Expression and Kinetic Characterization of PYCR3

Abstract

PYCRs are proline biosynthetic enzymes that catalyze the NAD(P)H-dependent reduction of Δ¹-pyrroline-5-carboxylate (P5C) to proline in humans. PYCRs - especially PYCR1 - are upregulated in many types of cancers and have been implicated in the altered metabolism of cancer cells. Of the three isoforms of PYCR, PYCR3 remains the least studied due in part to the lack of a robust recombinant expression. Herein, we describe a procedure for the expression of soluble SUMO-PYCR3 in Escherichia coli, purification of the fusion protein, and removal of the SUMO tag. PYCR3 is active with either NADPH or NADH as the coenzyme. Bi-substrate kinetic measurements obtained by varying the concentrations of both L-P5C and NADH, along with product inhibition data for L-proline, suggest a random ordered bi bi mechanism. A panel of 19 proline analogs was screened for inhibition, and the kinetics of competitive inhibition (with L-P5C) were measured for five of the compounds screened, including N-formyl-L-proline, a validated inhibitor of PYCR1. N-formyl-L-proline was found to be ten times more selective for PYCR1 over PYCR3. The SUMO-PYCR3 expression system should be useful for testing the isoform specificity of PYCR1 inhibitors.

Graphical Abstract

1. Introduction

The enzyme Δ¹-pyrroline-5-carboxylate (P5C) reductase (P5CR) catalyzes the last step of proline biosynthesis, the NAD(P)H-dependent reduction of L-P5C to L-proline (Fig. 1). Humans have three isoforms of P5C reductase, known as PYCR1, PYCR2, and PYCR3 (a.k.a. PYCRL, UniProt Q53H96). PYCR1 and PYCR2 are 85% identical in sequence and both are mitochondrial enzymes. PYCR3 is only 45% identical to PYCR1 and is cytosolic.

Fig. 1.

Reaction catalyzed by PYCRs.

PYCRs are targets of inhibitor discovery because of their roles in cancer metabolism and upregulation in many different cancers [1]. PYCR1 has been the focus of these efforts so far, but recent experimental studies and computational analysis of gene transcript data suggest that PYCR3 may also be important in some cancers and therefore potentially of interest in cancer biology research and drug discovery. In a genome wide analysis, the PYCR3 gene was found to be upregulated in taxol-resistant nasopharyngeal carcinoma [2]. A study of several cancer cell types found PYCR3 to be one of the top “candidate metabolism drivers” in response to tumor hypoxia [3]. Reduced expression of PYCR3 and ALDH18A1 were found in cancer cells that are dependent on extrinsic sources of L-proline for survival, and the knockdown of PYCR3 inhibited formation of colonies in these cancer cells [4, 5].

PYCR3 is the least characterized PYCR isoform. To our knowledge, only one biochemical study of PYCR3 has appeared in the literature. Ten years ago, De Ingeniis et al. reported a limited steady-state kinetic characterization of recombinant PYCR3 as part of a larger study of proline metabolism in human melanoma cells [6]. In contrast, several crystal structures of PYCR1 in complex with substrates, products, and inhibitors have been determined [7, 8], and PYCR1 has been the target of inhibitor screening campaigns [9-11]. Similarly, a crystal structure of PYCR2 has been determined, and extensive biochemical data on PYCR2 have been reported [12, 13]. The lack of biochemical data on PYCR3 may reflect challenges in producing the recombinant enzyme. Although both PYCR1 and PYCR2 are readily expressed as soluble enzymes in Escherichia coli without the need for solubility-enhancing protein fusions, PYCR3 exhibits very poor solubility in E. coli (vide infra). Indeed, De Ingeniis et al. expressed PYCR3 with a SUMO tag [6].

