Carotenoid biosynthesis and overproduction in Corynebacterium glutamicum

Background Corynebacterium glutamicum contains the glycosylated C50 carotenoid decaprenoxanthin as yellow pigment. Starting from isopentenyl pyrophosphate, which is generated in the non-mevalonate pathway, decaprenoxanthin is synthesized via the intermediates farnesyl pyrophosphate, geranylgeranyl pyrophosphate, lycopene and flavuxanthin. Results Here, we showed that the genes of the carotenoid gene cluster crtE-cg0722-crtBIYeYfEb are co-transcribed and characterized defined gene deletion mutants. Gene deletion analysis revealed that crtI, crtEb, and crtYeYf, respectively, code for the only phytoene desaturase, lycopene elongase, and carotenoid C45/C50 ɛ-cyclase, respectively. However, the genome of C. glutamicum also encodes a second carotenoid gene cluster comprising crtB2I2-1/2 shown to be co-transcribed, as well. Ectopic expression of crtB2 could compensate for the lack of phytoene synthase CrtB in C. glutamicum ΔcrtB, thus, C. glutamicum possesses two functional phytoene synthases, namely CrtB and CrtB2. Genetic evidence for a crtI2-1/2 encoded phytoene desaturase could not be obtained since plasmid-borne expression of crtI2-1/2 did not compensate for the lack of phytoene desaturase CrtI in C. glutamicum ΔcrtI. The potential of C. glutamicum to overproduce carotenoids was estimated with lycopene as example. Deletion of the gene crtEb prevented conversion of lycopene to decaprenoxanthin and entailed accumulation of lycopene to 0.03 ± 0.01 mg/g cell dry weight (CDW). When the genes crtE, crtB and crtI for conversion of geranylgeranyl pyrophosphate to lycopene were overexpressed in C. glutamicum ΔcrtEb intensely red-pigmented cells and an 80 fold increased lycopene content of 2.4 ± 0.3 mg/g CDW were obtained. Conclusion C. glutamicum possesses a certain degree of redundancy in the biosynthesis of the C50 carotenoid decaprenoxanthin as it possesses two functional phytoene synthase genes. Already metabolic engineering of only the terminal reactions leading to lycopene resulted in considerable lycopene production indicating that C. glutamicum may serve as a potential host for carotenoid production.


