Article Cite This: Biochemistry XXXX, XXX, XXX−XXX
pubs.acs.org/biochemistry
3‑Phosphoglycerate Transhydrogenation Instead of Dehydrogenation Alleviates the Redox State Dependency of Yeast de Novo L‑Serine Synthesis Nicole Paczia,† Julia Becker-Kettern,† Jean-François Conrotte,† Javier O. Cifuente,‡ Marcelo E. Guerin,‡,§ and Carole L. Linster*,† †
Luxembourg Centre for Systems Biomedicine, University of Luxembourg, L-4367 Belvaux, Luxembourg Structural Biology Unit, CIC bioGUNE Technological Park of Bizkaia, 48160 Derio, Vizcaya, Spain § IKERBASQUE, Basque Foundation for Science, 48013 Bilbao, Spain
Biochemistry Downloaded from pubs.acs.org by EASTERN KENTUCKY UNIV on 01/28/19. For personal use only.
‡
S Supporting Information *
ABSTRACT: The enzymatic mechanism of 3-phosphoglycerate to 3-phosphohydroxypyruvate oxidation, which forms the first step of the main conserved de novo serine synthesis pathway, has been revisited recently in certain microorganisms. While this step is classically considered to be catalyzed by an NAD-dependent dehydrogenase (e.g., PHGDH in mammals), evidence has shown that in Pseudomonas, Escherichia coli, and Saccharomyces cerevisiae, the PHGDH homologues act as transhydrogenases. As such, they use α-ketoglutarate, rather than NAD+, as the final electron acceptor, thereby producing D-2-hydroxyglutarate in addition to 3-phosphohydroxypyruvate during 3-phosphoglycerate oxidation. Here, we provide a detailed biochemical and sequence−structure relationship characterization of the yeast PHGDH homologues, encoded by the paralogous SER3 and SER33 genes, in comparison to the human and other PHGDH enzymes. Using in vitro assays with purified recombinant enzymes as well as in vivo growth phenotyping and metabolome analyses of yeast strains engineered to depend on either Ser3, Ser33, or human PHGDH for serine synthesis, we confirmed that both yeast enzymes act as transhydrogenases, while the human enzyme is a dehydrogenase. In addition, we show that the yeast paralogs differ from the human enzyme in their sensitivity to inhibition by serine as well as hydrated NADH derivatives. Importantly, our in vivo data support the idea that a 3PGA transhydrogenase instead of dehydrogenase activity confers a growth advantage under conditions where the NAD+:NADH ratio is low. The results will help to elucidate why different species evolved different reaction mechanisms to carry out a widely conserved metabolic step in central carbon metabolism.
T
the PSAT and PSP homologues, respectively.4 An alternative serine synthesis pathway (the glyoxylate pathway), deriving from tricarboxylic acid cycle intermediates, is repressed by glucose and induced by growth on ethanol or acetate in yeast.9,10 In this pathway, an alanine-glyoxylate aminotransferase (P43567 encoded by AGX1 in S. cerevisiae) generates glycine from glyoxylate9 (itself derived from isocitrate), which is further converted to serine by serine hydroxymethyltransferase (P37292 and P37291 encoded by SHM1 and SHM2, respectively, in S. cerevisiae) at the expense of 5,10-methylene-tetrahydrofolate.11 PHGDH enzymes can be categorized into at least three different classes termed type I, II, and III, according to the variable number of domains in their corresponding architectures (Figure 1).12−19
he serine biosynthesis pathway derived from glucose was described by Ichihara and Greenberg in the 1950s.1 They demonstrated the existence of an enzymatic system in rat liver extracts that uses the glycolytic intermediate 3-phosphoglycerate (3PGA) to produce L-serine. The pathway comprises three steps: an initial oxidation step, catalyzed by 3phosphoglycerate dehydrogenase (PHGDH), to convert 3PGA to 3-phosphohydroxypyruvate (PHP), followed by PHP transamination to 3-phosphoserine (3PSer) by the aminotransferase PSAT and a phosphoester hydrolysis carried out by phosphoserine phosphatase (PSP) to produce L-serine (reviewed, for example, in refs 2 and 3). This de novo serine synthesis pathway is well conserved across prokaryotic and eukaryotic species,1,2,4−6 suggesting its fundamental importance for many living cells, at least in the absence of exogenous serine. In Saccharomyces cerevisiae, four genes encode the enzymes of this pathway: SER3 (UniProtKB entry P40054) and SER33 (P40510), two paralogous genes that arose from the whole genome duplication event in the ancestral yeast species Kluyveromyces waltii,7 encode the PHGDH homologues (ScPHGDH).8 SER1 (P33330) and SER2 (P42941) encode © XXXX American Chemical Society
Special Issue: Future of Biochemistry: The International Issue Received: September 14, 2018 Revised: January 9, 2019
A
DOI: 10.1021/acs.biochem.8b00990 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 1. Structural architecture of the PHGDH family. (A) Bar diagram showing the overall domain organization of type I PHGDH protein sequences: substrate binding domain (SBD, orange), nucleotide binding domain (NBD, yellow), allosteric substrate binding (ASB, green) domain, and aspartate kinase−chorismate mutase−TyrA (ACT) domain (N, N-terminus; C, C-terminus). Crystal structure of one protomer of the paradigmatic type I MtPHGDH enzyme (PDB entry 3DDN). Different domains are colored in agreement with the bar diagram. The active site is indicated. (B) Type II architecture bar diagram and crystal structure of one protomer of the paradigmatic type II EcPHGDH (PDB entry 1YBA). (C) Type III architecture bar diagram and crystal structure of one protomer of the paradigmatic type III EhPHGDH (PDB entry 4NJO). (D) MtPHGDH homotetramer showing a pair of colored protomers (A and B) and two gray protomers (C and D). Each pair includes protomers in two different conformations. The images clearly reflect the prominent NBD−NBD and ACT−ACT interactions. This architecture also allows ASB−ASB and NBD−ASB interactions. (E) EcPHGDH homotetramer showing a colored protomer and three gray protomers. All protomers display the same conformation. The NBD−NBD and the ACT−ACT interactions are clearly depicted. (F) EhPHGDH dimer showing a colored protomer and one gray protomer. Both protomers in the same conformation are bound by NBD−NBD interactions.
