Control of Metabolite Flux during the Final Steps of Heme b Biosynthesis in Gram-Positive Bacteria

Abstract

The pathway for assembling heme ends with a unique set of enzymes in Gram-positive bacteria. Substrates for these reactions include coproporphyrin III, Fe(II), and H₂O₂, which are highly reactive and toxic. Because these bacteria lack membranous compartments, we hypothesized that metabolite flux may occur via a transient protein–protein interaction between the final two pathway enzymes, coproporphyrin ferrochelatase (CpfC) and coproheme decarboxylase (ChdC). This hypothesis was tested using enzymes from the pathogen Staphylococcus aureus and a corresponding ΔchdC knockout strain. The ultraviolet–visible spectral features of coproporphyrin III served as an in vitro indicator of a protein–protein interaction. A CpfC–ChdC K_D of 17 ± 7 μM was determined, consistent with transient complexation and supported by the observation that the catalytic competence of both enzymes was moderately suppressed in the stable complex. The ΔchdC S. aureus was transformed with plasmids containing single-amino acid mutants in the active site gate of ChdC. The porphyrin content and growth phenotypes of these mutants showed that K129 and Y133 promote the ChdC–CpfC interaction and revealed the importance of E120. Understanding the nature of interactions between these enzymes and those further upstream in the heme biosynthesis pathway could provide new means of specifically targeting pathogenic Gram-positive bacteria such as S. aureus.

Graphical Abstract

The recent discovery of the coproheme decarboxylase (ChdC) enzyme led to the elucidation of a new branch of the biosynthetic pathway for the essential cofactor, heme b.¹ Now known as the coproporphyrin-dependent branch (CPD), this portion of the pathway is almost exclusive to Gram-positive bacteria, deviating from the canonical, protoporphyrin-dependent branch (PPD) at the terminal three steps (Scheme 1, purple and blue, respectively). These branches differ in the order in which the porphyrin chemical transformations occur, the enzymes that catalyze them, and the identities of the intermediates, all of which are highly reactive and potentially toxic molecules (Scheme 1).^2,3 In particular, reduced porphyrinogens are pro-oxidants, while their oxidized porphyrin counterparts are acutely photoreactive. In addition, Fe(II) is well-known for catalyzing the Fenton reaction in which cellular H₂O₂ is converted to hydroxide anion and a hydroxyl radical, a potent initiator of DNA damage. Tight regulation of metabolite flux during heme biosynthesis is therefore essential to protect the cell.

Scheme 1. PPD and CPD Branches of Heme b Biosynthesis Differ in the Terminal Three Steps^a.

^aThe terminal steps of heme biosynthesis employ reactive and potentially toxic substrates. In contrast to the enzymes of the canonical pathway (PPD branch, blue), all three of which are membrane-associated and/or compartmentalized, the enzymes that catalyze the last three steps of the CPD branch (purple) are found in Gram-positive organisms that lack cellular membranes for restricting their locations within the cell.

Control mechanisms containing highly reactive substrates and intermediates have been described for organisms that use the PPD branch of heme biosynthesis (namely, eukaryotes and most Gram-negative bacteria).^3–5 In eukaryotes, the enzymes catalyzing the first five steps of the biosynthetic pathway (PgbS-UroD), which do not produce cytotoxic molecules, are all located in the cytoplasm. The last three enzymes (CgdC, PgoX, and PpfC) are localized to the mitochondrial inner membrane.² Compartmentalization keeps toxic intermediates within a single location and facilitates direct interaction between the terminal pathway enzymes. In particular, protoporphyrinogen oxidase (PgoX) and protoporphyrin IX (PPIX) ferrochelatase (PpfC) are bound to either side of the inner mitochondrial membrane, where they have been shown to pass the substrate and/or product directly to each other through a channel in PgoX. This allows delivery of cytotoxic Fe(II) from the mitochondrial matrix and rapid conversion of PPIX to heme, keeping the free PPIX concentration negligible and harmless to the cell.³ In a similar manner, Gram-negative bacteria can compartmentalize the same terminal step enzymes in their inner cell membrane and the periplasmic space.⁶

How metabolite flux is controlled in Gram-positive bacteria is unknown. Gram-positive bacteria have only a single membrane and a thick but porous peptidoglycan layer that does not allow formation of a canonical periplasmic compartment. In addition, the terminal enzymes of the CDP branch [coproporphyrinogen oxidase (CgoX), coproporphyrin ferrochelatase (CpfC), and coproheme decarboxylase (ChdC)] are not membrane-bound. Consequently, any mechanism for control over the cytotoxic metabolites associated with this pathway—coproporphyrinogen III, coproporphyrin III, and Fe(II)—must be distinct from the PPD branch.

Protein–protein interactions offer a potential means for controlling metabolite flux through metabolic pathways,^7,8 permitting direct delivery of substrates and/or products between the relevant enzymes. We hypothesized that direct metabolite channeling through protein–protein interactions could be used as an alternative to membrane compartmentalization within the cell for the CPD branch of heme biosynthesis in Gram-positive bacteria. We specifically focused our efforts at describing protein–protein interactions that occur between the terminal two enzymes, CpfC and ChdC, which catalyze Fe(II) metalation and decarboxylation reactions to yield heme. Here, using a variety of biochemical tools and characterization of in vivo phenotypes, we describe a CpfC–ChdC protein–protein interaction that appears to support heme assembly in the Gram-positive pathogen, Staphylococcus aureus.

MATERIALS AND METHODS

Reagents and Stocks.

Ferric coproporphyrin III chloride (coproheme III, Frontier Scientific) and coproporphyrin III dihydrochloride (Frontier Scientific) were obtained in 10–100 mg ampules and used to generate 5–10 mM stock solutions in dimethyl sulfoxide (DMSO). These stocks were further diluted in 50 mM Tris (pH 8) and 150 mM NaCl prior to use. The coproporphyrin and coproheme concentrations in these solutions were ascertained by performing a pyridine hemochrome assay for coproheme or an ε₅₄₈ of 16.8 mM⁻¹ cm⁻¹ in 0.1 M HCl for coproporphyrin.²²

Preparation of the Coproheme Decarboxylase (ChdC) and Coproporphyrin Ferrochelatase (CpfC) Enzymes from S. aureus.