Background
Carotenoids are yellow to red colored pigments originating from the terpenoid biosynthetic pathway. They are very abundant in plants and microorganisms, where they can have diverse functions such as photo protection or light harvesting molecules or as membrane stabilizers [1,2]. In the biosynthetic pathways of certain hormones (like retinoic acid, a hormone regulating the epidermal growth of mammals) they serve as precursors [3]. Carotenoids are also proposed to prevent cancer and reduce the risk of cardiovascular and Alzheimer disease due to their antioxidative properties [4][5][6]. Traditionally, terpenoids have been used in the feed, food and nutraceutical industries [1]. As the large-scale chemical synthesis of terpenoids is often difficult and/or costly due to their structural complexity [7] and as their isolation from natural sources usually does not yield sufficient quantities [8], microbial production processes offer a promising alternative.
Carotenoids are derived from the universal precursor isopentenyl pyrophosphate (IPP) and its isomer dimethylallyl pyrophosphate (DMPP) [9]. Enhancing cellular metabolic flux toward IPP and DMAPP is one strategy to improve rates and yield of microbial isoprenoid production [10,11]. There are two independent pathways leading to IPP: the mevalonic acid (MVA) pathway and the methylerythritol phosphate (MEP) pathway. The MVA pathway is found in eukaryotes (mammals, fungi, in the cytoplasm of plant cells), archaea, and a limited number of bacteria. Most bacteria as well as plant plastides synthesize IPP through the MEP pathway [1,12,13]. The MVA pathway requires acetyl-CoA as the primary educt, whereas the MEP pathway starts by condensation of pyruvate and glyceraldehyde 3-phosphate (GAP) [14,15].
In this study, gene-directed deletion mutagenesis was employed to decipher the functions of the genes present in the main carotenogenic gene cluster of C. glutamicum and in a second cluster encoding putative phytoene synthase and phytoene desaturase paralogs. Moreover, the potential of C. glutamicum to produce carotenoids was estimated by metabolic engineering of the conversion of GGPP to lycopene.
The cluster crtB2/crtI2-1/crtI2-2 has not yet been analyzed. While CrtB and CrtB2 share 49% identity, CrtI2-1 shares 49% identical amino acids with the 364 N-terminal amino acids of CrtI and CrtI2-2 63% identical amino acids with the 104 C-terminal amino acids of CrtI. Thus, it is not clear whether CrtI2-1 and CrtI2-2 function as a twosubunit phytoene desaturase or whether a frameshift mutation disrupted the gene downstream of crtB2.
A comparison of the sequenced genomes of corynebacteria ( Figure 1, Additional file 1: Table S1) revealed that C. glutamicum WT is the only species possessing two crtB and crtI like genes, while the organization of the large gene cluster is comparable in C. glutamicum WT, C. glutamicum R (and ATCC 14067 and S9114) and C. efficiens YS-314. In C. glutamicum R, no crtY e Y f is annotated as likely a G-> T mutation at position 814478 of the C. glutamicum R genome altered the start codon of an open reading frame coding for a protein with 99% amino acid identity to crtY e Y f of C. glutamicum WT to a leucine codon.
A second group of corynebacterial species (e.g. C. diphteriae, C. aurimucosum and C. pseudotuberculosis) only possess the clustered genes crtB and crtI (50 to 55% amino acid identity to the C. glutamicum enzymes; Additional file 1: Table S1). An intermediate situation is found in C. lipophiloflavum, which possesses a gene cluster with crtB, crtI, crtY e/f and crtEb, as well as in C. genitalium possessing crtB, crtI and crtY e/f but lacking crtEb (Additional file 1: Table S1). Members of a third group (C. kroppenstedtii, C. jeikeium, C. urealyticum as examples) also lack crtY e/f and crtEb orthologs, but possess crtB and crtI, however not clustered. Although the overall amino acid sequence identities of the crtB and crtI gene products are below 50% as compared to the respective CrtB and CrtI from C. glutamcium WT, their domain structure includes the crtI domain (TIGR02734) as well as an N-terminal NAD(P)-binding Rossmann-like domain (NCBI Domain structure). As an exception, C. variabile only possesses CrtI with an amino acid identity to CrtI from C. glutamicum WT of 58%.
The phylogeny of the crtI gene product (Additional file 2: Figure S1), which is present in all analysed corynebacteria, is congruent to the grouping of cornyebacterial species with respect to occurrence and clustering of crt genes as shown in Figure 1 and Additional file 1: Table S1.

Analysis of the transcriptional organization of the carotenogenic gene clusters
Annotation of the carotenogenic gene cluster of the C. glutamicum WT for the biosynthesis of decaprenoxanthin from the precursor GGPP suggests co-transcription of crtB, crtI, crtY e and crtY f and crtEb, while the upstream GGPP synthase gene crtE appears to be monocistronic. To characterize the transcriptional organization of this cluster RT-PCR experiments have been carried out. PCR analysis of cDNA synthesized from total RNA of the C. glutamicum WT using primer crtEb-rv (see Additional file 3: Table S2) revealed that the entire gene cluster is co-transcribed since fragments overlapping adjacent genes could be amplified in each case. A cDNA preparation without the addition of reverse transcriptase served as a negative control ( Figure 3).
Similarly, RT-PCR analysis of the small gene cluster revealed that crtB2, crtI2-1 and crtI2-2 are co-transcribed. Figure 3B displays the amplificate of a fragment overlapping crtB2 and crtI2-1 based on cDNA generated by reverse transcription using the crtI-rv primer. To determine the transcriptional start point (TSP) of crtE and crtB2, respectively, RNA was isolated from C. glutamicum WT grown in LB complex medium. By use of 5' RACE_PCR, the TSP of crtE was identified as a guanosine 114 nucleotides upstream of the first nucleotide of the ATG start codon. The three most conserved nucleotides of the consensus −10 hexamer of C. glutamicum promoters [27] can be found in the −15 to −10 region. The −39 to −34 region contains a sequence motif sharing four identical nucleotides to −35 consensus. The TSP of crtB2 was determined as a guanosin thirteen nucleotides upstream of the first nucleotide of the start codon GTG. The hexamer TAAAGT at position −13 to −8 relative to the TSP matches the three most conserved bases of the TANANT consensus sequence of the −10 region of C. glutamicum promoters [27]. At position −32 to −27 the hexamer TTGTCT was found, which resembles the key recognition motif for the −35 region of C. glutamicum promoters TTGNCA [27].

