Barriers to 3-Hydroxypropionate-Dependent Growth of Rhodobacter sphaeroides by Distinct Disruptions of the Ethylmalonyl Coenzyme A Pathway

Mutant analysis is an important tool utilized in metabolic studies to understand which role a particular pathway might have under certain growth conditions for a given organism. The importance of the enzyme and of the pathway in which it participates is discretely linked to the resulting phenotype observed after mutation of the corresponding gene. This work highlights the possibility of incorrectly interpreting mutant growth results that are based on studying a single unit (gene and encoded enzyme) of a metabolic pathway rather than the pathway in its entirety. This work also hints at the possibility of using an enzyme as a drug target although the enzyme may participate in a nonessential pathway and still be detrimental to the cell when inhibited.

ABSTRACT

Rhodobacter sphaeroides is able to use 3-hydroxypropionate as the sole carbon source through the reductive conversion of 3-hydroxypropionate to propionyl coenzyme A (propionyl-CoA). The ethylmalonyl-CoA pathway is not required in this process because a crotonyl-CoA carboxylase/reductase (Ccr)-negative mutant still grew with 3-hydroxypropionate. Much to our surprise, a mutant defective for another specific enzyme of the ethylmalonyl-CoA pathway, mesaconyl-CoA hydratase (Mch), lost its ability for 3-hydroxypropionate-dependent growth. Interestingly, the Mch-deficient mutant was rescued either by introducing an additional ccr in-frame deletion that resulted in the blockage of an earlier step in the pathway or by heterologously expressing a gene encoding a thioesterase (YciA) that can act on several CoA intermediates of the ethylmalonyl-CoA pathway. The mch mutant expressing yciA metabolized only less than half of the 3-hydroxypropionate supplied, and over 50% of that carbon was recovered in the spent medium as free acids of the key intermediates mesaconyl-CoA and methylsuccinyl-CoA. A gradual increase in growth inhibition due to the blockage of consecutive steps of the ethylmalonyl-CoA pathway by gene deletions suggests that the growth defects were due to the titration of free CoA and depletion of the CoA pool in the cell rather than to detrimental effects arising from the accumulation of a specific metabolite. Recovery of carbon in mesaconate for the wild-type strain expressing yciA demonstrated that carbon flux through the ethylmalonyl-CoA pathway occurs during 3-hydroxypropionate-dependent growth. A possible role of the ethylmalonyl-CoA pathway is proposed that functions outside its known role in providing tricarboxylic acid intermediates during acetyl-CoA assimilation.

IMPORTANCE Mutant analysis is an important tool utilized in metabolic studies to understand which role a particular pathway might have under certain growth conditions for a given organism. The importance of the enzyme and of the pathway in which it participates is discretely linked to the resulting phenotype observed after mutation of the corresponding gene. This work highlights the possibility of incorrectly interpreting mutant growth results that are based on studying a single unit (gene and encoded enzyme) of a metabolic pathway rather than the pathway in its entirety. This work also hints at the possibility of using an enzyme as a drug target although the enzyme may participate in a nonessential pathway and still be detrimental to the cell when inhibited.

INTRODUCTION

A typical actively growing cell contains about 50% carbon per unit of dry weight. The incorporation of carbon derived from a carbon source into cell constituents is referred to as carbon assimilation. The fact that for heterotrophs organic compounds usually serve not only as a source of carbon but also as a source of energy complicates the study of carbon metabolism and makes it difficult to discern the carbon converted to cell carbon (carbon assimilation) from the carbon that is metabolized to provide energy to the cell. The photoheterotrophic purple nonsulfur bacteria, like Rhodobacter sphaeroides, are notable exceptions to this rule. Being able to use light energy for ATP production and a large variety of organic carbon compounds as a single growth substrate (1), R. sphaeroides may use all carbon provided for the synthesis of cell constituents. This makes this bacterium an ideal organism to study carbon assimilation. Indeed, light energy conversion occurs at spatially distinct membrane invaginations within the cell, where cyclic electron transport photophosphorylation takes place (2, 3), whereas carbon assimilation occurs mostly in the cytoplasm. Thus, in photoheterotrophically grown R. sphaeroides, carbon assimilation is conceptually separated from energy metabolism.

Although all carbon provided is available for cell carbon biosynthesis under photoheterotrophic conditions, how much of the carbon is actually assimilated depends on the average oxidation state of the carbon in the substrate compared to the oxidation state of the cellular carbon. If the average carbon in the carbon source is more oxidized than the cell carbon (e.g., l-malate), CO₂ is released as a by-product. In contrast, CO₂ (or another electron acceptor) has to be provided if the average oxidation state of the carbon substrate is more reduced than that of the cell carbon (e.g., propionate or butyrate), or otherwise the cells will not grow (4). The fact that all electrons have to be accounted for in the conversion of the carbon source to cellular carbon plus possible products is referred to as redox balancing (4–6).

Almost 30 years ago, Neidhardt et al. introduced the concept of precursor metabolites in cell growth (7). Precursor metabolites are intermediates of central carbon metabolism that are the starting points for biosynthetic pathways that ultimately lead to cellular constituents. According to this concept, understanding the assimilation of carbon in a given organism means elucidation of the pathways required for the formation of all precursor metabolites from a given carbon source. These pathways may include peripheral pathways leading to an intermediate common to other substrates and to intermediate metabolic pathways that ultimately lead into central carbon metabolism (8). The entry point into central carbon metabolism determines which further routes are necessary to synthesize all precursor metabolites, including so-called anaplerotic reaction sequences that fill the pools of tricarboxylic acid cycle intermediates. For example, growth with substrates that enter central carbon metabolism on the level of acetyl coenzyme A (acetyl-CoA) requires a specialized reaction sequence to synthesize malate from acetyl-CoA unless a pyruvate synthase is present that catalyzes the reversible oxidation of pyruvate. The glyoxylate cycle is used by many organisms for acetyl-CoA assimilation, and it is an anaplerotic pathway that converts two molecules of acetyl-CoA to malate (9). R. sphaeroides does not have a functional glyoxylate cycle but uses the ethylmalonyl-CoA pathway instead (10). In this case, three acetyl-CoA and two CO₂ molecules are converted to two molecules of malate (Fig. 1B).