The PHGDH from Escherichia coli (EcPHGDH, P0A9T0) represents a paradigm of type II enzymes.12,15,16,18 Crystal structures of EcPHGDH reveal a homotetrameric assembly of the protomers that can be viewed as a dimer of dimers. The most important contributions to the oligomeric interface are mediated by the ACT−ACT and NBD−NBD interactions. Importantly, relative movements of all three domains are allowed by the connecting flexible hinges.12,15,16,18 The accepted oxidation mechanism involves deprotonation of the hydroxyl group of the 3PGA substrate by a histidine (His292 in EcPHGDH) whose reactivity is enhanced by the interaction with a glutamyl residue (Glu269 in EcPHGDH). Thus, a hydride from the hydroxyl group can be transferred to NAD+ in an aromatic bimolecular nucleophilic addition reaction.21,22 Certain enzymes from different categories [e.g., type II
All PHGDHs contain a catalytic core comprising two distinct Rossmann fold domains, separated by a deep cleft where the active site is located.20 The N-terminal substrate binding domain (SBD) is connected with the nucleotide binding domain (NBD) via short flexible hinges.16 Type I enzymes display two additional domains at the C-terminus of the catalytic core: (i) the allosteric substrate binding or ASB domain and (ii) the regulatory aspartate kinase−chorismate mutase−TyrA or ACT domain. Type II enzymes comprise the catalytic core and an ACT domain, whereas type III enzymes do not contain either of the two C-terminal domains. Type I and II protomers build into physiological and functional homotetrameric structures. In contrast, type III protomers assemble into dimers.21 B
DOI: 10.1021/acs.biochem.8b00990 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 2. Ser3 and Ser33, but not human PHGDH, act as 3PGA-αKG transhydrogenases. (A) De novo serine synthesis from 3PGA involves three steps, starting with oxidation of 3PGA to PHP, transamination of the latter to 3PSer by Ser1 (yeast) or PSAT1 (humans), and finally dephosphorylation of 3PSer to serine by Ser2 (yeast) or PSPH (humans). In yeast, the first step is catalyzed by two isozymes, Ser3 and Ser33, which carry out a 3PGA-αKG transhydrogenase reaction. Here, the first half-reaction consists of oxidation of 3PGA to PHP and involves reduction of the enzyme-bound NAD+ coenzyme to NADH. In the second half-reaction, αKG is reduced to D-2HG with concomitant regeneration of NAD+. By contrast, hPHGDH consumes external NAD+ for 3PGA oxidation and does not couple this reaction to αKG reduction. Both the yeast28 and human31 PHGDH enzymes can, however, reduce αKG to D-2HG in the presence of external NADH in the absence of 3PGA. (B and C) The complete reaction mixture (as described in Materials and Methods) contained 5 μg/mL Ser3, 40 μg/mL Ser33, or 0.2 μg/mL hPHGDH. For the human protein, HEPES (pH 7.1) was replaced with Tris-HCl (pH 9.0). For certain assays, single components were omitted as indicated. Enzymatic activities were determined by measuring time-dependent 2HG and 3PSer formation by LC−MS/MS. Values shown are means ± SDs of three independent replicates. The activities measured for Ser3, Ser33, and hPHGDH in the complete mixture corresponded to 0.063 ± 0.003, 0.006 ± 0.001, and 0.943 ± 0.094 μmol min−1 (mg of protein)−1 (3PSer formation) and 0.063 ± 0.003, 0.009 ± 0.001, and 0 μmol min−1 (mg of protein)−1 (2HG formation), respectively. Abbreviations: GLU, L-glutamate; D-2HG, D-2-hydroxyglutarate; αKG, α-ketoglutarate; 3PGA, 3phosphoglycerate; PHP, 3-phosphohydroxypyruvate; 3PSer, 3-phosphoserine; Ser, serine.
