Engineering an aldehyde dehydrogenase toward its substrates, 3-hydroxypropanal and NAD+, for enhancing the production of 3-hydroxypropionic acid

3-Hydroxypropionic acid (3-HP) can be produced via the biological route involving two enzymatic reactions: dehydration of glycerol to 3-hydroxypropanal (3-HPA) and then oxidation to 3-HP. However, commercial production of 3-HP using recombinant microorganisms has been hampered with several problems, some of which are associated with the toxicity of 3-HPA and the efficiency of NAD+ regeneration. We engineered α-ketoglutaric semialdehyde dehydrogenase (KGSADH) from Azospirillum brasilense for the second reaction to address these issues. The residues in the binding sites for the substrates, 3-HPA and NAD+, were randomized, and the resulting libraries were screened for higher activity. Isolated KGSADH variants had significantly lower Km values for both the substrates. The enzymes also showed higher substrate specificities for aldehyde and NAD+, less inhibition by NADH, and greater resistance to inactivation by 3-HPA than the wild-type enzyme. A recombinant Pseudomonas denitrificans strain with one of the engineered KGSADH variants exhibited less accumulation of 3-HPA, decreased levels of inactivation of the enzymes, and higher cell growth than that with the wild-type KGSADH. The flask culture of the P. denitrificans strain with the mutant KGSADH resulted in about 40% increase of 3-HP titer (53 mM) compared with that using the wild-type enzyme (37 mM).


Results and Discussion
Engineering the aldehyde-binding pocket of KGSADH. Initially, we targeted residues close to the active site of KGSADH. Since its structure was not available when we started the engineering, a homology model of the enzyme was constructed using the structure of human mitochondrial aldehyde dehydrogenase complexed with its aldehyde substrate, crotonaldehyde (PDB: 1O01) 40 , as template ( Figures S1 and S2). The complex structure is one of several aldehyde dehydrogenase structures having their aldehyde substrates in the active site [40][41][42] , which helps to predict residues interacting with the aldehyde molecule in the modeled KGSADH structure. The comparison of its sequence with those of other aldehyde dehydrogenase enzymes had predicted E253 and C287 as the catalytic residues 43 . The superimposition of the template and the modeled KGSADH structures supported this prediction ( Figure S1B). Based on the modeled structure, 10 positions close to the catalytic residues ( Figure S3) were selected and randomized individually, using the NNK (K: G or T) degenerate codon, producing 10 libraries. Each library was screened (approximately 200 colonies) using the criterion of rate of formation of NADH from NAD + , by monitoring the increase in A 340 . Substrate concentrations chosen for 3-HPA and NAD + were 1.5 mM and 1.0 mM respectively. The solution pH chosen for screening the libraries was 6. The rationale was to find variants tolerant to and/or showing an improved activity at low pH, because 3-HP production can lead to a decrease in cytoplasmic pH 1,44 . However, clones exhibiting higher activity than or comparable activity to the wild-type SCIENTIfIC REPORTS | 7: 17155 | DOI:10.1038/s41598-017-15400-x clone in cell lysate did not have any changes in the amino acid sequence (data not shown). The resistance to even small changes in the side chains suggests the critical roles of the 10 residues in catalysis and/or structure of KGSADH.
We then turned our focus to the residues in the aldehyde-binding pocket but having some distance to the catalytic residues. Based on the comparison of the template structure and the homology model of KGSADH, we selected 13 residues ( Fig. 2A). Two more residues (E213 and K273) were chosen based on a report on the engineering of an aldehyde dehydrogenase from Cupriavidus necator (GabD4) 29 . In this study, two residues (E209 and E269) of GabD4 had been randomized for engineering the enzyme. We selected E213 and K273 of KGSADH as target positions, based on the sequence alignment of KGSADH with that of GabD4 ( Figure S4). A total of 15 libraries were generated, in which the target residues were randomized using the NNK degenerate codon. Approximately 200 colonies were screened for each library. Eleven clones, the cell lysates of which exhibited similar or higher enzyme activity compared to that of the wild-type clone and included changes in the amino acid sequence, were purified and their kinetic properties of k cat and K m were characterized (Table S1). Several clones showed slightly higher activities (k cat /K m ) toward either 3-HPA or NAD + than the wild-type enzyme, but there was no variant having an enhanced activity for both the substrates. Interestingly, the N159V mutation significantly increased the K m value toward 3-HPA, resulting in a 5.7-fold reduction of k cat /K m , but showed an enhanced activity toward NAD + , largely due to the decrease in K m . The crystal structure of KGSADH complexed with NAD + was reported, while this manuscript was under preparation 45 . N159 is located close to the NAD + binding pocket, but does not directly interact with the substrate ( Figure S5). Actually, in the KGSADH structure paper, this residue was proposed to interact with 3-HPA. We suspect that the N159V mutation causes a relatively large structural perturbation in the substrate binding pockets.
