Functional Expression and Extension of Staphylococcal Staphyloxanthin Biosynthetic Pathway in Escherichia coli*

Background: The biosynthetic pathway for staphyloxanthin has previously been proposed to consist of five enzymes. Results: A sixth pathway enzyme, 4,4′-diaponeurosporen-aldehyde dehydrogenase, was identified using a synthetic module approach. Conclusion: The complete staphyloxanthin biosynthetic pathway consists of six enzymes in Staphylococcus aureus. Significance: This is the first report demonstrating the complete staphyloxanthin pathway. The biosynthetic pathway for staphyloxanthin, a C30 carotenoid biosynthesized by Staphylococcus aureus, has previously been proposed to consist of five enzymes (CrtO, CrtP, CrtQ, CrtM, and CrtN). Here, we report a missing sixth enzyme, 4,4′-diaponeurosporen-aldehyde dehydrogenase (AldH), in the staphyloxanthin biosynthetic pathway and describe the functional expression of the complete staphyloxanthin biosynthetic pathway in Escherichia coli. When we expressed the five known pathway enzymes through artificial synthetic operons and the wild-type operon (crtOPQMN) in E. coli, carotenoid aldehyde intermediates such as 4,4′-diaponeurosporen-4-al accumulated without being converted into staphyloxanthin or other intermediates. We identified an aldH gene located 670 kilobase pairs from the known staphyloxanthin gene cluster in the S. aureus genome and an aldH gene in the non-staphyloxanthin-producing Staphylococcus carnosus genome. These two putative enzymes catalyzed the missing oxidation reaction to convert 4,4′-diaponeurosporen-4-al into 4,4′-diaponeurosporenoic acid in E. coli. Deletion of the aldH gene in S. aureus abolished staphyloxanthin biosynthesis and caused accumulation of 4,4′-diaponeurosporen-4-al, confirming the role of AldH in staphyloxanthin biosynthesis. When the complete staphyloxanthin biosynthetic pathway was expressed using an artificial synthetic operon in E. coli, staphyloxanthin-like compounds, which contained altered fatty acid acyl chains, and novel carotenoid compounds were produced, indicating functional expression and coordination of the six staphyloxanthin pathway enzymes.

Synthetic biological approaches have been widely used for the redesign and reconstruction of biosynthetic pathways in heterologous hosts (1,2). The functional expression of redesigned pathway enzymes in a modular manner can improve the productivity and yield of target molecules and can demonstrate previously undefined functions of pathway enzymes (3)(4)(5)(6). Moreover, the modular expression of a pathway enzyme can be used as a tool for diversifying the structures of molecules such as carotenoids (7) using combinatorial biosynthesis and directed evolution in heterologous hosts (8 -10).
Carotenoids are compounds that belong to a class of isoprenoid derivatives found in many organisms, including photosynthetic and non-photosynthetic bacteria. The carotenoids found in bacteria and yeast have antioxidative functions and play roles in light harvesting, energy transfer, and the regulation of membrane fluidity (11)(12)(13). The protective functions of carotenoids against oxidative stress, singlet oxygen, and peroxy radicals promote the survival of pathogenic microbes during host immune responses (14,15).
The pathogen Staphylococcus aureus is a Gram-positive, gold-colored bacterium that is resistant to methicillin and all available ␤-lactam antibiotics (16). The antibiotic resistance of S. aureus has necessitated the development of new types of anti-infective drugs, such as virulence factor-specific drugs (14,17). The gold color of this pathogen is derived from the yelloworange carotenoid staphyloxanthin, a virulence factor. The chemical characterization of staphyloxanthin, combined with analysis of S. aureus mutants, enabled the elucidation of the staphyloxanthin biosynthetic pathway (18), which was thought to consist of five enzymes: 4,4Ј-diapophytoene synthase (CrtM), 4,4Ј-diapophytoene desaturase (CrtN), 4,4Ј-diaponeurosporene oxidase (CrtP), glycosyltransferase (CrtQ), and acyltransferase (CrtO) (see Fig. 1A) (19). CrtP introduces a terminal oxygen molecule into 4,4Ј-diaponeurosporene, which results from sequential activities of CrtM and CrtN, to form a carboxylic acid intermediate (4,4Ј-diaponeurosporenoic acid) via an aldehyde intermediate (20). Staphyloxanthin is finally synthesized by further modification of 4,4Ј-diaponeurosporenoic acid by glycosylation (CrtQ) and acylation (CrtO) at a terminal carboxyl group. Interestingly, CrtP was reported to function as both an oxidase and an aldehyde dehydrogenase (20).Here, we report the identification of a sixth enzyme, 4,4Ј-diaponeurosporen-aldehyde dehydrogenase (AldH), in the S. aureus staphyloxanthin biosynthetic pathway. With a complete and redesigned staphyloxanthin pathway, staphyloxanthin-like  This article contains supplemental Tables S1 and S2 and additional references. 1 To whom correspondence should be addressed. compounds and novel carotenoids were successfully produced in Escherichia coli for the first time.

