In Salmonella enterica, Ethanolamine Utilization Is Repressed by 1,2-Propanediol To Prevent Detrimental Mixing of Components of Two Different Bacterial Microcompartments

ABSTRACT

Bacterial microcompartments (MCPs) are a diverse family of protein-based organelles composed of metabolic enzymes encapsulated within a protein shell. The function of bacterial MCPs is to optimize metabolic pathways by confining toxic and/or volatile metabolic intermediates. About 20% of bacteria produce MCPs, and there are at least seven different types. Different MCPs vary in their encapsulated enzymes, but all have outer shells composed of highly conserved proteins containing bacterial microcompartment domains. Many organisms have genes encoding more than one type of MCP, but given the high homology among shell proteins, it is uncertain whether multiple MCPs can be functionally expressed in the same cell at the same time. In these studies, we examine the regulation of the 1,2-propanediol (1,2-PD) utilization (Pdu) and ethanolamine utilization (Eut) MCPs in Salmonella. Studies showed that 1,2-PD (shown to induce the Pdu MCP) represses transcription of the Eut MCP and that the PocR regulatory protein is required. The results indicate that repression of the Eut MCP by 1,2-PD is needed to prevent detrimental mixing of shell proteins from the Eut and Pdu MCPs. Coexpression of both MCPs impaired the function of the Pdu MCP and resulted in the formation of hybrid MCPs composed of Eut and Pdu MCP components. We also show that plasmid-based expression of individual shell proteins from the Eut MCP or the β-carboxysome impaired the function of Pdu MCP. Thus, the high conservation among bacterial microcompartment (BMC) domain shell proteins is problematic for coexpression of the Eut and Pdu MCPs and perhaps other MCPs as well.

IMPORTANCE Bacterial MCPs are encoded by nearly 20% of bacterial genomes, and almost 40% of those genomes contain multiple MCP gene clusters. In this study, we examine how the regulation of two different MCP systems (Eut and Pdu) is integrated in Salmonella. Our findings indicate that 1,2-PD (shown to induce the Pdu MCP) represses the Eut MCP to prevent detrimental mixing of Eut and Pdu shell proteins. These findings suggest that numerous organisms which produce more than one type of MCP likely need some mechanism to prevent aberrant shell protein interactions.

INTRODUCTION

Bacterial microcompartments (MCPs) are a functionally diverse group of protein-based organelles found in a wide range of bacterial species (1–4). MCPs consist of metabolic enzymes encapsulated within a protein shell, and their function is to optimize certain metabolic pathways by confining toxic and/or volatile metabolic intermediates (5, 6). Confinement helps to prevent carbon loss, increases metabolic flux, and mitigates cellular toxicity. Genomic analyses predict that MCPs are involved in seven or more metabolic process and are distributed among about 20% of bacteria (1–4). The best-studied MCPs are the carboxysomes of cyanobacteria and chemoautotrophs, which are used to enhance autotrophic CO₂ fixation via the Calvin cycle (6). Two other well-studied MCPs are used to optimize growth on 1,2-propanediol (1,2-PD) or ethanolamine (Pdu and Eut MCPs) (5). Both 1,2-PD and ethanolamine are important carbon sources in anaerobic environments, and the ability to degrade these compounds is thought to provide a competitive advantage to enteric pathogens, such as Salmonella, in the inflamed intestine (7–10). Among the species that contain MCP genes, nearly 40% encode multiple types of MCPs (1).

Despite their wide taxonomic distribution and varied functions, all MCPs have related protein shells (11–13). MCP shells are polyhedral in shape and range from about 80 to 150 nm in diameter. The faces of the shells are composed of a family of proteins that contain bacterial microcompartment (BMC), domains and the vertices are formed by conserved pentameric proteins known as bacterial microcompartment vertex (BMV) proteins (14, 15). The BMC domain-containing proteins that comprise the faces of the MCP shell have a flat hexagonal quaternary structure and interact edge to edge to form extended protein sheets (11, 13, 16–18). Most MCP shells are built from 3 to 10 different types of BMC domain proteins. The varied types of BMC domain proteins have complementary edges, suggesting that conserved edge-to-edge interactions play an important role in the assembly of MCP shells as functionally sophisticated mosaics of diversified BMC shell proteins (18).

Salmonella enterica expresses both Eut and Pdu MCPs (19, 20). The metabolic pathways, structure, and regulation of these MCPs have been the subject of a number of studies (5).The Pdu MCP consists of a protein shell that encapsulates 1,2-PD-degradative enzymes, and its function is to sequester a pathway intermediate (propionaldehyde) that is toxic and poorly retained by the cell envelope (19, 21–23). Production of Pdu MCPs is under the control of the regulatory protein PocR, and they are induced in the presence of 1,2-PD (24, 25). At the global level, expression of pdu genes is regulated by catabolite repression and anaerobiosis via the Crp/cyclic AMP (cAMP) and ArcAB systems (26–28). The second type of MCPs found in Salmonella are Eut MCPs, which are used for the degradation of ethanolamine as a sole carbon and nitrogen source (29). The function of these MCPs is to confine the acetaldehyde (an intermediate of ethanolamine degradation) to minimize carbon loss and cellular toxicity (30–33). In Salmonella, induction of the eut operon requires both ethanolamine and coenzyme B₁₂ (Ado-B₁₂), and this dual-effector regulation is mediated by the positive activator protein EutR (34). The eutR gene is transcribed from its own constitutive promoter and also by the main eut operon promoter, which is activated by ethanolamine and Ado-B₁₂ (35, 36) (Fig. 1.). This arrangement creates a positive feedback loop which helps to ensure that the EutR protein is present in a high enough concentration to compete for available B₁₂ (with the EutBC ethanolamine ammonia lyase) and maintain operon induction (35). In some genera, the regulation of eut expression has been shown to be dependent on a two-component regulatory system composed of EutV and EutW, which are an ANTAR-containing regulatory protein and a histidine sensor kinase, respectively (37–39).

