Correspondence: Gavin H. Thomas, Department of Biology (Area 10), University of York, PO Box 373, York YO10 5DD, UK. Tel.: +44 1904 328678; fax: +44 1904 328825; e-mail: firstname.lastname@example.org
The ability to use the sialic acid, N-acetylneuraminic acid, Neu5Ac, as a nutrient has been characterized in a number of bacteria, most of which are human pathogens that encounter this molecule because of its presence on mucosal surfaces. The soil bacterium Corynebacterium glutamicum also has a full complement of genes for sialic acid catabolism, and we demonstrate that it can use Neu5Ac as a sole source of carbon and energy and isolate mutants with a much reduced growth lag on Neu5Ac. Disruption of the cg2937 gene, encoding a component of a predicted sialic acid-specific ABC transporter, results in a complete loss of growth of C. glutamicum on Neu5Ac and also a complete loss of [14C]-Neu5Ac uptake into cells. Uptake of [14C]-Neu5Ac is induced by pregrowth on Neu5Ac, but the additional presence of glucose prevents this induction. The demonstration that a member of the Actinobacteria can transport and catabolize Neu5Ac efficiently suggests that sialic acid metabolism has a physiological role in the soil environment.
Bacteria that live in complex and changing environments have often evolved to utilize a wide range of potential nutrients that they are likely to encounter in their particular biological niche. For a range of human pathogens, the ability to utilize the sialic acids has received attention and is important for colonization and pathogenesis in many cases (Vimr et al., 2004; Severi et al., 2007; Almagro-Moreno & Boyd, 2009). Sialic acids are related 9-carbon nonulosonic acids that have important cellular functions in deuterostome animals, and the most common of these is N-acetylneuraminic acid (Neu5Ac or NANA) (Angata & Varki, 2002; Schauer, 2004). Many bacteria produce sialidases (also known as neuraminidases), which are secreted, and cleave off sialic acids from host cell surfaces and from the surfaces of other bacteria (Corfield, 1992). After release of sialic acids, some bacteria also then take up the Neu5Ac and use a series of catabolic enzymes to convert it to fructose-6-phosphate, pyruvate and ammonia (Vimr & Troy, 1985; Vimr et al., 2004). Some pathogens such as Haemophilus influenzae also use the transported sialic acid to decorate their own cell surface, which is an important mechanism for their persistence in the body (Bouchet et al., 2003).
Corynebacterium glutamicum is a Gram-positive, nonmotile bacterium that belongs to the phylum Actinobacteria. It was first isolated from soil in 1975 during a screen for glutamate-producing bacteria (Kinoshita et al., 1957). Because of its ability to produce high levels of glutamate and lysine, it has become a widely used organism in industrial biotechnology (Kumagai, 2000). Every year around 1.5 million tons of l-glutamate and 0.75 million tons of l-lysine are produced commercially using C. glutamicum (Kelle et al., 2005; Kimura, 2005). Besides glucose as a sole carbon source, it is able to utilize a wide range of other carbon sources, such as fructose, sucrose, gluconate, acetate, propionate, pyruvate, l-lactate and ethanol as well as the amino acids glutamate and serine (Cocaign et al., 1993; Peters-Wendisch et al., 1998; Claes et al., 2002; Netzer et al., 2004). The C. glutamicum ATCC 13032 genome is around 3.3 Mb and encodes metabolic pathways for utilization of a range of sugars, many of which have been well studied in relation to providing high outputs of l-amino acids (Kalinowski et al., 2003).
A recent phenotype array study of Rhodococcus opacus PD630, which included C. glutamicum ATC 13032 as a control organism, revealed that Neu5Ac can support growth of C. glutamicum. Upon further investigation, it appears that C. glutamicum has a potential set of genes that would allow it to transport and catabolize Neu5Ac as a sole carbon source (Holder et al., 2011). As sialic acid utilization is normally associated with animal commensal or pathogenic bacteria and the presence of these genes has not been detected in other recent analysis of sialic acid utilization genes in bacteria (Almagro-Moreno & Boyd, 2009), we wished to verify this novel finding and identify the gene(s) responsible for sialic acid uptake into this soil-dwelling actinobacterium.
Materials and methods
Bacterial strains, media and growth conditions
Escherichia coli DH5α was grown aerobically in 37 °C in Luria–Bertani medium. Corynebacterium glutamicum ATCC 13032 was cultivated aerobically at 30 °C in complex brain–heart infusion medium (BHI; Difco Laboratories) or in minimal CGXII medium (Elleling & Reyes, 2005), supplemented with 1% (w/v) glucose or other carbon sources as indicated. Growth of C. glutamicum was monitored at 600 nm. Kanamycin was added to culture when required at 25 μg mL−1 for C. glutamicum or 30 μg mL−1 for E. coli.