EXPERIMENTAL PROCEDURES
Cloning and Synthetic Module Construction-All cloning and carotenoid expression experiments were performed in the E. coli SURE strain except for using pBBR1MCS-2-derived vectors. E. coli XL1-Blue was used for cloning and expressing pBBR1MCS-2-derived vectors (21). Genes encoding CrtM, CrtN, CrtP, CrtQ, and CrtO from S. aureus ssp. aureus (KCTC 1928) were cloned into the constitutive expression vector pUCM (22). Four genes, encoding AldH (aldH), glycine betaine aldehyde dehydrogenase (gbsA), aldehyde dehydrogenase homolog (aldA), and aldehyde dehydrogenase family protein, were amplified from the genomic DNA of S. aureus KCTC 1928 with PCR primers that were designed according to corresponding gene sequences from S. aureus strain Newman (AP009351) and cloned into the pUCM vector. The aldH gene from S. aureus KCTC 1928 was cloned into pBBR1MCS-2, a vector that is compatible with pUCM and pACYC184 in E. coli. A putative aldH gene was amplified from the genomic DNA of Staphylococcus carnosus ssp. carnosus KCTC 3580 and cloned into pUCM. To reconstruct the staphyloxanthin biosynthetic pathway in E. coli, a synthetic module containing crtM, crtN, crtP, or crtQ was sequentially assembled into the pACYC184 vector to generate pACM-M SA -N SA , pACM-M SA -N SA -P SA , and pACM-M SA -N SA -P SA -Q SA . Briefly, a gene was subcloned from pUCM-X SA (where X is a pathway gene) into pACYC184  The following carotenoids were identified: 4,4Ј-diaponeurosporen-4-al (peak 1) and 4,4Ј-diapolycopene-dial (peak 2). The inset shows the recorded UV-visible absorption spectra for individual peaks (see Fig. 1A and Table 2 for details).

Functional Expression of Staphyloxanthin Pathway
by amplifying the gene together with a modified constitutive lac promoter. The staphyloxanthin gene cluster (crtOPQMN) from S. aureus was amplified with a forward PCR primer for crtO and a reverse PCR primer for crtN containing an XbaI restriction enzyme site at its 5Ј-end (supplemental Table S1). The PCR product was then digested with XbaI and cloned into the corresponding site in the pUC19 vector to facilitate controlled expression of the genes from a lac promoter. All plasmids and strains used in this study are listed in supplemental Table S2.
Culture Growth for Carotenoid Production-For carotenoid production, S. aureus strains (KCTC 1928, RN4220, and an aldH deletion mutant of RN4220) were cultivated for 24 h in the dark at 30°C in B-medium. For carotenoid production, recombinant E. coli SURE or XL1-Blue was cultivated for 36 h in the dark at 30°C with shaking at 250 rpm in Terrific broth medium (50 ml of medium in a 300-ml flask or 200 ml of medium in a 1-liter flask) supplemented with the appropriate antibiotics: 50 g/ml chloramphenicol, 100 g/ml ampicillin, and/or 30 g/ml kanamycin.
Isolation of Carotenoids-Carotenoids were extracted from cell pellets using 15 or 30 ml of acetone or methanol until all visible pigments were removed. When performing saponification, carotenoids were extracted using 15 ml of methanol containing 6% KOH and incubated for 14 h at 4°C. Colored supernatants were pooled after centrifugation (4°C and 4000 rpm) and concentrated into a small volume using an EZ-2 Plus centrifugal evaporator (Genevac Inc., Gardiner, NY). Five milliliters of EtOAc was added to the concentrated solution and reextracted after the addition of 5 ml of NaCl (5 N) solution for salting out. The upper organic phase containing carotenoids was collected, washed with distilled water, and completely dried using the EZ-2 Plus evaporator. Dried samples were stored at Ϫ70°C until analyzed.
