|Home | About | Journals | Submit | Contact Us | Français|
Caulobacter crescentus, which thrives in freshwater environments with low nutrient levels, serves as a model system for studying bacterial cell cycle regulation and organelle development. We examined its ability to utilize lactose (i) to gain insight into the metabolic capacities of oligotrophic bacteria and (ii) to obtain an additional genetic tool for studying this model organism, aiming to eliminate the basal enzymatic activity that hydrolyzes the chromogenic substrate 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-gal). Using a previously isolated transposon mutant, we identified a gene, lacA, that is required for growth on lactose as the sole carbon source and for turning colonies blue in the presence of X-gal. LacA, which contains a glucose-methanol-choline (GMC) oxidoreductase domain, has homology to the flavin subunit of Pectobacterium cypripedii's gluconate dehydrogenase. Sequence comparisons indicated that two genes near lacA, lacB and lacC, encode the other subunits of the membrane-bound dehydrogenase. In addition to lactose, all three lac genes are involved in the catabolism of three other β-galactosides (lactulose, lactitol, and methyl-β-d-galactoside) and two glucosides (salicin and trehalose). Dehydrogenase assays confirmed that the lac gene products oxidize lactose, salicin, and trehalose. This enzymatic activity is inducible, and increased lac expression in the presence of lactose and salicin likely contributes to the induction. Expression of lacA also depends on the presence of the lac genes, implying that the dehydrogenase participates in induction. The involvement of a dehydrogenase suggests that degradation of lactose and other sugars in C. crescentus may resemble a proposed pathway in Agrobacterium tumefaciens.
Caulobacter species inhabit diverse aquatic and soil environments, their ubiquitous distribution reflecting an ability to prosper despite low nutrient levels (47, 48). Unraveling the physiological adaptations that enable these aerobic chemoheterotrophs to thrive under nutrient-poor conditions will facilitate interpretation of microbial activities in oligotrophic environments, such as water bodies (which make up the majority of the earth's surface). Recent studies of Caulobacter have led to novel findings regarding the uptake and catabolism of various carbon sources (6, 14, 38, 44, 52). In addition, one particular member of the group, Caulobacter crescentus, has emerged as a prominent model for the study of bacterial cell cycle progression, asymmetric cell division, and organelle development (7, 8, 34, 49). Although factors that regulate the C. crescentus cell cycle appear to respond to changes in metabolic states (3, 22, 35), details of the connection remain nebulous. Investigation of C. crescentus's metabolic capabilities can improve understanding of bacterial cell cycle control as well as the biogeochemical contributions of oligotrophs.
We initiated an examination of C. crescentus's ability to utilize lactose for two reasons. (i) C. crescentus cells predominantly dwell in freshwater ponds, lakes, and rivers (47, 48). While carbon sources from the degradation of plant materials might be relatively abundant in such environments (25), lactose should be a rare nutrient source. Understanding how C. crescentus metabolizes lactose can reveal how it survives under conditions where the availability of any nutrient is capricious and ephemeral. (ii) C. crescentus has native β-galactosidase activity that turns its colonies blue on solid media containing the chromogenic substrate 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-gal) (32, 42). By identifying and deleting genes required for lactose metabolism, we can abolish the basal activity of X-gal hydrolysis, thus creating a genetic background that allows for blue/white screening of colonies. This additional genetic tool would facilitate the study of this model organism.
By mapping a previously isolated transposon mutant, we have identified a gene, lacA, that is essential for lactose utilization in C. crescentus. Deletion of lacA reduces the basal level of inducible β-galactosidase activity, allowing color-based distinction of wild-type and mutant strains on plates containing X-gal. This β-galactosidase activity also contributes to the hydrolysis of o-nitrophenyl-β-d-galactoside (ONPG), another chromogenic analog of lactose, and is sensitive to sodium dodecyl sulfate (SDS) and chloroform. Unexpectedly, lacA encodes a protein with an oxidoreductase domain and homology to the flavin subunit of a gluconate dehydrogenase. Sequence evaluation led to the identification of adjacent genes (lacB and lacC) as ones that encode the other two subunits of the dehydrogenase. Analysis using Phenotype MicroArrays revealed that lacA is also involved in the catabolism of three other β-galactosides (lactulose, lactitol, and methyl-β-d-galactoside) and two glucosides (salicin and trehalose). Enzymatic assays suggest that LacA indeed acts as a dehydrogenase, capable of oxidizing lactose, salicin, and trehalose. This dehydrogenase activity is inducible, and elevated expression of lacA in the presence of lactose and salicin, but not trehalose, contributes to the increase in activity. Moreover, lacA is required for the induction of both gene expression and dehydrogenase activity. We found that the lacB and lacC deletion strains exhibited characteristics similar to those of the lacA deletion mutant, in support of the assertion that products of the three genes act in concert as a dehydrogenase. The involvement of a dehydrogenase in the metabolism of varied sugars provides insight into C. crescentus's ability to scavenge for scarce nutrients. Based on a possible lactose degradation pathway in Agrobacterium tumefaciens, we propose that the dehydrogenase catalyzes the formation of 3-keto sugars.