Growth experiments in liquid and on plate
For liquid growth experiments with C. glutamicum, cells from starter cultures grown during the day in 5 mL of BHI medium were used to inoculate 10 mL of CGXII media supplemented with 1% (w/v) glucose for overnight growth. The overnight cultures were diluted to an OD600 of c. 1 in 10 mL CGXII medium supplemented with the appropriate carbon source, which for sialic acid (Neu5Ac) was 0.5% (w/v) unless indicated otherwise. The cultures were incubated under shaking at 30 °C, and growth was monitored every hour. For growth experiments on agar plates, cells were precultured as described previously and diluted to an OD600 nm of 0.5 in CGXII medium. From this, a twofold dilution series in CGXII was prepared and 2 μL of each dilution spotted onto CGXII agar plates containing the appropriate carbon source. Agar plates were incubated for 24–48 h at 30 °C.
Bacterial strains, plasmids and mutagenesis
For gene inactivation in C. glutamicum, a vector integration method was used (Elleling & Reyes, 2005; Jolkver et al., 2009). To produce an insertion in cg2937, an c. 500 bp fragment of the target gene was amplified by PCR using the oligonucleotides 5′-ACTCGCCGCAATTTCCTCCG-3′ and 5′-CGAGGCGTTCGCTGATGATG-3′ and cloned into the pDRIVE vector (Qiagen), which was verified by PCR. Transformation into C. glutamicum was performed using electroporation (2.5 kV, 5 ms), and positive colonies were selected on kanamycin-containing agar plates, and chromosomal integration was confirmed by PCR using primers binding c. 100 bp before the start codon (5′-GTCTGATGTCTGATGTATAT-3′) and the appropriate M13 primer for the pDRIVE vector (5′-AACAGCTATGACCATG-3′).
[14C] Neu5Ac uptake assays
Cells were precultured as described previously for the growth experiment unless annotated otherwise. Cells were washed three times in CGXII media containing no carbon source. Cells were diluted to an OD600 nm of 1 in CGXII media supplemented with 10 mM glucose and incubated on ice until the uptake assay was performed. Before the measurement, cells were incubated under stirring at 30 °C for energizing. The assay was started by adding 2 μM [14C] Neu5Ac. Every minute over 5 min, 100-μL samples were taken and cells were collected by rapid filtration (0.45 μm pore size; Millipore). Cells were washed with 5 mL CGXII medium, and the radioactivity retained on the filter discs was determined by liquid scintillation counting. The number of colony-forming units in each culture was also determined using serial dilutions onto BHI plates and incubation at 30 °C overnight followed by counting.
We identified orthologues of known sialic acid catabolism genes from E. coli using the ncbi blastp and identified similar sialic acid clusters across the Corynebacteriaciae using XBase (Chaudhuri et al., 2008) and the SEED (Overbeek et al., 2005).
Corynebacterium glutamicum can grow on sialic acid as a sole carbon source
To verify the initial phenotype array data, which suggested that C. glutamicum ATCC 13032 could grow on Neu5Ac (Holder et al., 2011), we grew the same strain on a minimal CGXII medium with 0.5 % Neu5Ac as the sole carbon source. No growth was observed in the absence of an added carbon source, but there was clear growth with Neu5Ac, albeit with a long lag phase compared with growth on glucose (Fig. 1a solid symbols). After 24 h growth, the final growth yield was very high with glucose, giving an OD600 nm of c. 17, while the value for Neu5Ac was c. 7. Hence, Neu5Ac can clearly serve as a sole carbon source for C. glutamicum.
Isolation of mutant strains of C. glutamicum with reduced lag on sialic acid
A frequently used method to improve growth of C. glutamicum on a particular metabolite is through selection of fast-growing mutant after serial subculture (Youn et al., 2008). Three cultures of C. glutamicum ATCC 13032 were grown in CGXII medium with 0.5% Neu5Ac and serially subcultured when each culture had reached an OD600 nm of at least 4. This was continued for each culture over 15 days, which was between 8 and 12 subculturing steps. The three resulting evolved strains, Ev1-3, all showed significantly reduced lag phases for growth on Neu5Ac (Fig. 1a open symbols and Supporting Information, Fig. S1), although the final growth yield with 0.5% Neu5Ac is similar to the wild-type strain (Fig. 1b). We investigated the concentration dependence of sialic acid growth in one of these strains, Ev1, and see a quantitative relationship between the starting Neu5Ac concentration and the final growth yield (Fig. 1d). During growth on 0.25% Neu5Ac, growth stops after around 9 h, presumably as the Neu5Ac has been consumed during growth (Fig. 1c).