With the emergence of PYCRs as cancer drug targets and the corresponding need for testing inhibitors for isoform specificity, we sought to develop a robust recombinant expression system for PYCR3. Herein we report the production of soluble, active PYCR3 expressed in E. coli, along with the measurement of steady-state kinetics and the inhibition by known PYCR1 inhibitors. Recombinant PYCR3 is active with either NADH or NADPH as the coenzyme. Analysis of bi-substrate kinetic data and product inhibition by L-proline is consistent with a random ordered bi bi mechanism. Screening of a panel of 19 proline analogs revealed that the validated PYCR1 inhibitor N-formyl-L-proline also inhibits PYCR3 but with 10-times lower affinity. The PYCR3 expression system reported here should be useful for assessing the isoform specificity of PYCR inhibitors.

2. Materials and methods

2.1. Expression and purification of SUMO-PYCR3

A plasmid encoding SUMO-PYCR3 with an N-terminal terminal His₆ tag and codons optimized for expression in E. coli was generated by Genscript. This plasmid was transformed into E. coli BL21(DE3) competent cells and incubated at 37°C overnight on LB agar plates with 50 μg/mL kanamycin. Starter cultures (5 mL) grown from selected colonies in LB were then used to seed 1L cultures of Terrific Broth media. After 3 hours of growth at 37°C with shaking at 225 rpm, the temperature was lowered to 18°C and protein expression was induced with 0.5 mM IPTG for 16-20 hours. Next, the cells were lysed via sonication in a buffer containing 50 mM HEPES/NaOH buffer, pH 7.5, 150 mM NaCl, and 10 mM imidazole, including Pierce protease inhibitor tablets (catalog number A32963). Cell lysate was centrifuged at 16,500 rpm (SS-34 rotor/Sorvall Evolution RC centrifuge) for 1 hour at 4°C. Immobilized metal ion chromatography (IMAC) was performed next by loading the crude extract from centrifugation onto a 25 mL Bio-Rad gravity-flow column containing 8 mL of Qiagen Ni²⁺-NTA Agarose resin (catalog number 30230), which had been equilibrated with 50 mM HEPES/NaOH buffer, pH 7.5, 150 mM NaCl, and 10 mM imidazole. Increasing amounts of imidazole (in the column equilibration buffer) were used to wash the resin (10 mM and 20 mM) and to elute the target protein (300 mM and 1000 mM). Elution fractions were dialyzed against a buffer containing 50 mM HEPES/NaOH buffer, pH 7.5, 150 mM NaCl, and 2% glycerol to prepare for SUMO tag cleavage by Ulp1. Fractions were analyzed via SDS-PAGE.

Ulp1 was expressed in E. coli BL21(DE3) cells from plasmid pFGET19_Ulp1 (Addgene #64697) as described for SUMO-PYCR3 above. Cells were incubated for 30 minutes at 37°C after resuspension and lysed via sonication in lysis buffer (50 mM Tris/HCl buffer, pH 8.0, 350 mM NaCl, 1 mM DTT, 10 mM imidazole, Pierce protease inhibitor tablets (no. A32963), 50 μg/mL DNAse, and 10 μg/mL RNase). Cell lysate was centrifuged at 16,500 rpm (SS-34 rotor/Sorvall Evolution RC centrifuge) for 1 hour at 4°C. The crude extract from centrifugation was loaded onto a gravity-flow Ni²⁺-NTA IMAC column equilibrated with 50 mM Tris/HCl buffer, pH 8.0, 350 mM NaCl, and 1 mM DTT. Successive washes were performed with the equilibration buffer supplemented with 10 mM, 25 mM, and 50 mM imidazole. Successive elutions were performed with the equilibration buffer supplemented with 250 mM and 500 mM imidazole. Fractions were analyzed via SDS-PAGE. Purified Ulp1 was dialyzed against 50 mM Tris/HCl buffer, pH 8.0, 100 mM NaCl, 1 mM DTT, and 5% glycerol overnight at 4°C. Ulp1 was concentrated to 5-7 mg/mL using centrifugal devices (MW cutoff of 10 kDa) and stored at −80°C.