Gene deletion and complementation analysis of the carotenogenic gene clusters in C. glutamicum
Gene-directed deletion mutants of C. glutamicum WT lacking crtB, crtI, crtEb, or crtY e Y f were constructed and characterized regarding carotenoid production. Besides the single deletion mutants, strain C. glutamicum ΔΔ lacking crtB, crtI, crtEb, and crtY e Y f as well as the putative paralogs crtB2, crtI2-1 and crtI2-2 was constructed. All strains showed growth rates of about 0.35 h -1 in CGXII minimal medium with 100 mM glucose as carbon source. Thus, growth was comparable to C. glutamicum WT. However, pigment accumulation differed between the various strains ( Figure 2). The different composition of carotenoids in the cell extracts could be demonstrated by HPLC analyses (Additional file 4: Figure S2, Additional file 5: Figure S3, Additional file 6: S4 and data not shown). The spectrophotometric analysis of the methanolic cell extracts of the C. glutamicum WT showed the characteristic absorption maxima at 415, 440, and 470 nm for the yellow Figure 2 Carotenoid biosynthesis in C. glutamicum ATCC 13032 and gene deletion and complementation analysis of carotenogenic genes. Cell pellets of C. glutamicum deletion mutants lacking one of the carotenogenic genes crtB, crtI, crtEb or crtY e Y f and the wild type and the corresponding complemented strains (right, EV: empty vector). The cells were grown in 50 ml CGXII medium with 100 mM glucose, inoculated to an OD 600 of 1 with a BHI overnight culture. The overexpression was induced at the beginning of the cultivation with 1 mM IPTG.
The multiple deletion strain C. glutamicum ΔΔ (Additional file 3: Table S2) was used for stepwise reconstruction of the decaprenoxanthin biosynthetic pathway. Expression of crtB and crtI in the white strain C. glutamicum ΔΔ entailed a pale pink cell color and accumulation of lycopene was observed in cell extracts. Additional expression of crtEb entailed an orange cell color and accumulation of flavuxanthin. When crtY e Y f was expressed additionally, a color comparable to that of the wild type was observed and the HPLC chromatograms of the cell extracts were comparable to those of the wild type. Thus, expression of crtB, crtI, crtEb, crtY e and crtY f in the multiple deletion strain was sufficient to allow for decaprenoxanthin biosynthesis.
This finding was supported by analysis of the single gene deletion strains. Each deletion mutant could be complemented by ectopic expression of the respective gene deleted in the chromosome (Figure 2). The mutant ΔcrtY lacking the final reaction in the synthesis of decaprenoxanthin, i.e. introduction of two ε-ionone groups into the acyclic flavuxanthin catalyzed by gene products of crtY e Y f , accumulated flavuxanthin and exhibited a pale orange to red color. In the absence of the penultimate enzyme reaction of decaprenoxanthin biosynthesis, i.e. prenylation of lycopene to flavuxanthin by lycopene elongase, in the mutant ΔcrtEb, lycopene accumulated and neither flavuxanthin nor decaprenoxanthin were observed (HPLC analysis of cell extracts not shown). Accordingly, mutants ΔcrtB lacking phytoene synthase and ΔcrtI lacking phytoene desaturase showed white cell color and ΔcrtI accumulated phytoene, which absorbs light at wavelengths below 300 nm. Taken together, our gene deletion and complementation analysis corroborates previous biochemical and transposon mutagenesis data and results from heterologous gene expression regarding the functions of the enzymes encoded by crtB, crtI, crtEb, crtY e and crtY f .
