The PrpF protein of Shewanella oneidensis MR-1 catalyzes the isomerization of 2-methyl-cis-aconitate during the catabolism of propionate via the AcnD-dependent 2-methylcitric acid cycle.

The 2-methylcitric acid cycle (2-MCC) is a common route of propionate catabolism in microorganisms. In Salmonella enterica, the prpBCDE operon encodes most of the 2-MCC enzymes. In other organisms, e.g., Shewanella oneidensis MR-1, two genes, acnD and prpF replace prpD, which encodes 2-methylcitrate dehydratase. We showed that together, S. oneidensis AcnD and PrpF (SoAcnD, SoPrpF) compensated for the absence of PrpD in a S. enterica prpD strain. We also showed that SoAcnD had 2-methylcitrate dehydratase activity and that PrpF has aconitate isomerase activity. Here we report in vitro evidence that the product of the SoAcnD reaction is an isomer of 2-methyl-cis-aconitate (2-MCA], the product of the SePrpD reaction. We show that the SoPrpF protein isomerizes the product of the AcnD reaction into the PrpD product (2-MCA], a known substrate of the housekeeping aconitase (AcnB]. Given that SoPrpF is an isomerase, that SoAcnD is a dehydratase, and the results from in vivo and in vitro experiments reported here, it is likely that 4-methylaconitate is the product of the AcnD enzyme. Results from in vivo studies using a S. enterica prpD strain show that SoPrpF variants with substitutions of residues K73 or C107 failed to support growth with propionate as the sole source of carbon and energy. High-resolution (1.22 Å) three-dimensional crystal structures of PrpFK73E in complex with trans-aconitate or malonate provide insights into the mechanism of catalysis of the wild-type protein.

Introduction

The 2-methylcitric acid cycle (2-MCC) (Fig 1) is widely distributed route of propionate catabolism in microorganisms. Originally identified in the fungus Candida lipolytica (Yarrowia lipolytica) [1], the 2-MCC has been characterized in Gamma-proteobacteria (e.g., Salmonella enterica, Escherichia coli) [2-4], actinobacteria (e.g. Mycobacterium tuberculosis, Mycobacterium smegmatis, Corynebacterium glutamicum) [5-9], and Beta-proteobacteria (e.g. Ralstonia eutropha, Burkholderia sacchari) [10, 11]. Most of the enzymes that comprise the 2-MCC are encoded as an operon [12]. In S. enterica (and many other bacteria), the operon consists of four genes, in the order prpBCDE. prpB encodes 2-methylisocitrate lyase (EC 4.2.1.99) [2, 13, 14]; prpC encodes the 2-methylcitrate synthase (EC 2.3.3.5) (2); prpD, encodes the 2-methylcitrate dehydratase (EC 4.2.1.79) [15]; and prpE encodes the propionyl-CoA synthetase (EC 6.2.1.17) [16, 17].

Fig 1

The 2-methylcitric acid cycle.

In this metabolic pathway propionate is oxidized to pyruvate with succinate as byproduct. Inset box–difference in operon structure of Salmonella enterica and Escherichia coli (Inset box A) compared to Shewanella oneidensis and Vibrio cholera (Inset box B). AckA–acetate kinase; Pta–phosphotransacetylase; AcnA/B–aconitase A, aconitase B.

There is one notable variation in bacterial prp operons. That is, some prp operons lack the 2-methylcitrate dehydratase prpD, but rather contain two other genes, acnD and prpF [15]. AcnD has aconitase-like activity [18], whilst PrpF has aconitate isomerase activity [19] (Fig 1). In S. oneidensis the sequence of prp genes is prpR prpB prpC acnD prpF compared to the S. enterica sequence prpR prpB prpC prpD prpE. In S. oneidensis, the prpE gene encoding propionyl-CoA synthetase is >1.5 Mbp away from the prp operon. AcnD and PrpF activities are necessary and sufficient to compensate for the lack of PrpD during growth with propionate of a prpD strain of S. enterica [18].

Structural and biochemical analyses of Shewanella oneidensis PrpF (hereafter SoPrpF), revealed a role of SoPrpF in the isomerization of trans-aconitate to cis-aconitate [19], leading us to propose that the role of SoPrpF in the 2-MCC was to change the stereochemistry of the S. oneidensis AcnD (hereafter SoAcnD) reaction product. That is, SoPrpF was proposed to have 2-methylaconitate isomerase activity [19].

Herein we specifically address the hypothesis that the S. oneidensis PrpF (SoPrpF) protein isomerizes the product of the SoAcnD dehydratase (putatively 4-methyl-cis-aconitate (4-MCA) to 2-methyl-cis-aconitate (2-MCA), the substrate of aconitase A, B. To facilitate this work, we used a S. enterica ΔprpD carrying the wild-type allele of the S. oneidensis acnD gene. Into the S. enterica ΔprpD / pacnD+ (S. oneidensis) strain we introduced a second plasmid encoding wild-type or variants of S. oneidensis PrpF (SoPrpF). In this heterologous system, PrpF functionality was assessed in vivo under conditions that demanded propionate utilization as the sole source of carbon and energy, or a combination of propionate and succinate, which would allow us to assess the effect of the accumulation of 4-MCA on growth in the presence of functional or dysfunctional PrpF proteins.

Materials and methods

Chemicals and bacteria culture media

All chemicals were purchased from Sigma Chemical Co. unless otherwise stated. 2-Methylcitrate was purchased from CDN Isotopes (Pointe-Claire, Canada). Authentic 2-methyl-cis-aconitate was custom synthesized by AsisChem (Cambridge, MA). The experiments reported here were performed in well-characterized S. enterica prp mutant strains. A previous report from our laboratory showed that this approach allowed us to assess S. oneidensis AcnD and PrpF functions in vivo [18]. E. coli strains used to overproduce recombinant proteins were grown in lysogeny broth (LB) [20, 21]. No-carbon essential (NCE) medium [22, 23] was used as minimal medium, and was supplemented with MgSO4 (1 mM) and methionine (0.5 mM). When added to rich medium, antibiotic concentrations were: ampicillin (100 μg/ml), kanamycin (50 μg/ml), and chloramphenicol (25 μg/ml). BactoTM Agar (Difco, 1.5% w/v) was used as medium solidifying agent.