FIG 1.

Biochemical pathways of R. sphaeroides linking 3-hydroxypropionate assimilation to central carbon metabolism. Precursor metabolites are represented in boxes. (A) Metabolic scheme showing possible routes by which 3-hydroxypropionate can enter central carbon metabolism at both the level of acetyl-CoA via an oxidative path and at the level of succinyl-CoA via a reductive path. An oxidative path has not been demonstrated for R. sphaeroides, and acetyl-CoA may also be formed by pyruvate oxidation. Acetyl-CoA can be used directly for the synthesis of cell constituents, such as fatty acids, or could enter the ethylmalonyl-CoA pathway. (B and C) The series of chemical reactions catalyzed by enzymes specific to the ethylmalonyl-CoA pathway and the reductive 3-hydroxypropionate assimilatory pathway of R. sphaeroides 2.4.1. Abbreviations: AcuI, acrylyl-CoA reductase; Ccr, crotonyl-CoA carboxylase/reductase; Ecm, ethylmalonyl-CoA mutase; Epi, ethylmalonyl-CoA/methylmalonyl-CoA epimerase; Mcd, methylsuccinyl-CoA dehydrogenase; Mch, mesaconyl-CoA hydratase; Mcl1, (3S)-malyl-CoA/β-methylmalyl-CoA lyase; Mcl2, (3S)-malyl-CoA thioesterase; PHA, polyhydroxyalkanoates; CBB cycle, Calvin-Benson-Bassham cycle or reductive pentose phosphate pathway.

R. sphaeroides is able to use 3-hydroxypropionate as the sole carbon source (11), and we have previously shown that a so-called reductive path via propionyl-CoA is required to convert 3-hydroxypropionate to the central carbon intermediate succinyl-CoA (Fig. 1C). A so-called oxidative path may also operate to provide acetyl-CoA, as was suggested for other bacteria (12, 13); however, an oxidative path, so far, has not been shown in the case of R. sphaeroides, and acetyl-CoA may also be formed by the oxidation of pyruvate. As succinyl-CoA does not need to be synthesized from acetyl-CoA, the anaplerotic ethylmalonyl-CoA pathway, refilling the pool of C₄ dicarboxylic acids of the tricarboxylic acid cycle, should not be required during growth with 3-hydroxypropionate. Here, we show that the ethylmalonyl-CoA pathway is, indeed, not essential for the assimilation of 3-hydroxypropionate by R. sphaeroides. Nevertheless, there is flux through the pathway, and a blockage late in the ethylmalonyl-CoA pathway leads to the buildup of CoA intermediates that can arrest growth. Our data show that even the blockage of the pathway that is not obligatory under certain conditions may result in growth inhibition. A possible participation of the ethylmalonyl-CoA pathway in redox balancing during 3-hydroxypropionate assimilation by R. sphaeroides is proposed.

RESULTS

3-Hydroxypropionate-dependent growth in the absence of the ethylmalonyl-CoA pathway.

Crotonyl-CoA carboxylase/reductase (Ccr) catalyzes the committed step of the ethylmalonyl-CoA pathway (Fig. 1B). Growth of an in-frame ccr deletion mutant with 3-hydroxypropionate as the carbon source seems to support the idea that, under these conditions, the ethylmalonyl-CoA pathway is not essential (Fig. 2A and Table 1). Surprisingly, however, 3-hydroxypropionate-dependent growth is abolished by a blockage of a later step in the pathway: an in-frame deletion of mch, the gene for mesaconyl-CoA hydratase, results in no growth with 3-hydroxypropionate (Fig. 2B). Introducing mch on a plasmid restored growth with 3-hydroxypropionate as the carbon source (Fig. 2B and Table 1). There are at least two possibilities for the apparent conundrum. One possibility is that the ethylmalonyl-CoA pathway is required but that another enzyme or other enzymes bypass the Ccr-catalyzed step in the RsΔccr23KB mutant (an R. sphaeroides strain with a mutation in ccr) during 3-hydroxypropionate-dependent growth although Ccr is essential for growth with acetate (11). Another possibility is that the Mch reaction participates in an additional unknown pathway. Mch (Rsp_0973) has two (R)-specific enoyl-CoA hydratase-like domains, and one of the domains may function outside the ethylmalonyl-CoA pathway during 3-hydroxypropionate-dependent growth. Both possibilities could be ruled out by the following experiment: an in-frame deletion of ccr was introduced into the RsΔmch49KB background, and the double mutant regained the ability to grow with 3-hydroxypropionate as the carbon source (Fig. 2C and Table 1). The fact that genes encoding two enzymes of the ethylmalonyl-CoA pathway can be deleted simultaneously and that growth with 3-hydroxypropionate as the carbon source is still possible shows that, for 3-hydroxypropionate assimilation by R. sphaeroides, the ethylmalonyl-CoA pathway is not essential. Reintroducing ccr on a plasmid into the RsΔccrΔmch24AF double mutant again resulted in a no-growth phenotype (Fig. 2C), just like that seen for the RsΔmch49KB single mutant (Fig. 2B). Therefore, an active Ccr enzyme is required for the mch deletion to exert its no-growth phenotype, suggesting that for the wild type, Ccr is active during growth with 3-hydroxypropionate although Ccr activity in cell extracts was below the detection limit (<5 nmol/min/mg) (11).

FIG 2.

Photoheterotrophic growth of wild type and mutants of Rhodobacter sphaeroides with 3-hydroxypropionate as the carbon source. (A) Growth of the wild type (squares) and RsΔccr23KB (triangles). (B) Growth of RsΔmch49KB (filled triangles), RsΔmch49KB (mch) (open triangles), and the wild type (mch) (squares). (C) Growth of RsΔccrΔmch24AF (filled circles), RsΔccrΔmch24AF (ccr) (left-half-filled circles), RsΔccrΔmch24AF (mch) (right-half-filled circles), and RsΔccrΔmch24AF (ccr mch) (open circles).

TABLE 1.