Expression and purification of SaChdC were carried out as previously reported.¹¹ The sequence encoding CpfC (also known as HemH) in S. aureus Newman was obtained from NCBI (accession number NC_009641.1 at 1924676–1925599 for hemH) and codon-optimized for heterologous expression in Escherichia coli. The cpf C-containing plasmid was transformed into Tuner (DE3) cells (Novagen). Heterologous expression was carried out in six 1 L flasks of Terrific Broth supplemented with the 50 mg/L kanamycin. Flasks were inoculated in a 1:100 ratio with a freshly saturated starter culture and grown at 37 °C until an optical density at 600 nm of 0.4–0.6 was reached. Protein expression was induced by the addition of 1 mM isopropyl β-D-1-thiogalactopyranoside (final concentration) for 16–18 h at 20 °C. Cell were harvested by centrifugation and subsequently lysed by sonication in buffer A [50 mM Tris (pH 8), 150 mM NaCl, and 5 mM imidazole]. The lysate was clarified by centrifugation at 45000g for 1 h, and the supernatants were loaded onto a nickel-nitrilotriacetic acid affinity column (Bio-Rad) equilibrated with buffer A. Protein was eluted by a 300 mL linear gradient from 0 to 100% buffer B [50 mM Tris (pH 8), 150 mM NaCl, and 500 mM imidazole] at 2 mL/min (AKTA Prime). The protein eluted around 40–60% buffer B. Pure fractions were identified via sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE), concen-trated, dialyzed into 50 mM Tris (pH 8, 150 mM NaCl), and stored at −80 °C.

Coproporphyrin Ferrochelatase (CpfC) Reactivity Assay.

A coproporphyrin–CpfC complex was preformed by adding 5 μM coproporphyrin to 20 μM CpfC in 50 mM Tris (pH 8) and 150 mM NaCl. For experiments with the CpfC–ChdC complex, 100 μM ChdC was then added and the solution allowed to incubate for 5 min. Fe(II) (20 μM), in the form of (NH₄)₂Fe(SO₄)₂·6H₂O, was used as a source of ferrous iron to initiate the ferrochelatase reaction. Spectral changes were monitored via ultraviolet–visible (UV–vis) spectroscopy (Cary60 instrument in kinetics mode at 25 °C) every 30 s for 10 min. Fe(II) stocks were generated just prior to use by weighing out the solid and introducing it into an anaerobic atmosphere (COY Chamber). The solid was resuspended in anaerobic water and diluted to a final concentration of 0.2 mM. The solution was placed in a septum-sealed vial to be used for experiments.

Coproheme Decarboxylase (ChdC) Reactivity Assay.

A coproheme–ChdC complex was preformed by adding 5 μM coproheme to 100 μM ChdC in 50 mM potassium phosphate (KP_i) buffer (pH 7.4). For experiments looking at the CpfC–ChdC complex, 20 μM CpfC was then added and the solution allowed to incubate for 5 min. These complexes were titrated with increasing concentrations (200 μM aliquots) of hydrogen peroxide (H₂O₂), and conversion of coproheme to heme was monitored by observing a Soret shift from 396 to 406 nm (Cary60 instrument, scanning mode, 25 °C). Stock solutions of H₂O₂ were prepared at a concentration of 20 mM, ascertained by titration with freshly prepared 0.02 M KMnO₄, as previously reported.¹¹

Determining a Coproporphyrin–CpfC–ChdC Complex Dissociation Constant and Protein–Porphyrin Dissociation Constants.

Values of K_D for equilibrium binding of coproporhyrin and coproheme to CpfC or ChdC were determined by fluorescence quenching (Cary). The tryptophan fluorescence was excited at 295 nm, and emission monitored at 340 nm. The percent change in fluorescence at 340 nm was plotted versus the concentration of added ligand. Plots were fitted with the Langmuir–Hill equation to determine the K_D.

$$\theta =\frac{{\left[\text{L}\right]}^n}{K_{\text{D}}+{\left[\text{L}\right]}^n}$$

where θ is the fraction of ligand binding sites occupied by the ligand, [L] is the ligand concentration, and n is the Hill coefficient describing cooperativity. K_D for equilibrium binding of ChdC to the coproporphyrin–CpfC complex was performed using UV–vis spectroscopy as described in the text. The Langmuir–Hill equation was again used to determine K_D.

Assessment of the Transfer of Coproheme from CpfC to ChdC.

Coproheme (10 μM) was bound to 20 μM CpfC and 5–200 μM ChdC subsequently added to the buffered solution [50 mM Tris (pH 8) and 150 mM NaCl]. After 30 min, the protein/porphyrin mix was loaded onto a 650 ENrich size-exclusion column (Bio-Rad) equilibrated with 50 mM Tris (pH 8) and 150 mM NaCl. The porphyrin/protein mixture was separated (0.5 mL/min). Fractions (1 mL) were collected and analyzed for coproheme and protein content by UV–vis (396 and 280 nm, respectively) and protein identity via SDS–PAGE.

Generation of S. aureus ChdC Mutants.

chdC was cloned from S. aureus Newman genomic DNA using primers chdC-F and chdC-R which include complementarity to the pOS1 P_lgt (lgt constitutive promoter) plasmid digested with XhoI and BamHI (NEB), and subsequently assembled using NEB HiFi Assembly according to the manufacturer’s recommendations. pOS1 P_lgtchdC was sequenced to confirm. Site-directed mutagenesis of the chdC gene was performed using the QuikChange XL Site-Directed Mutagenesis Kit (Agilent) and the primers specified in Table 1. Plasmids containing the confirmed mutation were transformed by heat shock or electroporation from E. coli into S. aureus cloning intermediate strain RN4220 before isolation and subsequent electroporation into the S. aureus ΔchdC strain.⁹ Plasmids containing E. coli strains were grown in LB or LB agar with 50 μg/mL carbenicillin. S. aureus strains were grown in TSB or TSB supplemented with 10 μg/mL chloramphenicol.

Table 1.

