much less preferrable for LreRihC.