Photoheterotrophic growth of Rhodobacter sphaeroides with 3-hydroxypropionate or succinate as the carbon source

Straina	Doubling time (h) withb :
	3-Hydroxypropionate	Succinate
Rhodobacter sphaeroides 2.4.1	5.5 ± 0.6 (1.6)	3.3 ± 0.4
WT (pBBR)	6.6 ± 0.4	3.5 ± 0.1
WT (ccr)	6.2 ± 0.5	3.8 ± 0.3
WT (ecm)	6.3 ± 0.3	4.0 ± 0.8
WT (mcd)	6.7 ± 0.7	3.6 ± 0.5
WT (mch)	6.7 ± 0.9	3.7 ± 0.4
WT (mcl1)	6.3 ± 0.3	3.6 ± 0.4
WT (ccr mch)	6.6 ± 0.6	3.8 ± 0.1
WT (yciA)	9.5 ± 0.4 (1.4)	4.5 ± 0.3
RsΔccr23KB	6.1 ± 0.4	3.1 ± 0.3
RsΔccr23KB (pBBR)	8.1 ±1.2	3.4 ± 0.3
RsΔccr23KB (ccr)	6.2 ± 1.1	3.2 ± 0.4
RsΔccr23KB (ccr mch)	7.2 ± 0.6	3.6 ± 0
RsΔecm47KB	NE (0.71)	3.8 ± 0.3
RsΔecm47KB (pBBR)	NE (0.41)	3.7 ± 0.6
RsΔecm47KB (ecm)	6.4 ± 0.4	3.9 ± 0.8
RsΔmcd11KB	NE (0.18)	3.7 ± 0.1
RsΔmcd11KB (pBBR)	NE (0.13)	3.5 ± 0.2
RsΔmcd11KB (mcd)	6.2 ± 0.7	3.9 ± 0.8
RsΔmch49KB	NG	3.1 ± 0.2
RsΔmch49KB (pBBR)	NG	3.5 ± 0.2
RsΔmch49KB (mch)	6.7 ± 0.9	3.6 ± 0.5
RsΔmch49KB (ccr mch)	7.4 ± 0.9	3.8 ± 0.1
RsΔmch49KB (yciA)	31 ± 3 (0.29)	4.7 ± 0.8
RsΔmcl1_4KB	NG	3.7 ± 0.1
RsΔmcl1_4KB (pBBR)	NG	3.8 ± 0.4
RsΔmcl1_4KB (mcl1)	6.3 ± 0.6	3.9 ± 0.6
RsΔccrΔmch24AF	7.5 ± 0.6	3.3 ± 0.5
RsΔccrΔmch24AF (pBBR)	8.2 ± 0.3	3.4 ± 0.4
RsΔccrΔmch24AF (ccr)	NG	4.1 ± 0.4
RsΔccrΔmch24AF (mch)	7.8 ± 0.9	3.8 ± 0.2
RsΔccrΔmch24AF (ccr mch)	6.8 ± 0.3	3.7 ± 0.4

^aWT, wild type.

^bStandard deviation was calculated using three or more biological replicates. The average OD₅₇₈ after 100 h of growth is given in parentheses. NE, nonexponential growth; NG, no growth.

Gradual decrease in growth by blocking consecutive steps in the ethylmalonyl-CoA pathway.

In order to investigate why a blockage late in the nonessential ethylmalonyl-CoA pathway results in the inability of R. sphaeroides to use 3-hydroxypropionate, genes encoding enzymes catalyzing consecutive steps of the ethylmalonyl-CoA pathway were deleted. Ethylmalonyl-CoA epimerase (Epi) was not investigated because the enzyme also catalyzes the epimerization of (2S)/(2R)-methylmalonyl-CoA, a step required for the reductive route for 3-hydroxypropionate assimilation (Fig. 1) (11). Ethylmalonyl-CoA mutase (Ecm) (14) catalyzes the carbon skeleton rearrangement to form (2S)-methylsuccinyl-CoA, which then is oxidized by methylsuccinyl-CoA dehydrogenase (Mcd) (15) to mesaconyl-CoA, the substrate of Mch (16) (Fig. 1B). The in-frame deletion mutants of both ecm and mcd resulted in nonexponential 3-hydroxypropionate-dependent growth (Fig. 3). The growth defect was more pronounced for the RsΔmcd11KB mutant than for the RsΔecm47KB mutant, as exemplified by a decrease in apparent growth yield for the mcd disruption mutant (Table 1). An in-frame deletion of mcl1 encoding (3S)-malyl-CoA/β-methylmalyl-CoA lyase, an enzyme that acts downstream of Mch, showed a no-growth phenotype with 3-hydroxypropionate, in the same manner as RsΔmch49KB mutant (Fig. 3). All mutants grew normally with succinate as the carbon source, and 3-hydroxypropionate-dependent growth for all mutants was restored by introducing the respective genes in trans (Table 1). The observation that the interruption of consecutive steps in the ethylmalonyl-CoA pathway leads to more pronounced deficiencies in 3-hydroxypropionate-dependent growth the further down in the pathway the blockage occurs is consistent with the idea that there is a buildup of intermediates that ultimately leads to complete growth inhibition.

FIG 3.

Photoheterotrophic growth of wild-type Rhodobacter sphaeroides and mutants, affecting the ethylmalonyl-CoA pathway, with 3-hydroxypropionate as the carbon source. Growth of the wild-type (squares), RsΔccr23KB (triangles), RsΔecm47KB (stars), RsΔmcd11KB (hexagons), RsΔmch49KB (inverted triangles), and RsΔmcl1_4KB (diamonds) strains was monitored. The growth of all mutants was restored by the introduction on a plasmid of the corresponding gene that had been deleted from the chromosome (Table 1).

The introduction of a thioesterase rescues 3-hydroxypropionate-dependent growth of ethylmalonyl-CoA pathway mutants.

