The unique Phe–His dyad of 2-ketopropyl coenzyme M oxidoreductase/carboxylase selectively promotes carboxylation and S–C bond cleavage

Abstract

The 2-ketopropyl-coenzyme M oxidoreductase/carboxylase (2-KPCC) enzyme is the only member of the disulfide oxidoreductase (DSOR) family of enzymes, which are important for reductively cleaving S–S bonds, to have carboxylation activity. 2-KPCC catalyzes the conversion of 2-ketopropyl-coenzyme M to acetoacetate, which is used as a carbon source, in a controlled reaction to exclude protons. A conserved His–Glu motif present in DSORs is key in the protonation step; however, in 2-KPCC, the dyad is substituted by Phe–His. Here, we propose that this difference is important for coupling carboxylation with C–S bond cleavage. We substituted the Phe–His dyad in 2-KPCC to be more DSOR like, replacing the phenylalanine with histidine (F501H) and the histidine with glutamate (H506E), and solved crystal structures of F501H and the double variant F501H_H506E. We found that F501 protects the enolacetone intermediate from protons and that the F501H variant strongly promotes protonation. We also provided evidence for the involvement of the H506 residue in stabilizing the developing charge during the formation of acetoacetate, which acts as a product inhibitor in the WT but not the H506E variant enzymes. Finally, we determined that the F501H substitution promotes a DSOR-like charge transfer interaction with flavin adenine dinucleotide, eliminating the need for cysteine as an internal base. Taken together, these results indicate that the 2-KPCC dyad is responsible for selectively promoting carboxylation and inhibiting protonation in the formation of acetoacetate.

Biological CO₂ fixation reactions are essential drivers of the global carbon cycle, drawing atmospheric carbon into biomass and thereby helping to modulate the global thermostat. CO₂-fixing carboxylases are biochemically diverse and taxonomically widespread (1).

2-Ketopropyl-coenzyme M oxidoreductase/carboxylase (2-KPCC) is a member of the well-characterized NAD(P)H/flavin adenine dinucleotide (FAD)–dependent disulfide oxidoreductase (DSOR) family and a highly unusual bacterial carboxylase (2, 3). DSOR enzymes contain a redox-active cysteine pair that catalyzes the two-electron and two-proton reduction of a disulfide substrate (4, 5, 6), where GSSG reduction, catalyzed by glutathione reductase (GR), is a paradigmatic example (Fig. 1B) (7, 8, 9, 10). An FAD cofactor relays the two electrons and one proton (net hydride, H−) from NADPH to the disulfide and then interacts electrostatically with the reduced Cys dithiol to maintain its reactive protonation state. The second proton needed for the reaction is relayed from a conserved active site histidine (11, 12). By contrast, 2-KPCC reductively cleaves the thioether bond of its 2-ketopropyl-CoM (2-KPC) substrate to yield an unstable enolacetone intermediate. This then attacks CO₂ to generate the acetoacetate product anion and CoM (Figs. 1C and 2) (13, 14, 15). In this way, 2-KPCC uniquely repurposes the DSOR scaffold to achieve carboxylase chemistry.

Figure 1.

Understanding how a DSOR catalyzes a carboxylation reaction.A, general reaction catalyzed by carboxylases and resulting carboxylate product. E represents an electrophile, and the red semicircle an electropositive zone in the enzyme. [1] The substrate is first activated by the enzyme to yield an unstable and deprotonated ketone/enolate nucleophile. [2] The ketone/enolate can then attack CO₂ (shown distorted toward sp² hybridization) to produce the carboxylation product, such as acetoacetate in 2-KPCC. [3] In an undesired side reaction, the ketone/enolate can attack H⁺ or O₂, producing the protonated product, such as acetone in 2-KPCC. B, reaction catalyzed by a representative DSOR, glutathione reductase (GR). The first structure depicts the active site in its NADPH-reduced catalytically active form. The hashed line indicates a charge transfer (CT) interaction between the active-site Cys_CT (87) and the oxidized FAD, giving rise to UV–visible features with the listed maxima. The His–Glu catalytic dyad, which acts as a proton shuttle, and the two protons required for the reaction, are highlighted in color. C, reaction catalyzed by 2-KPCC. The first structure illustrates its reduced, reactive form, which contains a single proton shared between the two cysteines and no CT species or resulting band in the UV–visible spectrum. The Phe–His catalytic dyad that distinguishes 2-KPCC from canonical DSORs is shown. The reactive proton, which is shared between the catalytic cysteines and is transferred to CoM, is highlighted in color. 2-KPCC, 2-ketopropyl-coenzyme M oxidoreductase/carboxylase; CoM, coenzyme M; DSOR, disulfide oxidoreductase; FAD, flavin adenine dinucleotide.