The function of the putative crtB paralogous gene crtB2 and of the putative crtI paralogous genes crtI2-1 and crtI2-2 has not yet been analyzed. As hardly any phytoene was detectable in ΔcrtB, but faint quantities of other carotenogenic intermediates were observed, CrtB appears to be the major phytoene synthase active under the chosen conditions. Similarly, the lack of the red chromophore lycopene in ΔcrtI indicated that CrtI is the only active phytoene desaturase. By contrast, a deletion mutant lacking the paralogous genes crtB2, crtI2-1 and crtI2-2 showed the same yellow phenotype as C. glutamicum WT and the cell extracts showed the identical elution pattern in the HPLC analysis. Moreover, deletion mutants lacking crtB2I2-1I2-2 and either crtB or crtI grew like wild type and showed the same white phenotype as the crtB and crtI single deletion mutants. Thus, the paralogous genes annotated as crtB2 and crtI2-1 and crtI2-2 are either not functional or not expressed (enough) under the chosen conditions. Complementation analysis of the deletion mutants ΔcrtB and ΔcrtI was chosen to test whether crtB2 and/or crtI2-1/2 encode functional enzymes. Overexpression of crtB2 almost completely complemented the crtB deletion and as HPLC analysis of extracts from C. glutamicum ΔcrtB(pEKEx3-crtB2) indicated accumulation of decaprenoxanthin crtB2 encodes a functional phytoene synthase (Figure 2, Additional file 5: Figure S3). By contrast, overexpression of crtI2-1/2 in C. glutamicum ΔcrtI did not restore the wild-type phenotype while overexpression of crtI did. Furthermore, while combined expression of crtB2 and crtI in C. glutamicum strain ΔΔ led to an accumulation of lycopene, the combined expression of crtB2 and crtI2-1/2 did not (Additional file 6: Figure S4). Thus, whereas no evidence for crtI2-1/2 encoding a phytoene desaturase was found, crtB2 encodes an enzyme active as phytoene synthase.
Enhancing lycopene production by overexpression of carotenogenic genes in the lycopene accumulating strain C. glutamicum ΔcrtEb The deletion of the gene crtEb entailed accumulation of lycopene to 0.03 ± 0.01 mg/g CDW in C. glutamicum. To enhance the production of lycopene we focused on improving conversion of GGPP to lycopene. Overexpression of the phytoene synthase gene crtB and/or the phytoene desaturase gene crtI in C. glutamicum ΔcrtEb (Additional file 3: Table S1) was tested. Whereas crtI overexpression showed no effect on lycopene production (0.02 ± 0.01 mg/g CDW), it could be shown that lycopene accumulation was increased two-fold when crtB was overexpressed (0.06 ± 0.01 mg/g CDW, Figure 4). However, combined overexpression of both genes did not increase the lycopene content significantly (0.04 ± 0.01 mg/g CDW).
Besides overexpression of crtB, also overexpression of crtE which codes for the geranylgeranyl pyrophosphatase catalyzing the condensation of IPP and DMPP eventually leading to GGPP (Figure 2), increased lycopene production ( Figure 4). As a consequence of overproduction of geranylgeranyl pyrophosphatase in C. glutamicum ΔcrtEb, lycopene accumulated to four-fold higher concentrations (0.12 ± 0.01 mg/g CDW). The combined overexpression of crtB and crtE resulted in about 25 fold higher lycopene accumulation (0.8 ± 0.1 mg/g CDW, Figure 4) as compared to C. glutamicum ΔcrtEb. The maximal lycopene concentration of 2.4 ± 0.3 mg/g CDW was achieved when all three enzymes, CrtE, CrtB and CrtI, were overproduced. Thus, after de-bottlenecking the CrtE reaction overexpression of crtB and crtI is beneficial for lycopene overproduction. The maximal lycopene accumulation was 80 fold higher than that of the empty vector control.
Lycopene production was associated with less biomass formation and slowed glucose consumption. In this regard the strain with the highest lycopene production, C. glutamicum ΔcrtEb(pVWEx1-crtE/pEKEx3-crtBI2), stood out. The cells reached the stationary phase after 32 h, exhausted glucose not before 54 h after inoculation and grew only to about half of the biomass concentration (3.7 ± 0.5 mg/ml CDW) as compared to the empty vector control (7.0 ± 0.2 mg/ml CDW).