Preparation of Carotenoids for LC/MS-For LC/MS analysis, dried samples were resuspended in 300 l of EtOAc, run through a silica column, and eluted stepwise with increasing amounts of acetone in a 9:1 hexane/EtOAc solvent. A 20-l aliquot of each fraction was subjected to TLC to confirm the purity of the compound in the fractions. The fractions were collected, concentrated using the EZ-2 Plus centrifugal evaporator, and subjected to a Varian 1200L LC/MS system.
Allelic Replacement-Allelic replacement of the aldH gene in S. aureus strain RN4220 (modification-positive, restrictionnegative) was performed according to the protocol described by Wyatt et al. (23). The erythromycin gene was amplified from pSAKON and fused with the flanking region of the aldH gene at both ends by overlapping PCR. The PCR product was purified and ligated into the pGEM-T Easy vector (Promega). Following sequencing, the erythromycin cassette was treated with NcoI and PstI and subcloned into the corresponding sites in pCL52.2, a temperature-sensitive E. coli/S. aureus shuttle vector. The resulting pCL-ALDKO plasmid was electrotransformed into S. aureus RN4220 (24) and incubated at 43°C on B-mediumagar plates containing 10 g/ml erythromycin (BM-agar Erm 10 plates) for 2 days. Colonies were restreaked onto BM-agar Erm 10 plates and incubated at 43°C for 2 days. A single colony was inoculated into 30 ml of B-medium without antibiotics and sequentially diluted (1:1000) into 30 ml of B-medium each day at 30°C for 5 days. After 5 days, 1:10 6 diluted cells were spread onto BM-agar Erm 10 plates, and colonies were visually screened. Deletion mutants that developed an orange color were confirmed by PCR and carotenoid profiling.
Analysis of Carotenoids-The initial TLC analysis was performed using a 9:1:1 hexane/EtOAc/acetone solvent system. A 9:1:3 hexane/EtOAc/acetone solvent was used for analyzing glycosylated C 30 carotenoids. A 10 -20-l aliquot of the crude extract or the collected polar fraction was applied to a ZORBAX Eclipse XDB-C18 column (4.6 ϫ 150 mm, 5.0 m; Agilent Technologies, Inc., Santa Clara, CA) and eluted under isocratic conditions with a solvent system (80:15:5 acetonitrile/methanol/isopropyl alcohol) at a flow rate of 1 ml/min using an Agilent 1200 HPLC system equipped with a photodiode array  (6). The compound corresponding to peak 3 was identified as 4,4Ј-diaponeurosporenoic acid, and the compound corresponding to peak 4 was identified as 4,4Ј-diapolycopene-4,4Ј-dioic acid (see Fig. 1A and Table 2 for details). detector. For structural elucidation, carotenoids were identified using a combination of HPLC retention times, UV-visible absorption spectra, and mass fragmentation spectra. Mass fragmentation spectra were monitored using both negative and positive ion modes in a mass range of m/z 200 -900 on the Varian 1200L LC/MS system equipped with an atmospheric pressure chemical ionization interface.
For further extension of the staphyloxanthin pathway in E. coli, the fourth gene, crtQ, was expressed by (i) a two-plasmid system (pUC19-CrtQ ϩ pACM-M SA -N SA -P SA or pUCM-CrtQ ϩ pACM-M SA -N SA -P SA ) and (ii) a one-plasmid system (pACM-M SA -N SA -P SA -Q SA ). In both cases, carotenoid aldehyde intermediates (4,4Ј-diaponeurosporen-4-al and 4,4Јdiapolycopene-dial) accumulated without being converted into the corresponding glycosylated carotenoid structures (Fig. 2B). The accumulation of carotenoid aldehyde intermediates was also observed in recombinant E. coli expressing the wild-type gene cluster (crtOPQMN) (Fig. 2C). These results indicate that non-functionality of CrtQ or the absence of the sixth pathway enzyme (AldH) may be attributable to the accumulation of carotenoid aldehyde intermediates. Therefore, we focused on the discovery of the missing AldH because, even though CrtP was reported to have dual functions as both an oxidase and AldH, no further oxidized carotenoid carboxylic acid intermediates such as 4,4Ј-diaponeurosporenoic acid were detected in recombinant E. coli expressing pACM-M SA -N SA -P SA .