Figure 2.

Carboxylation of 2-KPC by 2-KPCC to produce acetoacetate and coenzyme M (CoM). During turnover, NADPH is oxidized to NADP+. 2-KPC, 2-ketopropyl-CoM; 2-KPCC, 2-ketopropyl-coenzyme M oxidoreductase/carboxylase.

Catalytic efficiency in DSORs is reported to depend on a His–Glu pair known as the “catalytic dyad” (16). The conserved His shuttles protons to the substrates and stabilizes a charge transfer (CT) interaction between the FAD and one of the catalytic cysteines (Cys_CT) when the DSOR is in its reduced and reactive form. The Glu hydrogen bonds to the conserved His through its imidazole δ-N, orienting the side chain and lowering its pK_a (11, 12). In 2-KPCC, the His–Glu dyad is substituted by a phenylalanine and histidine (F501 and H506, respectively). We hypothesize that these amino acid substitutions are essential for making 2-KPCC an effective carboxylase. Prior work with the 2-KPCC F501H variant indeed showed that F501 serves an essential role in restricting protonation of the enolacetone intermediate (Fig. 1A, step [3]), as this substitution resulted in near-complete conversion of 2-KPC to the unproductive protonation product acetone (2).

Carboxylases operate via a shared mechanistic logic that defines canonical behavior (Fig. 1A). A nucleophilic substrate or an anionic substrate, often in the form of an enolate, is stabilized by the enzyme and serves as a nucleophile toward CO₂ (17). The CO₂ is either entrapped inside the protein as a molecule of solubilized gas or released on demand—from carboxybiotin or carboxyphosphate, for example—adjacent to the enolate. The sp-hybridized carbon of CO₂ is bent and polarized in the transition state that forms en route to the product carboxylate. Finally, an efficient carboxylase protects the reactive enolate from electrophiles that could potentially compete with CO₂ (18). For example, Rubisco, earth's most abundant carboxylase (19), shields its reactive enolate from O₂, a much more abundant electrophile; the fact that it does so imperfectly leads to Rubisco's observed oxygenated side products.

Here, the F501H and H506E single and F501H_H506E double variants, in which one or both residues of the 2-KPCC catalytic dyad are exchanged for the characteristic His–Glu pair of a typical DSOR, were used to test the hypothesis that F501 and H506 are responsible for endowing a typical DSOR with the essential attributes of a carboxylase. In conjunction with X-ray structural, spectroscopic, and catalytic data, we provide evidence that the noncanonical F501–H506 dyad of 2-KPCC promotes the canonical steps of carboxylation chemistry in Figure 1A by stabilizing the carboxylic acid product as well as the reactive protonation state of the active site.

Results and discussion

Members of the DSOR family of enzymes are best known for their roles in sulfur metabolism, where they reductively cleave disulfide bonds. We hypothesized that, despite being a member of the DSOR family, 2-KPCC would maintain much of the same chemical logic as other carboxylases in its catalytic conversion of 2-KPC and CO₂ to CoM and acetoacetate (Fig. 2). We further hypothesized that carboxylase chemistry would be encoded in the unique F501–H506 catalytic dyad of 2-KPCC, which replaces the canonical His–Glu of DSORs. These hypotheses were tested here using 2-KPCC with one or both residues of its dyad substituted by DSOR residues, with key results as follows:

F501 is optimally situated to block proton access to the reactive enolate intermediate

We began by testing the hypothesis that the phenylalanine in the dyad (F501) of 2-KPCC is chiefly responsible for excluding electrophilic protons from the 2-KPCC active site (Fig. 1A, step [3]). We previously showed that protonation of the enolacetone intermediate, resulting in toxic acetone, was strongly favored by the F501H substitution (2). The product ratio shifted from 0% acetone (WT 2-KPCC) to 90% acetone (F501H; Table 1). Notably, the overall rate of NADPH turnover remained relatively unchanged in the F501H variant relative to WT. Moreover, acetone formation by F501H was comparable in rate to the carboxylation reaction catalyzed by WT 2-KPCC and leading to acetoacetate. These results are consistent with kinetic analyses that showed that Cys–disulfide reduction by NADPH is rate limiting in 2-KPCC (3) and insensitive to the F501H substitution.

