Heterologous Production of Cyanobacterial Mycosporine-Like Amino Acids Mycosporine-Ornithine and Mycosporine-Lysine in Escherichia coli

ABSTRACT

Mycosporine-like amino acids (MAAs) are an important class of secondary metabolites known for their protection against UV radiation and other stress factors. Cyanobacteria produce a variety of MAAs, including shinorine, the active ingredient in many sunscreen creams. Bioinformatic analysis of the genome of the soil-dwelling cyanobacterium Cylindrospermum stagnale PCC 7417 revealed a new gene cluster with homology to MAA synthase from Nostoc punctiforme. This newly identified gene cluster is unusual because it has five biosynthesis genes (mylA to mylE), compared to the four found in other MAA gene clusters. Heterologous expression of mylA to mylE in Escherichia coli resulted in the production of mycosporine-lysine and the novel compound mycosporine-ornithine. To our knowledge, this is the first time these compounds have been heterologously produced in E. coli and structurally characterized via direct spectral guidance. This study offers insight into the diversity, biosynthesis, and structure of cyanobacterial MAAs and highlights their amenability to heterologous production methods.

IMPORTANCE Mycosporine-like amino acids (MAAs) are significant from an environmental microbiological perspective as they offer microbes protection against a variety of stress factors, including UV radiation. The heterologous expression of MAAs in E. coli is also significant from a biotechnological perspective as MAAs are the active ingredient in next-generation sunscreens.

INTRODUCTION

Mycosporine-like amino acids (MAAs) are commonly found in marine microbes exposed to high levels of UV radiation, including aquatic fungi, eukaryotic algae, and cyanobacteria (1–6). These small (<400 Da), colorless, water-soluble compounds are composed of a 4-deoxygadusol core conjugated with an amino acid or alcohol (7). Their absorption maxima typically range from 310 to 360 nm and their molar extinction coefficients (ε) range from 28,100 to 50,000 M⁻¹ cm⁻¹ (8, 9). Commonly termed “microbial sunscreens,” MAAs can dissipate UV energy as heat without generating free oxygen radicals (7).

While UV exposure is necessary to elicit the synthesis of MAAs (10), some studies suggest that other factors, including osmotic stress, may also be involved in MAA regulation (11). Indeed, MAAs are widely considered to be multifunctional compounds offering protection against oxidative, osmotic, and thermal stress as well as UV radiation (12–17). However, this may not be true for all MAAs (11). Due to their diverse biological properties, MAAs are promising candidates for pharmaceutical and cosmetic applications (3, 18). For example, shinorine, produced by the cyanobacterium Anabaena variabilis, has already been commercialized as an active ingredient in sunscreen creams (Helioguard 365 and Helionori) (14).

The biosynthesis of shinorine (Fig. 1) was recently elucidated via heterologous expression in Escherichia coli (19, 20) and is a model pathway for MAA biosynthesis. In A. variabilis ATCC 29413, the shinorine biosynthesis gene cluster consists of four genes: ava_3855 to ava_3858. The first step in the biosynthesis of shinorine is the conversion of sedoheptulose-7-phosphate (SH-7P) to 2-demethyl-4-deoxygadusol (DDG) by a DDG synthase (Ava_3858). Following this, DDG is converted to the mycosporine core compound 4-deoxygadusol (4-DG) by an O-methyltransferase (O-MT) (Ava_3857). Glycylation of 4-DG is catalyzed by a C-N ligase (Ava_3856), resulting in the production of mycosporine-glycine. Attachment of serine to mycosporine-glycine by a nonribosomal peptide synthetase (NRPS) (Ava_3855) results in the final product shinorine.

FIG 1.