All intermediates of the ethylmalonyl-CoA pathway are CoA esters (Fig. 1B). One possibility to prevent the buildup of intermediates of the pathway upon blockage at a late step, and possibly avoid the depletion of the free-coenzyme A pool, is to delete ccr and, therefore, prevent any flux through the pathway in the first place (Fig. 1 and 2C). In principle, another possibility would be the hydrolysis of the CoA thioester intermediates inside the cell to relieve growth inhibition. In order to test this possibility, the thioesterase YciA was introduced into the two R. sphaeroides mutants that showed no growth with 3-hydroxypropionate, RsΔmch49KB and RsΔmcl1_4KB. The thioesterase YciA has been shown to hydrolyze several CoA intermediates of the ethylmalonyl-CoA pathway to form the corresponding acids and free CoA (17). The expression of the yciA gene indeed rescued the growth of both mutants (Fig. 4), consistent with the idea that CoA depletion was the cause of the observed growth inhibition.

FIG 4.

Rescue of photoheterotrophic growth of the RsΔmch49KB and RsΔmcl1_4KB mutants with 3-hydroxypropionate by expression of the yciA gene encoding a thioesterase: RsΔmch49KB (yciA) (triangles), RsΔmcl1_4KB (yciA) (diamonds), and wild type (yciA) (squares).

Excretion of organic acids by strains of R. sphaeroides.

Although 3-hydroxypropionate-dependent growth of the RsΔmch49KB (yciA) and RsΔmcl1_4KB (yciA) strains was restored compared to growth of the mutants not carrying the thioesterase-encoding gene, the growth rate and also the growth yield of the RsΔmch49KB (yciA) strain were significantly lower than those of the wild-type (yciA) strain (Fig. 4 and Table 1). In the defined medium used, growth of the wild-type strain ceased when 3-hydroxypropionate was exhausted (Fig. 5). A decrease in growth yield for the RsΔmch49KB (yciA) strain compared to that of the wild type suggested that either less of the 3-hydroxypropionate was consumed or that not all of the carbon used was made available for cell synthesis. Therefore, the spent growth medium of the RsΔmch49KB (yciA) mutant was analyzed for 3-hydroxypropionate that remained and also for the presence of possible products formed (Fig. 5). Less than one-half of the carbon source supplied was consumed by the RsΔmch49KB (yciA) mutant before growth ceased, and roughly one-fifth of the supplied carbon was assimilated (Table 2). This is in contrast to carbon use by the wild-type strain for which most of the supplied carbon was assimilated; some CO₂ was formed for redox balance, as expected, because the carbon in 3-hydroxypropionate is more oxidized than the average cell carbon. Based on the ash-free biomass elemental composition of Rhodopseudomonas palustris, it was estimated that 11% of the carbon would be released as CO₂ for the wild type (11); the value of 8.3% of CO₂ released determined here is in good agreement with the theoretical value (Table 2). For the RsΔmch49KB (yciA) strain, the remaining carbon that was not assimilated was recovered in methylsuccinate, mesaconate, and CO₂. About one-fourth of the supplied carbon was recovered in mesaconate (Table 2), meaning that almost one-half of the 3-hydroxypropionate consumed was converted to mesaconate by the RsΔmch49KB (yciA) mutant strain. Only less than 1% of the supplied carbon was recovered in CO₂, consistent with the fact that the average carbon in mesaconate is more oxidized than cell carbon, and, therefore, less CO₂, as an additional oxidized product, is released to maintain redox balance. Mesaconate and methylsuccinate accumulated during growth, in parallel to 3-hydroxypropionate consumption, for the RsΔmch49KB (yciA) mutant, and apparently both organic acids were not reconsumed (Fig. 5).

FIG 5.

Analysis of organic acids in the spent medium during photoheterotrophic growth of different R. sphaeroides strains with 3-hydroxypropionate as the carbon source. (A) Growth of wild-type R. sphaeroides (left) and the wild-type (yciA) (right) strains was monitored by optical density (squares) along with the concentration of 3-hydroxypropionate and production of organic acids, as indicated on the figure (methylsuccinate was below the detection limit for the wild type). (B) RsΔmch49KB (yciA) growth (open triangles) and 3-hydroxypropionate, mesaconate, and methylsuccinate concentrations were monitored, as indicated on the figure.

TABLE 2.

Carbon balance after photoheterotrophic growth of different strains of Rhodobacter sphaeroides with 3-hydroxypropionate as the carbon source

Strain	Carbon content (%)a
	Biomassb	Total CO2c	Organic acid detected in supernatantd			Carbon balance
			MS	MC	3HP
Wild type	95 ± 1	8.3 ± 0.2	ND	ND	ND	103 ± 3
Wild type (yciA)	92 ± 2	8.5 ± 0.1	ND	0.06 ± 0.02	ND	101 ± 2
RsΔmch49KB (yciA)	19 ± 2	0.9 ± 0.1	2.0 ± 0.9	24 ± 3	56 ± 2	102 ± 2

^aCalculated as (moles of carbon/moles of initial carbon) × 100. Data represent three biological replicates with standard deviations.

^bBased on 22.4 g cell dry weight/mol C.

^cHeadspace CO₂ plus dissolved supernatant CO₂.

^dAmount detected after the final OD₅₇₈ was reached. MS, methylsuccinic acid; MC, mesaconic acid; 3HP, 3-hydroxypropionic acid; ND, not detectable (below the detection limit).

Carbon flux through the ethylmalonyl-CoA pathway during growth with 3-hydroxypropionate.

Interestingly, the wild-type (yciA) strain also formed mesaconate as a product during 3-hydroxypropionate assimilation (Fig. 5 and Table 2). It was therefore possible to directly show that carbon flux through the ethylmalonyl-CoA pathway does occur during photoheterotrophic growth of R. sphaeroides with 3-hydroxypropionate as the carbon source.

DISCUSSION

The ethylmalonyl-CoA pathway is not required for the synthesis of precursor metabolites from 3-hydroxypropionate.