All bacterial strains and plasmids used in this study are listed in Table Table1.1. C. crescentus strains were grown at 30°C in peptone-yeast extract (PYE) complex media (27) or M2 minimal media (47) with appropriate supplements, as previously published (12). When testing utilization, individual carbon sources were present at a final concentration of 0.2%. X-gal and isopropyl-β-d-thiogalactopyranoside (IPTG) were used at 20 μg/ml and 1 mM, respectively. Escherichia coli strains were grown at 37°C in LB media with the appropriate antibiotics (12, 43). Procedures for strain construction were described previously (12). Established protocols were used for (i) generalized transduction, as mediated by ΦCr30 (15); (ii) conjugation, to mobilize plasmids from E. coli (20); and (iii) generation of chromosomal deletions by homologous combination (54).
Standard techniques were used for DNA manipulation, electroporation, and transformation (1). Primers for this study are included in Table S1 in the supplemental material. DNA sequencing was done by Elim Biopharmaceuticals (Hayward, CA). To determine the location of the lac-101::Tn5 insertion, we performed arbitrarily primed PCR according to the methods of O'Toole and Kolter (46), except that primers Tn5 151out and Tn5 117out were used as Tn5-specific primers for the first and second rounds of PCR, respectively. Briefly, primers Tn5 151out and ARB1 were used for amplification from chromosomal DNA, and an aliquot of this first reaction mixture was used as the template for the second reaction, which involved Tn5 117out and ARB2. DNA sequences from the PCR were then matched to the C. crescentus genome with the Basic Local Alignment Search Tool (BLAST) (26).
pJC343 was constructed as follows: region 5′ of CC1634 was amplified with primers CC1634 −497F H and CC1634 14R B and then digested with HindIII and BamHI; region 3′ of CC1634 was amplified with primers CC1634 1722F B and CC1634 2253R E and then digested with BamHI and EcoRI; and the two regions were inserted into pNPTS138 digested with HindIII and EcoRI by triple ligation. pJC345 was constructed by inserting the Ω cassette (4) into the BamHI site of pJC343, between regions 5′ and 3′ of CC1634 (lacA). pJC389 was made by amplifying CC1634 (lacA) with primers CC1634 −34F X and CC1634 +14R S, digesting the fragment with XbaI and SacI, and inserting it into pCM62 digested with the same enzymes. pJC404 was made by amplifying the promoter region 5′ of CC1634 (lacA) with primers CC1634 −480F B and CC1634 49R P, digesting the fragment with BglII and PstI, and inserting it into pJC326 cut with the same enzymes; the fragment contains the first 17 codons of CC1634 (translationally fused to lacZ on the plasmid) as well as 480 bp upstream of the start codon (Fig. (Fig.1A).1A). pJC414 was constructed by amplifying the region 5′ of CC1632 with primers CC1631 591F SpeI and CC1632 5R HindIII and digesting with SpeI and HindIII; amplifying the region 3′ of CC1632 with primers CC1632 315F HindIII and CC1632 + 524R SphI and digesting with HindIII and SphI; and then inserting both fragments into pNPTS138 digested with SpeI and SphI. pJC415 was constructed by amplifying the region upstream of CC1635 with primers CC1634 1231F SphI and CC1635 17R HindIII and digesting with SphI and HindIII; amplifying the downstream region with primers CC1635 531F HindIII and CC1636 399R SpeI and digesting with HindIII and SpeI; and then inserting both fragments into pNPTS138 digested with SpeI and SphI. pJC416 and pJC417 were made by amplifying CC1632 with primers CC1632 −42F SphI and CC1632 +18R XbaI or amplifying CC1635 with primers CC1635 −18F SphI and CC1635 +15R XbaI, cutting the fragments with SphI and XbaI, and inserting each into the corresponding sites of pCM62. pJC418 was constructed by amplifying a 557-bp fragment that contains the entire CC1632 gene and a portion of CC1631 with primers CC1631 904F BglII and CC1632 +15R HindIII, cutting the fragment with BglII and HindIII, and inserting the fragment into pJBZ281 cut with BamHI and HindIII. pJC419 was constructed by amplifying a 581-bp fragment that contains the entire CC1635 gene with primers C1635 −18F BglII and CC1635 +16R PstI, cutting the fragment with BglII and PstI, and inserting the fragment into pJBZ282 cut with BamHI and PstI. pJC418 and pJC419 were constructed such that chromosomal integration of the plasmid places E. coli lacZ (with its own ATG start codon) 14 bp after CC1632 or 10 bp after CC1635.
Phenotype MicroArrays from Biolog were used according to the manufacturer's instructions and as previously described (5, 52), with modifications. Cells were grown overnight in M2 media plus 0.02% glucose and then washed and diluted in M2 media without carbon source, so that optical density at 570 nm (OD570) was 0.02. Each well of the PM1 and PM2A plates was inoculated with 145 μl of this cell suspension, and OD570 values of individual wells were measured with a microtiter plate reader after incubation at 30°C for 48 and 72 h. We performed four independent growth comparisons between wild-type and mutant strains. Consistent differences in OD570 values were recorded as variations in carbon source utilization.