Schematic diagram of shinorine biosynthesis and proposed biosynthesis of mycosporine-ornithine and mycosporine-lysine. 2-Demethyl-4-deoxygadusol (DDG) synthase catalyzes the formation of DDG via sedoheptulose-7-phosphate. Following this, O-methylation of DDG is catalyzed by an O-methyltransferase (O-MT) to form 4-deoxygadusol (4-DG), the core structure of mycosporines. Glycylation of 4-DG producing mycosporine-glycine (an oxomycosporine) is catalyzed by a C-N ligase. Finally, a serine residue is attached to mycosporine-glycine to produce shinorine. In this study, we propose that the two C-N ligases and/or d-Ala−d-Ala ligase catalyzes the addition of ornithine and lysine to 4-DG, forming compounds 1 and 2, respectively.

In the cyanobacterium Nostoc punctiforme ATCC 29133, MAA biosynthesis is similar to that of A. variabilis and involves homologous genes (NpR5598 to NpR5600). However, one exception is the last step (condensation of serine to mycosporine glycine), which is catalyzed by a d-Ala−d-Ala ligase-like protein (NpF5597) rather than a nonribosomal peptide synthetase (21).

In this present study, a new gene cluster (myl) putatively responsible for MAA biosynthesis was identified in the soil-dwelling cyanobacterium Cylindrospermum stagnale strain PCC 7417 via genome mining. The myl gene cluster is homologous to MAA gene clusters from A. variabilis, N. punctiforme, and Aphanothece halophytica and comprises five genes (mylA to mylE). This study describes the cloning and heterologous expression of mylA to mylE in E. coli. The structures of the resulting MAA expression products were elucidated via spectroscopic techniques.

MATERIALS AND METHODS

Bioinformatics.

NCBI BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) was used to mine the genome sequence of C. stagnale PCC 7417 using inferred MAA synthase peptide sequences from A. variabilis ATCC 29413 (ava_3855 to ava_3858) and N. punctiforme ATCC 29133 (NpR5598 to NpR5600 and NpF5597) (18, 19) as query sequences (NCBI accession numbers NC_007413 and NC_010628, respectively). A 6,236-kb putative MAA gene cluster comprising five genes was identified in C. stagnale PCC 7417 and designated mylA to mylE. This sequence was submitted to GenBank under the accession number KU376485. Multiple sequence alignments and identity scores were generated using Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/).

The percentage of rare codons in mylA to mylE was determined using the web-based tool Rare Codon Calculator (http://nihserver.mbi.ucla.edu/RACC).

DNA cloning.

The myl gene cluster was PCR amplified from C. stagnale PCC 7417 genomic DNA using the primers MAA_gcF and MAA_gcR. These primers incorporated NcoI and NdeI restriction sites at their respective 5′ ends to facilitate the ligation of the amplified myl cluster into pET-28b (Novagen) (see Table S1 in the supplemental material). PCRs were performed in a final volume of 50 μl and were composed of 1× HiFi buffer, 1 μM deoxynucleoside triphosphates (dNTPs), 20 pmol of each primer, 3% dimethyl sulfoxide (DMSO), 1.6 units of Velocity DNA polymerase (Bioline), and 200 ng of genomic DNA template. Thermal cycling was carried out in a Mastercycler (Eppendorf) as follows: initial denaturation (98°C for 3 min), followed by 2 cycles of amplification (98°C for 30 s, 54°C for 30 s, 72°C for 3 min 10 s), 28 cycles of amplification (98°C for 30 s, 61.5°C for 30 s, 72°C for 3 min 10 s), and a final extension step (72°C for 10 min). PCRs were analyzed by electrophoresis through 1% agarose gels and visualized under UV light in the presence of ethidium bromide. Amplicons were purified using the DNA Clean & Concentrator-5 kit according to the manufacturer's protocol (Zymo Research).

The purified myl gene cluster amplicon and pET-28b plasmid were digested with NdeI and NcoI enzymes according to the manufacturer's protocol (New England BioLabs). Digested PCR products were purified using the DNA Clean & Concentrator-5 kit (Zymo Research). The linearized pET-28b vector was purified using the Zymoclean gel DNA recovery kit (Zymo Research).