EcPHGDH,18 type I Mycobacterium tuberculosis PHGDH (P9WNX3), or MtPHGDH23] were shown to be allosterically inhibited by serine, the end product of the corresponding metabolic pathway, while for instance, the rat PHGDH homologue (O08651) was insensitive to serine (tested up to 5 mM).24 On the basis of amino acid sequence analysis, the yeast Ser3 and Ser33 enzymes are most similar to EcPHGDH (Figure S2 and refs 21 and 24), but whether they have similar enzymatic properties and are sensitive toward serine has not yet been studied. Very recently, the PHGDH homologues from Pseudomonas stutzeri (PsPHGDH, A4VGK3) and EcPHGDH were found to act as transhydrogenases rather than classical dehydrogenases.25,26 Both enzymes oxidize 3PGA to PHP with a concomitant reduction of α-ketoglutarate (αKG) to 2hydroxyglutarate (2HG) (see Figure 2A). We also provided evidence that the yeast PHGDH homologues act as trans-
hydrogenases instead of dehydrogenases.27 Here, we report a more detailed biochemical characterization of the yeast Ser3 and Ser33 enzymes, in comparison to human PHGDH (hPHGDH, O43175). We confirmed that both yeast enzymes are acting as transhydrogenases, while the human enzyme is a dehydrogenase. We show that both yeast homologues, but surprisingly also hPHGDH, tightly bind NADH, which is rapidly oxidized to NAD+ upon addition of αKG. In the presence of 3PGA and αKG, however, D-2HG is formed in a stoichiometric manner with respect to PHP only by the ScPHGDHs, intimately linking serine and 2-hydroxyglutarate metabolism in yeast. Furthermore, we found that serine as well as hydrated NADH derivatives inhibit the yeast transhydrogenase activities but not hPHGDH. Protein sequence and structural analyses showed that all of the PHGDH homologues with hitherto experimentally verified transhydrogenase activity cluster within the type II category of C
DOI: 10.1021/acs.biochem.8b00990 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Yeast Cultivation Experiments. Yeast cultivations were performed on solid agar plates, in microtiter plates, shaking flasks, or benchtop bioreactor systems, as described in detail in the Supplemental Methods. Liquid cultivations were inoculated with 0.1−2% (v/v) glycerol stocks prepared from exponentially growing cells, while cultivations on solid agar plates were inoculated from liquid precultures. Extraction of Intracellular Yeast Metabolites. Intracellular metabolites were extracted from biomass (yeast cells) generated either in microtiter plates (48-well) or in bioreactors, based on a sequential quenching−extraction approach;29 500 μL culture aliquots were pipetted into 1.5 mL of 60% (v/v) cold methanol (−60 °C). Cells were immediately pelleted by centrifugation (10 min, −10 °C, 13000g), and the supernatant was removed using a 2 mL syringe equipped with a needle. Pellets were stored at −80 °C until extracted. Intracellular metabolites were extracted by adding a volume equivalent to 200 times the biovolume of the cells (which was determined at each sampling point via Multisizer Z3 measurement) of both extraction fluid {50% (v/ v) methanol, 50% (v/v) TE buffer [10 mM TRIZMA (pH 7.0), 1 mM EDTA]; −20 °C} and chloroform (−20 °C) to each cell pellet. The resulting mixture was incubated at −20 °C for 2 h on a shaking device (Eppendorf shaker) and centrifuged for 10 min at 13000g and −10 °C. The upper phase of the two-phase system was filtered (0.22 μm, PTFE, 4 mm diameter, Phenomenex) and stored at −80 °C until the polar metabolites were analyzed. LC−MS(/MS) Analyses. Two different types of LC−MS(/ MS) analyses were performed. Intermediates of the serine synthesis pathway (3PGA, 3PSer, 2HG, and αKG) were analyzed using an adaptation of a previously described method27 (or the unmodified method for some samples). In the adapted method, target metabolites were separated by reversed phase liquid chromatography using an EVO C18 column (150 mm × 2.1 mm, 3.5 μm particle size, 100 Å pore size; Phenomenex) connected to a guard column (SecurityGuard ULTRA cartridge, UHPLC EVO C18 for 2.1 mm inside diameter columns; Phenomenex), at a column oven temperature of 55 °C and a flow rate of 0.2 mL/min in isocratic mode using 0.5% formic acid in water as the sole eluent. Metabolites were detected in negative ionization MRM mode, as described in detail in ref 27. An example chromatogram and the mass transition details for the adapted method are given in Figure S1A. For the analysis of NAD+ and NADH, a high-resolution accurate mass (HRAM) reversed phase LC−MS method using a Dionex UltiMate 3000 LC instrument coupled to a Q Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific) equipped with a heated electrospray ionization source was used as described previously30 or with the following changes. Metabolites were separated at a constant flow rate of 0.2 mL/min in gradient mode, where solvent A was 50 mM ammonium acetate (pH 7) and solvent B was acetonitrile, according to the following profile: 0−1 min, 0% B; 1−18 min, 0−5% B; 18−26 min; 5−100% B; 26−27 min, 100−0% B; 27−40 min, 0% B (see also Figure S1B). For the quantification of NAD(H)(X) in some samples, LC separation was performed on a Nexera UHPLC instrument (Shimadzu) using this same chromatography method followed by detection with an SPD-M20A photodiode array detector (Shimadzu) at 279 and 340 nm.
PHGDH enzymes and revealed a conserved CXC motif in the substrate binding site of transhydrogenase enzymes specifically. Expressing either of the two yeast isozymes or the human homologue in an L-serine auxotrophic ser3Δser33Δ strain, we found different growth behaviors on fermentative but not on respiratory carbon sources. The results suggest that a transhydrogenase activity instead of an NAD+-consuming dehydrogenase activity, to carry out the thermodynamically unfavorable 3PGA to PHP oxidation, provides a growth advantage under conditions conducive to low intracellular NAD+:NADH ratios.