β-Galactosidase activity, as encoded by the E. coli lacZ gene, was measured by the method of Miller (43); cells were grown in the appropriate media to mid-logarithmic phase and harvested for analysis. Each sugar being tested was present in the growth media at a final concentration of 0.2%, except for salicin, which was present at 0.02%. At least three independent triplicates were averaged for each measurement. To determine the endogenous enzymatic activity that hydrolyzes o-nitrophenyl-β-d-galactoside (ONPG) in C. crescentus, we modified Miller's β-galactosidase assay to exclude cell lysis with SDS and chloroform. Cells were grown in M2 plus 0.1% xylose, with or without 0.05% lactose, to late log phase or early stationary phase. They were then resuspended and diluted in M2 media without a carbon source, so that the OD600 of the dilution was approximately 0.08. ONPG was incubated directly with the cell suspension at a final concentration of 0.1 mg/ml at 30°C. If testing inhibition, sodium azide (1 mM), chloramphenicol (1 μg/ml), SDS (0.01%), or chloroform (5%) was added to the reaction mixture. When it turned visibly yellow, the cell suspension was centrifuged, and the supernatant's OD420 and OD550 were measured. Enzymatic activity was calculated according to Miller (43), except the reaction duration was in hours, not minutes. Wild-type NA1000 cells that were induced with lactose usually exhibited 45 to 50 units of activity. At least three independent measurements were averaged for each data point.
Dehydrogenase activity was measured in permeabilized cells spectrophotometrically, similar to methods previously described for membrane-bound gluconate dehydrogenases (41, 59). Cells were grown in M2 minimal media plus 0.2% xylose, with or without an inducer (0.1% lactose or 0.02% salicin), until mid- to late log phase. They were washed and resuspended in M2 to an OD600 of 0.15. For each 1.5-ml reaction mixture, we mixed 1 ml of cell resuspension, 150 μl 0.1 M sodium phosphate buffer (pH 6), 10 μl 10 mM 2,6-dichloroindophenol (DCIP), 10 μl 3 mM phenazine methosulfate (PMS), 150 μl 67 mM sugar (electron donor), 150 μl 1% Triton X-100, and 30 μl water. Reference reaction mixtures contained all components except cells or sugar (replaced with M2 medium or water instead). After incubation at 30°C, the reaction mixture was centrifuged, and the OD600 of the supernatant measured. Enzymatic activity was calculated as the initial rate of DCIP reduction, which resulted in the reaction mixture turning from blue to colorless, using the following formula: 1,000 × (OD600 of reference − OD600 of reaction mixture)/[elapsed time × (OD600 of cell suspension/1.5)], where elapsed time was measured in minutes. Reference reaction mixtures gave similar readings, regardless of whether cells or sugars were omitted. At least three independent measurements were averaged. We found that enzymatic activity was reduced in the absence of Triton X-100, in agreement with previous results indicating that certain detergents can stimulate membrane-associated enzymes (33, 41).
To investigate lactose metabolism in C. crescentus, we took advantage of transposon mutagenesis screens previously conducted by Ely and Croft, which had yielded a mutant unable to use lactose as a carbon source (16, 17). First, to confirm linkage to the mutant phenotype, we transduced the insertion mutation (lac-101::Tn5) into CB15 and its derivative NA1000, two commonly used and sequenced wild-type strains, selecting for kanamycin resistance conferred by the Tn5 transposon. The inability to use lactose was assessed subsequently by streaking the transductants onto M2 minimal media containing lactose as the sole carbon source; the defect in lactose utilization always cotransduced with the resistance marker (data not shown). In addition, all Kmr transductants formed white colonies on PYE rich media containing X-gal, as opposed to their wild-type parents, which formed blue colonies (data not shown). Because X-gal serves as a chromogenic substrate for monitoring the processing of lactose, these results indicated that the Tn5 insertion was indeed in a locus required for lactose catabolism.
Using arbitrarily primed PCR, we determined that the lac-101::Tn5 insertion disrupted CC1634, originally annotated to encode an oxidoreductase in the CB15 genome (45) (Fig. (Fig.1A).1A). In the more recently sequenced NA1000 genome, the same gene is annotated as a dehydrogluconate dehydrogenase (M. E. Marks, C. Teiling, L. Du, V. Kapatral, T. L. Walunas, and S. Crosson, unpublished data). Our independent analysis using BLAST (26) indicated that the CC1634 gene product has homology to the flavoprotein subunit of the biochemically characterized gluconate dehydrogenase from Pectobacterium (formerly Erwinia) cypripedii (59). Like its homolog, the gene product is predicted to be membrane bound and to contain a conserved protein domain that belongs to the widespread glucose-methanol-choline (GMC) oxidoreductase superfamily (10, 39, 60) (Fig. (Fig.1B).1B). Sequence evaluation using MEMSAT3 (28) suggested that at least one transmembrane segment exists in the CC1634 protein. We also found that the N terminus of CC1634 may contain a signal sequence suitable for the twin-arginine translocation (TAT) pathway, which allows export of folded proteins; in particular, substrates of the TAT pathway often consist of redox proteins and those that require incorporation of a cofactor, such as flavin adenine dinucleotide (FAD), for proper folding (13). To ascertain that the gene is involved in lactose utilization, we constructed a null deletion of CC1634. This in-frame deletion produced the same phenotypes as the lac-101::Tn5 mutation; ΔCC1634 mutants were unable to use lactose as a carbon source and formed white colonies on rich media containing X-gal (Fig. (Fig.1C).1C). Thus, CC1634 was named lacA for its involvement in lactose metabolism.