Table 1.

Specific activities and product analyses for WT 2-KPCC and dyad variants support key roles for each in specifying carboxylation reactivity

2-KPCC Enzyme	NADPH oxidation (nmol min−1 mg−1)a	Acetoacetate production (nmol min−1 mg−1)b	Acetone production (nmol min−1 mg−1)	Acetoacetate (%)
WT	517.0 (8.8)	284.5 (62)c	ND	100
F501H	452.7 (10.6)	22.1 (0.70)c	224 (51)c	10
H506E	326.0 (3.3)	16.2 (3.3)	27.1 (3.2)	37
F501H_H506E	189.7 (2.8)	ND	61.6 (7.1)c	0

Abbreviation: ND, no detectable activity.

^aOxidation of NADPH was monitored as a decrease in UV–visible absorbance at 340 nm via 2-KPCC–dependent 2-KPC oxidation of NADPH (60 mM KHCO₃, 5 mM 2-KPC, and 0.25 mg 2-KPCC).

^bCarboxylation of 2-KPC was monitored via the rate of decrease in UV–visible absorbance at 340 nm as described by the coupled assay (2, 21).

^cRates from Ref. (2).

To provide further insight into the validity of this interpretation, we determined the structure of the F501H variant and refined it to a resolution of 1.85 Å. The differences in the structure were restricted to the active-site region in proximity to the substitution. Crystals were grown in the presence of the substrate 2-KPC and, as in previous structures (14, 15, 20), was captured in the Cys–S–S–CoM disulfide state, which resulted from the reaction with 2-KPC during crystallization (Fig. 3). Moreover, His at the 501 position overlaps perfectly with the cognate dyad residue of GR, H467 (Fig. 3). This side chain occupies a central position between the Cys pair and the Cys-product mixed disulfide, where it has been shown to act as an essential active site acid–base (12). These structures strongly support the conclusion that the F501H variants, like WT GR, have an optimally positioned His that acts as a source of water-derived protons during turnover. This pathway for proton entry is effectively eliminated by the F501 side chain in WT 2-KPCC, which protects the enolacetone intermediate from water and/or side-chain bound H⁺ that could compete with CO₂ as an electrophile (Fig. 1A). This model supports the hypothesis that we put forth nearly 20 years ago to explain the differences observed in the catalytic dyad residues of 2-KPCC relative to GR.

Figure 3.

The His (or Phe) half of the dyads in 2-KPCC, F501H active-site variant, and GR occupy similar positions relative to the active-site cysteines in the mixed disulfide state. The dyad of WT 2-KPCC (left), F501H (middle), and GR (right) is shown with distances between the Phe (2-KPCC) or His (2-KPCC F501H variant and GR) to the catalytic cysteines. In each case, Cys_INT (C82) is shown making a mixed disulfide with the substrate. Protein Data Bank file of WT 2-KPCC: 2C3C; F501H variant: 7MGO; and GR: 1GRE. 2-KPCC, 2-ketopropyl coenzyme M oxidoreductase/carboxylase; GR, glutathione reductase.

Our recent work has suggested that the catalytic dyad differences observed in 2-KPCC go beyond that of simply limiting protonation in the promotion of carboxylation. We have shown that 2-KPCC operates through a different key reactive intermediate than GR (3) in which the reduced state of the redox active dithiol exists with a bridged single proton instead of the FAD–Cys_CT species observed in GR (Fig. 1, B and C). We can see in our structure of the F501H variant and the structure of GR that the interactions of His with the exchange thiol (Cys_INT (C82)) would promote the stabilization of the GR-type reduced reactive intermediate relative to the 2-KPCC intermediate through stabilizing the more formal thiolate. We have attributed these differences in reactive intermediates observed between 2-KPCC and GR as yet another mechanism to limit proton access and promoting carboxylation. This is very much in line with the results of the reactivity of the F501H variant having substantial reductive thioether bond cleavage but markedly suppressed carboxylation activity (Table 1).