An additional library was generated to investigate the combination of the mutations found in the aldehyde-binding site, and degenerate codons were used to replace a target position with multiple amino acids ( Figure S6). The theoretical size of the library was 27648, and approximately 8000 colonies were examined using the same screening method. Three variants (designated 104, 106, and 108), whose cell lysates exhibited high activity, were isolated, and their amino acid sequences were determined (Table 1). These enzymes were purified, and their catalytic properties were characterized with respect to 3-HPA and NAD + ( Table 2). One variant (108) exhibited enhanced catalytic activity as demonstrated by the k cat /K m value for 3-HPA. The K m of variant 108 decreased more than 2-fold, compared to that of the wild-type enzyme. However, two enzymes (104 and 106) had lower k cat / K m values than the wild-type enzyme toward 3-HPA. Interestingly, all three variants exhibited enhanced catalytic activities toward NAD + , in contrast to that toward 3-HPA, which was largely because of the increased k cat values.
Engineering the NAD + -binding pocket of KGSADH. Next, we targeted the NAD + binding site of KGSADH, and its K m for NAD + was determined as 0.21 mM (Table 2). While the aldehyde-binding sites of ALDHs are as diverse as their substrates, the binding site for NAD + of ALDHs has been known to be relatively conserved 46 . Therefore, we examined the interaction with NAD + , using the crystal structures of two ALDHs complexed with NAD + , whose K m values were 0.047 mM and 0.0019 mM respectively (PDB: 1A4Z, 1BXS) 47,48 . Nineteen positions were identified, whose side chains are involved in the interactions with NAD + , and the residues of KGSADH for the positions were identified based on the sequence alignment (Table S2). Then, the amino acid residues at these positions of KGSADH were compared with those of ALDHs showing low K m values toward NAD + 49-59 (Table S2). Seven positions (P211, A212, V235, Q238, L239, R334, and A337) (Fig. 2B), which show variations as compared with the sequences of other ALDHs showing low K m values, were selected and randomized, as shown in Figure S7. Even though some residues (P153 and Q160) in the NAD + -binding pocket of KGSADH are different from other ALDHs, they were tested earlier, when the residues close to the catalytic residues were targeted ( Figure S3), and therefore were not included for making the NAD + binding pocket libraries.
Two libraries were made using clones 104 and 106 as templates and were screened at low substrate concentrations (0.5 mM of 3-HPA and 0.3 mM of NAD + ), with the objective of finding variants with reduced K m values. Clones from both the libraries, which showed high activity in cell lysates, were isolated, and their amino acid sequences are shown in Figure S8. Clone 104-M5, exhibiting the highest activity, had two changes at positions R334 and A337; the two residues were mutated into Q and R respectively. Interestingly, two clones (clones 106-M4 and 106-M5) isolated from the 106 library also had the same changes (R334Q and A337R) in the targeted NAD + -binding site and another clone (clone 106-M8) showing a comparable activity to that of these two clones had one additional mutation at position 238 (Q238K) ( Figure S8B). Based on these results, we decided to introduce the two mutations, R334Q and A337R, into the 108 variant also. The three enzymes (104-QR, 106-QR, and 108-QR; the sequences are shown in Table 1) were purified and their catalytic properties, k cat and K m , toward 3-HPA and NAD + were characterized ( Table 2). The two mutations in the NAD + binding pocket resulted in 7-11-fold reduction of K m toward NAD + for the three variants, and unexpectedly the K m values for 3-HPA also decreased to a considerable extent. However, the k cat also decreased. Therefore, the improvement in k cat /K m was in the range of 2-4-fold. It is quite interesting that the mutations in the NAD + -binding pocket made such a dramatic effect on the binding of 3-HPA to the enzyme. The two mutations are located far from the 3-HPA-binding site. To examine the effect of the mutations R334Q and A337R in the absence of other mutations, these mutations were introduced into the wild-type enzyme, and the catalytic properties were characterized ( Table 2). Similar to the three variant enzymes, the wild-type enzyme harboring R334Q and A337R mutations, named WT-QR, showed a decrease in the K m values for both 3-HPA and NAD + , which suggests that the two mutations affect not only the NAD + binding pocket, but also the binding of 3-HPA. However, variants found in this study showed higher activities (k cat /K m ) than WT-QR, which indicates the contribution of other mutations, besides the two mutations in the NAD + binding site, to the improvement of catalytic efficiency.