For cloning purposes, restriction endonuclease NdeI was purchased from Fermentas (Glen Burnie, MD), and BamHI was purchased from Promega (Madison, WI). Cloning was performed in E. coli strain DH5α/F’ (New England Biolabs). Plasmids were introduced into S. enterica strains by electroporation. Cultures were grown in LB medium to an optical density (650 nm) ~0.6–0.8, 1.0 ml of culture was centrifuged at 18,000 x g using a Microfuge 18 Centrifuge (Beckman Coulter), cells were washed three times with 1.0 ml of ice-cold sterile water, and re-suspended in 100 μl of water. Plasmids were electroporated into cells using a Bio-Rad Gene Pulser (Hercules, CA) according to the manufacturer’s recommendations. Strains and plasmids used in this study are listed in Table 1.

Table 1

Strains and plasmids used in this study.

Strain	Relevant genotype	Source
E. coli strains
BL21(λDE3)	F- ompT hsdSb (rb- mb+) dcm gal l(DE3)	New England Biolabs
DH5α/F’	F’/endA1 hsdR17 (rk- mk+) supE44 thi-1 recA1 gyrA (Nalr) relA1 Δ(lacZYA-argF)U169 deoR [f80dlacD(lacZ)M15]	New England Biolabs
XL10-Gold® Ultracompetent cells	TetR Δ(mcrA)183Δ(mcrCB-hsdSMR-mrr)173 endA1 supE44 thi-1 recA1 gyrA96 relA1 lac Hte [F' proAB lacIqZΔM15 Tn10 (TetR) Amy CamR]a	Stratagene
S. enterica strains
TR6583	metE205 ara-9	K. Sanderson via J. Roth
Derivatives of TR6583
JE8250	prpD126::cat+	This work
JE8256	JE8250 / pPRP138	This work
JE8429	JE8256 / pPRP153	This work
JE9373	JE8256 / pPRP215 prpF255	This work
JE9374	JE8256 / pPRP216 prpF256	This work
JE9592	JE8256 / pPRP218 prpF257	This work
JE9593	JE8256 / pPRP219 prpF258	This work
JE11597	JE8256 / pPRP223 prpF259	This work
Plasmids
pPRP138	S. oneidensis acnD+ in pBAD30 bla+	[18]
pPRP153	S. oneidensis prpF+ in pBAD18Kan kan+	[18]
pPRP195	S. oneidensis acnD+ cloned into pET-15b bla+	This work
pPRP196	S. oneidensis prpF+ cloned into pTEV4 bla+	[19]
pPRP205	S. oneidensis acnD+ cloned into pTEV5	This work
pPRP215	S. oneidensis prpF+(prpF255 C107A) in pBAD18Kan kan+	This work
pPRP216	S. oneidensis prpF+(prpF256; encodes PrpFC107S) cloned into pBAD18-Kan kan+	This work
pPRP217	S. oneidensis prpF+(prpF255; encodes PrpFC107A) cloned into pTEV4 bla+	This work
pPRP218	S. oneidensis prpF+(prpF257; encodes PrpFK73A) cloned into pBAD18-Kan kan+	This work
pPRP219	S. oneidensis prpF+(prpF258; encodes PrpFK73E) cloned into pBAD18-Kan kan+	This work
pPRP220	S. oneidensis prpF+(prpF257; encodes PrpFK73A) cloned into pTEV4 bla+	This work
pPRP221	S. oneidensis prpF+(prpF258; encodes PrpFK73E) cloned into pTEV4 bla+	This work
pPRP223	S. oneidensis prpF+(prpF259; encodes PrpFK73M) cloned into pBAD18-Kan kan+	This work
pET-15b	Cloning vector pBR origin of replication bla+; fuses a thrombin-cleavable H6 tag to the N-terminus of the protein of interest	Novagen
pTEV4	Cloning vector; F1 origin of replication bla+; fuses a TEV-cleavable H6 tag to the N-terminus of the protein of interest	[24]
pTEV5	Cloning vector; F1 origin of replication bla+; fuses a TEV-cleavable H6 tag to the N-terminus of the protein of interest	[24]
pBAD30	Cloning vector; pACYC184 origin of replication, bla+; expression of the gene of interest under the control of the arabinose-inducible ParaBAD promoter	[46]
pBAD18-Kan	Cloning vector; pBR origin of replication, kan+; expression of the gene of interest under the control of the arabinose-inducible ParaBAD promoter	[46]

Site-directed mutagenesis

All site-directed mutations were introduced into targeted genes using the QuikChange® II XL Site-Directed Mutagenesis Kit (Stratagene); all manipulations of plasmids carrying wild-type or mutant alleles of genes of interest were performed in XL10 Gold® Ultracompetent E. coli cells (Stratagene).

Polymerase chain reaction (PCR)

Amplification conditions for site-directed mutagenesis were as follows: 95°C for 1 min, followed by 19 cycles of 95°C for 50 s, 60°C for 50 s, 68°C for 6 min 15 s, ending with 68°C for 7 min. All plasmids carrying mutant S. oneidensis prpF alleles were sequenced using two described primers [19]. DNA sequencing reactions were performed using BigDye (Applied Biosystems), were purified using CleanSEQ protocols (Agentcourt Biotechnology), and were resolved at the University of Wisconsin-Madison Biotechnology Center.