FIG 1.

Schematics of the pdu and eut operons of Salmonella. (A) Schematic representation of the genes of the pdu operon. The locations of relevant insertion mutations used in this study are indicated. (B) Schematic representation of the eut operon and its regulation by Ado-B₁₂ and ethanolamine. This schematic includes the locations of lacZ insertions relevant to this study. Additionally, the eut operon schematic displays the region of the eut promoter which was replaced with consensus sigma-70 promoter (P^c_eut).

Despite the substantial work on the regulation of individual MCPs, there have been no studies to investigate simultaneous expression of two MCP operons. This is of interest in Salmonella (and many other organisms), which produces Eut and Pdu MCPs, both of which are thought to contribute to growth of Salmonella in the intestinal environment and to promote pathogenesis (7–10). Moreover, structural and bioinformatic studies have shown that the BMC domain shell proteins of the Eut and Pdu MCPs share edge complementarity (11, 13, 16–18); hence, simultaneous expression might lead to detrimental shell protein interactions. In this work, we examine the ability of Salmonella to concurrently express and regulate assembly of both Eut and Pdu MCPs and the effects of concurrent expression on growth of Salmonella on 1,2-PD.

MATERIALS AND METHODS

Chemicals and reagents.

Antibiotics and vitamin B₁₂ (CN-B₁₂, B₁₂) were from Sigma Chemical Company (St. Louis, MO). Isopropyl-β-d-1-thiogalactopyranoside (IPTG) was from Diagnostic Chemicals Limited (Charlotteville, PEI, Canada). KOD DNA polymerase was from Novagen (Cambridge, MA). Choice Taq Blue Mastermix was from Denville Scientific (South Plainfield, NJ). Bacterial protein extraction reagent (B-PER II) was from Pierce (Rockford, IL). Other chemicals were from Fisher Scientific (Pittsburgh, PA).

Bacterial strains and growth conditions.

The bacterial strains used in this study are listed in Table 1. The rich media used were lysogeny broth (LB) (40), also known as Luria-Bertani/Lennox medium (Difco, Detroit, MI), and Terrific broth (TB) (MP Biomedicals, Solon, OH). The minimal medium used was no-carbon-E (NCE) medium (41). MacConkey lactose agar was from Difco (Detroit, MI), and green plates were prepared as described previously (42). Tryptone-yeast extract medium (TYE) contained 10 g tryptone, 5 g yeast extract, 8 g of NaCl, 15 g agarose, and 1 liter deionized water, and TYE supplemented with 10% sucrose was used to select for sucrose sensitivity.

TABLE 1.

Strains used in this study

Strain	Genotype	Source
LT2	Wild type
BE123	eutP171::MudJ	Lab collection
BE213	Δpdu675	Lab collection
BE287	LT2/pLac22	Lab collection
BE420	eut-38::MudA	Lab collection
BE510	eutP171::MudJ pduA201::Tn10d(Tet+)	Lab collection
BE1649	eut-38::MudA	Lab collection
BE2160	Δ2578852...2579975::sac cat	This study
BE2161	eut constitutive promoter replacement (Pceut)	This study
BE2162	LT2/pLac22-eutL	This study
BE2163	eut171::MudJ pduA201::Tn10d(Tet+) metE205 ara9 pocR15::Tn10d(Cam+)	This study
BE2164	eut171::MudJ pduA201::Tn10d(Tet+) metE205 ara9 pocR15::Tn10d(Cam+)/pLac22-pocR	This study
BE2165	eut171::MudJ pduA201::Tn10d(Tet+) metE205 ara9 pocR15::Tn10d(Cam+)/pLac22	This study
BE2166	LT2/pLac22-ccmO	This study
BE2167	Pceut eutP171::MudJ	This study
BE2168	Pceut eut-38::MudA	This study
BE2169	metE205 ara9 pocR15::Tn10d(Cam+)	Roth lab

Molecular methods.

Agarose gel electrophoresis, plasmid purification, PCR, restriction digests, ligation reactions, and electroporation were carried out using standard protocols as described previously (19, 43). Plasmid DNA was purified using Qiagen products (Qiagen, Chatsworth, CA) according to the manufacturer's instructions. Following restriction digestion or PCR amplification, DNA was purified using Promega Wizard PCR Preps (Madison, WI) or Qiagen gel extraction kits. For ligation of DNA fragments, T4 DNA ligase Instant Mastermix was used according to the manufacturer's instructions (New England BioLabs, Ipswich, MA).

Strain construction.