Substrate specificity and product inhibition of KGSADH enzymes. The wild-type KGSADH
showed enzymatic activity toward various aldehyde substrates, in addition to its original substrate, α-ketoglutaric semialdehyde 30,60 . To examine how the mutations in the KGSADH variants affected the substrate specificity, their activities toward several aldehyde compounds were compared (Fig. 3A). All the variants showed lower relative activity toward other aldehydes, compared to that toward 3-HPA, than the wild-type enzyme, which indicates that the wild-type enzyme is more promiscuous toward aldehyde substrates than the engineered ones. In particular, the 108-QR variant exhibited the highest specificity, which might be attributed to its lowest K m value toward 3-HPA, among all the tested enzymes. Interestingly, the WT-QR, which has mutations only in the NAD + binding site, also exhibited increased specificity toward 3-HPA. The results, in addition to the decreased K m for 3-HPA of WT-QR, suggest that the two mutations, R334Q and A337R, induce a structural change in the aldehyde-binding pocket, even though they are located away from the site. The KGSADH variants were characterized with respect to NADP + also, because the wild-type enzyme was shown to be active with this cofactor ( Table 2). The engineered enzymes had several-fold lower K m values toward NADP + and showed higher catalytic activity (k cat /K m ) than the wild-type enzyme. However, the improvement toward NAD + was higher than that for NADP + , and thus the activity of the engineered enzymes for NAD + relative to those engineered for NADP + was higher, compared to that of the wild-type enzyme.
Biological processes for the production of 3-HP, for which the engineered KGSADH variants are intended to be used, should generate high concentrations of 3-HP and NADH inside the recombinant microorganisms. Therefore, we investigated the reverse reaction of 3-HPA formation from 3-HP and product inhibitions by 3-HP and NADH. The highest concentration tested for 3-HP was 1 M, since the targeted titer of 3-HP production is approximately 100 g/L for developing an economically feasible process. The highest concentration chosen for NADH was 3 mM, considering the reported concentrations of NAD + and NADH 61 . The reverse reaction did not take place up to 1 M 3-HP and 3 mM NADH, as determined by monitoring the absorbance at 340 nm (data not shown). However, product inhibitions were observed with both 3-HP and NADH. While there was no detectable  Table 1. Amino acid sequences of KGSADH enzymes. *The amino acid residues different from those in the wild-type KGSADH are shown in bold.
inhibition by 3-HP up to 100 mM, the activities of KGSADH enzymes dropped approximately by 40% at 1 M 3-HP for all the tested KGSADH enzymes ( Figure S9). Interestingly, the wild-type KGSADH exhibited much higher inhibition by NADH than the engineered enzymes (Fig. 3B). The inhibition was observed even at 0.5 mM NADH for the wild-type enzyme. The activity decreased linearly with increasing NADH concentration. In contrast, the variants showed less than 10% decrease in activity up to 2 mM NADH. The residual activity at 3 mM NADH was approximately 60%, compared to that in the absence of NADH. The concentration of NADH is rarely over 2 mM in the cytoplasm of microorganisms 62 . Therefore, the engineered KGSADH variants are not likely to experience inhibition by NADH during 3-HP production by recombinant microorganisms.