The myl amplicon was ligated into the linearized pET-28b vector using T4 DNA ligase (New England BioLabs). Electro-competent E. coli (GB2005) cells were transformed with the resulting pET-28b_myl construct. Positive transformants were identified via colony PCR using a universal primer targeting the T7 promoter and SCREEN2-MAA-R (see Table S1 in the supplemental material). PCRs were performed in a final volume of 20 μl and reaction mixtures were composed of 1× PCR buffer, 1 μM dNTPs, 20 pmol of each primer, 1 unit of BioTaq DNA polymerase (Bioline), and 200 ng of plasmid DNA template (pET-28b_myl construct). Thermal cycling was carried out in a Mastercycler (Eppendorf) as follows: initial denaturation (94°C for 5 min) followed by 30 cycles of amplification (94°C for 30 s, 50°C for 30 s, 72°C for 3 min 10 s), and a final extension step (72°C for 10 min). Selected positive transformants were sequence verified using the Prism BigDye cycle-sequencing system and ABI 3730 DNA analyzer sequencer (Applied Biosystems).

Heterologous expression of Myl synthase and extraction of resulting MAAs.

The percentage of rare codons in the genes comprising the myl biosynthesis gene cluster was relatively high, ranging from 10 to 14%. Therefore, the expression strain, E. coli BL21(DE3) (Invitrogen), was transformed with the pRARE plasmid (carrying rare tRNA genes) as well as the pET-28b_myl construct (22).

Five hundred microliters of the expression and control strains, grown overnight at 37°C in lysogenic broth (LB), was used to inoculate 50 ml of LB medium supplemented with 50 μg · ml⁻¹ kanamycin and 34 μg · ml⁻¹ chloramphenicol. The cultures were incubated with shaking at 37°C until the optical density at 600 nm (OD₆₀₀) reached 0.5 to 0.6 (∼1.5 h). Cultures were then cooled to 18°C and induced with isopropyl-β-d-thiogalactopyranoside (IPTG) at a final concentration of 0.5 M, 0.8 M, or 1 M. Cultures were incubated with shaking at 18°C for a further 19 or 43 h. Uninduced control cultures lacking IPTG were grown under identical conditions.

Cells were harvested by centrifugation (3,220 × g, 4°C, 20 min), resuspended in 1 ml of methanol, and lysed on ice via sonication (3 20-s pulses with a 1-min recovery between each pulse, 30% amplitude; Digital Sonifier, Branson). The lysate was clarified by centrifugation (16,100 × g, 4°C, 30 min), and the supernatant was collected and evaporated to dryness under reduced vacuum. The desiccated residue was resuspended in 200 μl of water and centrifuged (16,100 × g, 4°C, 10 min) prior to high-performance liquid chromatography (HPLC) analysis.

High-performance liquid chromatography of E. coli extracts.

Cell extracts were analyzed on an HP Agilent series 1100 HPLC using an Agilent Zorbax SB-C18 (4.6 × 250 mm) column at 25°C. The binary solvent system was composed of water with 0.1% formic acid (solvent A) and methanol (solvent B). Elution was carried out using a flow rate of 1 ml · min⁻¹. The gradient profile was 1% B at time zero ramped linearly to 20% B at 7 min, 50% B at 9 min, and 80% B at 17 min and held for 5 min. UV absorption (310 nm) and UV spectra (210 to 400 nm) were monitored by a diode array detector.

Purification and structural elucidation of compounds 1 and 2.

Crude extracts were dissolved in 40 ml of water and extracted with an equal volume of dichloromethane. The aqueous layer was lyophilized, and the residue was dissolved in a small amount of methanol to remove salts. The methanol extract was evaporated to dryness, and the desiccated residue was dissolved in water. Water-soluble material was subjected to HPLC as described above. Fractions with UV absorbance maxima of 310 nm were collected and characterized using spectrometric techniques.

Mass spectrometry of compounds 1 and 2.