Not only an in-frame ccr deletion mutant but also a mutant which had an additional in-frame deletion of mch grew photoheterotrophically with 3-hydroxypropionate (Fig. 2). The conclusion, then, that the ethylmalonyl-CoA pathway is dispensable for 3-hydroxypropionate assimilation reemphasizes the requirement for a reductive path via propionyl-CoA for the synthesis of succinate and consequently the precursor metabolite oxaloacetate (Fig. 1) (11). The synthesis of all other precursor metabolites, including acetyl-CoA, may start from tricarboxylic acid cycle C₄ intermediates. It is not clear yet, which specific enzymes are required for the conversion of C₄ (either malate or oxaloacetate) to pyruvate or phosphoenolpyruvate (PEP); the genome of R. sphaeroides encodes two possible malic enzymes and a PEP carboxykinase. Once PEP is formed, all the required C₃ to C₆ sugar phosphates are synthesized via gluconeogenesis. Acetyl-CoA may be formed by oxidative decarboxylation of pyruvate by pyruvate dehydrogenase. Another possibility is the formation of acetyl-CoA directly from 3-hydroxypropionate via an oxidative path, possibly involving malonate semialdehyde as an intermediate (Fig. 1). Regardless, if acetyl-CoA is formed via propionyl-CoA, succinyl-CoA, and pyruvate or directly from 3-hydroxypropionate via malonate semialdehyde, the overall carbon and redox balance is the same:

$$3\text{-hydroxypropionate}+\text{HSCoA}\to \text{acetyl-CoA}+{\text{CO}}_{\text{2}}+\text{4}[\text{H}].$$

The ATP requirements and most likely also the electron-carrying cofactors, however, would differ between the two routes. An oxidative path leading directly from 3-hydroxypropionate to acetyl-CoA has not been established for R. sphaeroides, and the specifics of possible electron-carrying cofactors are not known. In summary, the current metabolic scheme for 3-hydroxypropionate assimilation by R. sphaeroides does not require flux through the ethylmalonyl-CoA pathway in order to synthesize all precursor metabolites.

CoA pool depletion as the likely cause of growth inhibition by distinct disruptions of the ethylmalonyl-CoA pathway.

The nongrowth phenotype with 3-hydroxypropionate of the in-frame mch-only deletion came as a surprise (Fig. 2). Interestingly, the introduction of the YciA thioesterase that acts on several CoA intermediates of the ethylmalonyl-CoA pathway caused growth recovery for the mch mutant and also the mcl1 mutant (Fig. 3). Furthermore, the growth yield for the mcd mutant improved when yciA was introduced (data not shown). The free acids of the key intermediates mesaconyl-CoA and methylsuccinyl-CoA were recovered in the spent medium for all three mutants in the presence of YciA (Table 2 and Fig. 5; also data not shown). We propose that the growth inhibition is due to the titration of free CoA and depletion of the CoA pool in the cell rather than to detrimental effects due to a specific metabolite accumulating and targeting another aspect of metabolism, based on the following observations. There is a gradual decrease in growth apparent by blocking consecutive steps in the ethylmalonyl-CoA pathway (Fig. 3) rather than a point in the pathway at which wild-type growth switches to a nongrowth phenotype. Therefore, the growth defect is not due to the deletion of a gene encoding a specific enzyme of the ethylmalonyl-CoA pathway resulting in the accumulation of a specific growth-inhibiting metabolite. Also, replotting the growth data for the in-frame ecm mutant on an arithmetic, instead of a semilogarithmic, scale reveals a straight line up to 70 h after inoculation. Linear growth is consistent with the dilution of an essential and stable cofactor for each doubling (18). We conclude that the essential cofactor is coenzyme A that becomes depleted, leading to growth inhibition. A similar conclusion was made by Schneider et al. concerning a Methylobacterium extorquens mutant strain (19). A propionyl-CoA carboxylase deletion strain was initially unable to grow with oxalate after a switch from succinate-containing medium, and several CoA esters accumulated inside the cell, including propionyl-CoA and mesaconyl-CoA, with a simultaneous decrease in the free-coenzyme A pool; however, growth ensued after the free-coenzyme A pool returned to wild-type levels even though the accumulation of CoA esters remained unchanged (19).

Carbon flux through the ethylmalonyl-CoA pathway during photoheterotrophic growth with 3-hydroxypropionate.

The introduction of the YciA thioesterase offers a very sensitive method to monitor actual carbon flux through the ethylmalonyl-CoA pathway due to the appearance of mesaconate in the spent medium and the high extinction coefficient of mesaconate at 230 nm (ε_230nm = 5,500 M⁻¹cm⁻¹) (20). Clearly, this tool can only be used under growth conditions where the ethylmalonyl-CoA pathway is not essential, as is the case for 3-hydroxypropionate-dependent growth. However, the extent of the flux through the ethylmalonyl-CoA pathway cannot be quantified by this method because the secretion of free acids depends on the specific activity of the thioesterase in the cell, the pool sizes of the corresponding CoA intermediates, and the extent to which the free acids are secreted by the cell or are able to accumulate within the cell. Using this method, carbon flux through the ethylmalonyl-CoA pathway during photoheterotrophic growth with 3-hydroxypropionate was demonstrated (Fig. 5B); however, no flux was detected during photoheterotrophic growth with succinate (data not shown).

A possible role for the ethylmalonyl-CoA pathway during 3-hydroxypropionate assimilation.

The fact that carbon flux occurs through the ethylmalonyl-CoA pathway during photoheterotrophic growth with 3-hydroxypropionate may simply mean that it is accidental: under the given conditions the pathway is not completely turned off, and, therefore, some acetyl-CoA is diverted through this route. Another possibility is that flux through the ethylmalonyl-CoA pathway is beneficial during 3-hydroxypropionate-dependent growth. In order to understand carbon assimilation from a given carbon source, it is important to know not only the nature of the pathways that allow for the synthesis of all precursor metabolites but also the relative quantitative requirements for each precursor metabolite have to be considered. Furthermore, CO₂ is an obligate intermediate (CO₂ is transiently produced and consumed) in the conversion of 3-hydroxypropionate to all precursor metabolites (Fig. 1); overall, though, a defined net amount of CO₂ is released from the cell because the average carbon in 3-hydroxypropionate is more oxidized than the cell carbon. Based on the cellular elemental composition of Rhodopseudomonas palustris (21), the net release of CO₂ would be 11% (0.5 per 4.5 carbons):

$$1.5\hspace{0.17em}{C}_3{H}_6{O}_3+0.72\hspace{0.17em}{NH}_3\to 4\hspace{0.17em}{CH}_{1.8}{N}_{0.18}{O}_{0.38}+0.5\hspace{0.17em}{CO}_2+1.98\hspace{0.17em}{H}_2\mathrm{O.}$$