Primers Used for Site-Directed Mutagenesis of ChdC

primer	sequence (5’ to 3’)
chdc-F	caattgaggtgaacatatgctcgagatgagtcaagcagccgaaac
chdc-R	aaacactacccccttgtttggatccttaagaaatcgcaaagaattgatc
Y113F SaChdC F	catcagatttaccagctaaaaaattgctcaattcaatgactgatac
Y113F SaChdC R	gtatcagtcattgaattgagcaattttttagctggtaaatctgatg
E120L SaChdC F	atgagggttctcataaggatctaaatcagatttaccagctaaataattgctca
E120L SaChdC R	tgagcaattatttagctggtaaatctgatttagatccttatgagaaccctcat
D121V SaChdC F	atgagggttctcataaggaacttcatcagatttaccagc
D121V SaChdC R	gctggtaaatctgatgaagttccttatgagaaccctcat
K129A SaChdC F	ggtaattctgggtataatcttgctgcgatatgagggttctcataaggatc
K129A SaChdC R	gatccttatgagaaccctcatatcgcagcaagattatacccagaattacc
Y133F SaChdC F	gtggtaattctgggaataatcttgctttgatatgagggttct
Y133F SaChdC R	agaaccctcatatcaaagcaagattattcccagaattaccac

Growth Curve Phenotype Analysis of S. aureus ChdC Mutants.

The S. aureus ΔchdC strains containing the specified ChdC protein variants were streaked on a TSA+ 10 μg/mL chloramphenicol plate and allowed to grow for 18–24 h. The E120L strain and the strain containing the empty vector (i.e., no ChdC) grew as SCVs. These strains were allowed to grow for an additional 24 h (48 h total) to obtain fully formed and visible colonies before proceeding; 3 mL of RPMI (Thermo Fisher Scientific, catalog no. 11879, without glucose), 1% casamino acids, 2 mg/mL glucose, and 10 μg/mL chloramphenicol were inoculated with a single colony from each strain, separately. This was done in triplicate for each strain. Hemin (2 μM) was added to the empty vector and E120L strains to obtain densities similar to that of the wild type (WT). The cultures were allowed to grow for 16–18 h at 37 °C and 250 rpm. In a 96-well plate, 200 μL of RPMI (catalog no. 11879, without glucose), 1% casamino acids, 4 mg/mL glycerol, 10 μg/mL chloroamphenicol, and 0 or 1 μM heme were inoculated in a 1:100 ratio with overnight cultures. OD₆₀₀ was measured using a VarioSkan plate reader every 30 min for 20 h [37 °C, discontinuous shaking (5 s on, 5 s off) at 120 rpm, slow speed]. Measurements were taken with the plate’s lid on and the perimeter of the 96-well plate covered in parafilm (one layer) to prevent evaporation.

Cellular Porphyrin Quantification of S. aureus ChdC Mutants by Liquid Chromatography–Mass Spectrome-try (LC–MS).

The procedure for making growth curves was followed up until inoculation in the 96-well plate, at which point in 250 mL flasks, 50–100 mL of RPMI (Thermo Fisher Scientific, catalog no. 11879, without glucose), 1% casamino acids, 4 mg/mL glycerol, and 10 μg/mL chloroamphenicol were inoculated in a 1:100 ratio with overnight cultures and incubated for 20 h at 37250 rpm. Cells were collected by centrifugation (5000 rpm, 8 min, 4 °C) and subsequently resuspended in 1 mL of a 1 M HCl/DMSO mixture [1:1 (v/v)] and transferred to Matrix B Lysis tubes (MP Biomedicals). The porphyrin extraction procedure optimized in our lab was followed as outlined by Choby et al.⁹ To normalize porphyrin content to colony forming unit (CFU) count, CFUs were enumerated by serial dilution and plating to TSA for each sample before cells were collected and lysed.

RESULTS

Coproheme Is Transferred from CpfC to ChdC at Increasing Relative Concentrations of ChdC.

Coproheme III (hereafter termed simply coproheme) is the product of CpfC and the substrate of ChdC. To determine whether and under what conditions coproheme transfer from CpfC to ChdC can be observed, 0, 5, 20, 50, or 200 μM ChdC was added to 10 μM coproheme that had been pre-equilibrated with 20 μM CpfC (binding of coproheme to CpfC was confirmed by UV–vis). Note that CpfC is a monomer, while ChdC is a homopentamer containing five coproheme binding sites. All ChdC concentrations given here refer to the reactive monomer. The porphyrin/protein mix was allowed to incubate for 30 min and then rapidly separated in their native states via analytical size-exclusion chromatography. In each chromatogram, the chromatogram peaks for the ChdC pentamer (147.5 kDa) and CpfC monomer (37 kDa) are well-resolved, with no indication of additional peaks suggestive of a stable, higher-molecular weight complex between the two proteins (Figure 1). However, partial transfer of coproheme from CpfC to ChdC was observed when the two protein active sites were present in a 1:1 CpfC:ChdC monomer stoichiometric ratio (i.e., 20 μM CpfC and 20 μM ChdC monomer, corresponding to five CpfC enzymes per ChdC homopentamer), and full transfer of coproheme to ChdC was observed at ChdC monomer concentrations of ≥50 μM.

Figure 1.

Coproheme transfer occurs from CpfC to ChdC at increasing concentrations of ChdC. ChdC (0, 5, 20, 50, or 200 μM) was added to a pre-equilibrated mixture of 10 μM coproheme and 20 μM CpfC. The mixture was incubated for 30 min before being loaded onto an analytical size-exclusion chromatography column. SDS–PAGE was used for identification of proteins (ChdC MW = 29.5 kDa; CpfC MW = 35 kDa) eluting from the SEC column. ChdC eluted between fractions 12 and 13, while CpfC eluted in fraction 14. Representative SEC chromatograms are shown for samples incubated with 5, 20, and 200 μM ChdC. The UV–vis signal was measured at both 280 nm (indicating protein content, solid lines) and 396 nm (indicating coproheme content, dashed lines). Complete transfer of copropheme from CpfC to ChdC occurred only at ChdC concentrations of ≥50 μM.

UV–Vis Spectroscopy Indicates a Coproporphyrin–CpfC–ChdC Interaction.