Complementation analysis verified that the lacA deletion was responsible for both the failure to grow on M2 lactose and the white color of colonies on plates with X-gal (Fig. (Fig.2).2). The same results were obtained in both the CB15 and NA1000 backgrounds. A plasmid carrying a copy of the lacA gene allowed the ΔlacA mutants to grow on M2 lactose and form blue colonies. Wild-type strains carrying the vector alone exhibited the same phenotypes, whereas ΔlacA mutants carrying the vector did not grow on M2 lactose and formed white colonies on rich media with X-gal. Finally, introduction of a plasmid carrying the lacZ gene from E. coli allowed the ΔlacA mutants to produce the distinctive blue color from X-gal but did not complement the defect in lactose utilization. This last result is not surprising, considering that E. coli lacZ encodes a β-galactosidase, while C. crescentus lacA appears to encode a distinct enzymatic activity. Because the presence of E. coli β-galactosidase in the C. crescentus cytoplasm should enable the breakdown of lactose into glucose and galactose, both of which can be used by C. crescentus (25, 31, 47), the failure of E. coli lacZ to support the growth of ΔlacA mutants on M2 lactose suggests that lacA is necessary for converting lactose into molecules that can be imported into the cytoplasm. This deduction dovetails with the finding that the lacA gene product shares the predominantly periplasmic membrane topology of P. cypripedii's gluconate dehydrogenase.
The gluconate dehydrogenase of P. cypripedii, as well as related enzymes purified from other bacterial species, consists of three protein subunits: (i) a flavoprotein that contains the primary dehydrogenase activity, (ii) cytochrome c, and (iii) a small subunit with unknown function (41, 59). In P. cypripedii, the three subunits are encoded by adjacent genes in the same operon. In C. crescentus, lacA (CC1634) encodes the flavoprotein subunit, while nearby genes appear to encode the other two subunits (45) (Fig. 1A and B). The gene immediately downstream of lacA, CC1635, whose putative start codon overlaps with the stop codon of lacA, encodes a protein that has homology to the smallest subunit of the P. cypripedii gluconate dehydrogenase (26). Computational analysis indicated that the CC1635 protein also contains a leader peptide that may serve as a twin-arginine translocation signal (13), allowing export into the periplasm. On the other side, two genes upstream of lacA, CC1632 encodes a cytochrome c family protein (39), which is predicted to be membrane bound as well (28). The intervening CC1633 gene encodes a hypothetical protein with a conserved domain of unknown function (39).
Based on sequence homology and their proximity to lacA, we postulated that CC1632 and CC1635 also contribute to lactose metabolism in C. crescentus. Indeed, in-frame deletions of CC1632 (named lacB) and CC1635 (lacC) prevented growth on minimal media containing lactose as the sole carbon source (Fig. (Fig.1C).1C). The ΔlacC mutant, like the ΔlacA mutant, also formed white colonies in the presence of X-gal, while the ΔlacB mutant formed blue colonies, albeit generally less blue than those formed by the wild type (Fig. (Fig.1C).1C). One possible explanation is that other cytochrome c family proteins (such as CC1210, which shares 35% identity with LacB ) can serve as part of the dehydrogenase, but with lower efficiency, such that lactose is processed insufficiently to support growth, whereas X-gal can be processed over longer time periods to turn the colonies blue. No similar substitutes appear to exist for LacA and LacC. lacC does not have close paralogs in the C. crescentus genome, and while there are at least four other genes that encode GMC oxidoreductases (CC0945, CC1278, CC1638, and CC2642), all of them encode proteins that share low homology with LacA (<24% similarity) (26, 45). Moreover, none of the other oxidoreductase genes has the same gene arrangements as those of lacA, lacB, and lacC.
We confirmed that LacB and LacC participate in lactose metabolism in C. crescentus by performing a complementation analysis similar to that described above for LacA (see Fig. S1 in the supplemental material). A plasmid carrying lacB or lacC supported growth of the ΔlacB or ΔlacC mutant, respectively, on lactose. The presence of either lacC or E. coli lacZ allowed the ΔlacC mutant to turn blue on X-gal. Because the ΔlacB mutant already appeared blue on X-gal, the presence of lacB or E. coli lacZ did not have an obvious effect on colony color for this strain. However, as for the ΔlacA mutant, lacZ did not restore lactose utilization in either the ΔlacB or ΔlacC mutant. In conjunction with information about homology and gene location, these phenotypic results suggest that, along with LacA, LacB and LacC form subunits of a dehydrogenase involved in lactose degradation. Since the three proteins appear to be parts of the same enzyme, our subsequent investigation focused primarily on one of the components, LacA.
While performing the complementation analysis, we also noticed that plasmid loss could be detected visually in the ΔlacA and ΔlacC mutants (Fig. (Fig.2;2; also see Fig. S1 in the supplemental material). Specifically, in the presence of tetracycline, which selected for maintenance of the complementing plasmid, ΔlacA strains carrying either lacA or lacZ on the plasmid formed only colonies that were completely blue on PYE rich media (Fig. (Fig.2).2). On the other hand, without tetracycline in the media, plasmid loss led to formation of white or partially white (sectored) colonies (Fig. (Fig.2).2). Wild-type strains carrying the vector or the complementing plasmids produced only blue colonies, regardless of whether tetracycline was present in the media (Fig. (Fig.22 and data not shown). We obtained analogous results with ΔlacC strains (see Fig. S1 in the supplemental material). The sectored colony phenotype seen in the ΔlacA and ΔlacC mutants due to plasmid loss may prove particularly useful in future genetic studies.