■
MATERIALS AND METHODS Chemicals and Reagents. Reagents, of analytical grade whenever possible, and synthetic complete dropout powder for yeast cultivations were acquired from Sigma-Aldrich, unless otherwise indicated. Liquid chromatography−mass spectrometry (LC−MS) grade solvents were obtained from VWR Chemicals. Yeast Nitrogen Base (YNB) was acquired from MP Biomedicals, and SC-Glc-Glu-Pro-Ser+YNB powder was obtained from Sunrise Science Products. Generation of Yeast Strains. The S. cerevisiae ser3Δser33Δ strain used in this study (isogenic to reference strain BY4741) was generated by mating of the corresponding single knockouts [obtained by polymerase chain reaction (PCR)mediated gene replacement], followed by sporulation and tetrad dissection as previously described.28 To generate rescue strains expressing the native forms of either SER3, SER33, or hPHGDH in this double-mutant background, the Gateway Cloning technology was used. For SER3 and SER33, the entry clones already existed from a previous study.28 For hPHGDH, the coding sequence for the native protein was PCR-amplified from a bacterial pET28-PHGDH expression plasmid (kindly provided by G. Bommer) with primers containing attB sites (forward primer, 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCATGGCTTTTGCAAATCTGCGGAAA-3′; reverse primer, 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTCTTAGAAGTGGAACTGGCCGGCTT-3′). The attB-flanked ORF was then shuttled into the pDONR221 vector (Invitrogen) using the Gateway BP Clonase II Enzyme mix (Invitrogen) according to the manufacturer’s instructions. All three inserts were further subcloned from the sequenceverified entry clones into yeast Destination Vector pAG416GPD-ccdB [AddGene ID 14148; containing a constitutive glyceraldehyde-3-phosphate dehydrogenase (GPD) promoter and the URA3 gene for nutrient selection] using the Gateway LR Clonase II Enzyme Mix (Invitrogen) to generate the desired expression clones. The purified expression plasmids were transformed into the ser3Δser33Δ mutant strain using the EZ-YEAST transformation kit (MP Biomedicals). Selective medium plates (YNB-agar supplemented with 2% glucose but no uracil) were incubated at 30 °C until colonies appeared. The plasmids and all yeast strains used in this study are listed in Table S1. Cultivation Media. Yeast cultivations were performed in chemically defined minimal medium, chemically defined rich medium (containing a subselection of amino acids), synthetic complete medium (containing all canonical amino acids, as well as adenine, myo-inositol, and p-aminobenzoic acid), or complex medium. Detailed compositions of media and an overview of which types of media were used for which experiments are given in the Supplemental Methods and Table S2, respectively. D
DOI: 10.1021/acs.biochem.8b00990 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
deviations of three independent replicates, unless otherwise indicated. The kinetic parameters for the transhydrogenase activities of Ser3 and Ser33 were calculated from at least three independent saturation curves fitted to the Michaelis−Menten V [S] equation without (v = K max+ [S] ) or with substrate inhibition
In all cases, absolute concentrations were calculated on the basis of external calibration curves established with the appropriate standard compounds dissolved in 50% methanol or in the matrix of the enzymatic assays that were performed. Protein Expression and Purification of Ser3, Ser33, and hPHGDH. Bacterial expression vectors based on pDest527 (AddGene ID 11518, IPTG-inducible T7 promoter, Nterminal His6 tag) and containing the coding sequences of either SER3 or SER33 were prepared as described previously.28 Bacterial expression plasmids pET28a(+)-hPHGDH (IPTGinducible T7 promoter, N-terminal His6 tag) and pET15bPSAT1, for expression of human PHGDH and human PSAT1 (Q9Y617), respectively, were kind gifts from G. Bommer and E. Van Schaftingen. All recombinant proteins were expressed in E. coli BL21(DE3) cells and purified by nickel-affinity chromatography followed by desalting on an Ä KTA protein purifier (GE Healthcare) as previously described.27 Protein concentrations were determined using the Bradford-based BioRad Protein Assay (Bio-Rad) according to the manufacturer’s instructions using bovine serum albumin as an external calibrator. The protein purity was estimated to >95% based on the band intensity of the proteins of interest compared to the sum of the intensities of all of the bands detected in the corresponding lanes in Coomassie Blue-stained sodium dodecyl sulfate−polyacrylamide gel electrophoresis gels. Purified protein fractions were stored at −80 °C in the presence of 10% glycerol and, for some Ser3 and Ser33 fractions, additionally 100 μg/mL bovine serum albumin. Coenzyme Characterization in Ser3, Ser33, and Human PHGDH Protein Preparations. Ser3, Ser33, and human PHGDH were incubated, at final concentrations of 320−580, 300−630, and 360−660 μg/mL, respectively, without or with αKG (200−250 μM) for 10 min (spectral analysis) or 30−60 min (HRAM LC−MS and HPLC analysis) at 30 °C, followed by heat denaturation of the proteins for 2 min at 95−99 °C and centrifugation