Mass spectra were collected on an LCQ Deca XP Plus mass spectrometer operating in positive ionization mode coupled to a Surveyor LC pump, autosampler, and PDA detector (200 to 600 nm; Thermo Fisher Scientific). Analytes were separated on a Phenomenex Luna C₁₈ column (3 μM, 2.1 × 150 mm) using a gradient of water (solvent A) against 0.2% ethanoic acid in methanol (solvent B). Elution was carried out using a flow rate of 0.8 ml · min⁻¹. The solvent program was as follows: 0% B at time zero ramped linearly to 10% B at 40 min, 100% B at 60 min, and 5% B at 61 min and held for 4 min.

Nuclear magnetic resonance spectroscopy of compounds 1 and 2.

Purified compounds were dissolved in deuterium oxide (Cambridge Isotope Laboratories) for nuclear magnetic resonance (NMR) analysis. One-dimensional (proton) and two-dimensional (correlation spectroscopy [COSY], heteronuclear multiple-quantum correlation [HMQC], heteronuclear multiple-bond correlation [HMBC]) NMR spectra were recorded on a Bruker Avance III 600 MHz spectrometer equipped with a cryogenic probe (operating at 600 MHz for ¹H and 150 MHz for ¹³C), locked to the deuterium signal.

Accession number(s).

A 6,236-kb putative MAA gene cluster comprising five genes was identified in C. stagnale PCC 7417. This sequence was submitted to GenBank under the accession number KU376485.

RESULTS AND DISCUSSION

Bioinformatic analysis of the myl gene cluster in C. stagnale PCC 7417.

The translated genome of C. stagnale PCC 7417 was mined for peptide sequences homologous to those from A. variabilis ATCC 29413 (ava_3855 to ava_3858) and N. punctiforme ATCC 29133 (NpR5598 to NpR5600 and NpF5597). A putative MAA biosynthesis (myl) gene cluster encoding five enzymes, MylA to MylE, was identified (Fig. 2; Tables 1 and 2). The inferred primary peptide sequences of MylA to MylE were aligned with homologous sequences from A. variabilis and N. punctiforme (see Fig. S1 to S4 in the supplemental material): MylA was homologous to dimethyl-4-deoxygadusol (DDG) synthases encoded by ava_3858 (73% identity) and NpR560 (88% identity); MylB was homologous to O-methyltransferases (O-MTs) encoded by ava_3857 (65% identity) and NpR5599 (88% identity); MylD and MylE were homologous to C-N ligases encoded by ava_3856 (63 and 65% identity, respectively) and NpR5598 (61 and 65% identity, respectively); and MylC was homologous to the d-Ala−d-Ala ligase encoded by NpF5597 (78% identity) (Table 2). An ava_9855 homologue encoding a nonribosomal peptide synthetase (NRPS)-like protein was not identified in the C. stagnale genome sequence.

FIG 2.

MAA biosynthesis gene clusters from cyanobacteria and an actinobacterium. Genes involved in the biosynthesis of mycosporines and MAAs in C. stagnale PCC 7417, N. punctiforme ATCC 29133, A. variabilis ATCC 29413, A. halophytica, and A. mirum DSM 43827. Gene families are indicated by arrows: black, DDG-synthase; gray, O-methyltransferase; black dashed, d-Ala−d-Ala ligase; white, C-N ligase; double-outlined (NRPS-like protein). Direction of arrows indicates direction of transcription. ap3858* is distal to the other genes in the A. halophytica MAA synthase gene cluster.

TABLE 1.

Putative mycosporine/MAA gene clusters in cyanobacteria and actinobacteria^a

Organism (accession no.)	DDG synthase	O-MT	C-N ligase	d-Ala–d-Ala ligase	NRPS	Transporter	Accession no.
Anabaena variabilis ATCC 29413	+	+	+		+		NC_007413
Aphanothece halophytica	+	+	+	+			AB854643, AB854644
Nostoc punctiforme ATCC 29133/PCC 73102	+	+	+	+			NC_010628
Nodularia spumigena CCY9414	+	+	+	+			NZ_AAVW00000000
Cyanothece sp. PCC 7424	+	+	+		+	+	NC_011729
Cyanothece sp. PCC 51142	+	+	+	+		+	NC_010546
Cyanothece sp. CCY0110	+	+	+	+		+	NZ_AAXW0000000
Cylindrospermum stagnale PCC 7417	+	+	++	+			NC_019757
Lyngbya sp. PCC 8106	+	+	+	+		+	NZ_AAVU00000000
Microcystis aeruginosa PCC 7806	+	+	+	+			AM778947
Microcoleus chthonoplastes PCC 7420	+	+	+	+			NZ_ABRS00000000
Crocosphaera watsonii WH 8501	+	+	+	+		+	NZ_AADV00000000
Trichodesmium erythraeum IMS101	+	+	+	++	++	+	NC_008312
Actinosynnema mirum (DSM 43827)	+	+	+	+			NC_013093