pH sensitivity, 3-HPA inactivation, and heat inactivation of KGSADH enzymes. The screening of the libraries were conducted under the slightly acidic conditions, with the objective of finding variants tolerant to and/or showing improved activity at low pH, and the catalytic properties of the engineered enzymes were also characterized at pH 6 (Table S3). In general, the k cat values decreased at low pH, whereas the K m values did not change much. Interestingly, for the two variants, WT-QR and 106-QR, the K m values for 3-HPA were lower at pH 6 than those at pH 8. The ratio of k cat /K m value at pH 6 vs. pH 8 of these two variants was higher than that of other enzymes (Table S3). The two mutations, R334Q and A337R, in the NAD + binding pocket might play a role in lowering the K m for 3-HPA at acidic pH. However, it is not clear why only WT-QR and 106-QR exhibited the decreased K m at pH 6, even though 104-QR and 108-QR have the same mutations. It is of note that 106-QR has two mutations near the aldehyde-binding site, whereas 104-QR and 108-QR have four mutations. Therefore, the structural difference between these two enzymes and WT-QR might be greater compared to that between WT-QR and 106-QR. The relative activities of the enzymes were determined in the pH range of 4-10 ( Figure S10). These relative activities reflect k cat rather than k cat /K m , considering the concentrations of the substrates used in this experiment (3 mM 3-HPA and 2 mM NAD + ). The KGSADH enzymes showed similar pH profiles, with the exception of at pH 10. The activities of all the variants increased up to pH 10; the wild-type KGSADH had the highest activity at pH 9. The variant enzymes were more resistant to alkaline conditions than the wild-type enzyme. 3-HPA is a highly reactive compound and reacts with side chains of various amino acids, including those of Cys, Lys, His, and Arg 63,64 . Chemical modifications of the side chains can result in the loss of enzyme activity. We characterized the inactivation kinetics of the KGSADH variants (Fig. 3C). When incubated with a high concentration of 3-HPA (20 mM), the engineered enzymes exhibited higher resistance to 3-HPA inactivation than the wild-type KGSADH: the half-life values (t 1/2 ) for the KGSADH variants, calculated using the fitted exponential decay, were at least 50% higher than that for the wild-type enzyme (Table S4). One possible reason for the resistance to alkaline condition and 3-HPA inactivation is the improved stability of the enzymes. We, therefore, compared the thermal stability of the enzyme variants ( Figure S11 and Table S5). Contrary to our expectation, wild-type enzyme showed the highest resistance to heat inactivation, which is largely related to the structural stability. These results suggest that the tolerance to basic pH and resistance to inactivation by 3-HPA of the engineered enzymes do not arise from the thermodynamic stability, but might be related to the remodeled substrate binding sites, considering their correlation to the enhanced activity, altered substrate specificity, and decreased product inhibition by NADH.
Structural modeling of the WT-QR variant. The results above indicated that the mutations in the NAD + binding site, R334Q and A337R, affect various catalytic properties of KGSADH; K m for both NAD + and 3-HPA, substrate specificity, product inhibition by NADH, pH sensitivity, and resistance to 3-HPA inactivation. In order to investigate the effect of the two mutations, the structure of WT-QR was modeled using Molecular Operating Environment (MOE), based on the crystal structure of KGSADH complexed with NAD + 45 (Figures S12 & S13). The two mutations, R334Q and A337R, result in new interactions with NAD + . In the wild-type enzyme, the guanidine group in the side chain of R334 interacts with the ribose ring close to the nicotinamide group via hydrogen bonding, but A337 does not show any interaction with NAD + (Figure S12B). However, Q334 and R337 of the WT-QR variant interact with the phosphate group close to nicotinamide, resulting in structural changes in the region where the mutations are located (residue 325-337) and also conformational changes of NAD + binding   Table 2. Kinetic characterization of KGSADH enzymes at pH 8*. *The experiments were repeated at least three times. **k cat and K m for 3-HPA were determined using 2 mM NAD + . ***k cat and K m for NAD + and NADP + were determined using 3 mM 3-HPA.
pocket ( Figure S12B). The new interactions could contribute to the reduced K m of WT-QR toward NAD + . Unexpectedly, in addition to the NAD + binding site, the substrate entry regions of aldehyde-binding pocket (residues 101-109, 265-290 and 436-442) of WT-QR exhibit changes in the backbone structure, compared to that of wild-type structure ( Figure S12C&S13). The peptide backbone for residues 441-445 moves inward, and thus the entrance size of aldehyde-binding pocket decreases, compared to that of wild-type enzyme ( Figure S12C), which might explain the decreased K m and the increased specificity toward 3-HPA. It is of note that the three  (Table 1). While the two proteins showed nearly identical peptide backbone structures in the region close to the catalytic residues, conformational differences in the side chains were observed ( Figure S12D), which might be associated with the decrease in k cat of the mutant enzyme. Even though the modeled structure of WT-QR provides interesting insights into the effects of the two mutations, the results need to be confirmed by experimental work. Currently, we are trying to determine the crystal structures of the engineered KGSADH variants.