Plasmid pPRP195

The S. oneidensis acnD gene was amplified from plasmid pPRP141 (18) using primers 5’–GTT ATG AGC ACA CAT ATG AAC ACC CAA TAT C– 3’ and 5’ -GAT ATA GGC GGG ATC CAT GTC GGC ATT GC-3’. The resulting DNA fragment (~2.5 kb) was extracted from the gel using the QIAQuick Gel Extraction kit (Qiagen), the fragment was digested with NdeI and BamHI, and ligated into plasmid pET-15b (bla) digested with the same enzymes; the resulting plasmid (pPRP195) was electroporated into E. coli DH5α/F’, and cells were plated onto LB + ampicillin medium.

Plasmid pPRP205

Plasmid pPRP195 was digested with NdeI and BamHI; the fragment containing the S. oneidensis acnD allele was extracted from the gel as described above. The fragment was ligated into plasmid pTEV5 [24] digested with the same enzymes. The resulting plasmid (pPRP205) was electroporated into E. coli DH5α/F’, and cells were plated onto LB + ampicillin medium.

Plasmids pPRP215-226

Plasmids pPRP153 [18] and pPRP196 [19] were used as templates to generate single-amino acid variants of the SoPrpF protein with the QuikChange® II XL Site-Directed Mutagenesis kit (Stratagene). Other information pertinent to the construction of these plasmids is summarized in Table 2.

Table 2

List of plasmids encoding PrpF variants used in this study.

Plasmid	Allele #	Primers 5’– 3’	Template	Protein Encoded
pPRP215	prpF255	GTGGATTGGAGTGGTAACGCGGGTAACTTAACAGCCGCCGGCGGCTGTTAAGTTACCCGCGTTACCACTCCAATCCAC	pPRP153	C107A
pPRP216	prpF256	GGATTGGAGTGGTAACAGCGGTAACTTAACAGCCGGCTGTTAAGTTACCGCTGTTACCACTCCAATCC	pPRP153	C107S
pPRP217	prpF256	GGATTGGAGTGGTAACAGCGGTAACTTAACAGCCGGCTGTTAAGTTACCGCTGTTACCACTCCAATCC	pPRP196	C107S
pPRP218	prpF257	CAACTTCTAGCACCAGCGCGACGGTTATACTGTCACGTGACAGTATAACCGTCGCGCTGGTGCTAGAAGTTG	pPRP153	K73A
pPRP219	prpF258	GCAACTTCTAGCACCAGCGAAACGGTTATACTGTCACACGTGTGACAGTATAACCGTTTCGCTGGTGCTAGAAGTTGC	pPRP153	K73E
pPRP220	prpF257	CAACTTCTAGCACCAGCGCGACGGTTATACTGTCACGTGACAGTATAACCGTCGCGCTGGTGCTAGAAGTTG	pPRP196	K73A
pPRP221	prpF258	GCAACTTCTAGCACCAGCGAAACGGTTATACTGTCACACGTGTGACAGTATAACCGTTTCGCTGGTGCTAGAAGTTGC	pPRP196	K73E
pPRP223	prpF259	CAACTTCTAGCACCAGCATGACGGTTATACTGTCACGTGACAGTATAACCGTCATGCTGGTGCTAGAAGTTG	pPRP153	K73M
pPRP225	prpF255	GTGGATTGGAGTGGTAACGCGGGTAACTTAACAGCCGCCGGCGGCTGTTAAGTTACCCGCGTTACCACTCCAATCCAC	pPRP196	C107A
pPRP226	prpF259	CAACTTCTAGCACCAGCATGACGGTTATACTGTCACGTGACAGTATAACCGTCATGCTGGTGCTAGAAGTTG	pPRP196	K73M

Isolation of proteins

SoPrpF protein was purified as described [19]. Variant SoPrpF proteins were purified using the Maxwell™ 16 system (Promega). Cultures were grown in LB + ampicillin at 37°C, and induced overnight with isopropyl-ß-D-thiogalactopyranoside (IPTG, 0.3 mM) at an OD650 of ~0.8. Recombinant PrpD, apo-AcnA and apo-AcnB proteins from S. enterica were purified from E. coli as His-tagged proteins as described [15]. SoAcnD protein was purified as follows. Plasmid pPRP205 (S. oneidensis acnD) was introduced into E. coli strain BL21 (λDE3) by electroporation selecting for ampicillin resistance on LB agar + ampicillin. Single colonies were used to inoculate 20 ml of LB + ampicillin; cultures were grown overnight at 37°C. Overnight cultures were used to inoculate two liters of LB + ampicillin, and cells were grown at 37°C until the culture reached an OD650 of ~0.7. At that point, expression of the plasmid-encoded S. oneidensis acnD gene was induced by the addition of IPTG (0.3 mM) to the medium, followed by an 18-h incubation period at 37°C. Cells were harvested by centrifugation at 8,000 x g at 4°C in 1-liter bottles using a JLA-8.l rotor and a Beckman/Coulter Avanti™ J-20 XPI centrifuge. Cells were broken by sonication (10 min, 50% duty, 5 s pulses, maximal setting) with a 550 Sonic Dismembrator (Fisher Scientific). Cell debris was removed by centrifugation at 39,000 x g for 20 min in a Beckman JA 25.5 rotor. The supernatant was filtered through a 0.45 μm filter (Thermo Fisher Scientific). Protein was isolated using Ni-chelate affinity chromatography on Novagen’s His Bind® Resin following the manufacturer’s protocols.