^a+, gene with this function is present in the genome; ++, two copies of a gene with this function are present in the gene cluster.

TABLE 2.

Bioinformatic analysis of MAA biosynthesis genes (mylA to mylE) in C. stagnale

Gene	Size (bp)	ORFa	Closest homologue(s)	Identity (%)	Function (NCBI accession no.)b
mylA	1,230	Cylst_1339	NpR5600 (Nostoc punctiforme PCC 73102)	88	DDG synthase (AFZ23628)
mylB	834	Cylst_1340	NpR5599 (Nostoc punctiforme PCC 73102)	88	O-Methyltransferase (AFZ23629)
mylC	1,380	Cylst_1342	ava_3856 (Anabaena variabilis ATCC 29413)	63	d-Ala–d-Ala ligase (AFZ23631)
mylD	1,269	Cylst_1343	NpR5598 (Nostoc punctiforme PCC 73102), ava_3856 (Anabaena variabilis ATCC 29413)	65	C-N ligase (AFZ23632)
mylE	1,029	Cylst_1341	NpF5597 (Nostoc punctiforme PCC 73102)	78	C-N ligase (AFZ23630)

^aORF, open reading frame.

^bPutative enzyme function based on homology (NCBI accession no. for peptide sequence).

Recent studies suggest that the first step in MAA biosynthesis, the production of mycosporine-glycine, is conserved in many producing organisms, including A. variabilis ATCC 29413 (Fig. 2), N. punctiforme ATCC 29133, A. halophytica, Actinosynnema mirum DSM 43827, and Microcystis aeruginosa sp. PCC 7806 (19, 21, 23–25). Several MAA-like compounds contain a second amino acid residue attached to the 1 position of mycosporine-glycine. Shinorine contains a serine residue at this position, whose attachment is catalyzed by an NRPS-like protein (Ava_3855) in A. variabilis or a d-Ala−d-Ala ligase in N. punctiforme (NpF5597) and M. aeruginosa (MysD) (19, 21, 25). In all organisms described, MAA synthase genes are organized collinearly with the proposed biosynthesis pathway, with the exception of NpF5597 in N. punctiforme, which is transcribed in the direction opposite to the mycosporine-glycine biosynthesis genes (Fig. 2). In A. halophytica, attachment of a second glycine residue in this same position, forming mycosporine-2-glycine, is catalyzed by a d-Ala−d-Ala ligase homologue (Ap_3855) (23). Similarly, in A. mirum, attachment of alanine or serine to form mycosporine-glycine-alanine and shinorine is catalyzed by a d-Ala−d-Ala ligase (24). Taken together, these data suggest that the fourth enzyme in the MAA biosynthesis pathway, a d-Ala−d-Ala ligase, is responsible for much of the diversity observed in MAA-like compounds.

In comparison, the MAA biosynthesis gene cluster in C. stagnale contains a rearrangement of the d-Ala−d-Ala ligase gene and comprises two C-N ligase genes (Fig. 2). Assuming collinearity, this rearrangement may result in the production of structurally different MAAs by C. stagnale (compared to those originating from mycosporine-glycine).

Heterologous expression of the myl gene cluster.