To generate tag-free PYCR3, purified SUMO-PYCR3 was mixed with Ulp1 (100 μL of 5 mg/mL Ulp1 per 10 mL of 5 mg/mL PYCR3). The mixture was incubated for 30 minutes at room temperature and loaded onto an IMAC column to separate tag-free PYCR3 from SUMO and SUMO-PYCR3. The IMAC column was equilibrated with 50 mM HEPES/NaOH buffer, pH 7.5, 150 mM NaCl, and 10 mM imidazole. The column was a 25 mL Bio-Rad column with 4 mL Qiagen Ni-NTA Agarose (catalog number 30230). The flowthrough was collected to obtain tag-free PYCR3. The column was washed with 50 mM HEPES/NaOH buffer, pH 7.5, 150 mM NaCl, and 20 mM imidazole and eluted with 50 mM HEPES/NaOH buffer, pH 7.5, 150 mM NaCl, and 1 M imidazole. The concentration of PYCR3 was determined with a Bradford assay as implemented in the Pierce Coomassie Plus kit using bovine serum albumin as a standard. Tag-free PYCR3 was dialyzed against 50 mM HEPES/NaOH buffer, pH 7.5, 150 mM NaCl and concentrated to 1-5 mg/mL using Sartorius Vivaspin Turbo 15 PES Centrifugal Concentrators (10kDa MWCO, 15 mL). PYCR3 was flash frozen in liquid nitrogen and stored at −80°C.

2.2. Enzyme activity assays

Kinetic measurements were performed with a BioTek Epoch 2 microplate spectrophotometer in a Corning 96-well plate at 26°C. Activity assays were performed in a buffer containing 50 mM HEPES/NaOH buffer pH 7.5 and 1 mM EDTA with a total volume of 200 μL. The decrease in absorbance at 340 nm was monitored and corresponds to the consumption of NAD(P)H in the reaction. D,L-P5C was synthesized as described previously [14] and stored in 1 M HCl at −20°C. A D,L-P5C stock solution (10-18 mM) was diluted into 1 M Tris/HCl buffer, pH 8.5 and neutralized to pH ~7 with 6 M NaOH. The concentration of L-P5C was assumed to be one-half that of D,L-P5C. In experiments when L-P5C was varied, L-P5C was transferred into the plate (40 μL) and then 160 μL of a master mix containing NAD(P)H, PYCR3, and buffer was added via a multichannel pipette to start the reaction. In experiments when NAD(P)H was varied, NAD(P)H was transferred into the plate (5 μL) and then 195 μL of a master mix containing L-P5C, PYCR3, and buffer was added via a multichannel pipette to start the reaction. The final concentration of PYCR3 in the assay was 0.1 nM. For inhibition assays, L-P5C was the variable substrate, and the inhibitor was transferred into the plate along with L-P5C, and then 150 μL of a master mix containing NADH, PYCR3, and buffer was added via a multichannel pipette to start the reaction.

Initial velocities were determined by linear regression of up to 30 minutes of absorbance data. The initial rates were converted to units of μM/sec using the negative of the slope and the extinction coefficient of NAD(P)H at 340 nM of 6220 M⁻¹ cm⁻¹. Initial rates were corrected for pathlength differences of each well.

Bi-substrate kinetic data, in which the concentrations of both L-P5C and NADH were varied, were analyzed by global fitting with Origin software to various kinetic models of bi bi reactions [15]: random ordered bi bi, Equation 1; compulsory ordered bi bi with the equilibrium assumption (hydride transfer in the central complex is rate-limiting), Equation 2; and compulsory ordered bi bi with the steady-state assumption (hydride transfer in the central complex is not rate-limiting), Equation 3.

$$v=\frac{V_{max}[A][B]}{\alpha K^A{K}^B+\alpha K^B[A]+\alpha K^A[B]+[A][B]}$$ (1)

$$v=\frac{V_{max}[A][B]}{K^A{K}^B+K^B[A]+[A][B]}$$ (2)

$$v=\frac{V_{max}[A][B]}{K^A{K}_m^B+K_m^B[A]+K_m^A[B]+[A][B]}$$ (3)

In these equations, A and B are the substrates, K^X is the dissociation constant for substrate X binding to the free enzyme, αK^X is the dissociation constant for substrate X binding to the preformed complex of the enzyme with the other substrate, and $K_m^X$ is the concentration of X that yields a velocity half V_max at fixed, saturating concentration of the other substrate. In the compulsory ordered bi bi mechanisms (Equations 2 and 3), substrate A binds to the enzyme before substrate B.