Heme and its porphyrin precursors have unique UV–vis absorption spectra. Each spectrum is unique to not only the porphyin itself but also the chemical environment in which it is found. As such, UV–vis reports on changes in the chemical environment (solvent vs a particular protein) or structure of a porphyrin. The spectral features of coproporphyrin III (hereafter termed simply coproporphyrin) and coproheme were analyzed alone and in the presence of a stoichiometric excess of either CpfC or ChdC (Figure 2). As expected, these signals were each distinct from one another, reflecting the different chemical environments of the coproporphyrin/coproheme (Figure 2 and Table S1). Next we analyzed the UV–vis spectrum of coproporphyrin and coproheme in the presence of both CpfC and ChdC, using protein concentrations where coproheme transfer was observed (20 and 100 μM, respectively). In the presence of both enzymes, the spectra for both coproporphyrin and coproheme exhibited distinct changes, which were most dramatic for coproporphyrin. In the presence of both enzymes, an intense, sharp Soret peak with a unique maximum absorbance at 418 nm was observed (Figure 2b and Table S1). This peak was markedly distinct from the Soret peaks measured for coproporphyrin bound to CfpC or ChdC alone, strongly suggesting that both CpfC and ChdC contribute to the coproporphyrin chemical environment. Additionally, the pattern of coproporphyrin-associated visible bands is distinct in the presence of both enzymes. Formation of an apparent complex is specific to the CpfC–ChdC pair, as substituting ChdC for any other enzyme in the biosynthetic pathway (e.g., CgoX) did not yield distinct spectral features (Figure S1).

Figure 2.

Coproporphyrin and coproheme UV–vis spectra suggest an interaction between CpfC and ChdC. UV–vis spectra were measured for (a) 5 μM coproporphyrin or (b) 5 μM coproheme in 50 mM Tris (pH 8) and 150 mM NaCl (green) and in the presence of 20 μM CpfC (red), 20 μM ChdC (blue), or 20 μM CpfC and 100 μM ChdC (black). The inset shows Q-bands of the porphyrin/heme in each of the different environments on an amplified scale. The strong distinctions between the coproporphyrin spectra suggested that UV–vis spectroscopy could be used to report on its chemical environment in a straightforward fashion.

The CpfC and ChdC Proteins Form a Ternary Complex with Coproporphyrin Exhibiting an Affinity Constant That Is Consistent with a Transient Protein–Protein Interaction.

The affinities of CpfC and/or ChdC for coproporphyrin and/or coproheme were measured titrimetrically using tryptophan fluorescence quenching (Figure S2). Fitting the data to Langmuir isotherms (see Materials and Methods) yielded the dissociation constants reported in Table S1 [50 mM Tris (pH 8) and 150 mM NaCl at 25 °C]. The inclusion of cooperativity via the Langmuir–Hill equation (n is the Hill constant) permitted better fits to the data measured for pentameric ChdC, with an n value of 1.4 (ChdC–coproporphyrin) or 1.5 (ChdC–coproheme). A value for n that is higher than unity suggests a moderate degree of cooperativity in porphyrin binding to the homopentameric ChdC, in which each monomer contains one porphyrin binding site. Moreover, the affinity of coproheme for ChdC is ~3-fold higher than for CpfC. The passage of coproheme from CpfC, where it is a reaction product, to ChdC, where it is a substrate, would therefore be expected to occur down a modest thermodynamic gradient if the proteins are not permitted to interact; however, the kinetic barriers to coproheme passage under biological conditions are not clear. Interestingly, ChdC binds coproporphyrin with an affinity that resembles that for coproheme. This similarity is observed despite the iron–histidine interaction that is missing when coproporphyrin binds ChdC, suggesting that multiple protein–porphyrin interactions make the most substantial contributions to the enzyme–substrate affinity. Formation of a CpfC–ChdC complex for specific passage of coproheme would offset the possibility of free coproporphyrin ending up unproductively bound to ChdC.

We next sought to determine the affinity of CpfC for ChdC by first measuring the affinity of ChdC for a preformed CpfC–coproporphyrin complex via UV–vis titration. (Note, the UV–vis spectra are best resolved for the protein–coproporphyrin rather than the coproheme complexes, as shown in Figure 2. The coproporphryin complexes were therefore monitored.) Coproporphyrin (5 μM) was incubated with excess CpfC (20 μM) to ensure complete binding; the amount of free coproporphyrin under these conditions, based on the K_D measured above, is 50 nM. The CpfC–coproporphryin complex was then titrated with increasing concentrations of ChdC, and the UV–vis absorbance was monitored as it gradually changed from the characteristic spectrum of the CpfC–coproporphyrin complex to the spectrum of the ChdC–CpfC–coproporphyrin ternary complex (Figure 3). The change in absorbance at 418 nm was plotted versus the concentration of ChdC. The data were fit to the Langmuir isotherm to obtain a K_D of 80 ± 1.5 μM. The affinity of a preformed ChdC–coproporphyrin complex (5 μM coproporphyrin and 20 μM ChdC; free coproporphryin concentration estimated to be 30 nM) for CpfC was then measured using the same method (shown in the blue inset), yielding a K_D of 40 ± 1.1 μm. Finally, a dissociation constant for the CpfC–ChdC complex without the porphyrin present (a species with no characteristic UV–vis spectrum) was computed using pairs of measured K_D values and the following relationships

$$K_{\text{D}}{\left(\text{ChdC}-\text{CpfC}\right)}_1=K_{\text{D}}\left(\text{CpfC}-\text{coproporphyrin}\right)\times K_{\text{D}}\left(\text{CpfC}-\text{ChdC}-\text{coproporphyrin}\right)=12\mu \text{M}$$

and

$$K_{\text{D}}{\left(\text{ChdC}-\text{CpfC}\right)}_2=K_{\text{D}}\left(\text{ChdC}-\text{coproporphyrin}\right)\times K_{\text{D}}\left(\text{CpfC}-\text{ChdC}-\text{coproporphyrin}\right)=22\mu \text{M}$$

A dissociation constant for the CpfC–ChdC complex was then estimated as the average of the two measured K_D values (17 ± 7.1 μM). A K_D in the micromolar range is considered relatively weak and consistent with a transient protein–protein interaction.⁷ The protein concentrations required for the CpfC–ChdC and/or CpfC–ChdC–porphyrin complexes to be observed stably under in vitro conditions are high yet relevant to the amounts required for transient docking in vivo, where macromolecules occupy ~40% of the cellular interior and macromolecular crowding promotes high local protein concentrations and enhances transient protein associations.^7,8 Moreover, the relative cellular concentration ratio of CpfC monomer to ChdC pentamer has been reported to be ~1:2–5.^9,10

Figure 3.