To check the stability of the evolved strains, we subcultured them on BHI medium and then from this further cultured them on CGXII with 1% glucose and then back onto CGXII 0.5% Neu5Ac, upon which the reduced lag phase observed initially was retained (data not shown). The decreased lag but unaltered growth properties suggests that the regulation of expression of the Neu5Ac uptake/catabolic genes is altered in these mutants.
Sialic acid uptake and catabolism is induced by growth on sialic acid
As there was a considerable lag in growth of the wild-type strain when pregrown in CGXII glucose media, we examined the effects of different pregrowth conditions for growth on CGXII Neu5Ac for both the wild-type strain and also for the Ev1 strain. Pregrowth of the wild-type strain in CGXII Neu5Ac yielded a reduced lag phase compared with pregrowth on CGXII glucose (Fig. 2a). In contrast, pregrowth in CGXII medium containing both Neu5Ac and glucose gave a similar growth lag as seen with glucose alone, suggesting that the presence of glucose has a dominant effect over the presence of Neu5Ac (Fig. 2a). When examining the potential of cells pregrown in the same conditions to take up [14C]-Neu5Ac, it is clear that uptake is only detectable in the cells that have been pregrown in CGXII Neu5Ac (Fig. 2c).
In contrast to the wild-type strain, the Ev1 strain exhibited similar growth lags on CGXII Neu5Ac, regardless of how the cells had been grown, suggesting that the repressive effect of glucose on expression of the sialic acid utilization genes was lost (Fig. 2a).
The cg2937 gene is essential for sialic acid utilization
The sialic acid cluster in C. glutamicum contains a likely ABC transporter for sialic acid, which is homologous to the SatABCD systems from Gram-negative Gammaproteobacterium Haemophilus ducreyi (Post et al., 2005). To test whether this system is also important in C. glutamicum, we constructed a disruption of cg2937, which encodes the substrate-binding protein of the transporter, using the pDRIVE system (Qiagen) in both the wild-type strain and the Ev1 background. Analysis of growth of the parental strain, Ev1 and their respective cg2937 disruptions in CGXII Neu5Ac medium, revealed that disruption of cg2937 results in a complete loss of growth (Fig. 3a and b). The same phenotype was observed on solid media (Fig. 3d). We examined [14C]-Neu5Ac uptake using Ev1 and Ev1 cg2937::pDRIVE where uptake was also completely abolished in the strain disrupted in cg2937 (Fig. 3c). Hence, we conclude that the cg2937-40 genes encode the sole sialic acid transporter in C. glutamicum.
Analysis of putative sialic acid utilization gene clusters in the Corynebacteria
Given the clear demonstration that the soil bacterium C. glutamicum has the ability to grow on sialic acid, we examined the distribution of the sialic acid transport and utilization genes within the genus Corynebacterium (Fig. 4). It is clear that the sialic acid genes are not unique to C. glutamicum, but are present in a number of other members of the genus Corynebacterium particularly in organisms that cause diseases in human and animals where genome sequences are available such as Corynebacterium diphtheriae (Cerdeno-Tarraga et al., 2003), Corynebacterium ulcerans (Trost et al., 2011) and Corynebacterium pseudotuberculosis (Trost et al., 2010). In every case, they have a SatABCD-like sialic acid transporter and the full set of genes needed for catabolism, namely nanA, nanE, nanK, nagA and nagB.
While C. glutamicum, C. diphtheriae, C. pseudotuberculosis and C. ulcerans all encode a sialidase on their genome, the predicted sialidase in C. glutamicum (cg2935) is the only one encoded within the main nan-cluster and is not a clear orthologue of the nanH sialidase seen in the other three organisms (marked as nanH in Fig. 4).
Sialic acid utilization has been well studied in a range of pathogens, and in this work, we demonstrate clearly that the soil bacterium C. glutamicum can transport and utilize Neu5Ac as a sole carbon source. Examination of the genome reveals what appears to be a fairly canonical sialic acid cluster containing a full set of genes including an ABC transporter that we have demonstrated is essential for uptake (Fig. 4). It is not clear why the presence of sialic acid utilization genes was not recognized in a previous study (Almagro-Moreno & Boyd, 2009), looking at the distribution of the nanAEK genes in bacteria.