Functional Expression of Staphyloxanthin Pathway
Identification of Enzyme Catalyzing Oxidation of Carotenoid Aldehyde to Carboxylic Acid in S. carnosus-To date, staphyloxanthin biosynthesis had been studied both in S. aureus and in the non-carotenogenic bacterium S. carnosus (19). Staphyloxanthin was heterologously produced in recombinant S. carnosus transformed with the S. aureus gene cluster encoding the five pathway enzymes (19). This indicates that wild-type S. carnosus expresses a carotenoid AldH, similar to the AldH expressed in S. aureus, that can catalyze the oxidation of carotenoid aldehyde (4,4Ј-diaponeurosporen-4-al) to a carboxylic acid intermediate (4,4Ј-diaponeurosporenoic acid). Therefore, we searched for AldH candidates in S. carnosus using the amino acid sequence of the S. aureus AldH and found a putative AldH enzyme in S. carnosus that shared 78% amino acid identity with the AldH of S. aureus. As expected, when the S. carnosus AldH was expressed in recombinant E. coli also expressing the pACM-M SA -N SA -P SA vector, 4,4Ј-diaponeurosporen-4-al was successfully converted into 4,4Ј-diaponeurosporenoic acid (Fig. 3C), as observed when the S. aureus AldH was expressed in recombinant E. coli expressing pACM-M SA -N SA -P SA . The AldH-mediated oxidation of 4,4Ј-diaponeurosporen-4-al into 4,4Ј-diaponeurosporenoic acid suggests that the AldH enzymes from S. carnosus and S. aureus have similar functions. This result explains how recombinant S. carnosus expressing only five staphyloxanthin enzymes can still produce staphyloxanthin without requiring the exogenous expression of the sixth S. aureus AldH. It is not clear why the genome of non-pigmented S. carnosus contains the aldH gene. The presence of aldH in the S. carnosus genome may be a clue to an evolutionary event and suggests that S. carnosus may lose a carotenoid biosynthetic pathway. Future studies investigating the regulation of AldH expression and its physiological function in S. carnosus are necessary.
Functional Verification of 4,4Ј-Diaponeurosporen-aldehyde Dehydrogenase (AldH) in S. aureus Using Allelic Replacement-To confirm the function of AldH in the original host (S. aureus), we replaced the aldH gene with an erythromycin resistance cassette (23). To do this, the modification-positive and restriction-negative S. aureus strain RN4220 was used instead of the multidrug-resistant S. aureus strain KCTC 1928. The carotenoid profile of S. aureus RN4220 was confirmed to be the same as that of S. aureus KCTC 1928. An aldH deletion mutant of S. aureus RN4220 was visually screened, and deletion of the aldH gene was verified by PCR (Fig. 5A). The aldH dele- tion mutant accumulated 4,4Ј-diaponeurosporen-4-al as a major product but did not produce 4,4Ј-diaponeurosporenoic acid or staphyloxanthin, as shown in the HPLC chromatogram (Fig. 5, B and D) and TLC plate (Fig. 5C). This accumulation of 4,4Ј-diaponeurosporen-4-al was consistent with that seen in recombinant E. coli expressing pACM-M SA -N SA -P SA -Q SA or pUC19-OPQMN SA (Fig. 2, B and C). Therefore, this mutational study confirms that AldH catalyzes the oxidation of 4,4Ј-diaponeurosporen-4-al to 4,4Ј-diaponeurosporenoic acid in the staphyloxanthin biosynthetic pathway in S. aureus.
Taken together, our data demonstrated that AldH is essential for staphyloxanthin formation and is the sixth enzyme in the staphyloxanthin biosynthetic pathway of S. aureus. Interestingly, unlike the aldH gene of Methylomonas sp. strain 16a, which is located between crtP and crtN in a gene cluster, the aldH gene of S. aureus is located 670 kilobase pairs from the staphyloxanthin biosynthetic gene cluster (Fig. 1B).