Affinity of the CpfC–ChdC complex that is consistent with a transient protein–protein interaction. A preformed coproporphyrin–CpfC complex was generated by equilibrating 5 μM coproporphyrin with 20 μM CpfC (red). ChdC was added titrimetrically (gray spectra, representing additions leading to final pre-equilibrated concentrations of 40, 80, 100, 200 and μM ChdC) until the spectral changes ceased and a final spectrum was obtained (black). The top inset shows the change in absorbance at 418 nm as a function of the titrated concentration of ChdC (red). This was fit to the Langmuir–Hill equation to determine a K_D for the coproporphyrin–CpfC complex with ChdC. In blue, similar data and a curve fit showing CpfC complexation with a preformed coproporphyrin–ChdC complex are shown (bottom inset, blue). These data were used to compute a dissociation constant for the two proteins, ChdC and CpfC.

Formation of a CpfC–ChdC Complex Does Not Enhance the Activity of Either Enzyme.

CpfC catalyzes the insertion of ferrous iron into coproporphyrin to yield coproheme (Scheme 1). The progress of the reaction was monitored by measuring the decrease in the Soret absorbance of the CpfC–coproporphyrin complex (405 nm) over time following addition of Fe(II) as the band shifts to 396 nm (Figure 4a). The reaction was complete within 8 min. To test whether complexation between CpfC and ChdC enhances the intrinsic reactivity of CpfC, 5 μM coproporphyrin bound to 20 μM CpfC was complexed with 100 μM ChdC (418 nm absorbance peak). Fe(II) (10 μM) was then added to start the reaction (Figure 4b). No spectral changes were observed. These results indicate that formation of the CpfC–ChdC–coproporphyrin complex does not enhance the intrinsic reactivity of CpfC and moreover inhibits it, suggesting that the coproporphyrin in this complex is not in a reactive conformation.

Figure 4.

CpfC–ChdC complex that does not enhance, but slightly impairs, the intrinsic reactivity of CpfC with coproporphyrin. (a) The CpfC–coproporphyrin complex (red line) was formed by incubating 5 μM coproporphyrin with 20 μM CpfC [50 mM Tris (pH 8) and 150 mM NaCl at 20 °C]. Fe(II) (10 μM) was added to initiate the metalation reaction, and spectra were measured every 30 s for 10 min (black). In the absence of ChdC, the reaction was complete by 8 min, yielding the characteristic spectrum of the ChdC–coproheme complex (purple, Table S1). (b) The CpfC–ChdC–coproporphyrin complex (black line) was formed by incubating 5 μM coproporphyrin, 20 μM CpfC, and 100 μM ChdC under the same conditions as in panel a. Fe(II) (10 μM) was added to initiate the metalation reaction, and spectra were measured over time. The characteristic spectrum of the coproporphyrin–CpfC–ChdC complex underwent little to no change following the addition of Fe(II) for ≤30 min (purple).

The CpfC–ChdC Complex Moderately Truncates the Intrinsic Reactivity of ChdC.

ChdC catalyzes the oxidative decarboxylation of coproheme to heme using H₂O₂ as the cosubstrate (Scheme 1).^11,12 To determine whether the CpfC–ChdC complex had an effect on the reactivity of ChdC, 5 μM coproheme was first mixed with excess (100 μM) ChdC to form a ChdC–coproheme complex. H₂O₂ was then titrimetrically added in increments of 100 μM (1 equiv relative to ChdC monomer), and the reaction progress monitored by UV–vis spectroscopy. In the absence of CpfC, the shift from a ChdC–coproheme complex (394 nm Soret peak) to a ChdC–heme complex (406 nm) was complete following addition of 10 equiv of H₂O₂ (Figure 5a). Next, the ChdC–coproheme complex was equilibrated with 20 μM CpfC to form a CpfC–ChdC–coproheme complex, and the experiment was repeated (Figure 5b). The presence of CpfC led to incomplete conversion of coproheme to heme upon addition of 10 equiv of H₂O₂ and did not yield the final product until 16 equiv of H₂O₂ was added. Finally, the rates of conversion of ChdC–coproheme and CpfC–ChdC–coproheme complexes to their respective heme complexes were monitored by time-resolved UV–vis, following addition of H₂O₂ in the amount needed to yield full conversion of the substrate to product. While the ChdC–coproheme complex was fully converted in 2.5 ± 0.5 min, the CpfC–ChdC–coproheme complex required 6.0 ± 0.5 min. These results indicated that the efficiency of the ChdC reaction with coproheme is somewhat inhibited by the presence of CpfC. It is possible that formation of a complex may hinder binding of substrates and/or release of products or catalytically important motions of the enzymes. From a biochemical standpoint, it is not clear why slowing the flux of substrates through the biosynthetic pathway would be useful or whether a slower ChdC reaction is merely a trade-off in return for more control over the biosynthetic pathway.

Figure 5.

CpfC–ChdC complex that moderately truncates the intrinsic reactivity of ChdC. (a) The ChdC–coproheme complex (forest green) was formed by incubating 5 μM coproheme with 100 μM ChdC (50 mM KP_i, pH 7.4, 20 °C). Ten equivalents of H₂O₂ relative to the ChdC monomer, previously shown to be sufficient for full conversion of the starting complex to the ChdC–heme complex (lime green), was titrimetrically added to observe the double-decarboxylation reaction (gray spectra, one spectrum per 2 equiv of H₂O₂). (b) The CpfC–ChdC–coproheme complex (purple) was formed by incubating 5 μM coproheme with 20 μM CpfC and 100 μM ChdC (50 mM KP_i, pH 7.4, 20 °C). H₂O₂ was titimetrically added as described above, monitoring spectral changes after each addition (gray spectra, one spectrum per 2 equiv of H₂O₂). The reaction went to completion after the addition of 16 equiv of H₂O₂.