Discussion
The synthesis of C50 carotenoids occurs in a restricted number of bacterial species. Decaprenoxanthin is the most abundant one and it is the predominant carotenoid of the yellow C. glutamicum. The gene deletion and complementation analysis along with the pathway reconstruction in the multiple deletion strain C. glutamicum ΔΔ corroborates the previous elucidation of decaprenoxanthin biosynthesis in C. glutamicum based on transposon mutants of the strain MJ233C [16] and on heterologous expression of genes of the crtE-cg0722-crtBIY e Y f Eb cluster in the non-carotenogenic host Escherichia. coli [17]. Furthermore, we have analyzed a hitherto uncharacterized putative second carotenogenic gene cluster of C. glutamicum, crtB2/crtI2-1/crtI2-2, regarding the C50 carotenoid production. For the second phytoene synthaselike gene, crtB2 (cg2672), annotated in the C. glutamicum genome [25] and postulated to be involved in the squalene synthesis [2], we provide evidence that crtB2 indeed codes for a functional phytoene synthase. Hence, C. glutamicum possesses two functional phytoene synthases, CrtB and CrtB2. The two other open reading frames in the small crt-cluster are annotated as N-and C-terminal units of a second phytoene desaturase, but experimental confirmation of a phytoene desaturase function could not be obtained. Within the genus Corynebacterium C. glutamicum ATCC 13032 is the only species that possesses a second set of crt genes. The GC content of 54 to 58% of the second crt cluster is similar to the overall GC content of the genome, whereas that of the larger cluster is slightly lower. The genes of the two phytoene synthase paralogs only share 51% identity on the nucleotide level and mobile genetic elements such as IS-elements could not be detected in the vicinity of the two clusters arguing against recent duplication or horizontal gene transfer events.
All genome-sequenced corynebacterial species possess a crtI ortholog and most (except C. variabile) also possess a crtB ortholog, either clustered with crtI or elsewhere in the genome. The phylogeny of the crtI gene product reflects the phylogeny of the species. Only the highly related species C. glutamicum and C. efficiens exhibit all genes necessary to form C50 carotenoids. For corynebacterial species lacking some of the crt genes it remains to be shown if and which carotenoids are synthesized. On the other hand, C. michiganense [21], C. erythrogenes [22], C. fascians [23] and C. poinsettiae [24] are known to synthesize carotenoids, but their genome sequences are unknown.
In this study it could be shown that the genes of the carotenoid gene cluster of C. glutamicum ATCC 13032 crtE-cg0722-crtBIY e Y f Eb are co-transcribed. Similarly, also the second cluster is transciptionally organized as an operon. Transcriptional regulation of both operons has not yet been reported. The in vivo activity of the crtB2 gene product appears low due either to low expression levels or to low catalytic activity as plasmid-borne overexpression was required to complement the phenotype of the deletion mutant lacking the paralog crtB. Currently, it remains unknown whether crtB2 expression is affected by environmental stimuli and if/how the function of the two paralogs is regulated.
The potential of C. glutamicum for overproduction of carotenoids is to our knowledge described here for the first time. The interest in production of carotenoids, which find application in a wide variety of products due to their antioxidative properties and their colors, by cost-effective, environmentally friendly microbial fermentation processes is steadily increasing. The carotenogenic C. glutamicum is generally recognized as safe (GRAS), can readily be metabolically engineered and has been safely used in the millionton-scale production of food-additives since more than 50 years [28]. Lycopene was chosen as a test carotenoid product as it may serve as a platform intermediate and as its red color serves as a simple read out. Lycopene is a commercial product obtained by fermentation with the fungus Blakeslea trispora [29] (Vitatene, Leon, Spain). Here we show that C. glutamicum overproduces lycopene if crtEb is deleted and that additional overexpression of the carotenogenic genes crtE, crtB and crtI boosted lycopene production 80 fold. The achieved lycopene concentration of 2.4 mg/g CDW is already comparable to that obtained with other popular biotechnological hosts like E. coli, for which e.g. a lycopene yield of 1.8 mg/g CDW was reported when the crtE, crtB and crtI genes of the plant pathogen Pantoea ananatis were overexpressed [20]. A higher lycopene concentration (6.6 mg/g CDW) could only be achieved in an E. coli strain overexpressing genes for IPP synthesis and carotenogenesis after a systematic screen identified three gene knockouts in the central carbon metabolism [30]. In E. coli harboring multiple modifications, i.e. carrying a plasmid with genes of the lycopene biosynthetic pathway (crtE, crtB and crtI) and a plasmid containing the entire heterologous MVA pathway as well as the IPP isomerase gene, idi, and overexpressing the endogenous dxs gene, a lycopene concentration of 6.8 ± 0.4 mg/g was obtained in batch culture [31]. Process engineering as fed-batch process with glycerol with 10 g/l glucose and 7.5 g/l arabinose boosted the lycopene concentration to 32 mg/g CDW [31].
The very good lycopene concentration obtained by C. glutamicum after engineering only the final three enzymatic steps of lycopene synthesis can likely be further enhanced by additional metabolic engineering of (a) IPP synthesis using the endogenous MEP pathway and/or the heterologous MVA pathway, (b) genomebased or computational approaches to identify target genes in the central metabolism or its regulation and (c) by process engineering using e.g. fed-batch protocols. Thus, C. glutamicum may serve as a suitable production host for lycopene and related carotenoids.
In addition, C. glutamicum is a natural producer of the relatively rare group of C50 carotenoids that feature strong antioxidative properties due to the multiple conjugated double bonds and the hydroxyl group [32][33][34]. The pharmaceutical potential of these C50 carotenoids is not yet well studied [35]. It is imaginable that decaprenoxanthin, its direct precursor flavuxanthin or the C50 carotenoid of Micrococcus luteus, sarcinaxanthin, could be of commercial interest. Notably, genes of C. glutamicum and of M. luteus have been used to engineer E. coli for the production of sarcinaxanthin [20]. Thus, the product range of structurally diverse C50 carotenoids could be accessible by engineered hosts including C. glutamicum.