Enzyme inhibition was characterized using the assay described above with L-P5C as the variable substrate, NADH fixed at 175 μM, and the inhibitor at five different concentrations (including zero). The data for each inhibitor were fit globally to the competitive model of inhibition (Equation 4). In Equation 4, [S] and [I] are the concentrations of L-P5C and the inhibitor, respectively, and K_i is the dissociation constant of the EI complex.

$$v=\frac{V_{max}[S]}{K_m(1+\frac{[I]}{K_i})+[S]}$$ (4)

2.3. Light scattering measurements

The particle size distribution of PYCR3 in solution was measured with light scattering using a Malvern Panalytical Zetasizer. Samples of SUMO-PYCR3 were measured after IMAC purification at 5 mg/mL. Samples of tag-free PYCR3 were measured after cleavage of the SUMO tag and the second IMAC step at 3 mg/mL. The measurements were performed at 25°C in 50 mM HEPES/NaOH buffer, pH 7.5, 150 mM NaCl, and 2% glycerol. PYCR1 was purified as described previously [11] and analyzed with light scattering at 10 mg/mL in 50 mM HEPES/NaOH buffer, pH 7.5, 300 mM NaCl, and 2% glycerol. All samples were filtered with 0.22 μm syringe filters and measured in 40 μL cuvettes. Measurements were taken for 60 seconds for all samples and repeated 3 times consecutively. The intercept of the correlation function was 0.948, 0.932, and 0.928 for SUMO-PYCR3, PYCR3, and PYCR1, respectively.

3. Results

3.1. Expression, solubility, and purification of SUMO-PYCR3 constructs

PYCR3 expressed as an N-terminally His-tagged protein in E. coli was highly insoluble. Given the widespread use of SUMO as a solubility enhancer [16] and the literature precedent for producing PYCR3 fused to SUMO [6], we engineered and tested eight SUMO fusion constructs. The eight constructs tested differed in the sequence of the His-tag, the linker between SUMO and PYCR3, and whether the N-terminus of PYCR3 was truncated. The best construct in terms of solubility and cleavage efficiency contains an N-terminal His6 tag, a linker of EAAAK, and PYCR3 truncated at Val11 (Fig. 2). The choice of linker was based on the literature on linkers used in fusion proteins [17, 18]. Truncation of the N-terminus of PYCR3 was motivated by the observation that PYCR3 has a unique 8-residue N-terminal extension compared to PYCR1 and PYCR2 (Supplementary Fig. S1).

Fig. 2.

SDS-PAGE analysis of the expression and purification of SUMO-PYCR3. (A) Results of IMAC with the lanes marked as follows: S, supernatant after cell disruption and centrifugation; P, insoluble pellet after cell disruption and centrifugation; F, flowthrough from IMAC; W, wash of IMAC resin with 20 mM imidazole; E1-E3, elution fractions from IMAC with 300-1000 mM imidazole. (B) Results of cleavage with Ulp1 after the first IMAC step: U, purified Ulp1; −, prior to addition of Ulp1; +, immediately after addition of Ulp1; 30, 30 minutes after the addition of Ulp1; F, flowthrough of the second IMAC column; W, wash of the second IMAC column with 10 mM imidazole; E, elution of the second IMAC column.

SDS-PAGE analysis of a large-scale purification is shown in Fig. 2. The fusion protein was evident in both the soluble and insoluble fractions after cell disruption (Fig. 2A, lanes S and P). Elution of the resin with high imidazole yielded a protein with the size expected for SUMO-PYCR3 (Fig. 2A, lanes E1-E3). After 30 minutes of incubation with Ulp1, bands with molecular weights corresponding to PYCR3 and SUMO appeared (Fig. 2B, lane “30”). The protein was applied to the IMAC column, and PYCR3 was observed in the flowthrough, as expected (Fig. 2B, lane F). Ulp1 and SUMO were present in the elution fraction, as expected (Fig. 2B, lane E). The yield of purified tag-free PYCR3 was approximately 50 mg from a 1-L culture.