The ChdC E120L, Y133F, and K129A Point Mutations Lead to Altered Levels of Heme Precursors.

Structural studies of ChdC in the presence and absence of bound substrate revealed a labile loop adjacent to, but outside of, the immediate binding site of the coproheme.^12,13 This structure, composed of residues ~110–135, is found in an open conformation in the absence of coproheme and hugging the active site in its presence (Figure 6). It has been hypothesized that the loop acts as an active site “gate”, serving as a regulator of substrate entry and product egress.^12–14 In addition, mobile elements like this “gate” have been implicated in facilitating transient protein–protein interactions.⁸ To determine whether the loop could form part of a substrate-receiving interface with CpfC, a series of point mutations to convert polar residues to nonpolar residues spanning the loop (N112L, Y113F, E120L, D121V, K129A, R131A, and Y133F) were generated (Figure 6). A chdC-deficient strain of S. aureus Newman (ΔchdC) was then transformed with a plasmid encoding each of these variants, and these strains were phenotypically characterized.

Figure 6.

ChdC active site gate that may play a role in a CpfC–ChdC protein–protein interaction. Juxtaposed structures of (a) apo-ChdC (tan, Protein Data Bank entry 1T0T) and (b) coproheme-bound ChdC (green, Protein Data Bank entry 5T2K, both structures from Geobacillus thermophilus) reveal a mobile active site loop (amino acids ~110–135) that closes in toward the active site to sequester coproheme (arrows in panel a indicate the direction of the change in the residue position upon coproheme binding). Solvent-exposed areas of the active site are colored gray. Point mutations were generated at the positions shown. The mutant proteins and the phenotype of a complemented ΔchdC strain were characterized. Note that the residue at position 129 is a lysine in the S. aureus homologue. The locations of residues 114–119 could not be crystallographically mapped and are missing in the 5T2K structure (b).¹²

It is well established that S. aureus heme auxotrophs, including a ΔchdC strain, exhibit growth that is significantly slower than that of WT, leading to so-called “small colony variants” on solid media and temporally delayed growth curves in liquid cultures.¹⁵ The ΔchdC growth defect can be reversed if the media are supplemented with exogenous heme, which can be internalized by the cell through its iron surface determinant (Isd) system or through passive diffusion (the latter not yet experimentally shown).^16,17 Of all of the mutants analyzed, only the N112L and E120L variants exhibited a growth defect comparable to that of the ΔchdC strain complemented with the empty vector (Figure 7a). The growth defects for both the E120L variant and the ΔchdC control were equivalently restored in the presence of 1 μM exogenous heme; interestingly, the growth defect of N112L was exacerbated under these conditions (Figure 7b). Notably, E120 and N112, like all of the residues comprising the proposed active site gate, are distant from residues already known to play key roles in catalysis and have no direct hydrogen bonding interaction connecting it either to these or to the coproheme itself (Figure 6).

Figure 7.

E120L and N112L chdC variants in S. aureus Newman exhibit growth defects. Representative growth curves for the ΔchdC strain complemented with an empty vector, WT chdC or chdC variants, were measured in (a) unsupplemented RPMI containing 1% casamino acids and 4 mg/mL glycerol or (b) the same medium supplemented with 1 μM heme. In the absence of added heme, the Y113F, D121V, K129A, R131A, and Y133F variants (gray lines) all grew like WT. The N112L, E120L, and ΔchdC strains (green, pink, and red, respectively) each exhibited growth delays and reduced growth yields. The defects in both E120L and ΔchdC could be reversed by the addition of exogenous heme to the growth medium. The N112L strain exhibited had an even longer lag phase in the presence of added heme.

To further examine the variant strains for phenotypic changes that would not produce a gross defect in growth, the concentrations of cellular porphyrins, including heme and its upstream porphyrin precursors uroporphyrin, coproporphyrin, and coproheme, were quantitatively profiled via LC–MS. Note that the reported numbers are averages of three biological replicates made on cell lysate solutions with concentrations well within the linear response range of the instrument. Hence, we expect that the large errors reflect both biological batch-to-batch variability and the minute quantities in which these analytes are present under biological conditions rather than instrument limitations. Four general phenotypes were observed. These included a WT-like pattern exhibiting small amounts of uroporphyrin, coproporphyrin, and coproheme and ~2–6 pmol of heme cfu⁻¹ (Table 2). The chdC-Y113F, -D121V, and -R131A variants fit within this category, having porphyrin quantities within ~2–3-fold of the values measured for WT. A second group included ΔchdC harboring an empty vector and the chdC-E120L variant, both of which exhibit moderately elevated uroporphyrin, coproporphyrin, and coproheme and substantially diminished (~0.04–0.05 pmol cfu⁻¹) cellular heme concentrations. Both of these strains had impaired growth that could be reversed by heme supplementation (Figure 7). Hence, these bear the expected characteristics of heme auxotrophs. A third phenotype, exhibited solely by chdC-N112L, had WT levels of heme precursors but a 7-fold increase in heme concentration. The reasons for the increased heme concentration are unclear but suggest the mutation leads to either faulty regulation of the pathway or impaired delivery of heme to a recipient chaperone protein. This strain’s impaired growth (Figure 7) may moreover be related to the impaired delivery of heme. Finally, a fourth group (chdC-Y133F and -K129A) exhibited elevations in uroporphyrin and especially coproporphyrin and coproheme concentrations that were larger than those in the WT group while still maintaining heme concentrations within the range of 2–6 pmol cfu⁻¹. Note that chdC-D121V is a borderline member of this group, but because its phenotype was less pronounced, it was included with the WT-like group. The fourth group is of greatest interest because it exhibits a phenotype consistent with expectations for protein–protein interaction mutants, namely, elevation of the substrates for both CpfC and ChdC (coproporphyrin and coproheme, respectively). Free diffusion of coproheme to ChdC might occur even if an interaction of this type were impaired, leading to the production of normal cellular heme levels. The reasons for elevation in the upstream intermediate uroporphyrin—in chdC-E120L, -D121V, -K129A, -R131A, and -Y133F variants, as well as in the E120L and ΔchdC strains—are unclear but suggestive of feedback between earlier and later intermediates along the biosynthetic pathway. This is consistent with the increased abundance of GtrR and increased terminal precursor abundance in the ΔchdC strain, as previously observed.⁹

Table 2.