Production of 3-HP from glycerol by a recombinant P. denitrificans strain on a flask scale.
We successfully engineered KGSADH for higher activity toward both 3-HPA and NAD + via decreasing the K m values. We were interested in investigating how the engineered enzymes performed in the production of 3-HP from glycerol by recombinant bacterial strains. The 108-QR enzyme showed the highest activity among the engineered enzymes, and we decided to compare the wild-type KGSADH with this variant. The P. denitrificans Δ3hpdH Δ3hibdhIV strain 65 was transformed with the pUCPK plasmid 25 for expressing glycerol dehydratase (DhaB), its reactivase (GdrAB), and KGSADH. The genes for DhaB and GdrAB are under the constitutive bla promoter 25 and that for KGSADH, under the constitutive 3hibdhIV promoter 63 . The culture experiments were conducted in duplicate for each enzyme, and the results were averaged. As shown in Fig. 4A and B, during the initial period of cultivation (3-12 hours), the 3-HP titer and the cell density were similar for the two recombinant P. denitrificans strains harboring wild-type KGSADH or 108-QR. However, in the late stage (15-24 hours), differences in 3-HP production and cell growth were observed. P. denitrificans with 108-QR exhibited a higher 3-HP titer (53 mM vs. 37 mM at 24 hours) and greater cell growth than P. denitrificans with the wild-type KGSADH (OD 600 = 2.1 vs. OD 600 = 1.6 at 24 hours) ( Fig. 4A and B). The concentrations of 3-HPA were compared between the two strains at the two culture points (12 and 24 hours) (Fig. 4C). P. denitrificans with 108-QR showed ~50% lower 3-HPA accumulation at both the time points. The 3-HPA toxicity has been shown to be related to the inactivation of cellular enzymes, in particular, DhaB and KGSADH 33,66,67 . The reduced accumulation of this intermediate in P. denitrificans with 108-QR, results in less reduction of the activity of the two enzymes during the culture (Fig. 4D and E). In summary, the results described above indicate that the higher enzyme activities via decreased K m enabled the

Conclusion
In this study, we engineered the aldehyde dehydrogenase of KGSADH to address the issues in the biological production of 3-HP from glycerol using recombinant microorganisms: toxicity of the intermediate, 3-HPA, and low NAD + regeneration efficiency in the late stages of culture. One of the approaches to tackle these problems could be improving the enzyme activity toward its substrates, 3-HPA and NAD + . KGSADH variants were isolated through screening various libraries targeting the substrate binding pockets. The engineered enzymes exhibited improved activity compared to the wild-type enzyme. In particular, the K m of the enzymes for both the substrates decreased significantly. In addition, the KGSADH variants had several interesting properties: higher substrate specificities for aldehyde and the reducing cofactor; less product inhibition by NADH; and greater resistance to inactivation by 3-HPA. One of the engineered enzymes (108-QR) was compared with the wild-type KGSADH for 3-HP production in a recombinant P. denitrificans strain. The microorganism harboring the variant enzyme exhibited lower accumulation 3-HPA, less inactivation of the two enzymes that produce 3-HP from glycerol, higher cell growth, and higher 3-HP production than that with the wild-type enzyme. These phenomena were more pronounced in the late stage of culture. We expect that the problems of 3-HPA toxicity and NAD + regeneration associated with developing economically feasible 3-HP production process could be addressed to some extent by using engineered aldehyde dehydrogenases. However, some of the properties of engineered enzymes can be further improved. These include the unintended decrease in k cat and thermal stability observed in this work, product inhibition by 3-HP, 3-HPA inactivation, and activity under acidic conditions.

Materials and Methods
Materials. Pfu

Construction of libraries.