Conclusion
The genes of the carotenoid gene cluster of C. glutamicum ATCC 13032 crtE-cg0722-crtBIY e Y f Eb are cotranscribed and encode the enzymes involved in the biosynthesis of the C50 carotenoid decaprenoxanthin. An alternative, functionally active phytoene synthase is encoded in the crtB2/crtI2-1/crtI2-2 operon leading to a certain degree of redundancy in carotenoid synthesis in C. glutamicum. The potential of C. glutamicum as production host for terpenoids in general was demonstrated by considerable lycopene production after engineering the terminal reactions leading to lycopene.

Bacterial strains, media and growth conditions
The strains and plasmids used in this work are listed in Additional file 3: Table S2. C. glutamicum ATCC 13032 was used as wild type (WT). Precultivation of C. glutamicum strains was performed in BHI or LB medium. For cultivation in CGXII medium [36] precultivated cells were washed once with CGXII medium without carbon source and inoculated to an initial OD 600 of 1. Glucose was added as carbon and energy source to a concentration of 100 mM. Standard cultivations of C. glutamicum were performed at 30°C in a volume of 50 ml in 500 ml flasks with two baffles shaking at 120 rpm. The OD 600 was measured in dilutions using a Shimadzu UV-1202 spectrophotometer (Duisburg, Germany). For cloning, E. coli DH5α was used as host and cultivated in LB medium at 37°C. When appropriate kanamycin or spectinomycin were added to concentrations of 25 μg/ml and 100 μg/ml, respectively. Gene expression was induced by adding 1 mM IPTG.

Recombinant DNA work
Plasmids were constructed in E. coli DH5α from PCRgenerated fragments (KOD, Novagen, Darmstadt, Germany) and isolated with the QIAprep spin miniprep kit (QIAGEN, Hilden, Germany). Oligonucleotides used in this study were obtained from Eurofins MWG Operon (Ebersberg, Germany) and are listed in Additional file 3: Table S1. Standard reactions like restriction, ligation and PCR were performed as described previously [37]. If applicable, PCR products were purified using the PCR purification kit or MinElute PCR purification kit (QIAGEN, Hilden, Germany). For transformation of E. coli the RbCl method was used [38] and C. glutamicum was transformed via electroporation [39] at 2.5 kV, 200 Ω and 25 μF. All cloned DNA fragments were shown to be correct by sequencing.

Determination of the transcriptional start point of crtE and crtB2
Total RNA was isolated from an exponentially growing culture of C. glutamicum WT as described previously [40]. Purified RNA was analyzed by UV-spectrometry in regard to quantity and quality and was stored at −20°C until use. 2 μg of total RNA were used to perform 5' rapid amplification of cDNA ends-PCR (5' RACE_PCR) basically as described previously [41] with use of crtE-rv and crtB2-rv primers, respectively, for reverse transcription. Both, individual C tailing and A tailing were performed and analyzed. RACE_PCR was performed with primers crtE-RACE and crtB2-RACE and either OligoT or OligoG. Sequencing of the generated PCR fragments was accomplished using the suitable RACE primers and gave identical results for C tailing and A tailing reactions.

Reverse transcription (RT) for the analysis of transcription units
Total RNA was isolated from an exponentially growing culture of C. glutamicum WT as described previously [40]. Purified RNA was analyzed by UV-spectrometry in regard to quantity and quality and was stored at −20°C until use. 2 μg of total RNA were used to perform reverse transcription to generate cDNA that was subsequently used as template for PCRs applying primer that bind at adjacent genes. The reverse transcription reactions were performed using SuperScript ™ II reverse transcriptase (Invitrogen, Karlsruhe, Germany), and the remaining RNA was removed by the use of RNase H (MBI Fermentas GmbH, St. Leon-Rot).