Porphyrin Profiles for ΔchdC Variant Strains of S. aureus Demonstrating the Effects of Changes in the Active Site Loop^a

S. aureus genotype	uroporphyrin (×10−15 mol cfu−1)	coproporphyrin (×10−15 mol cfu−1)	coproheme (×10−15 mol cfu−1)	heme (×10−15 mol cfu−1)	phenotype description
chdC-WT	26.1 ± 24	47.3 ± 64	13.3 ± 16	3780 ± 970	WT-likeb
chdC-N112L	29.8 ± 6.5	71.1 ± 18	1.40 ± 2.5	27200 ± 3500	heme accumulator
chdC-Y113F	24.5 ± 17	138 ± 150	7.50 ± 0.20	5500 ± 1100	WT-like
chdC-E120L	156 ± 50	75.5 ± 3.4	306 ± 18	53.4 ± 39	heme auxotroph-like
chdC-D121V	68.3 ± 62	181 ± 240	12.9 ± 1	6100 ± 4300	WT-like
chdC—K129A	88.2 ± 39	516 ± 370	32.6 ± 32	4100 ± 5200	potential interaction mutant
chdC-R131A	47.9 ± 51	109 ± 3.3	34.4 ± 4.7	2200 ± 1200	WT-like
chdC-Y133F	103 ± 82	385 ± 540	28.6 ± 38	2400 ± 3090	potential interaction mutant
chdC-empty	1940 ± 1100	264 ± 40	431 ± 340	40.0 ± 11	heme auxotroph-like

^aAll experiments were carried out with ΔchdC knockout strains of S. aureus Newman that were complemented as described in the table.

^bSee ref 9 for representative porphyrin profiles for WT S. aureus Newman.

Recombinant Protein Assays Confirm That the K129A and Y133F Mutations Impair the ChdC–CpfC Interaction.

The K129A, Y133F, and E120L ChdC protein variants were overexpressed in E. coli and purified via procedures outlined for the WT enzyme.¹¹ The enzymes were then tested for alterations in their catalytic activity, affinity for coproheme, and ability to interact with CpfC relative to WT ChdC. WT ChdC requires 10 equiv of H₂O₂ to convert one molecule of coproheme to heme,¹¹ where the reaction can be monitored by UV–vis spectroscopy (Figure 5) and/or high-performance liquid chromatography (HPLC). Like the WT enzyme, each of the protein variants examined here was able to convert coproheme completely to heme with 10 equiv of added H₂O₂ (data for E120L shown in Figure 8). HPLC profiles showing the conversion of coproheme to the stable 3-propionate intermediate (harderoheme) and then to product (heme) were identical for WT and each ChdC protein variant. This is in contrast to previously studied ChdC variants having mutations surrounding the reactive propionate of the bound coproheme, where a large excess of H₂O₂ was required to achieve at best partial conversion of the substrate; at the extreme, substituting the catalytically active Y145 with serine led to no observable substrate conversion.¹² Hence, we concluded that, as expected, altering residues on the active site gate had no notable effects on ChdC’s ability to convert the substrate to the product. To determine how these mutations affected ChdC/coproheme affinity, equilibrium dissociation constants (K_D) were measured for each protein variant using tryptophan fluorescence quenching. A K_D value equivalent to that of WT (0.5 ± 0.03 μM) was obtained for K129A (0.5 ± 0.05 μM), while Y133F and E120L had K_D values close to 3-fold greater than that of WT (1.7 ± 0.3 and 1.3 ± 0.07 μM, respectively), suggesting relatively modest reductions in their affinity for the substrate. Lastly, the ChdC variants were tested for their ability to interact with CpfC by assessing 418 nm peak formation upon the addition of each variant to the CpfC–coproporphyrin complex, as was previously done with WT ChdC (Figure 3). Similar to WT, ChdC-E120L formed a sharp peak at 418 nm upon titration with CpfC, though with a slightly lower absorbance (Figure 9). In contrast, K129A and Y133F did not display any significant changes at 418 nm up to 300 μM protein added [50 mM Tris (pH 8) and 150 mM NaCl]. This indicated that these two mutations appear to have largely abrogated the ChdC–CpfC interaction.

Figure 8.

E120L, Y133F, and K129A ChdC protein variants exhibit substrate turnover characteristics equivalent to those of WT. (a) Coproheme-bound E120L, Y133F, or K129A ChdC protein (5 μM) was each reacted with 10 equiv of H₂O₂ (50 mM KP_i, pH 7.4). Subsequent conversion of the ChdC–coproheme complex (dark green) to the ChdC–heme complex (light green) was monitored over time via UV–vis spectroscopy (Soret shift from 393 to 406 nm; spectra measured every 30 s after mixing are colored charcoal). Similar data were measured for each protein variant. Representative data measured for E120L are shown. (b) The substrate (coproheme), product (heme), and 3-propionate-containing stable intermediate (harderoheme) were quantified via HPLC following addition of increasing numbers of equivalents of H₂O₂ under conditions identical to those used in panel a. All mutations led to no change in the number of equivalents required to effect full turnover. Representative data for E120L are again shown.

Figure 9.

Y133F and K129A ChdC variants are largely unable to interact with CpfC. Coproporphyrin (5 μM) bound to 20 μM HemH (red spectrum, 405 nm Soret peak) was titrated with increasing equivalents of WT, E120L, Y133F, or K129A ChdC as shown [≤200 or 300 μM, in 50 mM Tris-HCl (pH 8) and 150 mM NaCl]. Changes at 418 nm were monitored via UV–vis following additions in increments of 20 μM ChdC (charcoal spectra). E120L ChdC formed a sharp and distinct 418 nm peak, just like WT, showing its ability to interact with ChdC. Y133F and K129A ChdC did not form a 418 nm peak, even upon addition of 300 μM ChdC.

Taken together, these results suggest that, as expected, alterations to the active site loop in ChdC did not affect catalytic activity. However, prior to binding, coproheme is delivered to ChdC directly by CpfC, Y133, and K129 on ChdC playing a major role in allowing this interaction to happen. Upon coproheme binding, this loop closes in toward the substrate, as was shown with the coproheme-bound crystal structure. E120 on this loop plays an important role in keeping the coproheme substrate in position for reactivity and upon heme formation N112 directs product delivery.