A strain that constitutively expresses the eut operon (BE2161) was constructed using the established sacB-cat method, which takes advantage of the lambda red recombinase from plasmid pKD46 (44, 45). The native eut promoter of S. enterica as well as a transposon located upstream of the eut operon (bases 2578852 to 2579975) were replaced with the sacB-cat cassette, selecting for chloramphenicol and ampicillin resistance and screening for sucrose sensitivity at 30°C. The sacB-cat cassette was replaced with a dual T7 and Salmonella consensus promoter by transformation with an oligonucleotide (AAATTAATACGACTCACTATAGGGGAATTGTGAGCGGATAACAATTCCCCTTTGACAGCTAGCTCAGTCCTAGGTATAATGACTAGAAATAATTTATGTTTACACTTTAATAA) and selection for growth on sucrose plates at 30°C. Transformants were then screened for chloramphenicol sensitivity and grown at 37°C to cure plasmid pKD46. The target DNA replacement was verified by DNA sequencing of chromosomal DNA amplified with primers 5′-CACCAGCGCGCCGCTAAATCTATCGAGAAAACCGATG-3′ and 5′-GATGCGGCGTTGGTGGAAAGTATTCGTAATGACCGG-3′.

A number of strains were constructed by P22 transduction as described previously (46). Strain BE2161, which contained the constitutive eut promoter (P^c_eut) was used as the recipient for transductions with strain BE123 (eut171::MudJ) or BE420 (eut38::MudA) as the donor, selecting for kanamycin or ampicillin resistance, respectively. The resulting strains were BE2149 (P^c_eut eut171::MudJ) and BE2150 (P^c_eut eut171::MudJ), both of which contain Mud insertions that form lacZ transcriptional fusions to the eut operon. Strain BE2163 [eut171::MudJ pduA201::Tn10d(Tet⁺) pocR15::Tn10d(Cam⁺)] was constructed using BE510 as the recipient and BE2147 as the donor, selecting for chloramphenicol resistance. The phenotypes of all strains constructed by P22 transduction were verified using MacConkey lactose plates and LB plates with appropriate antibiotics, and phage sensitivity was tested by H5 cross-streaking (42).

Gene cloning.

The pocR and eutL genes (gi numbers 1253557 and 1253978, respectively) were amplified by PCR using S. enterica chromosomal DNA as the template and a 1:1 mixture of Choice Taq Blue and KOD polymerases. The primers used were 5′-GCCGCCAGATCTATGATTTCTGCGAGCGCT-3′ and 5′-GCCGCCAAGCTTTCATAACGATGGAGGATGAGA-3′ for pocR and 5′-GCCGCCAGATCTATGCCTGCATTAGATTTAATTCGAC-3′ and 5′-GCCGCCAAGCTTTTACGCACGCTGGACAGGGTTA-3′ for eutL. BglII and HindIII restriction sites were built into the primers used for ligation into vector pLac22. Colony PCR with these primers was used to screen for plasmids with the desired inserts. All clones were verified by DNA sequencing. The ccmO gene (accession number AP008231) from Synechococcus elongatus PCC 6301 was similarly cloned into pLac22 via PCR except that the template was pet22 vector containing the ccmO gene, which was generously provided by the Yeates lab, and the primers used were 5′-GCCGCCAGATCTATGTCGGCTTCTCTTCCCG-3′ and 5′-GGCGGCTTCGAATCACTGATCATCACGAGGAT-3′.

Quantification of eut operon transcription.

lac operon fusions (formed by Mud insertions) were used to measure transcription of the eut operon. Appropriate strains were grown overnight in LB medium supplemented with 2% 1,2-propanediol. Following overnight incubation, 50 μl was inoculated into 3 ml of NCE minimal medium containing 50 μM ferric citrate and 1 mM magnesium sulfate and supplemented with CN-B₁₂, sodium succinate, and ethanolamine, with or without 1,2-propanediol added as indicated below. These cultures were grown overnight and analyzed using a β-galactosidase assay modified from that described by Miller (47) for measurement in a 96-well plate. Briefly, 20 μl of cells was added to 10 μl of BPERII (Thermo, Waltham MA) and incubated at 37°C for 5 min. Following incubation, 170 μl of buffer (100 mM Na₂HPO₄, 10 mM KCl, 1 mM MgCl₂, pH 8.0) containing 2.7 μl/ml of 2-mercaptoethanol was added, and the mixture was placed in a Biotek Synergy plate reader (BioTek, Winooski, VT) at 30°C. Forty microliters of o-nitrophenolgalactoside (4 mg/ml) was dispensed into each well, and the absorbance at 420 nm was recorded over time (t). The path length (L) for these assays (240-μl samples) was determined to be 0.83 cm. The optical density at 600 nm (OD₆₀₀) of each culture was recorded using a Cary 50 spectrophotometer, and β-galactosidase activity was determined in Miller units (47) using the following equation: (ΔA₄₂₀/Δt [min])(1/L [cm])(total assay volume [ml]/volume of cells added [ml])(1/assay volume [ml])0.22 = nmol/(min · OD₆₀₀).

Microcompartment purification.

Pdu and hybrid Pdu/Eut MCPs were isolated as described previously (48). Purified MCPs were stored at 4°C until used. Attempts to purify intact Eut MCPs were carried out similarly except that 1,2-PD was replaced in all media and buffers by ethanolamine at the same concentrations.

Growth curves.