As in many other bacteria, the lactose utilization pathway appears to be inducible in C. crescentus (32). For example, we observed colonies that were bluer on X-gal plates containing lactose than on those lacking lactose (data not shown). To assess the role of lacA in lactose utilization and induction, we modified Miller's β-galactosidase assay (43) to measure hydrolysis of ONPG, a colorimetric analog of lactose; cells were grown in M2 minimal media plus xylose with or without lactose, resuspended in M2 without carbon sources, and incubated directly with ONPG. ONPG hydrolysis activity in wild-type cells grown without lactose was about 50% of that in cells induced with lactose (Fig. (Fig.3A).3A). On the other hand, regardless of whether they were induced with lactose, ΔlacA strains exhibited enzymatic activities at about 30% of that in induced lacA+ cells. These results suggest that the enzymes required for lactose catabolism are inducible in C. crescentus. Furthermore, lacA is required for the induction.
Originally, we were puzzled by C. crescentus's ability to hydrolyze X-gal on solid media but not ONPG in Miller's β-galactosidase assay; wild-type cells exhibited minimal activity (<1 Miller unit) in the assay (data not shown). We suspected that the presence of SDS and chloroform in the assay inhibited enzymatic activity in C. crescentus. That was why we modified the β-galactosidase assay by excluding cell lysis, to allow measurement of the ONPG hydrolysis described above. To confirm this suspicion, we tested the effects of different metabolic inhibitors with our modified assay, by adding them to washed cells that were being incubated with ONPG. Indeed, the addition of SDS and chloroform (together or individually) diminished ONPG hydrolysis to the same level as that in the ΔlacA mutants (Fig. (Fig.3B3B and data not shown). In contrast, the addition of sodium azide or chloramphenicol, which inhibits electron transport or protein synthesis, respectively, did not reduce the enzymatic activity significantly (Fig. (Fig.3B).3B). Thus, in C. crescentus, SDS and chloroform appear to be detrimental to endogenous enzymes involved in lactose catabolism. Use of E. coli lacZ as a reporter in traditional β-galactosidase assays has been feasible because there is an insignificant level of intrinsic activity, even in wild-type strains, under these conditions.
Since the lacA gene product belongs to the family of GMC oxidoreductases, members of which catalyze diverse reactions and individually can often act on a number of different substrates (10, 60, 61), we asked if it is involved in the metabolism of other carbon sources. Phenotype MicroArrays (Biolog) were used to compare the growth of wild-type and ΔlacA strains on each of 190 different carbon sources. The lacA deletion prevented utilization of four β-galactopyranosides, lactose, lactulose, lactitol, and methyl-β-d-galactoside (Fig. (Fig.4A).4A). Somewhat surprisingly, ΔlacA mutants were also unable to use salicin, a β-glucopyranoside, and trehalose, an α-glucopyranoside (Fig. (Fig.4).4). However, the mutation did not affect the utilization of other glucosides, such as cellobiose or sucrose. We confirmed these findings by streaking the lac deletion strains on M2 minimal media containing select sugars as the sole carbon source (Fig. (Fig.4B4B and data not shown). When incubated at 30°C, wild-type strains formed visible colonies (1 mm in diameter) on glucose or xylose within 2 to 3 days; on lactose, lactulose, lactitol, or salicin within 5 to 7 days; and on trehalose or methyl-β-d-galactoside within 7 to 10 days. (C. crescentus was unable to use methyl-α-d-galactoside as a sole carbon source.) The ΔlacA strains did not grow if the carbon source present was one of those identified in our Phenotype MicroArray experiments. One notable exception was that the lacA gene was necessary in NA1000 but not in CB15 for growth on salicin. While the CB15 ΔlacA mutant grew on salicin, its colonies were smaller than those of the CB15 wild-type strain. Introduction of a plasmid carrying lacA was able to complement these defects in sugar utilization. Similar results were also obtained with ΔlacB and ΔlacC strains (Fig. (Fig.4B4B and data not shown). Therefore, the lac genes appear to contribute to the catabolism of distinct carbon sources, possibly by serving as a nexus in branching degradation pathways.
Based on sequence homology and the Phenotype MicroArray results, we hypothesized that the three lac gene products act in concert as a dehydrogenase for different sugars. To validate this idea, we assessed enzymatic activity by using protocols previously established for membrane-bound gluconate dehydrogenases (41, 59). Wild-type and lac deletion strains were grown in M2 minimal medium plus xylose, with or without lactose as the inducer. Dehydrogenase activities in harvested cells were then measured spectrophotometrically as the rate of reduction of DCIP, mediated via PMS, in the presence of equimolar concentrations of individual carbon sources (lactose, salicin, trehalose, xylose, or gluconate). If the carbon source served as an appropriate electron donor, electrons flowed from the flavin-dependent dehydrogenase to PMS, a carrier, to the terminal acceptor DCIP, which turned from blue to colorless when it was reduced.