Crystallization and structural determination of SoPrpFK73E

SoPrpFK73E was screened for initial crystallization conditions in a 144-condition sparse matrix screen developed in the Rayment laboratory (unpublished information). Single, diffraction quality crystals were grown by hanging drop vapor diffusion by mixing 2 μL of 28 mg/mL SoPrpFK73E in 2-amino-2-(hydroxymethyl)propane-1,3-diol hydrochloride buffer (Tris-HCl, 10 mM, pH 7.6) containing NaCl (50 mM) with 2 μL well solution containing 2-(N-morpholino)ethanesulfonic acid buffer (MES, 100 mM, pH 6.0) containing sodium malonate (136 mM), polyethylene glycol 4000 (PEG 4K, 15% w/v) at room temperature. Hanging droplets were nucleated after 24 h from an earlier spontaneous crystallization event using a cat’s whisker. Crystals grew to approximate dimensions of 200 X 200 X 400 μm within 3 days. The crystals were transferred directly to a cryoprotecting solution that contained MES buffer (100 mM, pH 6.0), sodium malonate (136 mM), PEG 4K (30% w/v) and vitrified by rapid plunging into liquid nitrogen. SoPrpFK73E crystallized in the space group P21 with unit cell dimensions of a = 51.8 Å, b = 103.4 Å, c = 78.1 Å and two chains in the asymmetric unit.

X-ray diffraction data were collected on a Pilatus detector at SBC Beamline 19-ID (Advanced Photon Source, Argonne National Laboratory, Argonne, IL) The X-ray data were processed and scaled using the HKL-2000 program that integrates data collection, data reduction, phasing and model building [25]. Relevant X-ray diffraction data collection statistics are presented in Table 3. The previously determined model for SoPrpF apo structure (PDB ID: 2PVZ) was used as the search model to solve the SoPrpFK73E apo structure via molecular replacement with the program Phaser [26]. Alternate cycles of manual model building and least squares refinement with the programs COOT [27] Refmac [28] and Phenix [29] reduced the R-factor to 16.5% for all X-ray data from 50–1.22 Å. Relevant refinement statistics are presented in Table 3.

Table 3

Data collection and refinement statistics.

Data collection	PrpFK73E
Space group	P21
Unit-cell	a = 51.8
parameters (Å)	b = 103.4
	c = 78.1
	ß = 104.47°
Wavelength	0.979
resolution range (Å)	50–1.22(1.24–1.22)a
reflections: measured	2973055
reflections: unique	227930
Redundancy	13.0 (6.1)
Completeness (%)	96.5 (84.5)
average I/	49.5 (2.4)
Rsym b (%)	5.3 (67.1)
Rwork c (%)	16.4 (23.2)
Rfree (%)	18.4 (28.6)
no. protein atoms	5907
no. water molecules	1013
Wilson B-value (Å2)	14.5
Average B factors (Å2)
PrpF monomer A	17.2 for 2956
PrpF monomer B	19.1 for 2923
Ligand	16.5 for 28
Solvent	29.2 for 1013
Ramachandran (%)
Most favored	98.0
Additionally	2.0
allowed
Generously	0.0
allowed
Disallowed	0.0
rms deviations
Bond Lengths (Å)	0.006
Bond angles (Å)	1.066
Chiral

aData in parentheses represent highest resolution shell.

bRsym = Σ|I(hkl)–I| / Σ|I(hkl)|, where the average intensity I is taken over all symmetry equivalent measurements and I(hkl) is the measured intensity for a given reflection.

cRfactor = Σ|F(obs)–F(calc)| / Σ|F(obs)|, where Rwork refers to the Rfactor for the data utilized in the refinement and Rfree refers to the Rfactor for 5% of the data that were excluded from the refinement.

Reactivation of aconitases

The Fe/S centers of S. enterica AcnA (hereafter SeAcnA), AcnB (hereafter SeAcnB), and SoAcnD were reactivated using described protocols without modifications [30, 31]. All solutions used were freed of dissolved O2 by degassing as described [32, 33].

High-performance liquid chromatography (HPLC)

Enzyme-dependent dehydration of citrate and 2-methylcitrate was performed in 1-ml reaction mixtures containing Tris-HCl buffer (90 mM, pH 8.0 at 25°C), citrate or 2-MC (5 mM), and SoAcnD (10 μg of reactivated protein) or S. enterica PrpD (hereafter SePrpD, 13 μg). Reaction mixtures were incubated for 1.5 h at 37°C. Reactions were stopped by the addition of 10 N H2SO4 to a final concentration of 5 mM. Particulate matter was removed from reaction products by filtration using a Spin-X® centrifuge tube filter (Costar), and products were resolved by HPLC using a Beckman/Coulter chromatograph equipped with an Aminex® HPX-87H HPLC organic acid analysis column (BioRad) equilibrated and developed isocratically with H2SO4 (5 mM). Elution of compound off the column was detected by monitoring the absorbance at 210 nm.

Kinetic analysis of S. enterica aconitases

All reactions were performed in Tris-HCl buffer (50 mM, pH 8.0 at 25°C) containing KCl (100 mM). Enzyme was incubated in buffer for 3–5 min before reactions were initiated by the addition of either cis-aconitate or 2-methy-cis-aconitate as substrate. Reactions were performed in triplicate, and their progress was monitored using a Perkin Elmer (Norwalk, CT) Lambda 40 UV/Vis Spectrometer at 240 nm; temperature in the cuvettes was maintained with a circulating water bath set at 37°C. Data collection and analysis was performed with Perkin Elmer UV Kinlab software. An extinction coefficient of 3800 M-1 cm-1 was used for cis-aconitate [34], and 4690 M-1 cm-1 was calculated for the chemically synthesized 2-methyl-cis-aconitate. Kinetic curves were each repeated three times, with each substrate concentration tested in triplicate. The reported values are the median of all three experiments. Data did not deviate more than 15% from the median value.

In vivo assessment of activity associated with variant SoPrpF proteins

Growth curves of S. enterica strains were performed in NCE medium supplemented with succinate (30 mM, pH 7.0 at 25°C) or propionate (30 mM, pH 7.0 at 25°C) as carbon and energy source. In both cases a low concentration of glycerol (1 mM) was added to accelerate the catabolism of succinate or propionate. Strains were grown overnight in LB medium supplemented with the appropriate antibiotic. Two microliters of an overnight culture were used to inoculate 198 μl of NCE medium in a 96-well microtiter plate. Growth was monitored using an EL808TM microplate reader (Bio-Tek Instruments) with the incubation chamber set at 37°C. Absorbance readings were recorded every 15 min at 630 nm with 850 s of shaking between readings. All cultures were grown in triplicate. Growth curves were plotted using Prism v4.0 software (GraphPad Software).