3.2. Light scattering measurements

The particle size distribution of PYCR3 in solution was measured with light scattering. The particle diameter distribution of SUMO-PYCR3 (5 mg/mL) measured after the first IMAC step and before cleavage of the SUMO tag exhibited a major peak at 138 nm and a smaller peak at 23 nm (Supplementary Fig. S2A). The major peak at 138 nm accounts for 90% of the scattered light intensity and presumably represents nonspecifically aggregated protein. The particle diameter distribution measured after cleavage of the SUMO tag and IMAC (3 mg/mL) also showed a bimodal distribution, with peaks at 21 nm (40%) and 219 nm (60%); however, the two peaks were closer in intensity suggesting a lower degree nonspecific aggregation (Supplementary Fig. S2B). The lower amount of aggregation could be due to removal of the SUMO tag or the lower protein concentration used, or both.

For comparison, we also measured the particle diameter distribution of PYCR1 (at 10 mg/mL). The distribution of PYCR1 exhibits a major peak at diameter of 20 nm (92 % of the scattered intensity) and a very minor peak at 2207 nm (8%) (Supplementary Fig. S2C). Previous analysis of PYCR1 with analytical ultracentrifugation showed that the enzyme exists in solution as a decamer at 6 mg/mL [7], and the crystal structures of PYCR1/2 and all other P5CRs show a pentamer-of-dimers decamer in the crystal lattice. Thus, the feature at ~20 nm in the particle size distributions of PYCR3 likely represents the classic P5CR decamer. These results suggest that PYCR3 forms the expected pentamer-of-dimers decamer but also tends to self-associate into soluble aggregates under the buffer conditions and protein concentrations used.

3.3. Steady-state kinetics of PYCR3

The catalytic activity of PYCR3 was measured using steady-state kinetic assays that monitored the disappearance of NAD(P)H. The initial rate exhibited hyperbolic dependence on the concentration of L-P5C with either NADPH or NADH fixed at 175 μM (Fig. 3A,B). Similarly, the enzyme displayed Michaelis-Menten behavior when NADPH or NADH was varied at fixed concentration of L-P5C (Fig. 3C,D). Higher maximum rates were obtained with NADH as the coenzyme compared to NADPH. At a fixed L-P5C concentration of 400 μM, the K_m for NADPH is ~4 times lower than that of NADH (Table 1).

Fig. 3.

Michaelis-Menten analysis of PYCR3. (A) PYCR3 with L-P5C as the variable substrate, NADH fixed at 175 μM, and PYCR3 at 0.1nM. (B) PYCR3 with L-P5C as the variable substrate, NADPH fixed at 175 μM, and PYCR3 at 0.1 nM. (C) PYCR3 with NADH as the variable substrate, L-P5C fixed at 400 μM, and PYCR3 at 0.1 nM. (D) PYCR3 with NADPH as the variable substrate, L-P5C fixed at 400 μM, and PYCR3 at 0.1 nM.

Table 1.

Apparent kinetic parameters for PYCR3 with NADH and NADPH as coenzyme.^a

Enzyme	Variable substrate	Fixed substrate	kcat (s−1)	Km (μM)	kcat/Km (M−1s−1)
PYCR3	L-P5C	NADH = 175 μM	146 ± 17	315 ± 65	4.6 ± 2.7 × 105
PYCR1b	L-P5C	NADH = 175 μM	51 ± 13	260 ± 63	1.9 ± 0.4 × 105
PYCR3	L-P5C	NADPH = 175 μM	48 ± 11	64 ± 48	7.5 ± 2.2 × 105
PYCR3	NADH	L-P5C = 400 μM	235 ± 5	192 ± 8	1.2 ± 0.6 × 106
PYCR3	NADPH	L-P5C = 400 μM	70 ± 2	45 ± 3	1.6 ± 0.6 × 106

^aMeasured at pH 7.5 and 26°C.