Expression of Complete Staphyloxanthin Pathway in E. coli and Characterization of Staphyloxanthin-like Compounds Produced in Engineered E. coli-After we had identified 4,4Ј-diaponeurosporen-aldehyde dehydrogenase as the sixth enzyme in the staphyloxanthin biosynthetic pathway, the complete pathway was then modularly reconstructed and expressed in E. coli using three compatible plasmids. Recombinant E. coli expressing pACM-M SA -N SA -P SA -Q SA ϩ pUCM-O SA ϩ pBBR-aldH SA changed color similarly to wild-type S. aureus and recombinant E. coli expressing pUC19-OPQMN SA ϩ pBBR-aldH SA (Fig.  6A). HPLC analysis of crude extracts of recombinant E. coli expressing pACM-M SA -N SA -P SA -Q SA ϩ pUCM-O SA ϩ pBBR-aldH SA detected two dominant peaks (peak 9 and 10 in Fig. 6B) and a few small peaks. Although the two dominant peaks had different retention times compared with staphyloxanthin (peak 8 in Fig. 6B), their UV-visible absorption spectra were very similar to that of staphyloxanthin (Fig. 6C) (Fig. 7, A and B). These fragmentation patterns were very similar to those of staphyloxanthin isolated from S. aureus (Fig. 7C). Saponification of the two compounds confirmed that structures 9 and 10 contained glucose and a fatty acid moiety with an ester bond (Fig. 8). Together, these results confirm that structures 9 and 10 are tetradecanoyl-glucosyl-4,4Ј-diaponeurosporenoic acid and hexadecanoyl-glucosyl-4,4Ј-diaponeurosporenoic acid, respectively, structurally similar to staphyloxanthin except for the addition of a different fatty acid moiety (Fig. 1A). The fatty acid moieties of these new structures are myristic acid and palmitic acid, whereas that of staphyloxanthin is 12-methyltetradecanoic acid (Fig. 1A). This difference in fatty acid moieties between staphyloxanthin and the staphyloxanthin-like structures 9 and 10 could be explained by differences in the availability of cellular fatty acids between E. coli and S. aureus. Branched saturated fatty acids such as 12-methyltetradecanoic acid are limited in E. coli but not in S. aureus (19,27). 12-Methyltetradecanoic acid, the fatty acid moiety of staphyloxanthin, accounts for 47% of the total fatty acids found in S. aureus at 37°C (28). Therefore, formation of staphyloxanthin-like compounds instead of staphyloxanthin in E. coli can be explained by the broad specificity of the acyltransferase CrtO; myristic acid (C 14:0 ) and palmitic acid (C 16:0 ), two abundant saturated fatty acids in E. coli, were utilized as substrates for CrtO instead of 12-methyltetradecanoic acid. The broad substrate specificity of CrtO was further confirmed by identification of small amounts of additional staphyloxanthin-like compounds with different lengths of fatty acids (asterisks in Fig. 8A) in the crude extracts of E. coli expressing pACM-M SA -N SA -P SA -Q SA ϩ pUCM-O SA ϩ pBBR-aldH SA .

DISCUSSION
To date, the biosynthesis of staphyloxanthin had been exclusively studied in S. aureus and S. carnosus. The alternative S. carnosus system had been successfully used for the identification of the five pathway enzymes (CrtO, CrtP, CrtQ, CrtM, and CrtN); however, it failed to elucidate the missing sixth pathway enzyme, AldH from S. aureus. Although the aldH gene is not located near the pathway gene cluster in S. aureus, we identified its function by coexpressing the aldH gene with synthetic staphyloxanthin pathway modules in E. coli. An aldH deletion mutant abolished staphyloxanthin formation and caused accumulation of an aldehyde intermediate, confirming the role of AldH in the oxidation of aldehyde intermediates (4,4Ј-diaponeurosporen-4-al) to carboxylic acid intermediates (4,4Ј-diaponeurosporenoic acid) in the complete staphyloxanthin biosynthetic pathway in S. aureus. For the first time, we succeeded in reconstructing the complete staphyloxanthin biosynthetic pathway in E. coli. The six staphyloxanthin pathway enzymes were functionally active and coordinated in heterologous E. coli. Staphyloxanthin-like structures were produced instead of staphyloxanthin due to differences in the types of fatty acids available for staphyloxanthin synthesis in E. coli versus S. aureus. Furthermore, the redesigned staphyloxanthin pathway produced novel carotenoids, which were not detected in S. aureus, proving that synthetic modules of biosynthesis are a powerful methodology for generating structural diversity of biochemicals. Finally, engineered E. coli cells expressing the synthetic staphyloxanthin pathway could supply pathway intermediates (Table 2) for biological studies and represent a good model system for investigating the role of staphyloxanthin or staphyloxanthin-like compounds in the virulence of S. aureus.