Growth curves were measured in a Biotek Synergy plate reader (BioTek, Winooski, VT) as described previously (48). Growth curves were plotted and analyzed using Igor Pro software (WaveMetrics Inc., Lake Oswego, OR).

Protein methods.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was carried out using Invitrogen NuPAGE gels and Invitrogen NuPAGE electrophoresis cells according to the manufacturer's instructions (Invitrogen, Green Island, NY). Coomassie brilliant blue R-250 was used to stain proteins following electrophoresis, and protein concentrations were measured using Bio-Rad protein assay reagent (Bio-Rad, Hercules, CA).

Propionaldehyde and ethanolamine determination.

To measure propionaldehyde production during growth on 1,2-PD, appropriate strains were first grown overnight in rich medium. Cultures were pelleted, resuspended in NCE, and inoculated into 100 ml of NCE medium supplemented with 50 μM ferric citrate, 1 mM magnesium sulfate, and 0.6% 1,2-PD to a starting OD₆₀₀ of ∼0.015. Cultures were grown at 37°C with shaking at 275 rpm and were sampled at timed intervals. Cells were removed by centrifugation, followed by filtration with 0.22-μm Millex-GV syringe filters (Millipore Corporation). Propionaldehyde was determined by high-performance liquid chromatography (HPLC) on a Varian Prostar instrument (Agilent Technologies, Santa Clara, CA) as described previously (23).

To measure ethanolamine consumption, appropriate strains were cultured as described above in minimal medium supplemented with ethanolamine as indicated. Ethanolamine was determined by HPLC in the W. M. Keck Metabolomics Facility at Iowa State University using the o-phthaldialdehyde (OPA) fluorescence method of amine detection (49, 50). Buffer A consisted of 10% methanol–90% sodium phosphate (pH 7.3), and buffer B was 80% methanol–20% sodium phosphate (pH 7.3). Samples were derivatized in an in-loop reaction by mixing 6 μl of sample with 6 μl of OPA buffer (12.25 mg OPA, 312.5 μl methanol, 6 ml 0.4 M potassium tetraborate, pH 9.5) and incubating at room temperature for 3 min. The mixture was injected onto a 250- by 4-mm, 5-μm Hypersil ODS column (Thermo Scientific, Waltham, MA) and eluted with a 5-ml linear gradient from 50% buffer B to 100% buffer B followed by 5 ml of 100% buffer B at a constant flow rate of 1 ml/min. Excitation of the fluorescent derivative was at 337 nm, and emission was monitored at 454 nm. Quantitation was done by comparison to a standard curve.

MS/MS protein identification.

Protein bands of interest were excised from an SDS-polyacrylamide gel. In-gel digestion was performed using the Genomics Solution ProGest (Chelmsford, MA) following the manufacturer's procedure with some modifications. After digestion, the solution from the ProGest was dried down. The tryptic peptides were then dissolved in 2 μl of 5-mg/ml α-cyano-4-hydroxycinnamic acid in 50% acetonitrile and 0.1% trifluoroacetic acid. Matrix-assisted laser desorption ionization–time of flight tandem mass spectrometry (MALDI-TOF MS/MS) analyses were performed using a QSTAR XL quadrupole TOF mass spectrometer (AB/MDS Sciex, Toronto, Canada) equipped with an oMALDI ion source. The mass spectrometer was operated in the positive-ion mode. Mass spectra were acquired over m/z 600 to 2200. After every regular MS acquisition, MS/MS was performed against most intensive ions. The molecular ions were selected by information-dependent acquiring (IDA) in the quadrupole analyzer and fragmented in the collision cell. All spectra were processed by MASCOT (MatrixScience, London, United Kingdom) database search. Peak lists were generated by Analyst QS (AB/MDS Sciex, Toronto, Canada) and were used for MS/MS ion searches.

RESULTS

1,2-PD represses transcription of the eut operon.

Prior studies showed that transcription of the eut operon of S. enterica is induced by the combination of ethanolamine and Ado-B₁₂ (35, 36). Here, we looked at the effects of 1,2-propanediol (1,2-PD) on the induction of the eut operon using two lacZ reporter genes (Table 2). The eutP171::MudJ element, which forms a lacZ fusion to the second gene in the eut operon (Fig. 1), is polar and prevents degradation of ethanolamine. In contrast, the eut-38::MudA forms a lacZ fusion just downstream of the eut operon and is wild type for ethanolamine degradation (35). For both of these reporters, supplementation of growth medium with 1,2-PD significantly reduced induction of eut operon (Table 2). Induction of the upstream transcriptional reporter (eutP-lacZ) was repressed from 72 Miller units of β-galactosidase activity to undetectable levels, and induction of the downstream reporter (eut-38-lacZ) was repressed 7-fold (from 49 to 7 U). No detectable β-galactosidase was produced from the eutP-lacZ or the eut-38-lacZ reporter in the absence of either ethanolamine or B₁₂.

TABLE 2.

Effect of 1,2-PD on transcription of the eut operon

Strain	Relevant genotype	β-Galactosidase activitya (Miller units) with:
		No addition	1,2-Propanediol
BE123	eutP171-lacZ	72 (13)	ND
BE420	eut-38-lacZ	49 (14)	7 (1)
BE510	eutP171-lacZ pduA201::Tn10d(Tet+)	63 (10)	3 (1)

^aND, none detected. The growth medium was NCE minimal medium containing 50 μM ferric citrate and 1 mM magnesium sulfate and supplemented with CN-B₁₂, sodium succinate, and ethanolamine with or without 1,2-propanediol. Values in parentheses indicate standard deviations.