When incubated with lactose or salicin, wild-type cells exhibited the largest increases in dehydrogenase activities, 6- to 8-fold higher in induced versus uninduced cells (Fig. (Fig.5A).5A). Incubation with trehalose led to an intermediate level of increase in activity, approximately 3-fold, when cells were induced, but the induced activity associated with trehalose was relatively low compared to that associated with lactose or salicin (8 versus 29 or 31 units of activity, respectively) (data not shown). Finally, when incubated with gluconate or xylose, wild-type cells, whether induced or uninduced, showed low levels of dehydrogenase activity in this particular assay, similar to those of uninduced wild-type cells incubated with any carbon source (Fig. (Fig.5A5A and data not shown). The inability of gluconate to stimulate DCIP reduction is consistent with a previous observation that C. crescentus cannot utilize gluconate for growth (25). Parallel patterns of activity levels were obtained when cells were induced with salicin instead of lactose (data not shown).
In contrast to wild-type cells, ΔlacA mutants exhibited only basal levels of dehydrogenase activity, regardless of induction or the carbon source being tested (Fig. (Fig.5A5A and data not shown). These basal levels were similar to those seen when no sugar was present in the assay, whether wild-type or mutant strains were used (data not shown). Furthermore, deletion of lacC abolished dehydrogenase activity to the same levels as those seen in the ΔlacA mutants (see Fig. S2 in the supplemental material). While deletion of lacB also reduced enzyme activity, the levels were slightly higher than those observed in ΔlacA and ΔlacC strains and similar to that observed in uninduced wild-type cells (see Fig. S2 in the supplemental material). This slightly higher residual activity in ΔlacB cells may again be attributed to possible substitution by other cytochrome c family proteins (see above). Nevertheless, results from these enzymatic assays suggest that lacA, -B, and -C together encode an inducible dehydrogenase that oxidizes specific carbon sources, such as lactose and salicin, or their derivatives.
We conducted complementation analysis to confirm that lacA encodes specific dehydrogenase activity. Introduction of a plasmid carrying lacA under the control of a constitutive promoter increased the levels of dehydrogenase activity in the ΔlacA mutant, under both uninduced and induced conditions, above corresponding levels in the wild-type strain (Fig. (Fig.5B).5B). The plasmid vector alone had no such effect on enzymatic activity in either the wild-type or mutant strain. Thus, lacA appears to be responsible for the specific dehydrogenase activity detected by our assay. Intriguingly, introduction of the complementing plasmid, which constitutively expressed lacA, into the wild-type strain did not alter activity levels. One possible explanation for the difference in the complementing plasmid's effects on wild-type and mutant strains is that cis elements at the endogenous lacA locus may contribute to the regulation of dehydrogenase activity; this regulation would remain intact in the wild-type strain but be lost in the ΔlacA mutant (see Discussion).
Our results indicated that the lac genes are required for the inducible catabolism of different sugars. To obtain better understanding of how the metabolic pathway is regulated, we examined lacA expression under different conditions. A reporter construct containing E. coli lacZ translationally fused to the first 17 codons of lacA, as well as the 480-bp region upstream (Fig. (Fig.1A),1A), was placed on a plasmid and introduced into C. crescentus to assess variation in expression levels by the use of Miller's β-galactosidase assay. First, wild-type cells carrying the plasmid were grown in different media, with or without lactose for induction (Fig. (Fig.6A).6A). Expression levels were lowest when cells were grown in PYE complex media, and the presence of lactose did not significantly increase expression. Expression levels in M2 minimal media plus glucose (M2G) without the inducer were similar to those in PYE, but the addition of lactose increased expression ~2-fold. Finally, while expression levels in M2 plus xylose (M2X) without lactose were only slightly higher than, if not similar to, uninduced levels in PYE and M2G, the presence of lactose in M2X produced the highest level of expression, five times the basal level in PYE. These results suggest that lacA expression is dictated by a complex hierarchy of nutrient preference; the availability of glucose and other carbon sources in M2G and PYE may limit the degree of induction by lactose.
Having determined that induction is detectable in M2 minimal media but not in PYE rich media, we next assessed the ability of different compounds to serve as inducers (Fig. (Fig.6B).6B). In M2G, salicin led to an almost 4-fold increase in expression, compared to lactose's 2-fold increase. On the other hand, trehalose and IPTG (commonly used as an inducer for the lac operon in E. coli) did not produce any detectable changes in expression (data not shown), whereas addition of xylose to M2G actually reduced the level of expression slightly. Deletion of lacA prevented induction by lactose or salicin but did not alter the reduced level of expression in the presence of xylose (Fig. (Fig.6B).6B). Deletion of lacB or lacC also prevented induction of this PlacA-lacZ reporter construct by lactose (data not shown). Thus, the lac genes are required for lactose and salicin to serve as inducers for promoting lacA expression. In addition, our results indicate that the levels of other nutrients, such as glucose and xylose, can also affect lacA expression, possibly via catabolic repression, as previously suggested (32, 42, 51). Expression of lacB and lacC is likely under similar types of control. When we inserted E. coli lacZ (with its own start codon) into the C. crescentus chromosome immediately after lacB or lacC, we also observed induction of lacZ expression in the presence of lactose, to the same extent as that seen for the PlacA-lacZ reporter (see Fig. S3 in the supplemental material). Like metabolic genes in other systems, expression of the lac genes appears to be under complex regulation to ensure optimal fitness with changing resource availability.