In vitro assessment of activity associated with variant SoPrpF proteins

To analyze the activity of SoPrpF variants, the product of the SoAcnD enzyme was synthesized as follows. Reactivated SoAcnD was incubated overnight at 30°C with 2-MC (1 mM) in Tris-HCl buffer (50 mM, pH 8.0) containing KCl (100 mM). SoAcnD protein was removed from the mixture by the addition of His-MagTM Agarose Beads (Novagen) resuspended in Tris-HCl buffer (50 mM, pH 8.0) containing KCl (100 mM). His-MagTM beads were removed, and reaction mixtures were pooled. SoPrpF protein (100 ng) was added to the reactions, samples were taken as a function of time, and were resolved by HPLC as described above.

Results

SoAcnD and SePrpD synthesize different methylaconitate isomers

We previously showed that SoPrpF has isomerase activity that can convert cis-aconitate to trans-aconitate [19]. Given that SoPrpF isomerase activity is needed to restore growth of a S. enterica prpD strain with propionate, we surmised that SePrpD and SoAcnD must synthesize two different methylaconitate isomers, and that SoPrpF isomerizes the SoAcnD product into the SePrpD product, so that SeAcnB, the next enzyme in the 2-MCC, can synthesize 2-methylisocitrate (Fig 1).

To test this hypothesis, we determined whether SePrpD and SoAcnD proteins synthesized different isomers of aconitate and methylaconitate. To do this, we incubated SePrpD and SoAcnD with citrate or 2-methylcitrate (each at 5 mM) in 1-ml reaction mixtures. After a 1.5-hr incubation period, the reactions mixtures were acidified with H2SO4 to a final concentration of 5 mM, and a 100 μl sample of the reaction was resolved by HPLC. When incubated with citrate, SePrpD (Fig 2A) and SoAcnD (Fig 2B) converted citrate (retention time ~8 min) into cis-aconitate (retention time ~7.2 min). When incubated with 2-methylcitrate, SePrpD synthesized a product that eluted off the column as a broad peak centered at ~13.8 min (Fig 2C). On the other hand, SoAcnD synthesized a product that eluted as a sharp peak at ~8 min (Fig 2D). The different retention times and chromatographic behavior indicated to us that the dehydration product of SePrpD and SoAcnD were different compounds. Analysis of chemically synthesized 2-methyl-cis-aconitate revealed a peak that matched the product of the SePrpD reaction (Fig 2C), indicating that SePrpD synthesized 2-methyl-cis-aconitate from 2-methylcitrate.

Fig 2

HPLC analysis of PrpD and AcnD reaction products.

Reactions products when SePrpD used citrate (panel A) or 2-methylcitrate (panel B) as substrate. Reaction products when SoAcnD used citrate (panel B) or 2-methylcitrate (panel D) as substrate. Reaction mixtures were resolved by HPLC using a BioRad Aminex® HPX-87H Organic Acids Analysis column developed isocratically with 5 mM H2SO4 as the mobile phase. Elution was monitored at 240 nm.

2-Methyl-cis-aconitate is a substrate of SeAcnA and SeAcnB

The HPLC chromatograms of the products of the SePrpD and SoAcnD reactions suggested that the SePrpD product, 2-methyl-cis-aconitate, was the substrate that aconitases rehydrated to yield 2-methylisocitrate agreeing with previously reported data [35, 36]. Since SoAcnD cannot support growth with propionate in the absence of SoPrpF, we surmised that the product of SoAcnD was not a substrate of aconitases, but the product of the SoPrpF reaction was.

Kinetic analysis of the S. enterica aconitases indicated that both, SeAcnA and SeAcnB, used 2-methyl-cis-aconitate as substrate to yield 2-methylisocitrate, albeit at a significantly slower rate than cis-aconitate, the TCA cycle intermediate (Table 4). The catalytic efficiencies of SeAcnA and SeAcnB were 40 and 60 lower when 2-methyl-cis-aconitate was the substrate than when cis-aconitate was the substrate, respectively. When 2-methyl-cis-aconitate ws the substrate, K values ranged from 180 to 229 μM and V values ranged from 50 to 60 μM min-1. The numbers reported here are representative of three kinetic studies.

Table 4

Kinetics of AcnA and AcnB with cis-aconitate and 2-methyl-cis-aconitate.

Isomer/Enzyme used	Km (μM)	Vmax (μM min-1)	kcat (min-1)	kcat / Km (min-1 μM-1)
cis-aconitate
AcnA	22 ± 2	86 ± 2	3739 ± 87	170
AcnB	17 ± 3	21 ± 1	2091 ± 100	123
2-methyl-cis-aconitate
AcnA	210 ± 35	50 ± 3	708 ± 42	3.4
AcnB	208 ± 39	10 ± 1	396 ± 40	1.9

All kinetic reactions were performed in triplicate and individual kinetic curves were also repeated in triplicate. Reported values represent median values from all three experiments. No value differed more than 15% from the median value.

SoPrpF isomerizes the SoAcnD reaction product to 2-methyl-cis-aconitate

We tested whether SeAcnA and SeAcnB used the SoAcnD reaction product as substrate (data not shown). The latter was synthesized from 2-MC using SoAcnD. After the SoAcnD enzyme was removed from the reaction mixture, either SeAcnA or SeAcnB was added, and the change in absorbance at 240 nm was monitored. Neither SeAcnA nor SeAcnB used the SoAcnD reaction product as substrate. Purified SoPrpF protein (100 μg) was added to the reaction mixture, and the reaction was allowed to proceed for one hour at 30°C. Addition of SeAcnA resulted in specific activities of approximately 14 ± 1 nmol min-1 μg-1 of protein, while addition of SeAcnB resulted in specific activities of 48 ± 4 nmol min-1 μg-1 of protein. Additionally, HPLC analysis of the products of the SoPrpF reactions indicated that SoPrpF converted the SoAcnD reaction product into 2-methyl-cis-aconitate over time (Fig 3), which can then be used as a substrate by the aconitase enzymes.