Because 1,2-PD is known to induce expression of the pdu operon (24, 25), we tested whether expression of pdu genes was required for repression of the eut operon by 1,2-PD. To do this, we looked at the effect of 1,2-PD on eut transcription using a eutP-lacZ reporter gene in a strain that has a polar insertion in the first gene of the pdu operon [pduA::Tn10d(Tet⁺)] (Table 2). The results showed that the pduA::Tn10d(Tet⁺) insertion had little effect on the repression of eut transcription by 1,2-PD (Table 2). The observed β-galactosidase activities from the eutP-lacZ reporter were 3 Miller units in the presence of 1,2-PD and 63 units in its absence. This indicated that the proteins encoded in the pdu operon as well as the metabolism of 1,2-PD were unnecessary for repression of the eut operon by 1,2-PD. Furthermore, we note that 1,2-PD repressed expression of eutP-lacZ, which is located in the second gene of the eut operon and is polar (Fig. 1). This indicated that repression of eut by 1,2-PD did not require proteins or functions encoded in the eut operon (with the possible unlikely exception of the first gene in the operon, eutS, which encodes a BMC domain shell protein.).

Engineering constitutive expression of the eut operon.

Above we showed that 1,2-PD blocked induction the eut operon. To investigate the rationale for 1,2-PD repression of the eut operon, we constructed a strain that constitutively expresses the eut operon by replacing the native eut promoter with a Salmonella sigma-70 consensus promoter (P^c_eut). To verify constitutive expression, we combined P^c_eut with a lacZ reporter and measured eut expression. The results showed that Salmonella strain BE2167 (P^c_eut eutP171-lacZ) expressed the eut operon at similar levels in the absence or presence of 1,2-PD (135 versus 130 Miller units, respectively) and that induction no longer required ethanolamine or B₁₂. These studies indicated that the native eut promoter was required for repression of the eut operon by 1,2-PD. They also provided a tool to investigate the consequences of eut expression on the growth of Salmonella on 1,2-PD, as further described below.

Growth on 1,2-PD is severely impacted by the constitutive expression of the eut operon.

Strain BE2161 (P^c_eut) was used to assess whether expression of the eut operon affected growth of Salmonella on 1,2-PD and/or the function of the Pdu MCP. Because this strain constitutively expresses the eut operon, we were able to investigate the effects of eut expression on 1,2-PD degradation without adding ethanolamine, which would complicate interpretation of the results by providing an additional growth substrate. Growth of BE2161 (P^c_eut) on 1,2-PD was measured at limiting B₁₂ (20 nM) or at saturating B₁₂ (100 nM). These conditions were chosen because our prior studies showed that faster growth on 1,2-PD at limiting B₁₂ and propionaldehyde toxicity at saturating B₁₂ are characteristic phenotypes of strains in which the Pdu MCP is damaged or disrupted by mutation (22, 51). Hence, these growth conditions would allow us to determine whether the Pdu MCP is functioning normally in vivo when the pdu and eut operons are coexpressed. The results showed that at saturating B₁₂ (100 nM), growth of the P^c_eut strain on 1,2-PD was severely inhibited, even more so than that of the control strain, which is unable to form the shell of the Pdu MCP due to deletion of two major shell proteins PduB and PduB′ (BE213, ΔpduBB′) (Fig. 2). The control strain showed modest growth inhibition from about 12 to 20 h, whereas the P^c_eut strain grew poorly from about 15 to 30 h. Prior studies showed that growth inhibition of the BE213 ΔpduBB′ mutant (and a variety of other shell protein mutants) during growth on 1,2-PD results from propionaldehyde toxicity due to disruption of the MCP shell (51). Hence, these studies indicated that coexpression of the eut and pdu operons resulted in the formation of Pdu MCPs with disrupted shells. We also measured growth of strain BE2161 (P^c_eut) on 1,2-PD at limiting B₁₂ and found that it grew faster than the wild type, although this phenotype was less pronounced than for the ΔpduBB′ mutant, where the MCP shell is disrupted by mutation (22) (Fig. 2). However, this result also suggests a disrupted MCP shell, since various shell disruption mutations which affect the Pdu MCP result in differing degrees of increased growth on 1,2-PD (51). Toxicity is not seen at limiting B₁₂ levels because of reduced production of propionaldehyde by B₁₂-dependent diol dehydratase. Thus, overall, growth tests at both limiting and saturating B₁₂ suggested that expression of the eut operon during growth of Salmonella on 1,2-PD resulted in disruption of the shell of the Pdu MCP. This raised the question of how expression of the eut operon interfered with the proper function of the Pdu MCP.

FIG 2.

Growth of selected Salmonella strains on 1,2-PD. (A) Growth of the wild-type (closed circles), P^c_eut (squares), and ΔpduBB′ (open circles) strains on 1,2-PD with limiting B₁₂ (20 nM). (B) The same strains grown on 1,2-PD with saturating B₁₂ (100 nM). The ΔpduBB′ strain is a control known to be defective for the assembly of the Pdu MCP.