We have determined that, in C. crescentus, lacA, lacB, and lacC are required for the utilization of lactose and three other β-galactosides, as well as salicin and trehalose, two glucosides. They appear to act as subunits of a dehydrogenase that oxidizes these sugars or their derivatives. When cells are grown in the presence of the appropriate inducer, such as lactose, the dehydrogenase activity encoded by the lac genes is elevated, concomitant with a rise in β-galactosidase activity, as detected by ONPG hydrolysis. The increase in dehydrogenase activity is attributable, at least in part, to an increase in transcription in the presence of the inducer. In the lac deletion mutants, neither gene expression nor enzymatic activity can be induced. Thus, the Lac dehydrogenase appears necessary for processing substrates into physiologically relevant inducers inside the cell. Taken together, these results indicate that lacA, -B, and -C encode a critical component of the metabolic pathways that dissimilate lactose and other sugars, and its dehydrogenase activity contributes to the substrate-dependent regulation of these pathways.
Two lines of evidence support the assertion that lacA, -B, and -C encode subunits of a dehydrogenase. First, analysis of gene sequence and arrangement indicated that LacA, -B, and -C are homologous to components of the biochemically characterized gluconate dehydrogenase from P. cypripedii (59). Second, enzymatic assay results suggest that they act as a dehydrogenase that specifically oxidizes lactose, salicin, and, to a lesser extent, trehalose. The addition of these sugars stimulated the ability of permeabilized, induced wild-type cells to reduce DCIP, whereas no such activity was detected in the lac deletion mutants or when other sugars were used. While C. crescentus contains xylose dehydrogenase activity (47, 52), our assay appears to detect only the enzymatic activity encoded by the lac genes.
Results from complementation analysis further bolster LacA's role as a subunit of the dehydrogenase, as the introduction of a plasmid that constitutively expressed lacA into the mutant cells raised enzymatic activity above that of wild-type cells. Confoundingly, the introduction of the same plasmid into wild-type cells did not produce a congruent increase in activity. One possible explanation is that lacA encodes only one of three subunits of the dehydrogenase and expression of all three subunits has to be increased in the correct ratio for enhanced activity. In wild-type cells, all levels of the regulation are intact, preventing excess expression. On the other hand, in the ΔlacA mutant, some requirements of the regulation, such as cis elements within the lacA locus that affect expression of the downstream CC1635 gene, may be lost, allowing an increase in expression of the other subunits and consequent enhancement in enzymatic activity.
Although LacA, -B, and -C apparently work together as a dehydrogenase, the exact function of the enzyme in the utilization of lactose and other sugars remains speculative. In bacteria, lactose metabolism typically involves one of two well-defined pathways (9, 29, 36, 58): (i) lactose may be imported by a specific permease into the cytoplasm and split by β-galactosidase into glucose and galactose, which are converted to glucose-6-phosphate for glycolysis; or (ii) lactose may be taken up by a phosphoenolpyruvate (PEP)-dependent phosphotransferase system (PTS), forming lactose phosphate, which is subsequently hydrolyzed by phospho-β-galactosidase into glucose and d-galactose 6-phosphate. Neither pathway appears to involve a dehydrogenase. However, a dehydrogenase/oxi- doreductase participates in a proposed pathway for lactose degradation in A. tumefaciens (9, 23, 56): (i) hexopyranoside:cytochrome c oxidoreductase catalyzes the oxidation of lactose to generate 3′-ketolactose; (ii) 3-ketolactose hydrolase then breaks down 3′-ketolactose into glucose and 3-keto-β-d-galactose; and (iii) a hypothetical, as-yet-unidentified enzyme converts 3-keto-β-d-galactose to β-d-galactose. While hexopyranoside:cytochrome c oxidoreductase has been purified and characterized, 3-ketolactose hydrolase activity has been only partially defined, and the corresponding genes for these enzymes have not been determined.
Despite the absence of assigned genes for sequence comparison, various pieces of evidence suggest that the Lac dehydrogenase of C. crescentus serves the same function as the hexopyranoside:cytochrome c oxidoreductase of A. tumefaciens (55). First, as for the C. crescentus dehydrogenase, one of the enzymatic assays for the A. tumefaciens oxidoreductase employs DCIP as an electron acceptor. Second, hexopyranoside:cytochrome c oxidoreductase uses FAD as a cofactor. Third, it can oxidize a number of different sugars, including those identified in this study. Fourth, three consecutive genes in A. tumefaciens encode products that are homologous to C. crescentus LacA, -B, and -C: Atu4377 (561 amino acids [aa]) is 56% identical to LacA, Atu4378 (186 aa) is 27% identical to LacC, and Atu4379 (cytochrome c2; 131 aa) is 35% identical to LacB (21, 26, 57). Fifth, the combined predicted molecular weight of Atu4377 (63,350) and Atu4378 (20,085) is similar to that of purified hexopyranoside:cytochrome c oxidoreductase (85,000 ± 7,700 without cytochrome c) (55). Therefore, homologs of LacA, -B, and -C in A. tumefaciens likely form subunits of the hexopyranoside:cytochrome c oxidoreductase. We propose that C. crescentus uses the same pathway as that prescribed for A. tumefaciens to degrade lactose. Other sugars are similarly converted to their corresponding 3-keto derivatives and subsequently hydrolyzed, as suggested for a candidate sucrose degradation pathway in A. tumefaciens (9, 50).