An actual percentage of 8.3% was determined in this study (Table 2), suggesting that the average cell carbon for R. sphaeroides is slightly more oxidized than that for R. palustris, but it is still more reduced than the average carbon in 3-hydroxypropionate. A balanced carbon distribution model at steady state for the central carbon metabolism of R. sphaeroides during photoheterotrophic growth with 3-hydroxypropionate was constructed (see Fig. S1 in the supplemental material). For the relative precursor requirements, values were used that were determined for Escherichia coli grown aerobically with glucose (7, 8). If a net release of 10% CO₂ from 3-hydroxypropionate is assumed, the model predicts that 5% of the starting carbon would pass through the Ccr-catalyzed step (Fig. S1). Based on the molar growth yield of 60 g of dry weight per mole of 3-hydroxypropionate consumed (11), a doubling time of 5.5 h (Table 1), and a 5% carbon flux through the Ccr step, the minimal required Ccr activity would be 3.5 nmol/min/mg. Therefore, the low flux predicted by the model is consistent with a Ccr activity that is less than the detection limit of 5 nmol/min/mg in cell extracts (11); the model, however, is also consistent with the fact that Ccr is active during photoheterotrophic growth with 3-hydroxypropionate, as shown here.

The following considerations may shed light on the possible function of the ethylmalonyl-CoA pathway during 3-hydroxypropionate assimilation. At first sight, the ethylmalonyl-CoA pathway seems to be redundant to a reductive path that ultimately also leads to tricarboxylic acid C₄ intermediates. The formation of malate from 3-hydroxypropionate by the reductive path via propionyl-CoA can be written as follows:

$$3\text{-hydroxypropionate}+{\text{CO}}_{\text{2}}\to \text{malate.}$$

Another possibility to form malate is the oxidation of 3-hydroxypropionate to acetyl-CoA and CO₂ either via malonate semialdehyde or pyruvate. The ethylmalonyl-CoA pathway then converts acetyl-CoA to malate but reuses only a fraction of the CO₂ that was initially released by the oxidation of 3-hydroxypropionate to form acetyl-CoA, according to the following equations:

$$3\times 3\text{-hydroxypropionate}+3\text{\hspace{0.17em}HSCoA}\to \text{3 acetyl-CoA}+3\hspace{0.17em}{\text{CO}}_{\text{2}}+12\hspace{0.17em}[\text{H}]$$

$$3\text{\hspace{0.17em}acetyl-CoA}+2{\text{\hspace{0.17em}CO}}_{\text{2}}+3{\text{\hspace{0.17em}H}}_{\text{2}}\text{O}\to \text{2 malate}+3\text{\hspace{0.17em}HSCoA}$$

$$3\text{-hydroxypropionate}+{\text{H}}_{\text{2}}\text{O}\to \mathrm{⅔}\text{\hspace{0.17em}malate\hspace{0.05em}\hspace{0.17em}}+\mathrm{⅓}{\text{\hspace{0.17em}CO}}_{\text{2}}+4[\text{H}]$$

The net CO₂ (and reductant) released from the conversion of 3-hydroxypropionate via acetyl-CoA can be used by the reductive pathway. We therefore propose that during growth with 3-hydroxypropionate, two routes that lead from 3-hydroxypropionate to malate operate simultaneously: (i) the reductive path and (ii) oxidation of 3-hydroxypropionate to acetyl-CoA in combination with the ethylmalonyl-CoA pathway. Carbon flux through both routes (one consuming CO₂, the other releasing CO₂) allows a balance which ensures that the precise net amount of CO₂ is released from the cell.

MATERIALS AND METHODS

Materials.

3-Hydroxypropionate was purchased from Tokyo Chemical Industries (TCI) America (Portland, OR). An aliquot of 25 ml of the 2.8 M 3-hydroxypropionic acid solution was neutralized on ice to a pH between 6.5 and 7.0 using 3 M NaOH (about 23 ml) and diluted to a final concentration of 1 M using double-distilled H₂O (ddH₂O). Concentrations of the 3-hydroxypropionate stock before and after neutralization/dilution were determined by an enzymatic assay using propionyl-CoA synthase (11). The solution was then filter sterilized using a low-binding polyethersulfone 0.22-µm-pore-size membrane (Millipore) and injected into N₂-sparged, stoppered, screw-cap (Hungate) tubes at volumes between 2 and 5 ml for storage at −20°C. All primers used in the study were obtained from Sigma-Aldrich (St. Louis, MO) and are listed in Table S1 in the supplemental material.

Bacterial strains and growth conditions.

Rhodobacter sphaeroides 2.4.1 (DSMZ 158) and its derivative strains were grown anaerobically in the light (3,000 lx) at pH 6.8 and 30°C in minimal medium (22) (with the exception that 0.5 mg liter⁻¹ of cobalt chloride hexahydrate instead of 0.05 mg liter⁻¹ was used). The medium was supplemented with a 10 mM concentration of the carbon source; for carboxylic acids, the sodium salts were used. Medium plates included 2.5% agar. Oxygen in liquid medium was removed and replaced with N₂ by repeated vacuuming and sparging. Growth in liquid cultures was estimated based on the optical density at 578 nm (OD₅₇₈). For growth studies, cells were pregrown anaerobically in 5 ml of minimal medium containing 10 mM sodium succinate (OD₅₇₈ of ∼1.5), and 0.1 ml of cells that were in stationary phase for about 5 to 10 h was transferred into Hungate tubes with 4 ml of minimal medium and the appropriate carbon source (initial OD₅₇₈ of 0.03 to 0.05). R. sphaeroides strains carrying plasmids were grown in the presence of 25 µg ml⁻¹ spectinomycin. Medium used for the isolation of single-crossover mutant strains contained kanamycin (20 µg ml⁻¹). Escherichia coli strains DH5α, S17-1, and SM10 were grown in lysogeny broth (LB) at 37°C with ampicillin (100 µg ml⁻¹), spectinomycin (50 µg ml⁻¹), or kanamycin (50 µg ml⁻¹) as needed.