Mutations were introduced into the KGSADH gene, using the assembly PCR method. The pQE80L/KGSADH plasmid 43 was used as template. The primers used are shown in Table S6. The PCR products from primers 1 and 2 were ligated into the pQE80L plasmid, using the BamH1 and Pst1 sites. The resulting KGSADH libraries were transformed to E. coli DH10β. Primers 1-32 were used for constructing single-site variant libraries for the aldehyde-binding pocket, and primers 1, 2, and 33-42 for combining the beneficial mutations from the single-site variant libraries. Primers 1, 2, and 43-58 were used for the NAD + binding site library.

Screening of KGSADH libraries.
Oxidation of 3-HPA to 3-HP is coupled with the reduction of NAD + to NADH. Increase in absorbance at 340 nm (λ max of NADH = 340 nm) was used to monitor the reaction catalyzed by KGSADH. Single colonies were picked from each library and inoculated into 96-well plates (U type). Cells were grown overnight in 100 µL 2xYT medium containing 200 µg/mL of ampicillin at 37 °C and then transferred into the same, but fresh media. When the OD 600 approached 0.5, KGSADH expression was induced by adding 1 mM IPTG. After 6 h, the cells were harvested by centrifugation at 3,134 g and 4 °C for 45 min, and the cell pellets were stored at −80 °C until further use. The cell pellets were resuspended in 100 µL extraction buffer (100 mM phosphate buffer, pH 8, containing 10% B-PER, 100 µg/mL lysozyme, and 5 U/mL DNase), and incubated at 25 °C for 1 h. The cell lysates were centrifuged at 3,134 g and 4 °C for 45 min. Ten microliters of the supernatant was diluted with 170 µL activity assay buffer (100 mM phosphate buffer, pH 6, containing 1 mM EDTA, 1 mM DTT, and 1 mM β-mercaptoethanol), and the activity assay was initiated by adding the substrates, 3-HPA and NAD + . The rate of 3-HP formation was determined via the ΔA/min at 340 nm. To compensate the enzyme concentrations from different cell growth, the 3-HP production rate (ΔA/min) was divided by OD 600 for analyzing the results.
Expression and purification of KGSADH enzymes. The wild-type and mutant KGSADH enzymes in the pQE80L plasmid were expressed in E. coli DH10β cells. Cells were grown in 2xYT medium at 37 °C until OD 600 reached 0.5, and then the expression of enzyme was induced by 1 mM IPTG. After 6 h, the cells were harvested by centrifugation at 9,300 g and 4 °C for 15 min, and the cell pellets were stored at −80 °C until further use. The KGSADH enzymes with the N-terminal His 6 -tag were purified using the Ni-NTA resin following manufacturer's instructions. The protein buffer was exchanged with 10 mM phosphate buffer, pH 7.4, containing 150 mM NaCl, using the centrifugal ultrafiltration filter. The purified proteins were stored in the storage buffer (20 mM potassium phosphate, pH 7.5, containing 1 mM EDTA, 1 mM DTT, and 50% (v/v) glycerol). The concentrations of the purified proteins were determined by measuring the absorbance at 280 nm, using extinction coefficients calculated from the ProtParam site (http://web.expasy.org/protparam/).
Characterization of KGSADH enzymes. k cat and K m values of the wild-type and variant enzymes were determined by measuring the reaction rate at 25 °C, by monitoring the increase in A 340 (ε = 6,220 M −1 cm −1 ).