Fig 3

SoPrpF isomerizes the SoAcnD product into 2-methyl-cis-aconitate.

Reactivated SoAcnD was incubated with 2-MC, once SoAcnD was removed from the reaction, PrpF was added (100 μg) and samples were removed at the indicated times stopped by the addition of H2SO4 to 5 mM. Reactions were resolved by HPLC using an Aminex HPX-87H Organic Acid column equilibrated and isocratically developed with 5 mM H2SO4; reaction products were monitored at 240 nm. A. void; B. 4-methyl-cis-aconitate C. 2-methyl-cis-aconitate.

Analysis of variants to gain insights into the mechanism of SoPrpF catalysis

Previous crystallographic data identified residues C107 and K73 in the active site of SoPrpF as likely to be involved in catalysis. Residue C107 is formally equivalent to the catalytic base of diaminopimelate epimerase [19, 37, 38], and residue K73 is in a structurally equivalent position to the catalytic glutamate of the phenazine biosynthetic protein PhzF from Pseudomonas fluorescens [39]. We investigated the involvement of these two residues in SoPrpF catalysis, using site-directed mutagenesis to introduce amino acids changes. Five SoPrpF variants were constructed. Residue C107 was changed to either Ala or Ser, and residue K3 was changed to Ala, Glu, or Met. Plasmid pPRP153 (prpF in pBAD18-Kan) was used in in vivo complementation studies using a S. enterica prpD strain to assess the effect of specific substitutions on PrpF function. Substitutions at C107 (Fig 4A) and at K73 (Fig 4B) resulted in proteins that failed to support growth with propionate as a sole carbon and energy source in a S. enterica ΔprpD strain harboring a plasmid expressing S. oneidensis acnD. However, when growing on medium containing succinate and propionate (30 mM each), a much less stringent test for propionate utilization, both proteins with substitutions at C107 supported propionate catabolism (Fig 4C), albeit at a slower growth rate than the wild-type prpF allele, suggesting the presence of proteins with lower activity that could have been caused by misfolding. In contrast, none of the variants with substitutions at residue K73 supported growth under either condition (Fig 4D) suggestive of proteins that either lacked catalytic activity or were misfolded. We note that when grown on medium containing both succinate and propionate, the final cell density of the cultures was approximately 1.3–1.5 A630 units, which was substantially higher than the final density of approximately 1.0 A630 unit when growing with propionate alone, suggesting that the strains containing the mutant alleles encoding SoPrpF variants used propionate and succinate as carbon and energy sources.

Fig 4

Growth behavior analysis of strains synthesizing variants of PrpF.

Growth curves were performed in NCE minimal medium supplemented with propionate (30 mM) (Panels A, B) or propionate + succinate (30 mM ea.) (Panels C, D). Plasmids encoding PrpF variants with substitutions at position C107 failed to restore growth of strains JE9373 (prpD / pAcnDWT pPrpFC107A) or JE9374 (prpD pAcnDWT pPrpFC107S) with propionate (Panel A: PrpFC107S, open diamonds; PrpFC107A, open triangles, respectively), but did restore growth with propionate + succinate, albeit at a slower rate (panel C, diamonds, triangles, respectively). In contrast, substitutions at position K73 failed to compensate for the absence of PrpD on either propionate (panel B: PrpFK73A, open triangles; PrpFK73E, open diamonds, PrpFK73M, open inverted triangles), or succinate + propionate (Panel D: PrpFK73A, open triangles; PrpFK73E, open diamonds; PrpFK73M, open inverted triangles).

To assess the residual level of isomerase activity associated with variant SoPrpF variants, appropriate mutations were introduced into plasmid pPRP196 (prpF+ in pTEV4); proteins containing the C107A or K73A substitutions were not overproduced and could not be purified. The remaining variants were isolated, and their activity was quantified. None of the variants had any detectable activity. While this may be expected for substitutions of K73, the proposed catalytic residue, it is unclear why the C107S mutant had no activity, especially since the variants were active in vivo. More studies may need to be undertaken to fully understand this result.

Analysis of the tertiary and quaternary structures of the catalytically inactive SoPrpFK73E variant

To further understand the roles of residues K73 and C107, we solved the three-dimensional crystal structure of Apo-SoPrpFK73E, which crystallized in space group P21 with cell dimensions a = 51.8 b = 103.4 c = 78.1 Å, and contained two monomers per asymmetric unit. The structure was determined at 1.22Å resolution by molecular replacement using apo-SoPrpF (PDB ID: 2PVZ) as a search model (Table 3). The two monomers in the asymmetric unit are related by a non-crystallographic twofold axis. The two monomers are highly similar where the rms difference between 381 α-carbon atoms is 0.13Å. Given the similarity between the two monomers, all of the discussion of the structure of a single protein chain is based on that of subunit A.

SoPrpF assembles to form a homodimer and buries 2400Å2 of surface area per monomer, which represents 15% of each monomer’s total surface area (Fig 5A). While the SoPrpFK73E variant structure can be superposed closely with both the apo-PrpFWT and trans-aconitate bound structures with rms differences of 0.6 Å and 0.4 Å over 385 α-carbon respectively, an alignment of the N-terminal domains from residues 5–185 reveals that the C-terminal domain of the SoPrpFK73E structure rotates into a more open conformation. The C-terminal domain rotates ~4.4° between the extremes provided by the apo (closed) and aconitate (open) bound forms. This rotation occurs around the segments that connect the N-terminal and C-terminal domain (E180-N183 and I379-M380) although there are negligible conformational changes associated with these residues since they represent the fulcrum point for the rotation. As such these residues cannot be viewed as a flexible hinge.