H506 defines a likely site for CO₂ binding and conversion to the carboxylate moiety of the acetoacetate product

CO₂ binding, deviation toward sp² hybridization, and development of negative charge in the ultimate carboxylation product (acetoacetate) are expected to be promoted by electropositive side chains in the active-site pocket (Fig. 1) (14, 20). In previous structures, we observed the binding of molecules assigned as anionic ligands derived from the crystallization solution in a small pocket that is flanked by Arg365 and His506, where His506 is at the analogous position of the Glu residue at the second position of the GR catalytic His–Glu dyad. We hypothesized, in line with the functional differences associated with the first residue of the dyad (Phe in 2-KPCC at the position of His in GR) that the second difference (His in 2-KPCC relative to Glu in GR) is also mechanistically relevant in having a role in promoting the unique carboxylation activity of 2-KPCC. Specifically, the anion-binding pocket may stabilize the developing charge incurred during carboxylation, thereby facilitating carboxylation and acetoacetate formation.

To address this potential role, we determined the structure of the F501H_H506E variant and refined it to a resolution of 1.8 Å (Fig. 4). In comparing the structure of the variant and WT 2-KPCC with that of GR, we observe that the analogous residue in the GR dyad, E472, is not in the same position as 506 in 2-KPCC, as shown in WT 2-KPCC in Figure 3. In GR, the E472 carboxylate side chain is positioned within hydrogen bonding distance to H468 and acts to modulate its pK_a. The different position of E506 in 2-KPCC is a likely result of differences in the isomeric state of the Pro in 2-KPCC (trans) versus GR (cis). In addition, the position of the Pro in the short region between the catalytic dyad residues in 2-KPCC (F₅₀₁LNPTH₅₀₆) versus GR (H₄₆₇PTSSE₄₇₂) results in differences in the backbone conformation. These differences presumably are important for the formation of an anion-binding pocket at this position in 2-KPCC. The unique position, dramatic difference in electrostatic character, and open cavity supported by the H506 residue in WT 2-KPCC all point toward a possible role in promoting carboxylation through a site that would stabilize the developing charge incurred during the formation of the product acetoacetate.

Figure 4.

Replacing the dyad to mimic DSORs changes the anionic pocket of 2-KPCC. The structure of 2-KPCC is shown with one dimer colored gray and the other cyan. The active site, which resides at the dimer interface, is shown in the black boxed area. A close up of the anionic pocket in the WT active site is shown in the top gray boxes, whereas the F501H_H506E variant is shown in the bottom two gray boxes. The residues from one dimer are shown in cyan carbons, whereas the other dimers are shown in gray carbons. The substituted His at 501 and Glu at 506 are shown in purple carbons. Although the orientation of the DSOR-like His–Glu aligns with the WT Phe–His dyad, the active-site pocket size has decreased near the E506 residue. Protein Data Bank file of WT 2-KPCC: 2C3C; F501H_H506E variant: 7MGN. 2-KPCC, 2-ketopropyl coenzyme M oxidoreductase/carboxylase; DSORs, disulfide oxidoreductases.

We further examined the hypothesis that the 2-KPCC H506 residue is involved in stabilizing the developing charge in the formation of the anionic acetoacetate product in two ways, focusing on the H506E and F501H single and the F501H_H506E double variants. First, we considered the impact of the H506E substitution on the product outcome. The H506E variant exhibited a dramatic shift toward acetone as the primary reaction product (63%; Table 1). If H506 is the site of CO₂ binding, polarization, and acetoacetate formation, the H506E substitution could favor acetone production by electrostatically blocking CO₂ binding and/or acetoacetate formation (Fig. 1A, step [2]). In the absence of a carboxylation reaction, the enolacetone intermediate is expected to linger until water-mediated protonation could occur. In fact, acetone was produced at an approximately 10-fold slower rate in 2-KPCC H506E relative to F501H. Consistent with the slower rate of acetone production, NADPH turnover was ∼1.5-fold slower in H506E versus WT. Substrate turnover and product formation by the F501H_H506E double variant bore the hallmarks of each individual mutation. This variant exhibited no detectable acetoacetate production, instead making entirely acetone. Furthermore, the rate of acetone production was double that of the H506E single variant, consistent with the proposed active-site acid role for the H501.

As a second, more direct test of the role of H506 as an acetoacetate binding site, we examined acetoacetate as a possible product inhibitor. First, the initial rate of NADPH turnover (v_i) was measured as a function of variable 2-KPC concentration (1–5 mM) at a series of fixed concentrations (0–10 mM) of acetoacetate using WT 2-KPCC. The plots of v_i versus [2-KPC] could each be fit to a rectangular hyperbolic function, consistent with the Michaelis–Menten model: v_i = V_max[S] ([S] + K_M)⁻¹. The value of V_max decreased as a function of increasing [acetoacetate], whereas the K_M (2-KPC) remained approximately the same (Fig. 5).