DISCUSSION AND CONCLUSIONS

The assembly of reactive cofactors requires a high level of coordination among groups of enzymes. Heme biosynthesis in eukaryotes makes use of the compartmentalization afforded by the mitochondrion and its intermembrane space to organize the enzymes and metabolites involved in the final three steps of the pathway. Such organization prevents the escape of toxic or reactive intermediates as well as the spontaneous formation of dead-end products. While Gram-negative bacteria use the outer and inner membranes flanking the periplasmic space in a similar fashion, it is unclear how Gram-positive bacteria achieve analogous control in the absence of membrane-bound compartments.

We examined the final two steps of heme biosynthesis in S. aureus, the metalation of coproporphyrin III (catalyzed by CpfC) and the double decarboxylation of the metalated coproheme product (catalyzed by ChdC), in order to determine whether these might be influenced by a protein–protein interaction. UV–vis spectroscopy demonstrated an interaction between the two proteins that appears to be stable on the UV time scale in the presence of coproporphyrin or coproheme. The speed of UV–vis reactivity measurements (10¹⁵ electronic absorbance events per second) and the micromolar affinity of the ChdC–CpfC complex suggest the proteins remain only transiently docked. This conclusion is in keeping with analytical SEC results, which showed no evidence of a stable complex, and the observation that CpfC’s coproporphyrin metalation activity is inhibited in the presence of ChdC.

On the basis of prior reports in the literature and our calculated K_D value of 17 μM, we have categorized the CpfC–ChdC interaction as a weak/transient protein–protein interaction.^7,8 These interactions are difficult to observe at the low enzyme concentrations commonly used to conduct biochemical kinetic experiments. However, transient protein–protein interactions are very relevant to the cellular environment where macromolecules occupy ~40% of the interior, which yields macromolecular crowding.^7,8 Macromolecular crowding increases the local concentration of enzymes and enhances protein association. For example, effective concentrations ranging from 32 μM to 1.4 mM have been reported for enzymes in the glycolytic pathway in E. coli.¹⁸ The high protein concentrations required to observe the CpfC–ChdC interaction described here are in accordance with the concentrations at which transient protein–protein interactions might occur in the cell. A transient protein–protein interaction is likely to be sufficient for the transfer of the product of CpfC to ChdC. In fact, we speculate that a stronger complex might even be detrimental for the CpfC–ChdC complex if the ChdC protein serves a secondary role as a trafficking/delivery agent for the heme b product.

An active site gate on ChdC offered an obvious potential site for interaction with CpfC during the hand-off of coproheme between the two enzymes. Several mutations along this gate had no impact on the in vitro interaction between CpfC and ChdC in the presence of coproporphyrin. The same mutations likewise either had little observable effect on a chdC strain complemented with chdC point mutants (Y113F, D121V, and R131A) or led to growth and molecular phenotypes consistent with either heme auxotrophy (E120L) or disrupted heme homeostasis (N112L). Analysis of the ChdC active site solvent accessibility in its apo versus holo form (Figure 6) shows that the position of N112 plays a major role in the “open” versus “closed” conformation of the ChdC active site. The N112L phenotype suggested that, after heme formation, this portion of the active site loop could mediate and direct product delivery, perhaps via interactions between ChdC and protein regulators of the heme biosynthetic pathway, such as HemX, or a heme-delivering chaperone.^9,17 The E120L phenotype is especially intriguing. In the presence of H₂O₂, free coproheme does not undergo decarboxylation to form heme; rather, it degrades. If the active site gate in the E120L mutant fails to remain closed, the enzyme’s ability to retain coproheme in the active site might be impaired. If this were true, this mutation could lead to heme auxotrophy. By contrast, the ChdC-K129A and -Y133F mutations both led to accumulation of coproporphyrin (the substrate of CpfC) and coproheme (the substrate of ChdC), suggesting that they might interfere with a CpfC–ChdC interaction, which UV–vis measurements supported.

While the active site of ChdC has a gating loop, the CpfC from the Gram-positive bacterium Bacillus subtilis has a solvent-exposed porphyrin binding site. This is in contrast to the human ferrochelatase (Homo sapiens PpfC), whose binding site is shielded by an “active site lip”²¹ (Figure 10). The open binding site of CpfC is flanked by a series of outstretched amino acid side chains that include R31, R33, E36, E38, E187, K188, N225, and D228. Any of these might help mediate the docking interaction with ChdC. Prior work demonstrated that the S54A mutant of this gene reduced the growth rate of B. subtilis and resulted in the accumulation of coproporphyrin III in the growth medium.¹⁹ It was suggested that S54 is involved in substrate reception or delivery of the enzymatic product.

Figure 10.

CpfC has a solvent-exposed porphyrin binding site with flanking polar residues that may interact with ChdC. Juxtaposed structures of (a) B. subtilis CpfC bound to N-methylmesoporphyin (purple, Protein Data Bank entry 1C1H) and (b) H. sapien PfpC bound to protoporphyrin IX (only one monomeric unit shown, pink, Protein Data Bank entry 2QD1). CpfC has a largely solvent-exposed active site that may allow for porphyrin transfer with ease upon interaction with ChdC. This is unlike the PPD branch ferrochelatase enzyme, PpfC, which is a membrane-bound dimer containing a characteristic “active site lip” (light pink) that moves toward and completely occludes the active site in the presence of porphyrin.

Protein–protein interactions are important for facilitating the stepwise assembly of many natural products. Until now, an interaction between the terminal enzymes of the CPD heme biosynthesis pathway had not been found. The interaction surface offers a locus for the development of allosteric inhibitors that could inhibit the production of essential heme and lead to the accumulation of toxic heme precursors in the CPD pathway of pathogens such as S. aureus.²⁰

ABBREVIATIONS

PPD

porphyrin-dependent

CPD

coproporphyrin-dependent

ChdC

coproheme decarboxylase

CpfC

coproporphyrin ferrochelatase

PgoX

protoporphyrinogen oxidase

PpfC

protoporphyrin ferrochelatase