Fig 5

Crystal structure of SoPrpFK73E at 2.35Å resolution.

A. Ribbon representation of the SoPrpFK73E homodimer (light blue = monomer A, red = monomer B). B. Overlay of SoPrpFK73E (blue), SoPrpF in complex with trans-aconitate (white, RCSB accession number 2PW0) and apo-SoPrpFWT (tan, RCSB accession number 2PVZ). The superposition was performed by aligning residues 1–185 with the program Superpose [40]. Figs 5 and 6 were prepared with the program Pymol; DeLano Scientific LLC, Palo Alto, CA.

Fig 6

Active site comparisons of SoPrpFWT and SoPrpFK73E.

A. Detailed view of the catalytic active site residues, which contact trans-aconitate in the SoPrpFWT protein determined in the presence of trans-aconitate (RCSB accession number 2pw0). B. Detailed view of the residues lining the active site in the SoPrpFK73E variant with malonate bound. The electron density map was calculated with coefficients of the form F−F where the ligand was omitted from the final phase calculation refinement and contoured to 4.5 σ. C. The superposition of active site residues with trans-aconitate and malonate included for reference.

A comparison of the three structures showed that the trans-aconitate bound structure was in the most open conformation, while the SoPrpFK73E structure was intermediate and the apo-SoPrpFWT structure was in the most closed conformation (Fig 5B). This was not surprising as a glycerol molecule, derived from the cryoprotectant used in the structural determination, was bound in the active site of the apo-SoPrpFWT structure. Glycerol is a smaller ligand than the true substrate and thus allows the C-terminal domain to move closer to the N-terminal domain since it is not impeded by the larger trans-aconitate molecule. Similarly, a malonate molecule derived from the crystallization solution was bound in the active site of the SoPrpFK73E structure. Because malonate is larger than glycerol and smaller than trans-aconitate, the C-terminal domain adopted an intermediary conformation. Thus, the structure of the apo-SoPrpFK73E variant protein provided an estimate of the conformational freedom available to the N- and C-terminal domains of PrpF. Attempts to obtain substrate complexed SoPrpFK73E crystals by co-crystallization or soaking into apo crystals with either trans-aconitate or 2-methylcitrate were not successful. This difficulty in co-crystallization was not surprising since binding of trans-aconitate into the SoPrpFK73E active site would position a substrate carboxyl moiety within 3 Å of the new glutamate carboxyl group of SoPrpFK73E variant, thus creating an unfavorable interaction (Fig 6).

Phylogenetic analysis of PrpF homologues

To provide some perspective of the wide distribution of PrpF homologues in nature, we performed a limited phylogenetic analysis of 70 microbes containing a total of 86 PrpF homologues using the MUltiple Sequence Comparison by Log-Expectation (MUSCLE) software. For this purpose, one hundred sequences were selected and aligned (Fig 7). Proteins displaying the shortest distances from the root (clusters at the top and bottom of the tree, respectively) were mostly found in operons with genes related to propionate catabolism. Proteins that are more divergent were more likely to be found in alternate genetic contexts. In Fig 7, proteins from Gram-positive species are indicated with squares and proteins from fungi are indicated with circles. When an organism possesses multiple copies of the gene, the additional copies that are not found to be associated with propionate catabolic genes are marked with triangles. As can be noted in the tree, proteins found in organisms that are more distant evolutionarily from the Gamma-proteobacteria, and proteins not involved in propionate catabolism tend to be more distant in the phylogenetic tree as well.

Fig 7

Phylogenetic tree of selected PrpF homologues in nature.

Phylogenetic tree of selected PrpF homologues. Eighty-six sequences were selected and aligned using ClustalW software. Proteins towards the top of the tree are mostly found in operons with genes related to propionate catabolism. Proteins that are more divergent are more likely to be found in alternate genetic contexts. The PrpF protein of Shewanella oneidensis and the E. coli K12 PrpF homologue are highlighted in red. Distance is shown in substitutions per 100 residues.

Discussion

In this paper, we show that the PrpF protein of S. oneidensis (SoPrpF) has isomerase activity that converts the product of the SoAcnD reaction into 2-methyl-cis-aconitate, which can then be converted into 2-methylisocitrate by aconitase. The SoAcnD product is likely 4-methyl-cis-aconitate. On the basis of the analysis of the crystal structures of SoPrpFWT and SoPrpFK73E, we propose that SoAcnD dehydrates 2-methylcitrate into 4-methyl-cis-aconitate. We established the order of the 2-MCC in Shewanella oneidensis and, by inference, in other bacteria that contain acnD and prpF homologues in their prp operons. That is, in organisms that use the AcnD/PrpF enzymes instead of PrpD, 2-methylcitrate is dehydrated by AcnD to 4-methyl-cis-aconitate, which is then isomerized to 2-methyl-cis-aconitate by PrpF, and rehydrated to 2-methylisocitrate by aconitases (Fig 1).

Rationale for the use of the AcnD/PrpF system

It is unclear why some bacteria such as S. oneidensis and V. cholera use AcnD and PrpF to synthesize 2-methyl-cis-aconitate, while others use a PrpD homologue. This work does not address this interesting question, but it does provide experimental data to support the assignment of a biochemical activity for the SoPrpF enzyme.

We note that a previous report suggested that the formation of 2-methyl-cis-aconitate must occur via a unique syn elimination of water [3]. To date, a detailed analysis of the mechanism of SoAcnD or any of its homologues has not been reported. Thus, for the sake of this discussion, we will assume that the SoAcnD protein, like most aconitases, is limited to anti β-eliminations [41]. Therefore, like other aconitases, SoAcnD would remove the pro-R proton from the oxaloacetate-derived carbon of citrate (C-4), not from the carbon derived from acetate (C-2) [42]. In 2-methylcitrate, carbons 3, 4, and 5 are the carbons derived from oxaloacetate, and as such, carbons 3 and 4 should be where SoAcnD dehydrates the substrate. The C2 of 2-methylcitrate is derived from propionate, and should not participate in the SoAcnD-catalyzed dehydration reaction. An anti ß-elimination of the pro-R proton at C4 should result specifically in the formation of 4-methyl-cis-aconitate by SeAcnD.