^bFrom Christensen et al.[11]

Bi-substrate kinetic analysis was performed in which the initial rate was measured as a function of both L-P5C and NADH concentrations. Two data sets were collected, one with L-P5C as the variable substrate and NADH fixed at four different concentrations, and another with NADH as the variable substrate and L-P5C fixed at four different concentrations. Double-reciprocal plots of the two data sets show intersecting lines consistent with a sequential substrate binding mechanism (Fig. 4D,E). The data were fit globally to the random ordered bi bi model depicted in Fig. 4A (Equation 1, kinetic parameters in Table 2) and the equilibrium ordered model shown in Supplementary Fig. S3A (Equation 2, kinetic parameters in Table 3). The fits of the data to the random model (Fig. 4B,C) were noticeably better than those of the ordered model (Supplementary Fig. S3) based on visual inspection of the fit quality as well as the adjusted R² and AIC metrics (Tables 2 and 3). The data were also fit to the steady-state ordered model (Equation 3). Because of the similarity of Equations 1 and 3, the goodness-of-fit to the steady-state ordered model is identical to that shown in Fig. 4 for the random model; the resulting kinetic parameters for the steady-state ordered model are listed in Table 4.

Fig. 4.

Bi-substrate kinetic analysis. (A) Diagram of the sequential random mechanism. (B) Initial rate as a function of L-P5C concentration for various fixed NADH concentrations. The curves are from global fitting to Equation 1. (C) Initial rate as a function of NADH concentration for various fixed L-P5C concentrations. The curves are from global fitting to Equation 1. (D) Double-reciprocal plot for the data in panel A. (E) Double-reciprocal plot for the data in panel C.

Table 2.

Kinetic parameters for the random ordered bi bi mechanism (Equation 1).^a

[Substrate] on abscissa	α	kcat (s−1)	KP5C (μM)	KNADH (μM)	Adj. R2	AIC
L-P5C	2.4 ± 2.1	3300 ± 300	110 ± 70	48 ± 32	0.98133	−304
NADH	2.2 ± 1.5	2200 ± 300	118 ± 58	31 ± 14	0.98672	−302

^aMeasured at pH 7.5 and 26°C.

Table 3.

Kinetic parameters for the equilibrium compulsory ordered bi bi mechanism (Equation 2).^a

[Substrate] on abscissa	Substrate binds first	kcat (s−1)	KP5C (μM)	KNADH (μM)	Adj. R2	AIC
L-P5C	L-P5C	2300± 130	1500± 700	27 ± 12	0.95806	−280
NADH	NADH	1400 ± 130	215 ± 75	79 ± 29	0.97033	−282

^aMeasured at pH 7.5 and 26°C.

Table 4.

Kinetic parameters for the steady-state compulsory ordered bi bi mechanism (Equation 3).^a

[Substrate] on abscissa	Substrate binds first	kcat (S−1)	KP5C (μM)	KNADH (μM)	$K_m^{P5C}(\mu \mathrm{M})$	$K_m^{NADH}(\mu \mathrm{M})$	Adj. R2
L-P5C	L-P5C	3326 ± 314	110 ± 70	N/A	267 ± 64	117 ± 26	0.98133
NADH	NADH	2171 ± 296	N/A	31 ± 14	257 ± 66	67 ± 19	0.98672

^aMeasured at pH 7.5 and 26°C.

Although it is not possible to distinguish between the random (Equation 1) and steady-state ordered (Equation 3) mechanisms based on the fitting of bi-substrate kinetic data alone, product inhibition can provide additional insight into the likely mechanism. As shown in the next section, the product L-proline displays a competitive inhibition pattern with L-P5C. This result is consistent with the random ordered bi bi mechanism [15]. Based on fitting the kinetic data to this model, k_cat is estimated to be 2000 – 3000 s⁻¹, and the dissociation constants for L-P5C and NADH binding to the free enzyme are ~114 μM and ~40 μM, respectively (Table 2). The α value of ~2 indicates that the binding of first substrate inhibits the binding of the second substrate. Note that similar values of k_cat, K^P5C, and K^NADH are obtained by fitting to the steady-state compulsory ordered model (Table 4).