Construction of the markerless in-frame deletions and complementation.

The suicide plasmid pKB23 employed for the markerless in-frame deletion of mch (rsp_0973) was constructed by amplifying 42 bp of the 5′ mch coding region plus 1,445 bp directly upstream (primers deltamchup_for1/rev1) and 45 bp of the 3′ coding region plus 1,546 bp directly downstream (primers deltamchdown_for1/rev1) of mch by PCR and cloning the products in tandem into pK18mobsacB (23). The resulting plasmid, pKB23, contains an in-frame deletion of mch of 942 bp. The remaining open reading frame encodes a 30-amino-acid peptide.

The suicide plasmid pKB15 employed for the markerless in-frame deletion of ccr (rsp_0960) was constructed by amplifying 123 bp of the 5′ ccr coding region plus 1,497 bp directly upstream (nested PCR using primers UP_ccr_5′/3′ and delta_KpnI_5′/delta_ccr_XbaI_3′) and 36 bp of the 3′ ccr coding region plus 1,483 bp directly downstream (nested PCR using primers ccrdown_for2/rev2 and deltaccrdown_for1/rev1) of ccr by PCR and cloning the products in tandem into pK18mobsacB. The resulting plasmid, pKB15, contains an in-frame deletion of ccr of 1,131 bp. The remaining open reading frame encodes a 54-amino-acid peptide.

The suicide plasmid pKB22 employed for the markerless in-frame deletion of ecm (rsp_0961) was constructed by amplifying 12 bp of the 5′ ecm coding region plus 1,434 bp directly upstream (primers deltaecmup_for1/rev1) and 51 bp of the 3′ coding region plus 1,773 bp directly downstream (deltaecmdown_for1/rev1) of ecm by PCR and cloning the products in tandem into pK18mobsacB. The resulting plasmid, pKB22, contains an in-frame deletion of ecm of 1,896 bp. The remaining open reading frame encodes a 22-amino-acid peptide.

The suicide plasmid pKB4 employed for the markerless in-frame deletion of mcd (rsp_1679) was constructed by amplifying 6 bp of the 5′ mcd coding region plus 1,487 bp directly upstream (primers Δmcd_EcoRI/XbaI_5′ and Δmcd_KpnI_3′) and 111 bp of the 3′ coding region plus 1,439 bp directly downstream (primers deltamcd_KpnI-5′ and deltamcd_XbaI_3′) of mcd by PCR and cloning the products in tandem into pK18mobsacB. The resulting plasmid, pKB4, contains an in-frame deletion of mcd of 1,530 bp. The remaining open reading frame encodes a 40-amino-acid peptide.

The suicide plasmid pKB11 employed for the markerless in-frame deletion of mcl1 (rsp_1771) was constructed by amplifying 18 bp of the 5′ mcl1 coding region plus 1,460 bp directly upstream (primers deltamcl1up_for1/rev1) and 111 bp of the 3′ coding region plus 1,367 bp directly downstream (deltamcl1down_for1/rev1) of mcl1 by PCR and cloning the products in tandem into pK18mobsacB. The resulting plasmid, pKB11, contains an in-frame deletion of mcl1 of 823 bp. The remaining open reading frame encodes a 44-amino-acid peptide.

The mutant strains RsΔmch49KB, RsΔccr23KB, RsΔecm47KB, RsΔmcd11KB, and RsΔmcl1_4KB were generated by mating R. sphaeroides 2.4.1 with E. coli S17-1 transformed with the respective suicide plasmid, and single and double crossovers were isolated as described previously (24). The double mutant RsΔccrΔmch24AF was isolated by mating RsΔccr23KB with E. coli S17-1 transformed with pKB23. The resulting mutant strains were genotyped by sequencing PCR products derived from chromosomal DNA and primers (listed in Table S1) that amplified the entire flanking regions where crossovers may have occurred.

For complementation studies pBBR1MCS-derived plasmids (25) and the BioBrick method (26) were used. The XbaI and SpeI restriction sites were removed from pBBRsm2MCS5 (11) by cutting the plasmid with XbaI/SpeI and self-ligating the product to obtain pBBRsm2(SJC)D.

For the complementation of the RsΔmch49KB mutant, a 1,116-bp fragment was amplified from R. sphaeroides 2.4.1 genomic DNA with primers mchcomp_for1/rev1 and inserted into pSC54. pBBRsm2(SJC)d-derived pSC54 is the source for the nptII promoter, which was obtained using primers PnptII_F1:5′/R1:5′ and plasmid pK18mobsacB (23) as the template. The rrnB terminator was added by inserting the KpnI/XbaI fragment of pSC16 to obtain pSC24. pSC16 was constructed by ligating the 294-bp downstream region of one of the two 23S rRNA genes present in R. sphaeroides (27) (primers rrnB_terminator_F1/R1; template, R. sphaeroides 2.4.1 genomic DNA) into pUC19.

For the complementation of the RsΔccr23KB mutant, a 1,347-bp fragment was amplified with primers RSccrcomp_SpeI_F2/NdeI_R2 and inserted into pUC19 using KpnI/NdeI restriction sites. To the resulting plasmid the rrnC terminator was added by inserting a KpnI/XbaI fragment of pSC19. pSC19 was constructed by ligating the 294-bp downstream region of the second 23S rRNA gene (primers rrnC_terminator_F1/R1; template, R. sphaeroides 2.4.1 genomic DNA) into pUC19. Finally, the KpnI/NdeI fragment of the resulting plasmid was added to plasmid pSC45, resulting in pSC65. pBBRsm2(SJC)d-derived pSC45 is the source for the tetA promoter and was constructed using primers cp_PtetAacuI_ F/R and plasmid pMA5-1 (28) as the template.

For the complementation of the RsΔecm47KB mutant, a 2,009-bp fragment was amplified with primers ecm_compSJC_F/_R and inserted into KpnI/NdeI-cut pSC54 to obtain pSC94.