The only member of the genus Corynebacterium, where sialic acid biology has been previously studied, is in C. diphtheriae. A sialidase was first isolated from this pathogen in 1963 (Warren & Spearing, 1963; Moriyama & Barksdale, 1967) and, remarkably, NanA activity was also identified shortly afterwards (Arden et al., 1972). Interestingly, the same study demonstrated that both sialidase and NanA (N-acetylneuraminate lyase) activities were also observed in C. ulcerans and Corynebacterium ovis (now C. pseudotuberculosis; Arden et al., 1972), which agrees with the presence of both nanH and nanA genes in all of these sequenced genomes (Fig. 4). More recently, the NanH sialidase from C. diphtheriae KCTC3075 has been characterized (Kim et al., 2010), which is the orthologue of DIP0543 C. diphtheriae NCTC 13129 (Fig. 4), confirming that this organism has both normal sialidase but also a reported trans-sialidase activity (Mattos-Guaraldi et al., 1998). Other data suggested that C. diphtheriae harbours sialic acid on its cell envelope (Mattos-Guaraldi et al., 1999). This could originate from the trans-sialidase activity moving sialic acid directly from host sialoglycans onto the bacterium's surface. Both the lack of association of the sialidase with production of the diphtheria toxin (Warren & Spearing, 1963; Moriyama & Barksdale, 1967) and the lack of a need for uptake for cell surface modification were perhaps the reason why no study has ever examined the capability of C. diphtheriae to use sialic acid as a nutrient in vivo, which this study would suggest it is capable of.
While the identity of a sialidase in C. diphtheriae has been known for nearly 50 years, the presence of a sialidase in C. glutamicum has not been suspected. This enzyme is not orthologous to the sialidases seen in C. diphtheriae, C. ulcerans and C. pseudotuberculosis, but rather is most similar to Arthrobacter sp. The essential nature of the ABC transporter in the same gene cluster as the sialidase (cg2935) suggests that Cg2935 is a sialidase; however, this will need experimental confirmation.
This study presents the first evidence for an active transporter for Neu5Ac in an actinobacterium and increases the range of bacteria that appear to use an ABC transporter for this purpose. Other bacteria where sialic acid transporters have been characterized use tripartite ATP-independent periplasmic (TRAP) transporters (Severi et al., 2005; Mulligan et al., 2009, 2012; Chowdhury et al., 2012), classical secondary transporters of the major facilitator superfamily (Martinez et al., 1995; Mulligan et al., 2012) or sodium solute symporter family (SSS) transporters (Severi et al., 2010), although perhaps significantly all these are Gram-negative bacteria. The only clear work on sialic acid transport in Gram-positives are from Streptococcus pneumoniae, which also uses an ABC transporter (Marion et al., 2011a, b).
The reduced growth lag when cells are precultured in the presence of sialic acid suggests either the presence of a sialic acid-specific activator or the inactivation of a repressor. Given the presence of a GntR-family transcription factor in the cluster (cg2936), it is probable that the presence of Neu5Ac or one of its catabolic product acts as the ligand to cause depression of the cluster. The additional observation is that derepression by Neu5Ac is not seen in the presence of glucose, suggesting a catabolic repression-type mechanism is in operation. The mechanisms of catabolite repression are not well studied in C. glutamicum, and in many cases, different carbon sources are co-metabolized. A known exception is the repression of ethanol metabolism by glucose (Arndt & Eikmanns, 2007). The global regulator GlxR, which controls expression of some catabolic operons, does not appear to have a binding site in or around the sialic acid cluster (Kohl et al., 2008; Toyoda et al., 2011), making the mechanism of glucose repression unclear.
We propose two potential explanations for the ability of C. glutamicum to use sialic acid so well as a nutrient. The first is that these genes are evolutionary remnants of a previous life of this bacterium in close association with a mucosal surface, the type of environments where the vast majority of bacteria that use sialic acids live. We consider this explanation a weak one as unless this association was very recent, the genes would not be intact and would have pseudogenized or been removed from the genome. It is also clear that the clusters in the pathogenic Corynebacteria are slightly differently organized to those in the soil bacterium C. glutamicum, suggesting some active gene transfer within the soil niche and suggesting a positive selection for the retention and regulation of these genes. The second explanation is that sialic acid is actually an important source of nutrients in the soil. This is supported by the fact that sialidases have in fact been characterized from other nonpathogenic soil Actinobacteria such as Micromonospora viridifaciens and Arthrobacter ureafaciens (Saito et al., 1979; Gaskell et al., 1995), but these have only been studied biochemically and structurally with no analysis on their physiological role in these environments. This study demonstrates that a nonpathogenic soil-dwelling, sialidase-positive actinobacterium can use sialic acid efficiently as a nutrient. The soil is a highly variable and complex environment and one could imagine potential sources of sialic acid and other nonulonosinic acids from other organisms in this niche such as Aspergillus sp., which are known to have sialic acids on their surface (Wasylnka et al., 2001) and perhaps other fungi in this niche, and so we favour this explanation being more likely. Also, soil bacteria are likely to encounter patches of rich organic material like decomposing animals, which would also contain sialic acid.
We would like to thank Dr Jason Holder for sharing unpublished data and the BBSRC for support on our research on bacterial sialic acid transporters.