3.4. Inhibition of PYCR3 by proline analogs

Nineteen proline analogs were screened against PYCR3 (Fig. 5). These compounds are a subset of a library of 27 compounds screened against PYCR1 previously [11]. The product L-proline appears to be a weak inhibitor. In the single-point assay, the activity was lowered to 60% in the presence of 5 mM L-proline (Fig. 5, compound 17). Compounds 2, 10, 16, and 19 showed the highest inhibition, decreasing catalytic activity to 50% or lower compared to the control assay with no inhibitor (Fig. 5).

Fig. 5.

Inhibition of PYCR3 by proline analogs. The bars represent the initial rate with L-P5C at 200 μM, NADH at 175 μM, and the proline analog at 5 mM. The error bars represent the standard deviation from three technical replicates. The data are normalized to the rate of PYCR3 in the absence of an inhibitor.

L-proline and the four compounds that decreased catalytic activity by at least 50% were studied using competitive kinetic assays with L-P5C as the variable substrate at fixed NADH concentration of 175 μM. Global fitting of the data to the competitive model of inhibition (Fig. 6) yielded K_i values of 1 – 6 mM (Table 5). Thus, all the compounds are weak inhibitors of PYCR3. Compound 2 is notable as being a validated inhibitor of PYCR1 with a K_i of 100 μM [11]. The K_i of 2 against PYCR3 is ~10-times higher, 1.2 mM (Table 5). These results suggest that 2 is significantly more selective for PYCR1 over PYCR3.

Fig. 6.

Inhibition of PYCR3 by (A) compound 2, (B) compound 10, (C) compound 16, (D) compound 19, and (E) L-proline with L-P5C as the variable substrate and NADH fixed at 175 μM. (F) Double reciprocal plot for the inhibition by L-proline. The curves in panels A-E are from global fits of the data to the competitive model of inhibition. The derived inhibition constants, K_i, are listed in Table 5.

Table 5.

Apparent competitive (with L-P5C) inhibition constants^a

Compound	Ki (mM)
2	1.2 ± 0.1
10	3.5 ± 0.3
16	2.8 ± 0.3
19	0.76 ± 0.06
17	5.7 ± 0.6

^aMeasured at pH 7.5 and 26°C with [NADH] fixed at 175 μM.

4. Discussion

We described a SUMO fusion approach for generating soluble, active PYCR3, along with characterization of enzymatic activity and inhibition. Bi-substrate kinetic data obtained by varying L-P5C and NADH are consistent with both the sequential random mechanism (Equation 1) and the steady-state ordered mechanism (Equation 3). Product inhibition can help distinguish between the two mechanisms. The double reciprocal plot for L-proline inhibiting PYCR3 at varying L-P5C concentration shows a pattern of straight lines with intersecting y-intercepts, consistent with L-proline inhibiting competitively with L-P5C (Fig. 6F). This result is expected for the random ordered mechanism, whereas noncompetitive or uncompetitive inhibition by L-proline would be expected for compulsory ordered mechanisms [15]. Thus, the available data are consistent with PYCR3 following a random ordered bi bi mechanism.

The random ordered bi bi mechanism assumes that the presence of the coenzyme in the active site does not prevent the binding of L-P5C, and vice versa. A structural feature that would be consistent with random binding is that the coenzyme and L-P5C have separate, nonoverlapping entranceways to their respective binding pockets. To test this idea, we made a model of the decamer of PYCR3 by applying D5 symmetry to the AlphaFold model of PYCR3 obtained from UniProt (Q53H96) [19]. The coenzyme NADPH and a proline analog (1, mimicking L-P5C) from the ternary complex structure of PYCR1 (PDB ID 5UAV) were included in the model. Then, the MOLEonline server was used to calculate tunnels from the L-P5C site in the presence of bound NADPH, and from the NADPH site when 1 was bound [20]. The calculations revealed separate entranceways for the coenzyme and substrate to their respective binding sites (Supplementary Fig. S4). The coenzyme is predicted to enter via a wide tunnel that runs along the C-termini of the β-strands of the Rossmann fold domain. The predicted tunnel from the L-P5C site flows in the opposite direction through an opening in the oligomer interface between two dimers of the decamer. This analysis supports the random ordered bi bi mechanism for PYCR3.