The myl gene cluster was cloned and heterologously expressed in E. coli to determine the compound(s) produced by the encoded pathway. Analytical-scale HPLC of the methanolic extracts of positive transformants showed significant differences in the HPLC profiles between IPTG-induced cultures and uninduced control cultures (Fig. 3). Increasing the IPTG concentration and duration of fermentations had no effect on the expression profiles (see Fig. S5 and S6 and Table S2 in the supplemental material). Four peaks at retention times (t_R) of 5.6 min, 6.4 min, 7.3 min, and 8.1 min were increased compared to those of the controls. Peaks I and II (t_R = 5.6 and 6.4 min, respectively) had UV absorbance maxima of 310 nm, similar to those of other MAA compounds, including mycosporine-glycine, mycosporine-serinol, mycosporine-glutaminol, mycosporine-glutaminol, and glutamicol-glucoside (26–31). Peaks III and IV (t_R = 7.3 and 8.1 min, respectively) were not studied in detail because their expression rates were very low. Purification of peaks I and II via semipreparative HPLC afforded compounds 1 and 2, respectively, which were characterized using liquid chromatography-mass spectrometry (LC-MS) and NMR.

FIG 3.

HPLC chromatograms (310 nm) and UV spectra of E. coli expression host extracts. Chromatogram of methanolic extracts from induced (solid line) and uninduced (dashed line) E. coli BL21(DE3) (pET-28b_myl) and chromatogram of extract from E. coli BL21(DE3) (pET-28b) negative control (dotted line). Four major peaks (I to IV) were overexpressed in the IPTG-induced strain compared to those in the controls. UV spectra (210 to 400 nm) for peaks I to IV are shown in the upper right inset.

Mass spectral analysis of compounds 1 and 2 identified molecular ions m/z 303.2 and 317.2 [M + H]⁺, respectively, both with absorbance maxima of 310 nm (see Fig. S7 to S10 and Table S6 in the supplemental material). LC-tandem MS (LC-MS/MS) analysis of compound 1 identified two molecular ions, m/z 267.01 (loss of 2H₂O from ion 303.2) and m/z 235.16 (loss of CH₂OH). Again LC-MS/MS analysis of m/z 267.01 identified a molecular ion m/z 235.16 (loss of CH₂OH), and LC-MS/MS analysis of m/z 235.16 identified a molecular ion m/z 191.14 (loss of CO₂). Similarly LC-MS/MS analysis of compound 2 identified a molecular ion m/z 281.08 (loss of 2H₂O). These masses had no match to currently known MAAs (27, 31, 32). This suggested that the myl gene cluster expressed in E. coli is responsible for the production of novel MAAs. Therefore, NMR analysis was performed to further characterize compounds 1 and 2.

NMR analysis (Table 3; see also Fig. S11 to S18 in the supplemental material) suggested that compounds 1 and 2 were both mixtures; however, chemical shifts characteristic of mycosporines were observed in both proton spectra. Deconvolution of the NMR spectrum for compounds 1 and 2 was assisted by two-dimensional NMR experiments. Careful analysis of the COSY, HSQC, and HMBC spectra of compound 1 allowed the assignment of two structural fragments corresponding to the cyclohexene ring and an amino acid side chain. Assembly of these two fragments yielded a novel compound, mycosporine-ornithine, compound 1 (Fig. 4). The NMR spectrum of compound 2 showed high similarity to that of compound 1; however, the amino acid fragment contained an additional methylene bridge. The remaining chemical shifts were assigned to the cyclohexene ring and were nearly identical to those of compound 1. Assembly of these two fragments resulted in mycosporine-lysine (Fig. 4). Compounds 1 and 2 both contained oxo-carbonyl structures. Unlike many other MAAs, they do not contain amino-acylated side chains on carbon 3 and only contain a single amino acid side chain attached to carbon 1. Numerous precedents exist for such compounds. Mycosporine-glutaminol has been isolated from the fungus Trichothecium roseum, lichen Degelia plumbea, and cyanobacterium Leptolyngbya sp. (27, 28). Mycosporine-glutaminol and glutamicol glucosides were found in Rhodotorula yeast species, microcolonial fungi, and Leptolyngbya foveolarum (29–31). Mycosporine-taurine has been reported from the freshwater cyanobacterium Synechocystis sp. and marine sea anemones (33, 34). Mycosporine-serinol was found in cyanolichens Lichina pygmaea and Lichina confinis and different species of Peltigera (26, 27). However, whether the cyanobacterial symbionts or the fungal hosts are the true producers of mycosporine-serinol is currently unknown. The pathways responsible for the biosynthesis of these compounds have yet to be elucidated.