Concurrent expression of the pdu and eut operons results in hybrid microcompartments.

Given the high sequence similarity and edge complementarity of BMC shell proteins from different MCP systems, a possible rationale for repression of the eut operon by 1,2-PD is to prevent association of Eut and Pdu shell proteins which might lead to hybrid MCPs with impaired function. To investigate this possibility, wild-type Salmonella and strain BE2161 (P^c_eut) were grown on 1,2-PD, and Pdu MCPs were purified by our standard procedure and analyzed by SDS-PAGE (21). Interestingly, extra protein bands were observed in MCPs isolated from strain BE2161 (P^c_eut) compared to the wild type (Fig. 3). Select bands were analyzed by MS/MS. Both Eut proteins and Pdu proteins were identified, including EutL, EutE, and EutC as well as PduC and PduB (Table 3). The molecular masses of the bands observed by SDS-PAGE also suggested the presence of PduA, PduJ, PduT, PduE, PduD, and PduP in MCPs purified from the P^c_eut strain BE 2161 (21). A possible explanation for these results is that Eut and Pdu MCPs copurified. Alternatively, coexpression of the eut and pdu operons might have resulted in the formation of hybrid MCPs consisting of both Eut and Pdu proteins. To distinguish between these possibilities, isolated MCPs were assayed for ethanolamine-ammonia lyase activity. The results showed no measureable activity in purified MCPs as opposed to 0.27 μmol min⁻¹ mg⁻¹ in crude extracts. Thus, although the EutC subunit of ethanolamine ammonia lyase was identified in purified MCPs by MS/MS (see above), the EutB subunit (which could not be detected by SDS-PAGE) was likely absent and hence there was no measureable enzymatic activity. This supported the interpretation of a hybrid MCP containing a subset of Eut proteins. Additionally, by using the same purification protocol, we were unable to purify Eut MCPs from strain BE2161 (P^c_eut) under identical conditions except that the Pdu MCP was not induced (no added 1,2-PD), which is contrary to the idea that Eut and Pdu MCPs copurified following coinduction of the eut and pdu operons.

FIG 3.

SDS-PAGE analysis of microcompartments isolated from wild-type Salmonella and a strain constitutive for expression of the eut operon (P^c_eut BE2161). The labeled bands for MCPs purified from the P^c_eut strain were identified by MS/MS. For MCPs purified from wild-type Salmonella, the Pdu proteins present are indicated by letter. These proteins were identified based on the established banding pattern of proteins in the Pdu MCP. M.W., molecular weight.

TABLE 3.

Proteins present in Pdu MCPs purified from a P^c_eut strain identified by MS/MS

Fragment(s) identified	Identity assigned
RPVNQDGFVK, DNPVQIAADAAEGAWRGFDEQETTVAVAR	PduC
AWIGVENPH, QAGLNVGTPFFVR	EutC
VIASVNDGFAR, QAMVEVVYGR, AACNAFTDAVLDIAR	EutL
QLAIHAIR, IAAAIGLNVPDQTR	EutE
ATNTEVVSIELPR	PduB

A single non-Pdu shell protein is sufficient to disrupt the function of the Pdu MCP.

The studies described above indicated that coinduction of the pdu and eut operons led to formation of hybrid Pdu-Eut MCPs that were functionally impaired due to the detrimental mixing of Eut and Pdu shell proteins. To test this hypothesis, we examined the effects of expression of two single Eut shell proteins (EutL or EutS) on the function of the Pdu MCP by measuring growth on 1,2-PD at a limiting (20 nM) B₁₂ concentration. Under these conditions, faster growth occurs when the shell of the Pdu MCP is disrupted (51). The results showed that strain BE2162, which expresses the EutL shell protein from a plasmid, grew much faster than wild-type Salmonella and similarly to the control strain in which the shell of the Pdu MCP is broken by a pduBB′ deletion mutation (BE213) (Fig. 4). We also measured growth of strain BE2167 (which constitutively expresses the EutS shell protein from the chromosome) on 1,2-PD minimal medium at limiting B₁₂. Growth of this strain was substantially faster than that of the wild type but not as fast as that of the control where the shell of the Pdu MCP is disrupted (BE213) (22). However, this is still indicative of a disrupted MCP shell, since the degree to which growth is increased by shell mutations varies with the extent of damage (51) (Fig. 4A).

FIG 4.

Effect of a single Eut shell protein on growth of Salmonella on 1,2-PD. (A) Strains: wild type (closed circles), pLac22-eutL (open circles), pLac22-no insert (closed squares), ΔpduBB′ (open squares) (control known to be defective for assembly of Pdu MCPs), and P^c_eut eutP171::lacZ (triangles) (insert which is polar on all the genes of the eut operon except the shell protein gene, eutS). (B) Strains: wild type (closed circles), pLac22-ccmO (open circles), ΔpduBB′ (open squares), and pLac22-no insert (closed squares). pLac22 is an expression plasmid that allows protein production in response to IPTG. The strains used in this experiment were LT2, BE213, BE 287, BE2161, BE2162, BE2166 (see Table 1 for complete genotypes).