While the model above is tantalizing, questions remain. For example, how is 3′-ketolactose transported into the cytoplasm, and do the other enzymes in the proposed pathway and their corresponding genes actually exist in C. crescentus? Identification of additional genes required for utilization of lactose and other sugars would help clarify the pathway. One approach is to exploit the genetic difference between the CB15 and NA1000 strains. Deletion of the lac genes prevented NA1000 from growing on salicin as the sole carbon source, whereas it only weakened growth of CB15. The genomes of both strains have been completely sequenced, and a limited number of genetic variations exist between the two (45; Marks et al., unpublished). Future analysis of these genetic variations should reveal which is responsible for the phenotypic difference between CB15 and NA1000 Δlac strains when grown on salicin. This and other prospective means of identifying more genes involved, such as additional transposon mutagenesis screens, will help clarify the biochemical pathways for catabolism of salicin, lactose, and other sugars.
The ecological niche that Caulobacter occupies may explain why the Lac dehydrogenase is involved in the catabolism of different sugars as well as why the induction of its expression and enzymatic activity is relatively modest. First, Caulobacter may have developed enzymes with versatile substrate specificities because it lives in low-nutrient-concentration environments, and the ability of a single gene to play multiple roles would be an economical way to digest the diverse and rare nutrients that it encounters. Intriguingly, a number of sugar oxidases and dehydrogenases, from plant-associated fungal species in particular, have been characterized biochemically, and their ability to oxidize cellobiose as well as lactose draws tantalizing parallels with LacA's involvement in the utilization of salicin and lactose (2, 24, 30, 37, 61). Like LacA, fungal cellobiose dehydrogenases contain a flavoprotein domain of the GMC oxidoreductase superfamily; in addition, they contain a cytochrome domain due to an ancestral gene fusion event (60, 61). Caulobacter and fungal species may have acquired enzymes with similar functions and substrate versatility because they both thrive in ecological niches where plant-derived carbon sources, such as salicin and cellobiose, are readily available. The same logic applies to A. tumefaciens, a plant pathogen.
Second, although induction of lac expression and dehydrogenase activity is clearly observable in C. crescentus, the level of induction appears modest in comparison to other metabolic systems, both in other bacteria and in C. crescentus, including that used for xylose catabolism (the xyl system) (25, 42, 53). Poindexter (48) has noted that, for the majority of sugar substrates, the quantitative response of the Caulobacter cell to the sudden availability of a new substrate is not impressive and has suggested that, in dilute environments where a nutrient is likely to be available for only a short time at irregular intervals, dramatic shifts to dedicated pathways may be maladaptive; low-level maintenance of different catabolic systems at all times would be more advantageous. This type of physiological response may be particularly applicable to nutrients such as lactose that are rarer and considered secondary by Caulobacter. This environmental adaptation would also help explain the catabolic repression of lacA transcription; Caulobacter would not invest in enhancement of pathways for other carbon sources if favored nutrients are present, yet there is no complete shutdown of expression. Tempered responses to the presence of particular nutrients may have evolved as a consequence of Caulobacter's natural environment. Unraveling the complex regulation of such responses would help elucidate how an oligotroph controls its metabolic flux to achieve optimal fitness. One avenue of investigation would be identification of the transcriptional factors that control lac expression. Inspection of genomic sequences surrounding lacA, -B, and -C readily yielded potential regulators; CC1627 encodes a regulator of the LacI family, while CC1640 encodes a regulator of the BlaI family. Characterization of these and other potential regulators of lac expression will further our understanding of how metabolic capacity is controlled in C. crescentus.
Finally, while regulation of lac expression and the metabolic role of the Lac dehydrogenase await further investigation, identification of the gene has immediate applicability for genetic studies of C. crescentus. By removing the background level of X-gal hydrolysis, the ΔlacA mutation facilitates use of visual assays, such as blue/white screening on solid media (43). For example, mutants with enhanced expression of a lacZ reporter construct can be detected as colonies that turn bluer on X-gal plates than do those of the parental strain. We have also used the visual observation of plasmid loss to confirm that specific genes are essential for viability in C. crescentus. Figure Figure22 illustrates such a visual test of growth requirement. In the presence of tetracycline, the plasmid-borne tetA gene, which confers resistance to the antibiotic, is required for growth. Hence, all colonies are completely blue because, to be viable, cells need to maintain the plasmid that carries both tetA and lacZ. However, in the absence of selection, cells can lose the plasmid. This simple and inexpensive assay has diverse applications, the extent of which relies solely on the user's imagination.
This work was initiated while J.C.C. was a postdoctoral fellow in Lucy Shapiro's group, a position funded in part by National Institutes of Health (NIH) grants GM32506 and F32 G067472. Partial funding for this work came from San Francisco State University. B.H.A. was supported by an NIH MARC Fellowship (T34-GM08574). J.C.C. is supported by award number SC2GM082318 from the National Institute of General Medical Sciences.
We thank M. R. K. Alley for pointing us to the appropriate strain and references, Bert Ely for providing the lac-101 mutant, and Nathan Hillson for drawing the chemical structures. We are grateful to Sean Crosson, Melissa Marks, and Craig Stephens for communicating protocols and unpublished results. Nathan Hillson and Craig Stephens also provided helpful comments.
Published ahead of print on 26 February 2010.
†Supplemental material for this article may be found at http://aem.asm.org/.
‡Dedicated to the memory of Mei-Long H. Chen.