for 15 min at 16100g and 4 °C to room temperature. The ultraviolet−visible absorption spectra of deproteinized samples were measured in a SPECORD 210 PLUS spectrophotometer (Analytik Jena), recording the absorbance in 1 nm steps from 200 to 600 nm with a slid width of 1 nm and a scanning speed of 50 nm/s in High Precision Cells (Quartz SUPRASIL, 10 mm light path, Centre 8.5). The identity of the soluble coenzyme in the deproteinized samples was confirmed by HRAM LC−MS, and coenzyme quantification was achieved via HPLC as described above. For quantifying the release of the coenzyme over time from the enzymes, samples were filtered without prior heat treatment on Vivaspin 500 concentrators (10 kDa molecular weight cutoff, VWR) and the deproteinized filtrate was analyzed by HPLC. Enzymatic Assays. The transhydrogenase activity of Ser3, Ser33, and hPHDGH was determined by incubating the enzyme at 30 °C in a reaction mixture containing 25 mM HEPES (pH 7.1), 50 μM 3PGA, 25 μM αKG, 500 μM NAD+, 250 μM L-glutamate, 50−100 μg/mL PSAT1, 1 mM DTT, 1 mM MgCl2, and 5 μM PLP, unless otherwise indicated. The reaction was started by adding 5−10 μg/mL Ser3, 15−40 μg/ mL Ser33, or 0.2−20 μg/mL hPHGDH. For the latter protein, the transhydrogenase assay mixture contained 25 mM Tris (pH 9.0) [instead of 25 mM HEPES (pH 7.1)]. Reactions were stopped by heat inactivation for 5 min at 95 °C. The formation of 2HG and 3PSer was assessed using the LC−MS/ MS method described above. Values are means and standard
m
[v =
Vmax[S]
(
K m + [S] 1 +
[S] Ki
)
] with GraphPad Prism (version 7.04) and
correcting for a “no substrate” (i.e., no 3PGA) control (where v is the velocity at substrate concentration [S], Vmax is the maximum velocity, Km is the substrate concentration at halfmaximal velocity, and Ki is the dissociation constant for substrate binding assuming that two substrate molecules can bind to the enzyme). The 3PGA dehydrogenase activity of hPHGDH was measured spectrophotometrically at 340 nm to follow NADH formation as previously described,27 except that NAD+ was added at a final concentration of 1.5 mM instead of 50 μM. Initial reaction velocities were determined using an extinction coefficient ε340 of 6220 M−1 cm−1 for NADH, and kinetic parameters were calculated using GraphPad Prism as described above for substrate inhibition enzyme kinetics, correcting for a “no substrate” control. For inhibition assays, (S)-NAD(P)HX, (R)-NAD(P)HX, and cyclic NAD(P)HX as well as NAD(P)+ or L-serine were added to the reaction mixture at final concentrations ranging from 0 to 10000 μM. NAD(P)HX derivatives were prepared and purified as described previously,27 and enzymatic activity measurements performed as described above. To determine IC50 and cooperativity (represented by the Hill slope) for serine inhibition, reaction velocities (Y) were fit to a nonlinear regression model (‘Dose Response−Inhibition’ analysis in GraphPad Prism selecting [inhibitor] vs response with variable Top − Bottom slopes): Y = Bottom + (where Top and Bottom i Hill slope y X 1 + jjjj Hill slope zzzz { k IC50
represent upper and lower reaction velocity plateaus, respectively, and X is the serine concentration). Bottom was constrained to 0 as control reactions run in the absence of the 3PGA substrate were used for background correction of the activity measurements.
■
RESULTS AND DISCUSSION Ser3 and Ser33, but Not Human PHGDH, Act as 3PGA-αKG Transhydrogenases. Ser3 and Ser33, currently annotated as 3PGA dehydrogenases, are the only functional PHGDH homologues in S. cerevisiae. Indeed, while single ser3Δ and ser33Δ mutants were able to grow in the absence of serine, this was not the case for a ser3Δser33Δ double mutant.8,28 We previously reported the detection of a pronounced NADH-dependent αKG reductase “side activity” of both Ser3 and Ser33, resulting in the formation of D-2HG.28 Despite extensive attempts, we failed, however, to detect a robust 3PGA dehydrogenase activity with these enzymes (based on the detection of NADH formation).28 Only when we added αKG to the reaction mixture were we finally able to measure a higher 3PGA oxidase activity, which was timedependent, independent of the addition of external NAD+, and stoichiometrically coupled to D-2HG production.27 We concluded that the yeast Ser3 and Ser33 enzymes catalyze a transhydrogenase reaction, rather than a classical dehydrogenase reaction, which was also recently demonstrated to be the E
DOI: 10.1021/acs.biochem.8b00990 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 3. Saturation curves for Ser3, Ser33, and human PHGDH with 3PGA as the varied substrate. (A and B) Activities of purified recombinant Ser3 (2−5 μg/mL) and Ser33 (28 μg/mL), respectively, were determined as described in Materials and Methods by measuring 2HG formation using LC−MS/MS. NAD+ (500 μM) was included for Ser3 activity measurements. (C) The activity of purified recombinant PHGDH (1.9 μg/mL) was measured spectrophotometrically by monitoring NADH formation. Values were corrected by subtraction of a “no substrate” control. Symbols and connecting lines represent independent replicates and nonlinear regression curves fit to the data.