To determine the kinetic constants for 3-HPA, the enzyme concentration was adjusted to 20 nM in 100 mM phosphate buffer (pH 8 or pH 6) containing 1 mM EDTA, 1 mM DTT, 0.2% BSA (w/v), and 2 mM NAD + . The reactions were initiated by the addition of varying concentrations of 3-HPA. For the k cat and K m values for NAD + or NADP + , the enzymes were diluted with the buffer without NAD + or NADP + , and the reactions were initiated by the addition of 3 mM 3-HPA and varying the concentrations of NAD + or NADP + . The results were analyzed by nonlinear regression for the Michaelis-Menten equation, using Origin software (Northaempton, MA, USA). The relative activity of the enzyme toward various aldehyde substrates was determined by measuring the activity in 100 mM phosphate buffer, pH 8, containing 1 mM EDTA, 1 mM DTT, 0.2% BSA (w/v), 1.5 mM NAD + , and 1 mM aldehyde (3-HPA, propionaldehyde, butyraldehyde, valeraldehyde or hexaldehyde) at 25 °C 43 . The inhibition of KGSADH enzymes by 3-HP was determined by measuring the activity in 100 mM phosphate buffer, pH 8, containing 1 mM EDTA, 1 mM DTT, 0.2% BSA (w/v), 2 mM NAD + , and 1-1000 mM 3-HP. NADH inhibition was assayed in the same buffer containing 3 mM 3-HPA and 0-3 mM NADH. The effect of pH on the enzyme activity was investigated using actetate buffer (pH 4 and 5), phosphate buffer (pH 6-8), and glycine buffer (pH 9 and 10). The reactions were conducted in 100 mM buffer containing 1 mM EDTA, 1 mM DTT, 0.2% BSA (w/v), 2 mM NAD + , and 3 mM 3-HPA, at 25 °C. The inactivation of KGSADH by 3-HPA was investigated by measuring the residual activity after incubating the enzyme with 20 mM 3-HPA for varying lengths of time. The enzyme activity was determined in 100 mM phosphate buffer, pH 8, containing 1 mM EDTA, 1 mM DTT, 0.2% BSA (w/v), and 2 mM NAD + , at 25 °C. Heat inactivation of the enzymes was determined by measuring the residual activity after incubating the enzymes at 65 °C for varying lengths of time. The enzyme activity was determined in 100 mM phosphate buffer, pH 8, containing 1 mM EDTA, 1 mM DTT, 0.2% BSA (w/v), 2 mM NAD + , and 3 mM 3-HPA at 25 °C. The half-life (t 1/2 ) values for inactivation by 3-HPA or heat treatment were determined by fitting the data to exponential decay (y = ae bx ) using Origin software.
Structural modeling of KGSADH enzymes. The three-dimensional structure of KGSADH was constructed by homology modeling, using the x-ray structure of human mitochondrial aldehyde dehydrogenase (PDB ID: 1O01) as the template 40 . The model was built using Molecular Operating Environment (MOE) and evaluated by PROCHECK and ProSA online structure analysis 69,70 . The protein structure was visually examined using PYMOL viewer (http://www.pymol.org).
The crystal structure of KGSADH complex with NAD + (PDB ID: 5X5U) and apo form (PDB ID: 5X5 T) were retrieved form the Protein Data Bank and investigated for structural abnormalities and missing residues. The unbound solvent molecules and other nonstructural residues were removed to clean the structure for further structural analyses. The In silico structure of WT-QR complexed with NAD + was generated via the residue scan module within MOE under the LowMode ensemble, in which the LowModeMD Search method generates mutant conformations using ~1 ps run of molecular dynamics simulation at constant temperature, followed by all-atom energy minimization under Amber10:EHT force field, as explained elsewhere 71 . Pymol and MOE were used for visual inspections and generating structural figures.
Flask scale production of 3-HP. Recombinant Psuedomonas denitrificans was grown in an Erlenmeyer flask for comparing the performance of wild-type KGSADH with that of 108-QR. P. denitrificans Δ3hpdH Δ3hibdhIV 65 , which does not degrade the produced 3-HP, was used as the host. The dhaB-gdrAB and kgsadh genes were cloned under the constitutive bla 25 and 3hibdhIV 63 promoters, respectively. Experiments were carried out in a working volume of 100 mL in a 250 mL non-baffled Erlenmeyer flask at 37 °C, on an orbital shaker incubator at 200 rpm. Modified M9 medium was used, which contained (per liter) 100 mmol potassium phosphate, 0.25 g MgSO 4 ·7H 2 O, 1.0 g NaCl, 1.0 g NH 4 Cl, 1 g yeast extract, 2.0 g tryptone, 2.5 g glucose, 2.5 g L-glutamate, 30 mg kanamycin, 25 mg cobalt chloride, and 100 mmol glycerol. Inoculum culture was conducted under the same conditions as the main culture, except that glycerol was not added in the medium. Inoculum was cultured for 9 h to reach late exponential growth phase and added at ~0.01 OD 600 . The DhaB and KGSADH activities in the crude cell extract were measured using previously reported methods 20,72 . The 3-HPA concentrations were determined indirectly by measuring 1,3-propanediol using HPLC after enzymatic conversion of the aldehyde using alcohol dehydrogenase.