We also tested whether production of a shell protein from a more distantly related MCP would impair the function of Pdu MCPs. The carboxysome shell protein CcmO from Synechococcus elongatus PCC 6301 was produced in Salmonella from plasmid pLac22, and growth on 1,2-PD at a limiting (20 nM) B₁₂ concentration was measured. We observed phenotypes similar to those obtained by expressing Eut shell proteins, namely, higher growth rates consistent with a disrupted MCP shell (51) (Fig. 4B).

Expression of the eut operon during growth on 1,2-PD leads to high levels of propionaldehyde and toxicity.

Prior studies showed that when the shell of the Pdu MCP is disrupted by mutation, propionaldehyde accumulates to toxic levels in the culture medium during growth on 1,2-PD at saturating B₁₂ (51). We therefore measured propionaldehyde production and 1,2-PD consumption during growth of wild-type Salmonella as well as strain BE2161 (P^c_eut) on 1,2-PD with 20 nm B₁₂ (Fig. 5). Analyses showed that strain BE2161 secreted 8 times more propionaldehyde into the culture medium than wild-type Salmonella (11 mM compared to 1.3 mM) (Fig. 5A). Further, growth of BE2161 and consumption of available 1,2-PD were severely inhibited compared to those for wild-type cells, indicating propionaldehyde toxicity as was observed for mutations which break the shell of the Pdu MCP (51) (Fig. 5B). In contrast to the growth studies performed in 96-well plates (Fig. 2), growth inhibition was not overcome in these studies that used glass shake flasks. This is consistent with prior investigations that showed that growth inhibition is reduced in 96-well plates due to absorption of propionaldehyde by the plastic plate (51).

FIG 5.

Propionaldehyde production and 1,2-PD consumption during growth of selected Salmonella strains on 1,2-PD. (A) Propionaldehyde production (open symbols) and OD₆₀₀ (closed symbols) for the wild-type (circles) and P^c_eut (squares) strains. (B) 1,2-PD consumption (open symbols) and OD₆₀₀ (closed symbols) for the wild-type (circles) and P^c_eut (squares) strains.

1,2-PD is preferred over ethanolamine as a growth substrate for Salmonella.

The observation that 1,2-PD repressed induction of the eut operon suggested that Salmonella consumes 1,2-PD in preference to ethanolamine. This was tested by growing wild-type Salmonella in medium containing 1,2-PD, ethanolamine, and B₁₂ and monitoring by HPLC the amounts of 1,2-PD and ethanolamine consumed (Fig. 6). 1,2-PD was rapidly degraded and completely consumed in about 40 h. In contrast, only a small amount of ethanolamine was consumed early in the growth curve, after which its concentration remained constant. These results indicate that Salmonella uses 1,2-PD in preference to ethanolamine, likely reflecting the regulatory scheme in which 1,2-PD represses eut induction. The experiments shown here were performed using 20 nM B₁₂, and the results shown are representative of 3 trials.

FIG 6.

Carbon source preference of Salmonella during growth in the presence of ethanolamine (circles) and 1,2-PD (squares). Triangles, OD₆₀₀.

The regulatory protein PocR is responsible for repression of eut by 1,2-PD.

Prior studies showed that the PocR protein induces the pdu operon in response to 1,2-PD (24, 25). It is well known that prokaryotic transcription factors can act both as repressors and activators of gene expression. For these reasons, we investigated a possible role for PocR in repression of the eut operon by 1,2-PD. A pocR null mutant carrying a eutP-lacZ transcriptional reporter (strain BE2163) did not show repression of the eut operon by 1,2-PD; it produced 79 Miller units of β-galactosidase activity in the absence of 1,2-PD and 96 units in its presence under conditions that induce the eut operon (Table 4). We also showed that production of PocR from plasmid pLac22 reestablished repression of the eut operon by 1,2-PD. In strain BE2164 [eutP-lacZ pduA201::Tn10d(Tet⁺) pocR15::Tn10d(Cam⁺)/pLac22-pocR] the eutP-lacZ reporter produced 78 Miller units of β-galactosidase in the absence of 1,2-PD and 25 Miller units in the presence of 1,2-PD (without induction of pocR by IPTG). With the addition of IPTG (0.1 mM), strain BE2164 produced 17 and 3 Miller units of β-galactosidase activity in the presence or absence of 1,2-PD, respectively. In contrast, plasmid without insert showed no complementation. Induction of pocR using IPTG resulted in significant eut repression in the absence of 1,2-PD. This was most likely due to aberrantly high PocR levels, as overexpression of prokaryotic transcription factors often leads to activity in the absence of coeffectors. Therefore, based on the above-described studies, we conclude the PocR is required for repression of the eut operon by 1,2-PD.

TABLE 4.

Complementation of a pocR mutant for repression of the eut operon^a

Strain	Relevant genotype	1,2-PD	Miller units (SD) with:
			No IPTG	0.1 mM IPTG
BE2163	eutP171-lacZ pduA201::Tn10d(Tet+) pocR15::Tn10d(Cam+)	−	79 (1)	71 (7)
		+	96 (6)	90 (6)
BE2164	eutP171-lacZ pduA201::Tn10d(Tet+) pocR15::Tn10d(Cam+)/pLac22-pocR	−	78 (8)	17 (1)
		+	27 (1)	3 (1)
BE2165	eutP171-lacZ pduA201::Tn10d(Tet+) pocR15::Tn10d(Cam+)/pLac22-no insert	−	78 (ND)	77 (2)
		+	73 (16)	90 (6)

^aND, none detected. The growth medium was NCE minimal medium containing 50 μM ferric citrate and 1 mM magnesium sulfate and supplemented with CN-B₁₂, sodium succinate, and ethanolamine with or without 1,2-propanediol.