case for the E. coli25 and Pseudomonas26 PHGDH homologues (Figure 2A). To further characterize this transhydrogenase activity, we incubated recombinant purified Ser3, Ser33, or hPHGDH, which has been reported to act as a dehydrogenase,31 in a reaction mixture containing 3PGA, αKG, and NAD + (complete mix). We also added human phosphoserine transaminase (PSAT1) and glutamate to the assay to pull the thermodynamically unfavorable 3PGA oxidation reaction toward PHP formation.32 Using a targeted LC−MS/MS method, we quantified 3PGA, αKG, 2HG, and 3PSer (resulting from PSAT transamination of PHP) in these incubations (see Figure 2A). In the complete mixture, 3PSer formation was detected in the presence of all three enzymes [0.063 ± 0.003 μmol min−1 (mg of protein)−1 with Ser3, 0.006 ± 0.001 μmol min−1 (mg of protein)−1 with Ser33, and 0.943 ± 0.094 μmol min−1 (mg of protein)−1 with hPHGDH] (Figure 2B). For both Ser3 and Ser33, 2HG was measured in the reaction mixture in amounts similar to that of 3PSer, suggesting that the formation of 2HG was involved in the main reaction [Figure 2C; 0.063 ± 0.003 and 0.009 ± 0.001 μmol min−1 (mg of protein)−1 for Ser3 and Ser33, respectively]. However, the presence of hPHGDH did not lead to the formation of such stoichiometric amounts of 2HG under the conditions tested. The reactions including hPHGDH were run at pH 9.0 (as opposed to pH 7.1 for the yeast enzymes), because we had previously successfully measured the dehydrogenase activity of the human enzyme at this pH. However, we also determined
2HG concentrations after incubations at pH 7.1 and 8.25 and tried different 3PGA concentrations (50−100 μM) and hPHGDH concentrations (≤20 μg/mL). Under none of these conditions were we able to detect 2HG concentrations of >0.6 μM [whereas 3PSer concentrations reached 5.9 and 11.9 μM in the incubations at pH 8.25 and 9.0, respectively, with 20 μg/mL hPHGDH, while no 3PSer was detected at pH 7.1 (data not shown)], indicating that the human PHGDH protein does not couple 3PGA oxidation with 2HG formation, which would be expected for a transhydrogenase. The very low 2HG formation activity measured with this enzyme likely results from its αKG reductase side activity31 and co-purified NADH, as we will show below. As expected, in the absence of the 3PGA substrate, no 2HG and/or 3PSer formation was detected with any of the enzymes (Figure 2B,C). When NAD+ was omitted from the transhydrogenase reaction mixture, 3PSer formation was no longer detected in the incubations containing hPHGDH, while in the presence of Ser3 and Ser33, 99 and 91% of the 3PSer forming activity, respectively, were preserved compared to that of the complete mixture containing NAD+ (Figure 2B). This clearly showed that the activity of the two yeast enzymes does not depend on externally added NAD+ and presumably uses the firmly bound coenzyme to catalyze the reaction. The activity of the human enzyme, however, depends on the addition of free NAD+ to oxidize 3PGA, as already observed by others.31 The slight decrease in transhydrogenase activity observed in the absence of externally added NAD+ for the yeast enzymes may be explained by a slow release of the enzyme-bound coenzyme F
DOI: 10.1021/acs.biochem.8b00990 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry Table 1. Kinetic Properties of Yeast Ser3 and Ser33 and Human PHGDHa Ser3 transhydrogenase activity Ser33 transhydrogenase activity hPHGDH dehydrogenase activity
Km (μM)
kcat (s−1)
kcat/Km (s−1 M−1)
Ki(3PGA) (μM)
181 ± 76 68 ± 30 39 ± 5
0.23 ± 0.03 0.07 ± 0.03 2.2 ± 0.3
(1.4 ± 0.5) × 103 (1.1 ± 0.1) × 103 (5.6 ± 0.8) × 105
89 ± 29 5200 ± 2000
a The 3PGA-αKG transhydrogenase activity of Ser3 and Ser33 was measured by assessing 2HG formation using LC−MS/MS, and the 3PGA dehydrogenase activity of hPHGDH was measured spectrophotometrically by monitoring NADH formation. The 3PGA substrate concentration was varied from 0 to 1 mM. For both types of activities, values were corrected by subtracting a “no 3PGA substrate” control value. Kinetic parameters were estimated using nonlinear regression fitting with GraphPad Prism (version 7.04). The values shown are means ± SDs obtained from three independent saturation curves.
observed variable effects of external NAD+ addition for different enzyme preparations (from virtually no effect to more or less pronounced stimulatory effects). While the precise reasons remain elusive and while we cannot formally exclude the possibility that a co-purifying protein may in some preparations interfere with the enzymatic assay, loss of enzymatic activity during storage at −80 °C and a tendency for protein precipitation at low salt concentrations most certainly contributed to these variations. We found that addition of external NAD+ to Ser3 activity assays led to a somewhat better reproducibility and therefore included the coenzyme at a final concentration of 500 μM for determining the kinetic parameters of this isozyme. A relatively high and variable background activity measured in the absence of externally added αKG (as described above due to αKG production in the coupling reaction by PSAT) also made it difficult to determine reliable values for the kinetic parameters of Ser3 and Ser33 when varying the αKG substrate concentration. On the basis of the results obtained with low αKG concentrations (and a 3PGA concentration fixed at 500 and 50 μM for Ser3 and Ser33, respectively), we found an average Km value of ∼10 μM for both enzymes. However, as with the 3PGA substrate, Ser33, but not Ser3, was