Figure 5.

Mixed noncompetitive inhibition of 2-KPC consumption is observed between WT 2-KPCC and acetoacetate. Each curve was fit to the Michaelis–Menten equation, and apparent values for k_cat and K_M for each inhibitor concentration are tabulated. Changes in both parameters are consistent with a mixed noncompetitive inhibition pattern, where the acetoacetate product inhibitor binds both the free enzyme and enzyme–substrate complex at a site distinct from the location of substrate binding. We propose that this site is formed in part by the positive charge at H506. 2-KPC, 2-ketopropyl coenzyme M; 2-KPCC, 2-ketopropyl coenzyme M oxidoreductase/carboxylase.

If acetoacetate binds adjacent to H506, we reasoned that the H506E substitution should severely distort the electrostatic nature of the binding pocket, reduce or eliminate acetoacetate binding, and eliminate the inhibitory effects of acetoacetate. In fact, acetoacetate had no effect on catalysis by the H506E variant, even at concentrations up to 20 mM. These data further support a model where H506 is the site of enolacetone attack on CO₂ to form acetoacetate.

2-KPCC can use either its native or a DSOR-like reduced form to cleave the C–S bond of 2-KPC

As described previously, F501 is an essential component of the 2-KPCC dyad, because it protects the enolacetone intermediate from reacting with a proton to make the toxic product, acetone. The DSOR-like F501H substitution contributes an acidic proton to the 2-KPCC active site, promoting this unwanted reaction.

The residue at position 501 serves a second crucial function in DSORs, including 2-KPCC, in maintaining the reduced enzyme in its reactive proton tautomer state (Fig. 1, B and C) (3). Specifically, the reduced and reactive form of DSORs is doubly protonated, at Cys_INT and at the histidine from the catalytic dyad (Fig. 1B). The latter interacts electrostatically with Cys_CT, stabilizing the thiolate–FAD–CT interaction. By contrast, the hydrophobic F501 side chain in 2-KPCC stabilizes the Cys_CT (C87) thiol (Cys_CT-SH), so that a CT is only observed at very high pH (pK_a = 9.4 for the C82A variant) (3). At neutral pH, the reactive form of 2-KPCC has a single proton in its active site, which is shared between the Cys_INT (C82) and Cys_CT (C87) and fully transferred to the latter immediately prior to the reaction with 2-KPC (Fig. 1C) (3).

We hypothesized that 2-KPCC F501H could, like H487 in GR, stabilize a CT interaction with the FAD in the reduced state (Fig. 1B, hashed line). To test this, we measured the UV–visible spectra of the C82A_F501H double variant as a function of pH (Fig. 6). This variant lacks the Cys_INT (C82); hence, all effects of pH on the active site will be confined either to the Cys_CT (C87) or to H501. As with the C82A protein, a single pH transition from an FAD to an FAD-CT species was observed as the pH increased, though with a pK_a shifted from 9.4 (C82A) to 8.2 (C82A_F501H). This indicates that the DSOR-like CT species is selectively stabilized by the F501H substitution.

Figure 6.

The F501H substitution favors a DSOR-like active-site electronic structure. The C82A_F501H double variant forms a Cys_CT (C87)-flavin charge transfer (CT) adduct at alkaline pH with a pK_a of 8.2, which is attributed to deprotonation of Cys_CT (C87). The acidic spectrum (pH 6.5) is shown in red, and the alkaline spectrum (pH 10.5) is shown in blue. The alkaline form is a CT species in which the Cys_CT (C87) interacts with the flavin. Intermediate pH spectra (pH 7.5, 8.5, and 9.5) are in gray. Inset, fitting the change in absorbance at 550 nm to the equation for a sigmoidal curve gives an apparent pK_a of 8.2, which is significantly lower than the analogous transition for WT (pK_a = 9.4). This suggests that H501, like the His in the canonical DSOR dyad, stabilizes the FAD/Cys_CT (C87) CT interaction. DSOR, disulfide oxidoreductase; FAD, flavin adenine dinucleotide.