TABLE 3.

NMR data for compounds 1 and 2

Compound	Position	δC	δH (mult, J [Hz])	COSY	HMBC (2,3,4JCH)
1 (mycosporine-ornithine)	1	159.2
	2	129.8
	3	184.4
	4	42.9	2.32 (dd, 1.6, 17.1)	4b	2, 3, 5, 6, 7
			2.59 (d, 17.2)	4a	2, 3, 5, 6, 7
	5	72.1
	6	32.7	2.707 (d, 17.6)	6b	1, 2, 4, 5
			2.81 (d, 17.1)	6a	1, 2, 4, 5, 7
	7	67.8	3.48 (d, 0.9)		4, 5, 6
	8	59.1	3.49 (s)		2
	9	42.0	3.34 (t, 7.3)		1, 10, 11
	10	27.4	1.85 (m)		9, 11, 12, 13
	11	25.5	1.63 (m)		9, 10
	12	54.3	3.7 (m)	11	11, 10, 13
	13	174.3
2 (mycosporine-lysine)	1	159.5
	2	129.9
	3	184.5
	4	42.8	2.31 (d, 17.4)	4b	2, 3, 5, 6, 7
			2.57 (d, 17.3)	4a	2, 3, 5, 6, 7
	5	72.0
	6	32.7	2.69 (d, 17.4)	6b	1, 2, 4, 5
			2.79 (d, 17.3)	6a	1, 2, 4, 5, 7
	7	67.7	3.46 (s)		4, 5, 6
	8	59.1	3.47 (s)		2
	9	42.4	3.29 (t, 7.3)	10	1, 10, 11
	10	29.3	1.58 (m)	9, 11	9, 11, 12
	11	21.6	1.36 (m)	10, 12	9, 12, 13
	12	30.1	1.81 (m)	11, 13	4, 10, 11, 13
	13	54.6	3.66 (m)	12	4, 10, 11
	14	174.7

FIG 4.

Chemical structures of mycosporine-ornithine (compound 1) and mycosporine-lysine (compound 2). Solid arrows represent HMBC correlations. Bold bonds represent COSY correlations.

Due to their structures, the biosyntheses of mycosporine-ornithine (compound 1) and mycosporine-lysine (compound 2) are proposed to be similar to that of other MAAs, generated via the addition of ornithine/lysine to 4-deoxygadusol, which is a common precursor in MAA biosynthesis (34). However, in the absence of gene deletion studies, it is difficult to assign a substrate to the C-N ligases or to the d-Ala−d-Ala ligase and hence fully elucidate the pathway.

This study is, to our knowledge, the first report of mycosporine-ornithine (compound 1). Mycosporine-lysine (compound 2) has been previously reported in a variety of cyanobacterial species, including Gloeocapsa sp. isolated from a limestone quarry wall in Calafell (Catalonia, Spain), Calothrix sp. isolated from Rheinfall (Schaffhausen, Switzerland), and Nostocal sp. (Diplocolon sp.) isolated from granitic outcrops in Catavina (Baja California, Mexico) (35, 36). In these studies, alkaline hydrolysis was used to indirectly determine the structure of mycosporine-lysine (35–37). However, this study represents the first direct spectral analysis-guided structural elucidation of mycosporine-lysine.

The ubiquity and diversity of MAAs in cyanobacteria may in some part explain the success of these tenacious photosynthetic microbes, which have persisted on Earth for more than three billion years. Here, we have demonstrated the heterologous expression of two MAAs, mycosporine-lysine and the novel mycosporine-ornithine. Future studies are likely to reveal additional novel MAA structures and biosynthesis pathways in cyanobacteria, including those of the next generation of biological sunscreens.