Since SoAcnD is likely to yield 4-methyl-cis-aconitate, the presence of an isomerase, like SoPrpF, becomes necessary, as the accumulation of 4-methyl-cis-aconitate could inhibit aconitase. Isomerization of 4-methyl-cis-aconitate to 2-methyl-cis-aconitate then allows aconitase to convert the latter to 2-methylisocitrate, which is cleaved by the 2-methylisocitrate lyase yielding succinate and pyruvate (Fig 1). The greater activity of SeAcnB on enzymatically-derived 2-methyl-cis-aconitate when compared to SeAcnA suggests AcnB is the primary aconitase utilized in propionate catabolism, which is in agreement with our previously reported data that indicated SeAcnB was the primary aconitase involved in propionate catabolism [15].

Modifications to the 2-methylcitric acid cycle

The identification of SoPrpF as a 4-methyl-cis-aconitate isomerase resolves the issue of why SoAcnD and SoPrpF proteins are required to restore propionate catabolism in a ΔprpD strain in S. enterica (18). Fig 1 reflects the findings of this work, in that it shows the sequence of reactions catalyzed by AcnD and PrpF.

Why AcnD, and not any of the other aconitases?

It is unclear why any organism would dedicate an aconitase to the 2-MCC, given that several aconitases already exist in the cell. Studies of the conformation of mitochondrial aconitases suggest that the methyl group of 2-methylcitrate may sterically interfere with this compound entering the active site correctly [43]. We suggest that the active site of SoAcnD may be different enough to use 2-methylcitrate as a substrate.

Structure-function analysis of the SoPrpF active site

In our previous work we suggested that active site of PrpF utilized a single catalytic lysine (Lys 73) to catalyze a proton extraction and a conformational rearrangement around the C2-C3 bond [19]. Data presented here indicates that the reaction instead involves an allylic rearrangement and bond migration mechanism of an aconitate isomerase from Pseudomonas putida described by Klinman and Rose that may have been a PrpF homologue [44]. However, using an allylic rearrangement mechanism would require two catalytic residues and PrpF appears to only have one. The most likely candidate for a second catalytic residue would be C107, which is formally equivalent to the catalytic residue of diaminopimelate epimerase. A recent publication suggested that 4-methyl-cis-aconitate would bind in a manner that would allow the C107 to function as a second catalytic residue [45]. Our mutational analysis sheds light on this possibility. If C107 were required for function, complete loss of enzyme activity in variants with substitutions in C107 would be expected. In Fig 4A, we present in vivo evidence in support for a critical role for C107. When propionate was used as the sole source of carbon and energy, the SoPrpFC107S/A variants failed to support growth of a ΔprpD strain expressing SoAcnD. Although SoPrpFC107S/A variants retain activity (Fig 4C), it is insufficient to support growth with propionate as a carbon and energy source. Further work is needed to better understand the catalytic mechanism of SoPrpF.

Significance of the broad distribution of PrpF homologues

The widespread distribution of genes encoding PrpF homologues in prokaryotes and eukaryotes, reveals the importance of isomerization in cell physiology. While the majority of the bacterial homologues appear to be part of operons along with other propionate utilization genes, many of them are present in operons encoding genes of unknown function. These prpF homologues suggest the possibility of double-bond isomerization being involved in many metabolic pathways.

In summary, we report data in support the following conclusions: i) PrpF isomerizes 4-methyl-cis-aconitate into 2-methyl-cis-aconitate; ii) the product of the AcnD reaction with 2-methylcitrate is most likely 4-methyl-cis-aconitate; iii) the aconitase proteins of S. enterica (i.e., AcnA, AcnB) only rehydrate 2-methyl-cis-aconitate to 2-methylisocitrate; and iv) the proposed catalytic lysine (Lys 73) is absolutely required for activity of PrpF, and while substitutions of C107 do not abolish enzyme activity, PrpFC107 variants do not support growth with propionate.

Conclusions

Work reported in this paper advances our understanding of the 2-methylcitric acid cycle responsible for the conversion of the short-chain fatty acid propionate to pyruvate. The best-characterized sequence of reactions of the pathway involves the PrpD enzyme, which dehydrates 2-methylcitrate to 2-methyl-cis-aconitate. However, some microorganisms have replaced PrpD with two proteins, namely AcnD and PrpF, whose functions are not well understood. In vivo and in vitro evidence presented in this paper support to the conclusion that in an AcnD/PprF-dependent 2-methylcitric acid cycle, AcnD likely generates 4-methyl-cis-aconitate, which is isomerized by PrpF into 2-methyl-cis-aconitate. Additional work is needed to confirm the identity of the SoAcnD product. Structural analysis of an inactive variant of PrpF provides insights into its mechanism of function. A better understanding of PrpF activity will be of value to other investigators researching the function of PrpF-like isomerases, which are widely distributed among prokaryotes.

Kinetics of aconitase product formation.

Reaction mixtures contained aconitase A (AcnA) and aconitase B (AcnB) with cis-aconitate (Panel A: AcnA; Panel B: AcnB) or 2-methyl-cis-aconitate (Panel C: AcnA; Panel D: AcnB) as substrate. Product formation (either isocitrate from cis-aconitate; 2-methylisocitrate from 2-methy-cis-aconitate) was monitored as a decrease in absorbance at 240 nm over time (min; y-axes) as a function of substrate concentration (µM; x axes). The increase in the rate at which A240 decreased is what is plotted. Detailed assay conditions are described under Materials and methods.

(PDF)

Click here for additional data file.