We then asked whether the CT species in the F501H variant is capable of reacting with 2-KPC to form the mixed disulfide intermediate. To answer this question, we examined the reaction between reduced 2-KPCC and bromoethanesulfonate (BES), a 2-KPC surrogate that covalently modifies Cys_INT (C82) via alkylation when Cys_INT (C82) is in its deprotonated form (Fig. 7) (21). Alkylation, which occurs when the active-site proton shifts from a shared position between Cys_INT (C82) and Cys_CT (C87), was detected via inactivation of the enzyme: that is, a fully modified enzyme exhibits no further reactivity toward NADPH, 2-KPC, and CO₂. In the plot of percent activity versus pH (Fig. 8), the inflection point is interpreted as the pK_a for Cys_INT (C82), since deprotonation of this residue leads synchronously to alkylation with BES. For WT 2-KPCC, this inflection point occurs near pH 8 (3), well below the pK_a describing formation of the CT species and consistent with Cys_CT (C87) acting as the base toward Cys_INT (C82) (Fig. 1C). For the F501H substitution (single and double variants), the inflection point undergoes an alkaline shift of approximately 0.5 to 1 pH unit. Hence, the F501H substitution has a modest but measurable effect on the pK_a for Cys_INT (C82), slightly suppressing its reactivity. We speculate that, because the F501H substitution stabilizes the Cys_CT (C87)-FAD CT interaction (pK_a = 8.2; Fig. 6), the Cys_CT (C87) is not available to act as a base toward the Cys_INT (C82) proton. Rather, the H501 side chain of the F501H variant may instead act as the base toward Cys_INT (C82), and the pK_a near pH 9 in Figure 8 may be due to the deprotonation of H501, which catalyzes deprotonation of Cys_INT (C82) in turn. Together, these data indicate that the F501H substitution stabilizes a DSOR-like protonation state in the reduced/reactive form of 2-KPCC, in which the Cys_CT (C87) thiolate engages in a CT interaction with the FAD. In spite of being distinct from WT, this form of the protein is nonetheless reactive with 2-KPC. We propose that the histidine in F501H is capable of acting as a base toward Cys_INT (C82), much like in a normal DSOR, removing the proton from the interchange thiol and catalyzing C–S bond cleavage within 2-KPC. This obviates the need for using Cys_CT (C87) as an internal base.

Figure 7.

Reaction of BES with Cys_INT(C82) to form the alkylated complex. BES, bromoethanesulfonate.

Figure 8.

The F501H variant reacts via the DSOR-like form of the active site. Irreversible inactivation of 2-KPCC (WT and variants) by BES was monitored as a function of pH. Here, the measured pK_a obtained by fitting a sigmoidal curve to the data is for the Cys that reacts with substrate (Cys_INT (C82)), and the recipient base is Cys_CT (C87). For the WT protein (red circles), this pK_a is close to 8. For F501H (blue circles) and F501H_H506E (green circles), the pK_a shifts closer to 9, again suggesting that protonation of Cys_CT (C87) is destabilized in both variants. 2-KPCC, 2-ketopropyl-coenzyme M oxidoreductase/carboxylase; BES, bromoethanesulfonate; DSOR, disulfide oxidoreductase.

Conclusion

The carboxylation activity of 2-KPCC is unique among the DSOR family of enzymes. Although 2-KPCC exists with the conserved overall architecture and key elements important for Cys disulfide–dependent reductive bond cleavage, there are key elements that differ from other members of the family. We have proposed that these differences are important for carboxylation activity, most importantly the differences in the residues that constitute the catalytic dyad. The canonical DSOR catalytic dyad is key to stabilizing a specific thiol-reactive intermediate and providing protons for product formation. For 2-KPCC, coupling reductive cleavage with carboxylation adds an additional carboxylase-specific layer of criteria that needs to be satisfied, including the protection of the stabilized enolate intermediate from other attacking electrophiles like protons. Here, we provide compelling new insights through structural and biochemical analyses that the lack of conservation in the 2-KPCC catalytic dyad is of key mechanistic importance in allowing 2-KPCC to maintain its function in reductive cleavage and carboxylation activity. The substitution of Phe in 2-KPCC for the His observed in other members of the DSOR family promotes the formation of a different reactive intermediate and limits protons as competing electrophiles at the active site, suppressing the formation of the unproductive product acetone and promoting carboxylation. The substitution of His in 2-KPCC for Glu observed in other members of the DSOR family, in addition, helps form an anion binding pocket that stabilizes the developing charge incurred during acetoacetate formation.