strongly inhibited by higher concentrations (>10 μM) of αKG (not shown). These results show that, despite the high level of sequence identity between Ser3 and Ser33 (92%), these enzymes display some differences in their kinetic properties, but more work is needed to elucidate the differential roles that Ser3 and Ser33 may play in amino acid metabolism. Interestingly, using the SPELL search engine,33 it appears that the level of SER33 expression is highly correlated with genes involved in sulfur amino acid metabolism (note that cysteine is synthesized from serine), whereas the level of SER3 expression is correlated with genes involved more generally in amino acid metabolism. A comparison across 15 proteomic data sets contained in the PaxDb database suggests that the Ser3 protein is on average more abundant than the Ser33 protein in yeast cells.34 We have previously described the presence, in yeast, of a cytosolic hydroxyacid-ketoacid transhydrogenases (P39976, encoded by DLD3)28 that is responsible for D-2HG reconversion to αKG. The catalytic efficiency for D-2HG transhydrogenation by Dld3 (≈3.6 × 104 s−1 M−1)28 is >25 times higher than the catalytic efficiency found here for Ser3 and Ser33, indicating that the D-2HG produced during serine synthesis from 3PGA should be metabolized efficiently by Dld3, at least if the enzymes are expressed at similar levels. We also determined the kinetic properties of the dehydrogenase activity of hPHGDH in the presence of varying concentrations of 3PGA, using a previously described spectrophotometric assay monitoring NADH formation at
during storage and/or the assay. Omitting PSAT1 from the mixture led to a drastic decrease in activity in all cases (Figure 2B), showing that coupling 3PGA oxidation to the subsequent 3PSer transamination step remains beneficial for detecting this reaction, also when it proceeds via the transhydrogenase reaction. When αKG was excluded from the reaction mixture, the transhydrogenase activity of Ser3 and Ser33 decreased only partially to 64 and 41%, respectively, compared to the activity measured in the complete reaction mixture (Figure 2B,C). The dehydrogenase activity of hPHGDH (reflected by the 3PSer forming activity) was not affected by the absence of αKG. The residual activity measured with the yeast enzymes, which is expected to completely depend on αKG for 3PGA oxidation, can be explained by the formation of αKG (at a rate-limiting concentration) by the coupling enzyme PSAT1 (see Figure 2A). Catalytic amounts of PHP, necessary to start the PSAT transamination reaction, can presumably be formed in the system lacking externally added αKG because of NAD+ molecules firmly bound to and therefore co-purified with the Ser3 and Ser33 proteins. Similar observations were recently reported for the E. coli PHGDH homologue, where the transhydrogenase activity was reduced by ∼50% in the absence of αKG (measured on the basis of inorganic phosphate formation when PSP was additionally included in the reaction mixture25). Taken together, these data demonstrate that, in contrast to hPHGDH, both yeast enzymes are independent of supplemented NAD+ for oxidizing 3PGA to PHP, supposedly because they can recycle a tightly bound redox coenzyme during a second-half reaction, where reduction of αKG to D2HG occurs. While hPHGDH has been shown to be able to reduce αKG to D-2HG in the presence of NADH,31 we found no evidence that the human enzyme can couple αKG reduction to 3PGA oxidation in the absence of the supplemented coenzyme. Kinetic Characterization of Ser3, Ser33, and Human PHGDH. Having established that yeast Ser3 and Ser33 act as transhydrogenases while hPHGDH acts as a dehydrogenase, we next compared the kinetic properties of all three enzymes in the presence of varying 3PGA concentrations. We observed substrate inhibition by 3PGA at concentrations of >100 μM for Ser33 but not for Ser3 (Figure 3A,B). When this is taken into account, Ser33 displayed on average an affinity for 3PGA ∼3-fold higher than that of Ser3 whereas the turnover rate of Ser33 was ∼3-fold lower than that of Ser3, resulting in a very similar catalytic efficiency for both yeast paralogs (Table 1). It should be noted that for some preparations of purified recombinant Ser3 and Ser33, we observed transhydrogenase activities ≤5-fold higher than those shown in Figure 3. We also G
DOI: 10.1021/acs.biochem.8b00990 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 4. Ser3, Ser33, and human PHGDH tightly bind NADH, which is oxidized to NAD+ in the presence of αKG. (A) Ser3 (6.8 μM), Ser33 (5.4 μM), and hPHGDH (5.9 μM) were incubated in the absence or presence of 250 μM αKG for 60 min at 30 °C followed by heat inactivation (2 min at 95 °C), removal of precipitated protein by centrifugation, and coenzyme quantification by HPLC in the supernatant. (B) Ser3 (10 μM) and Ser33 (11.5 μM) were incubated in the absence or presence of 200 μM αKG for 10 min at 30 °C and treated similarly as described for panel A to record absorbance spectra in the deproteinized supernatants. (C) Ser3 (5.7 μM), Ser33 (5.6 μM), and hPHGDH (11.0 μM) were incubated at 30 °C, and the released coenzyme was measured at the indicated times by HPLC after protein removal by filtration. NAD+ and NADH concentrations were normalized to the total NAD(H)(X) concentration measured after heat denaturation (5.6, 5.8, and 6.4 μM for Ser3, Ser33, and hPHGDH, respectively). Values shown in panels A and C are means ± SDs (sometimes too small to be visible) of three independent replicates.