Overexpression of carotenogenic genes from C. glutamicum
Plasmids harboring a carotenogenic gene allowed its IPTG-inducible overexpression and were based on pEKEx3 [42] or pVWEx1 [43], respectively. Amplification of a carotenogenic gene by polymerase chain reaction (PCR) from genomic DNA of C. glutamicum WT, which was prepared as described [44], was carried out using the respective primers (Additional file 3: Table S1). The amplified products were cloned into the appropriately restricted pEKEx3 or pVWEx1 plasmid DNA.
Deletion of carotenogenic genes in C. glutamicum WT For deletion of a carotenogenic gene, the suicide vector pK19mobsacB was used [36]. Genomic regions flanking a carotenogenic gene were amplified from genomic DNA of C. glutamicum WT using the respective primer pairs A/B and C/D as indicated in Additional file 3: Table S1. The PCR products were purified and linked by crossover PCR using the respective primer pair A/D (Additional file 3: Table S1). The either SmaI or BamHI restricted purified PCR product was cloned into pK19mobsacB resulting in the construction of the respective deletion vector (Additional file 3: Table S1). Targeted deletion of a carotenogenic gene via two-step homologous recombination using the respective deletion vector was carried out as described previously [26]. For the first recombination event integration of the vector into one of the flanking regions was selected via kanamycin resistance. Integration of the deletion vector into the genome results in a sucrose sensitivity because of the sacB gene product levansucrase. Selection for clones that have excised the deletion vector in a second recombination event was carried out via sucrose-resistance. Deletion of a carotenogenic gene was verified by PCR analysis of the constructed mutant using the respective primer pair E/F (Additional file 3: Table S1).

Extraction of carotenoids from bacterial cell cultures
To extract carotenoids from the C. glutamicum strains 20 ml aliquots of the cell cultures were centrifuged at 10,000 × g for 15 min, and the pellets were washed with deionized H 2 O. The pigments were extracted with 10 ml methanol:acetone mixture (7:3) at 60°C for 80 min with thorough vortexing every 20 min. When necessary, several extraction cycles were performed to remove all visible colors from the cell pellet.

Analysis of carotenoids
The extraction mixture was centrifuged 10,000 × g for 15 min and the methanol supernatant was transferred to a new tube. The absorption spectra of the various extracts were measured at wavelengths between 400 and 800 nm using the UV-1202 spectrophotometer (Shimadzu, Duisburg, Germany). High performance liquid chromatography (HPLC) analyses of the C. glutamicum extracts were performed on an Agilent 1200 series HPLC system (Agilent Technologies Sales & Services GmbH & Co. KG, Waldbronn), including a diode array detector (DAD) for UV/visible (Vis) spectrum recording. Quantification of carotenoids was performed using the extracted wavelength chromatogram at peak λ max , 450 nm for decaprenoxanthin and carotenoids with corresponding UV/Vis profiles and 470 nm for lycopene and corresponding carotenoids. Lycopene from tomato (Sigma, Steinheim, Germany) was used as standard. It was dissolved in chloroform according to its solubility and diluted in methanol. The HPLC protocol comprised isocratic elution for 25 min using a flow rate of 1.4 ml/min, a mobile phase composition of methanol/methyl tert-butyl ether/ethyl acetate (5:4:1) and a column system consisting of a precolumn (10x4 mm MultoHigh 100 RP18-5, CS Chromatographie Service GmbH, Langerwehe, Germany) and a main column (ProntoSIL 200-5 C30, 250x4 mm, CS Chromatographie Service GmbH, Langerwehe, Germany). 100 μl extract was injected.

Computational analysis
To compare the crt genes from C. glutamicum ATCC 13032 with other sequenced corynebacteria, the amino acid sequences of the respective genes were obtained from CoryneRegNet database (http://www.coryneregnet. de/). Sequence comparisons were carried out using BLASTP (www.ncbi.nlm.nih.gov/blast/Blast.cgi, [45]) and CLUSTALW [46] and the identification of potential orthologs and paralogs to C. glutamicum ATCC 13032 proteins was achieved by pairwise reciprocal BLAST analysis [47]. Species names, gene identifier and accession numbers are given in Additional file 1: Table S2. A phylogenetic tree was constructed using the neighbor joining method [48] with 1,000 bootstrap replicates.