For the complementation of the RsΔmcd11KB mutant, a 1,694-bp fragment was amplified with primers mcd_compSJC_F/_R and inserted into KpnI/NdeI-cut pSC54 to obtain pSC97.

For the complementation of the RsΔmcl1_4KB mutant, a 1,004-bp fragment was amplified with primers mcl1_compSJC_F/_R and inserted into KpnI/NdeI-cut pSC54 to obtain pSC100.

For the complementation of the RsΔccrΔmch24AF mutant, the KpnI/XbaI fragment of pSC65 was inserted into SpeI/KpnI-cut pSC24 to obtain pSC25. The expression of the ccr gene is driven by the tetA promoter, and that of the mch gene is driven by the nptII promoter.

For expression of the thioesterase YciA from E. coli, the yciA gene was amplified using primers yciA_F/_R and pCM160_RBS-YciA-Ec-Mex as a template (a kind gift by M. Buchhaupt). The 433-nucleotide (nt) product was cloned into pSC75 to obtain plasmid pSC113. pSC75 is the source of the rrnB promoter (27) and was constructed using primers rrnB_promoter_Up/Dn and chromosomal R. sphaeroides 2.4.1 DNA as the template.

The pBBRsm2(SJC)D plasmid (empty vector control), pSC24, pSC25, pSC65, pSC94, pSC97, pSC100, and pSC113 were independently transferred to the wild-type, RsΔmch49KB, RsΔccr23KB, RsΔccrΔmch24AF, RsΔecm47KB, RsΔmcd11KB, and RsΔmcl1_4KB strains by mating the R. sphaeroides strains with E. coli SM10 transformed with the respective plasmid. All plasmids and strains used in this study are listed in Tables S2 and S3.

Product analysis by HPLC.

For the detection and quantification of organic acids present in the spent medium as a result of photoheterotrophic growth of R. sphaeroides, samples (0.6 ml) were taken at different time points during growth and centrifuged at 16,000 × g for 3 min to remove the cells, and 0.5 ml of the supernatant was frozen at −20°C until further analysis. Samples were thawed and centrifuged at 16,000 × g for 3 min. The supernatant (100 to 150 µl) was acidified by addition of 3 M sulfuric acid to a final concentration of 0.03 M and centrifuged at 16,000 × g for 3 min, and the supernatant (30 to 100 µl) was used for analysis. Organic acids were analyzed on a Shimadzu Prominence high-performance liquid chromatography (HPLC) system with dual-wavelength detection (210 and 230 nm) using two different columns.

Ion exclusion HPLC was performed using a Rezex ROA-organic acid H⁺ (8%) 250- by 4.6-mm (Phenomenex) column that was preceded by a Carbo-H+ cartridge. Organic acids were separated isocratically in 2.5 mM sulfuric acid at 60°C with a flow rate of 0.5 ml min⁻¹. Peak retention times for known standards of the following organic acids were determined (in minutes): phosphoenolpyruvic (4.0), 3-phosphoglyceric (4.1), oxalic (4.1), pyruvic (4.2), citric (4.6), glyoxylic (5.3), (3S)-malic (5.3), succinic (5.9), ethylmalonic (6.0), methylsuccinic (6.3), (S)-lactic (6.4), 3-hydroxypropionic (6.5), (R)-3-hydroxybutyric (6.5), acetoacetic (6.6), acetic (7.0), mesaconic (7.3), propionic (7.9), butyric (9.1), and crotonic (10.3) acid. Limits of detection for mesaconic, methylsuccinic, and 3-hydroxypropionic acid were 0.5 µM, 50 µM, and 50 µM, respectively.

Because there was an overlap of the 3-hydroxypropionic acid and methylsuccinic acid peaks for the ion exclusion column, reverse-phase HPLC was performed using an Alltima C₁₈ 250- by 4.6-mm (5-µm pore size) column (Hichrom) preceded by an Alltima 5-µm-pore-size C₁₈ guard cartridge to determine methylsuccinic acid concentrations. Organic acids were separated via a gradient between buffer B (80% ACN–20% 0.1% phosphoric acid, pH 2.5) and buffer A (0.1% phosphoric acid, pH 2.5) at 30°C with a flow rate of 0.8 ml min⁻¹. The gradient was as follows: 0.06% B for 6 min, 0.06% B to 40% B for 35 min, 40% B to 75% B for 5 min, 75% B to 0.06% B for 7 min, and 0.06% B for 12 min to reequilibrate the column. Retention times of acid standards that were analyzed were as follows (in minutes): 3-hydroxypropionic acid, 7.5; mesaconic acid, 22.3; and methylsuccinic acid, 23.8.

Gas chromatography CO₂ detection.

Quantification of CO₂ present in the headspace and liquid growth cultures was performed by gas chromatography (GC) using a splitless Shimadzu GC-14A with a thermal conductivity detector. All syringes and centrifuge tubes were stored in an N₂-filled chamber prior to use to ensure that no CO₂ was introduced during sampling. The headspace was directly sampled from the growth culture bottles. To determine dissolved CO₂ in the growth medium, 1.5 ml of the culture was withdrawn, the cells were removed by centrifugation at 16,000 × g for 3 min, 1 ml of the supernatant was injected into a N₂-sparged, stoppered, screw-cap Hungate tube containing 0.2 ml of 37% hydrochloric acid, and the tube was intermittently shaken for 1 h. One milliliter of the headspace was applied to a 60/80 Carboxen-1000 column (Supelco) using helium carrier gas at a column temperature of 190°C. A CO₂ standard curve was generated by acidifying known amounts of solid sodium bicarbonate (0.3 to 1.5 mg) with 1 ml of 37% hydrochloric acid in N₂-sparged Hungate tubes and sampling of the headspace. Retention time for CO₂ was 7.8 min. The limit of detection for CO₂ was 4 µM.

Determination of the dry weight.

To determine cell dry mass, the cells from 50 ml of the growth culture were collected by centrifugation at 13,000 × g for 5 min and washed three times with deionized water. The resulting cell pellets were lyophilized overnight, and the masses were determined using an analytical balance. The analysis was performed in triplicates.