Experimental procedures

Expression and purification of native 2-KPCC and amino acid–substituted variants

Escherichia coli BL21(DE3)pLysS or BL21(DE3) cells were transformed with the pBAD plasmid harboring the WT or corresponding variant 2-KPCC gene from Xanthobacter autotrophicus Py2. The construct contained a His affinity tag, a thioredoxin tag, and a tobacco etch virus cleavage site. Cells were plated on LB agar + kanamycin (25 μg/ml) plates and grown overnight. A single colony from the plate was used to inoculate a 5-ml overnight culture in LB + kanamycin (25 μg/ml). The 5 ml overnight culture was used as the inoculum for 500 ml of ZYP-rich media + kanamycin (25 μg/ml) in a baffled flask. Cells were grown at 37 °C with agitation at 225 rpm until the cultures reached an absorbance of 0.6 to 1.0. The temperature was reduced to 25 °C, l-arabinose was added to 0.02%, and the cells were grown for an additional 16 to 18 h. The cells were pelleted by centrifugation, frozen, and stored at −80 °C. The cell pellet was resuspended in four volumes of lysis buffer (20 mM Tris, pH 8.0, 300 mM NaCl, 5 mM imidazole, and protease inhibitor tablet [Pierce]) and thawed at 30 °C. All subsequent treatments were performed on ice or at 4 °C. The resuspended cells were lysed via multiple rounds of sonication (Branson Ultrasonifier). Cell lysates were clarified via centrifugation at 75,000g for 45 min. Clarified lysates were loaded onto a nickel–nitrilotriacetic acid resin column (Bio-Rad) via a Bio-Rad FPLC, washed with lysis buffer, and eluted using a 0.005 to 0.4 M imidazole gradient in lysis buffer at 3 ml/min. The eluted 2-KPCC was diluted sixfold into buffer A (20 mM Tris, pH 6.5, and 5% w/v glycerol), applied to a DEAE-Sepharose ion-exchange column (Bio-Rad), and eluted using a 0 to 1 M NaCl gradient in buffer A. Fractions were screened using SDS-PAGE. Pure 2-KPCC protein was pooled and run on a desalting column (Bio-Rad) in 20 mM Tris, pH 7.4, 10% glycerol, and 200 mM NaCl. Protein was concentrated using a 30 K molecular weight cutoff spin filter concentrator. Total protein concentration was determined using a bicinchoninic protein assay (Thermo Scientific), whereas flavin concentration was determined from its UV–visible absorbance at 450 nm using an ε450 of 11,828 M⁻¹ cm⁻¹. All concentrations of protein cited in the text refer to flavin-containing protein.

Crystal growth and structure determination

The crystals for this study were obtained by the vapor diffusion method. The purified F501H and F501H_H506E 2-KPCC (20 mg/ml) were incubated with substrate 2-KPC or CoM (10 mM final concentration), respectively, for 10 min prior to addition of an equal amount of precipitating solution containing 0.17 M ammonium acetate, 0.085 M trisodium citrate, pH 5.6, 27% PEG 4000, and 15% glycerol. Crystals in the form of clusters of lathes appeared after 1 week of incubation. They belong to the monoclinic space group with one dimer per asymmetric unit and cell dimensions a = 87.05 Å, b = 60.21 Å, c = 104.86 Å, and α = γ = 90°, and β = 100.46°. Data were collected from single plate flash-cooled crystals with a continuous flow of liquid nitrogen at 100 K at Stanford Synchotron Radiation Lightsource beamline 12-2 equipped with a PILATUS detector with 0.15° oscillation.

The diffraction images were indexed, integrated, and scaled using HKL2000 (22) or XDS (23). The F501H and F501H_H506E 2-KPCC structures were solved by the molecular replacement method using the mixed-disulfide crystal structure of 2-KPCC (Protein Data Bank [PDB] code: 2C3C) as a search model with Phaser of the CCP4 program suite (24).

The solutions were refined and improved by phenix.refine. Model building was subsequently completed manually using Coot software (The University of Oxford) (25). Figures were prepared using PyMol (Schrodinger) (26). F501H and F501H_H506E 2-KPCC structures were deposited in the PDB with code 7MGO and 7MGN, respectively. The data processing and refinement statistics for these structures are given in Table S1.

Monitoring UV–visible changes in the flavin species in 2-KPCC variant as a function of pH

All pH titrations were carried out anerobically in a Coy anerobic chamber. Samples contained 20 μM 2-KPCC, 100 μM NADP⁺, 100 mM glycine/Tris/phosphate buffer and were conducted using a procedure similar to that described by Kofoed et al. (27). Samples were loaded into gas tight cuvettes, and the reaction was monitored using an Agilent Cary UV–visible NIR spectrophotometer.