340 nm27 (Table 1). At 3PGA concentrations of >500 μM, a slight substrate inhibition was observed (Figure 3C). The apparent Km for 3PGA determined with our recombinant hPHGDH preparation (39 μM) was 7−12 times lower than those found by others (260 μM31 and 480 μM25), and the turnover number was 6−29 times higher [2.2 s−1 in this study vs 0.075 s−1 (ref 31) and 0.38 s−1 (ref 25)]. Different assay conditions (e.g., temperature, buffer composition, pH, salt content, and coupling reaction used) and/or the presence or absence of N-terminal tags, which have been described to affect hPHGDH activity,35 certainly at least partially explain these discrepancies. Using our assay conditions, but changing the pH to 8.25 or 7.1, we found ≈90% or no residual activity, respectively, with our hPHGDH preparation (data not shown). Under those artificial in vitro conditions, the human enzyme was ≈400-fold better (at pH 9), at the catalytic efficiency level, in oxidizing 3PGA than the yeast Ser3 and Ser33 enzymes (at pH 7.1). However, the former depends on the availability of free NAD+ and a probably high NAD+:NADH ratio at the neutral intracellular pH to push the dehydrogenase reaction in the physiological, thermodynamically unfavorable direction. By contrast, the yeast enzymes are independent of unbound NAD+ and rely on αKG. As suggested by Grant,25 the production of αKG by the PHP transamination reaction generates a self-sustaining cycle in terms of the αKG cosubstrate needed to fuel the first step of L-serine biosyn-
thesis. In budding yeast, one has to take into account the additional recycling of this α-ketoacid by Dld3, which leads to D-lactate formation (at the expense of pyruvate) and the coupling with the mitochondrial electron transfer chain at the level of cytochrome c.28 To gain insight into the reasons leading to the different reaction mechanisms used by the human and yeast PHGDH enzymes, we next investigated the identity and reactiveness of the potentially bound coenzyme for all three enzymes. Ser3 and Ser33 Tightly Bind NAD(H) but Slowly Release NAD+ over Time. Like transhydrogenases homologous to NAD-dependent dehydrogenases in other species, we hypothesized that Ser3 and Ser33 tightly bind the NAD coenzyme to function as an intermediate electron acceptor during 3PGA to PHP conversion. To test this hypothesis, we incubated Ser3 and Ser33, but also hPHGDH, in the absence or presence of αKG, denaturated the proteins by heating to release the coenzyme, and, after centrifugation, analyzed the supernatants by HRAM LC−MS and HPLC for small molecule identification and quantification, respectively. For all conditions and proteins, we could identify the release of the NAD coenzyme and the reduced and oxidized forms summed to an amount that indicated tight binding of this coenzyme in a stoichiometry corresponding to 0.5:1 for the human enzyme subunit and a stoichiometry that ranged from 0.6:1 to 1:1 for H
DOI: 10.1021/acs.biochem.8b00990 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 5. Ser3 and Ser33 are inhibited by L-serine and NADHX but not NADPHX. Ser3 (10 μg/mL) or Ser33 (20 μg/mL) was incubated in the absence or presence of (A and B) 0.1 and 1 μM NAD(P)(HX) or (C and D) the indicated L-serine concentrations, and reaction velocities were determined on the basis of 2HG formation as described in Materials and Methods. Means ± SDs (n = 3; except for Ser33 in the presence of 1 μM NAD+, where n = 2) are shown in panels A and B, and three independent replicate curves are shown in panels C and D. The activities measured in the absence of tested inhibitors ranged from 0.02 to 0.22 μmol min−1 (mg of protein)−1 for Ser3 and from 0.008 to 0.02 μmol min−1 (mg of protein)−1 for Ser33 and were determined for each inhibition assay independently.
the yeast enzyme subunits (the 1:1 stoichiometry was observed more frequently for the latter). In the absence of αKG, the reduced coenzyme (NADH) was predominantly bound to the enzymes, representing 83, 73, and 85% of the total coenzyme content for Ser3, Ser33, and hPHGDH, respectively (Figure 4A). When the enzymes had been preincubated with αKG, NAD+ was exclusively found (hPHGDH) or was in large excess over NADH (Ser3 and Ser33), showing that the bound coenzyme can be oxidized by all three enzymes in the presence of this α-ketoacid (Figure 4A). Consistent with these observations, we detected 2HG in the samples that had been incubated in the presence of αKG at concentrations expected from the amount of protein-bound NADH that had been oxidized to NAD+ (Figure 4A). In addition to LC(−MS) analysis, spectrophotometric analysis of similar samples further confirmed the coenzyme identity showing the characteristic absorption maximum of NADH at 340 nm, which disappeared after incubation with αKG (Figure 4B). Our results corroborate previous findings suggesting that the E. coli PHGDH homologue25 and also the D-2HG forming E. coli PdxB nicotinoenzyme36 preferably, but substoichiometrically, bind NADH, while other nucleotide binding sites (in the protein oligomers) contain NAD + . A lower coenzyme:subunit ratio of 0.5 was reported for the P. stutzeri PHGDH homologue,26 possibly due to partial coenzyme release during the purification procedure. The finding of copurified NAD(H) also with the human PHGDH enzyme was unexpected for us. However, others have independently suggested that this is indeed the case on the basis of the high A260/A280 ratio of human PHGDH preparations.19
We also investigated potential coenzyme release over time for the yeast and human proteins. Enzyme preparations were incubated at 30 °C, and after filtration, deproteinized aliquots were analyzed by HPLC. We detected increasing amounts of coenzyme in the filtrates, reaching, after 24 h, 5, 16, and 13% of the total coenzyme bound to Ser3, Ser33, and hPHGDH, respectively, as determined after heat denaturation (sum of NAD+, NADH, and NADHX) (Figure 4C). As for enzyme activity and its sensitivity to external NAD+, the amount of cofactor released under the conditions tested, especially for the Ser3 enzyme, was variable depending on the protein preparation used. The coenzyme released over time corresponded virtually exclusively to NAD+ (NADH represented