DISCUSSION

S. enterica produces Pdu and Eut MCPs for the degradation of 1,2-PD and ethanolamine. The function and regulation of these MCPs have been investigated individually, but no prior studies have investigated their coinduction. Here, we showed that 1,2-PD prevents induction of the eut operon and that Salmonella uses 1,2-PD as a carbon source in preference to ethanolamine. The results indicated that a key reason for 1,2-PD repression of the eut operon is to prevent detrimental interactions among Pdu and Eut shell proteins. Using a Salmonella strain, BE2161, engineered to express the eut operon constitutively (P^c_eut), we showed that coexpression of eut and pdu resulted in the production of hybrid MCPs that contained a mixture of Eut and Pdu proteins. We also showed that coexpression of Pdu and Eut MCPs resulted in 1,2-PD growth phenotypes characteristic of Pdu MCPs with disrupted shells, including growth inhibition due to propionaldehyde toxicity (Fig. 2), which has been shown to result in increased DNA damage (23). The results also showed that expression of a single eut shell protein (EutS or EutL) impaired the function of the Pdu MCP. Furthermore, it is known that the BMC domain proteins tile edge to edge (to form the protein sheets from which the shell is assembled) and that the edges of pdu and eut shell proteins are highly conserved (18). Hence, we propose that the coexpression of the Pdu and Eut MCPs results in the formation of nonfunctional hybrid MCPs due to incorrect interactions among Pdu and Eut BMC domain shell proteins and that this is the main rationale for 1,2-PD repression of the eut operon. Presumably, the mixing of BMC domain proteins from the Eut and Pdu systems disrupts the overall architecture of the shell and prevents shell closure. It is also of note that our experiments indicate that pdu induction takes precedence over eut induction, which in turn leads to preferential metabolism of 1,2-PD over ethanolamine when both carbon sources are present. Prior studies have indicated that both ethanolamine and 1,2-PD are important carbon sources in anaerobic environments such as the intestines of mammals. The results presented here suggest the 1,2-PD would be preferentially used in niches where both are present.

Our results indicate that the PocR regulatory protein mediates repression of the eut operon by 1,2-PD. Previous studies have shown that PocR is a positive regulator of the pdu and cobalamin (B₁₂) biosynthesis (cob) operons, activating both operons in response to 1,2-PD (24, 25). Previous reports have also indicated that the reason the eutR gene is located at the end of the eut operon is that autoinduction is needed to ensure that EutR will be produced at levels sufficient to compete for available Ado-B₁₂ with the B₁₂-dependent ethanolamine ammonia lyase, which catalyzes the first step of ethanolamine degradation (35). Thus, it is possible that PocR exerts control over the expression of the eut operon via available Ado-B₁₂ levels inside the cell. Although induction of the pdu operon would in turn lead to production of the PduCDE protein complex, which would further compete for available Ado-B₁₂, our expression results from strain BE 510 [pduA201::Tn10d(Tet⁺)] (Table 2) suggest that Ado-B₁₂ availability is not the key mechanism for repression by PocR. Therefore, we must conclude that the mechanism by which PocR prevents induction of the eut operon is uncertain and will require further investigation.

The results of this study also raise the question of how diverse species control the expression of multiple MCPs. Recent genomic analyses showed that nearly 40% of MCP-encoding species have genes for multiple MCP systems (1). Furthermore, the BMC domain proteins that form the shells of MCPs are homologous across the known types of MCP systems, including the Pdu, Eut, Grp, and carboxysome systems (5, 6), and are understood to assemble via conserved edge complementarity. Given the widespread homology of MCP shell proteins and the BMC domain present in these proteins, it is likely that a myriad of other bacterial species which have the capacity to express multiple MCP systems (1) also have some mechanism to regulate proper assembly and function of these MCPs.

While about 40% of MCP-encoding species have the capacity to express more than one type of MCP, cyanobacteria and other carboxysome-containing taxa only have one MCP type per genome. This is somewhat inconsistent with a predicted high rate of horizontal gene transfer among MCP loci based on phylogenetic analyses (1). However, carboxysomes are used to enhance autotrophic CO₂ fixation by the Calvin cycle, and this process is essential to the lifestyles of many organisms that produce carboxysomes. Hence, a horizontal gene transfer that interferes with carboxysome function would be extremely detrimental. In these studies, we showed that the CcmO shell protein from the β-carboxysome, which is distantly related to Salmonella Pdu proteins, interfered with the function of the Pdu MCP (Fig. 3B). Given this information, it is reasonable to infer that production of a nonnative shell protein in the context of a carboxysome would interfere with carboxysome function. This would present a significant barrier, i.e., the inability to grow at ambient CO₂ concentrations (52), to the introduction of new MCPs into a carboxysome-containing genome. Given the key metabolic role of the carboxysome and the potential detrimental effects of shell protein mixing, we propose that this strong negative selection pressure might block horizontal transfer of MCPs to genomes containing carboxysomes. On the other hand, however, obligate autotrophy (which occurs in many species that make carboxysomes) might be incompatible with acquisition of MCPs used for heterotrophic growth.