BES inactivation

All reactions were conducted under nitrogen (∼2% H₂) in an anerobic chamber (Coy Lab Products). Solutions of 100 mM buffers were made at varying pH from Tris base, K₃PO₄, and glycine. DTT and BES from the concentrated stocks were added to each solution to give final concentrations of 5 mM BES and 10 mM DTT. A control sample that lacked BES was run in parallel. The samples were allowed to incubate at 20 °C for 4 h. This incubation time was previously shown (3) to result in a range of enzyme reactivity with BES, from <10% to nearly stoichiometric, depending on the pH. It was concluded that increasing pH therefore shifts the proton tautomerization state, from a position where it is shared between Cys_INT/Cys_CT and Cys_CT. To remove unreacted BES, samples were treated with DOWEX resin as previously described (27). Each sample was assayed for residual activity via addition of the BES-treated enzyme to a 5 mM 2-KPC and 0.1 mM NADPH solution made in 100 mM GTP buffer at pH 7. The initial rates of NADPH consumption were monitored at 340 nm using an Agilent 8453 spectrometer equipped with a diode array detector and compared with the sample control. Triplicate technical replicates were carried out for WT, F501H, and F501H_H506E 2-KPCC.

Kinetic/functional assays and product analyses

Acetoacetate production was measured using the coupled β-hydroxybutyrate dehydrogenase (β-HBDH) assay as previously described (2, 21) using a BMG Labtech CLARIOstar microplate reader in reaction volumes of 200 μl. Under excess β-HBDH, acetoacetate produced by 2-KPCC is coupled to NADH oxidation by the reverse reaction of β-HBDH. The reaction mixture contained 0.25 mg 2-KPCC, 10 mM DTT (as the reducing agent instead of NADPH), 60 mM KHCO₃, 0.2 mM NADH, and the reaction was initiated by injecting 5 mM 2-KPC. Reactions were conducted in a 100 mM Tris at pH 7.4 buffer. Rates of NADH oxidization, observed by the decrease in the absorbance at 340 nm, are proportional to acetoacetate production.

The rate of NADPH oxidation was measured as the time-dependent decrease in the absorbance at 340 nm on an Agilent Cary UV–visible NIR spectrophotometer in reaction volumes of 500 μl. The reaction contained 0.2 mM NADPH, 0.25 mg 2-KPCC, 60 mM KHCO₃, and the reaction was initiated by adding 5 mM 2-KPC. Reactions were conducted in a 100 mM Tris at pH 7.4 buffer. Assays were analyzed in triplicate.

The off-pathway product acetone was quantified via 100 μl headspace injections on a ThermoFisher GC Trace 1300 using buffer degassed on a vacuum/nitrogen gas manifold in series with a CO₂ scrub KOH trap. The standard assay contained 100 mM Tris at pH 7.4, 10 mM DTT, 0.25 mg 2-KPCC, and 5 mM 2-KPC in a 10 ml crimp-sealed vial. After 50 min incubation at 30 °C and 180 rpm, KHCO₃ (60 mM) was added via syringe as a source of CO₂ substrate. The difference in rates of activities measured by the GC and the UV–visible coupled assay from the NADPH oxidation assay (Table 1) can be attributed to the use of the non-natural reductant DTT as opposed to the physiological reductant NADPH.

Acetoacetate inhibition assays were performed at varying concentrations of acetoacetate (0–10 mM), monitoring 2-KPC–dependent oxidation of NADPH at 340 nm. The standard assay mixture contained 100 mM Tris at pH 7.4, 0.5 mM NADPH, 60 mM KHCO₃, and 0.25 mg 2-KPCC. Assays were initiated by addition of substrate 2-KPC and monitored with a BMG Labtech CLARIOstar plate reader in reaction volumes of 200 μl.

Kinetic constants (K_M and V_max) were calculated by fitting initial rate data to the Michaelis–Menten equation using Kaleidagraph software (Kaleidagraph).

Data availability

X-ray crystal structure data that support the findings of this study have been deposited in the Worldwide PDB. The PDB accession codes for the F501H and double F501H_H506E 2-KPCC variants are 7MGO and 7MGN, respectively. All other data are contained within this article and supporting information.

Supporting information

This article contains supporting information.

Conflict of interest

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

Edited by Ruma Banerjee