Recombinant PYCR3 exhibits catalytic activity with L-P5C as the substrate and either NADH or NADPH as the coenzyme. Kinetic parameters for PYCR1 measured by our group at room temperature with P5C varying and NADH fixed at 175 μM are available for direct comparison with the data reported here [11] (Table 1). The k_cat of PYCR3 of 150 s⁻¹ is about three times that of PYCR1, and the catalytic efficiencies of the enzymes for L-P5C under these conditions are within a factor of ~2. Thus, the catalytic activity of PYCR3 is not markedly different from PYCR1, at least under these specific conditions.

Our kinetic results may be compared to a previous study of PYCR3. De Ingeniis et al. reported a limited steady-state kinetic analysis of recombinant PYCR3, also expressed as a SUMO fusion protein [6]. They measured kinetic parameters for varying L-P5C at fixed saturating concentration of NAD(P)H, and for varying NAD(P)H at saturating L-P5C concentration. In agreement with their data, we obtained higher k_cat values with NADH as the coenzyme compared to NADPH. Also, we found that the product L-proline is a very weak inhibitor with an estimated K_i of 6 mM, consistent with their estimate of 8 mM. A difference between the two studies is that our catalytic efficiencies are 2-20 times higher. It is possible that this reflects differences in the specific details of assay implementation, the synthesis and quantification of P5C, and measurement of protein concentration. We suggest that the level of agreement between the two studies is reassuring, considering they were performed a decade apart by different groups using different protein constructs.

At least half of the soluble PYCR3 produced with our method appears to be nonspecifically aggregated when assayed by light scattering at a concentration of ~3 mg/mL, which complicates the characterization by biophysical and structural methods. We note there is literature precedent for aggregation of the target protein following cleavage from the fusion partner, and it has been noted that the addition of fusion tags often leads to a new set of problems, including issues with tag removal and whether the proteins retain their native structure and activity following removal of the tag [21, 22]. In our case, aggregation was observed prior to cleavage of the SUMO tag. A possible remedy is the use of detergents to disrupt aggregation. Initial attempts using Octyl β-glucopyranoside, CHAPS, and NDSB-201 were unsuccessful, but perhaps a more extensive survey of detergents and buffer conditions could reduce aggregation levels to allow structural studies. Nevertheless, the current protocol does produce soluble PYCR3 that exhibits desirable biochemical characteristics, such as hyperbolic dependence of enzyme activity on substrate concentration, kinetic parameters similar to those reported previously, and inhibition by known PYCR1 inhibitors.

Finally, the main motivation for creating a PYCR3 expression system is to test the isoform specificity of PYCR1 inhibitors. Here, we tested the inhibition of PYCR3 by L-proline and several proline analogs, including N-formyl-L-proline (compound 2), a PYCR1 inhibitor that has been validated by demonstrating the kinetic mechanism of inhibition using purified enzyme, the physical mode of binding using X-ray crystallography (PDB ID 6XP0), and activity in cancer cells showing that 2 phenocopies the PYCR1 gene knockdown [11]. 2 appears to be a 10-fold better inhibitor of PYCR1 than PYCR3. Our results suggest that it may be possible to develop inhibitors that are selective for PYCR1 over PYCR3.

Highlights.

• PYCRs are highly expressed in cancer cells and the target of inhibitor discovery

• A bacterial expression system for PYCR3 fused to SUMO was developed

• The kinetic parameters for PYCR3 were measured

• The inhibition of PYCR3 by a panel of 19 proline analogs was tested

• N-formyl-L-proline is ten times more selective for PYCR1 over PYCR3

Abbreviations:

IMAC

immobilized metal ion chromatography

P5C

Δ¹-pyrroline-5-carboxylate

P5CR

Δ¹-pyrroline-5-carboxylate reductase

PDB

Protein Data Bank