Characterization of carotenoids in Rhodothermus marinus

Abstract Rhodothermus marinus, a marine aerobic thermophile, was first isolated from an intertidal hot spring in Iceland. In recent years, the R. marinus strain PRI 493 has been genetically modified, which opens up possibilities for targeted metabolic engineering of the species, such as of the carotenoid biosynthetic pathway. In this study, the carotenoids of the R. marinus type‐strain DSM 4252T, strain DSM 4253, and strain PRI 493 were characterized. Bioreactor cultivations were used for pressurized liquid extraction and analyzed by ultra‐high performance supercritical fluid chromatography with diode array and quadropole time‐of‐flight mass spectrometry detection (UHPSFC‐DAD‐QTOF/MS). Salinixanthin, a carotenoid originally found in Salinibacter ruber and previously detected in strain DSM 4253, was identified in all three R. marinus strains, both in the hydroxylated and nonhydroxylated form. Furthermore, an additional and structurally distinct carotenoid was detected in the three strains. MS/MS fragmentation implied that the mass difference between salinixanthin and the novel carotenoid structure corresponded to the absence of a 4‐keto group on the ß‐ionone ring. The study confirmed the lack of carotenoids for the strain SB‐71 (ΔtrpBΔpurAcrtBI’::trpB) in which genes encoding two enzymes of the proposed pathway are partially deleted. Moreover, antioxidant capacity was detected in extracts of all the examined R. marinus strains and found to be 2–4 times lower for the knock‐out strain SB‐71. A gene cluster with 11 genes in two operons in the R. marinus DSM 4252T genome was identified and analyzed, in which several genes were matched with carotenoid biosynthetic pathway genes in other organisms.

R. marinus has recently been a target for genetic engineering, enabling gene knock-out and the introduction of heterologous genes (Bjornsdottir, Fridjonsson, Kristjansson, & Eggertsson, 2007). This gives way for further opportunities, including genetic engineering for the production of carotenoid derivatives. (Bjornsdottir, Fridjonsson, Hreggvidsson, & Eggertsson, 2011) The possibility of production of natural or altered carotenoids highlights the importance of the understanding of the carotenoid biosynthetic pathway. A few genes of the pathway have previously been determined through knock-out trials in R. marinus strain SB-71. Strain SB-71 (ΔtrpBΔpurAcrtBI'::trpB) has two carotenoid biosynthetic genes (crtB and crtI) knocked out, which normally catalyze two units of geranylgeranyl pyrophosphate to phytoene and phytoene to lycopene, respectively (Bjornsdottir et al., 2011).
These mutations resulted in white colonies in contrast to its natural red color due to the absence of carotenoids.
Carotenoids and carotenoid glycosides are of interest due to their use as coloring agents in foods and chemical pre-cursors to fragrances and flavor industries (Dembitsky, 2005). It has also been shown that consumption of vegetables with carotenoids such as ß-carotene has health-beneficial properties due to the conversion into vitamin-A.
Even though R. marinus type-strain DSM 4252 T is the most studied of the strains in the genus, there are no reports on its carotenoid content. Moreover, characterization of the natural carotenoids and the carotenoid biosynthetic pathway of PRI 493 are required for the continuation of the genetic engineering and thus the production of carotenoid derivatives in R. marinus.
The aim of this study is to analyze and compare the carotenoids present in the aforementioned R. marinus strains and comparing these to the previously characterized monocyclic carotenoid glucoside ester, found in strain DSM 4253. In addition, the antioxidant capacity of the cell extracts is determined and correlated with the carotenoid biosynthetic pathway of R. marinus, which is proposed in this paper.
The bioreactor had a total working volume of 0.5 L and parameters were set to: 65°C, 1 vvm aeration, and pH 7. The stirrer speed was cascaded with a pO 2 level at 40%. Growth was monitored through offline OD measurements at 620 nm. At the end of the exponential phase, the cells were harvested by centrifugation (Thermo Scientific Sorvall Lynx 4000 centrifuge) at 5500 g for 15 min at 10°C. The pellets were washed with a 0.9% (w/v) NaCl solution three times at the original liquid volume before freezing and subsequent freeze-drying (LABCONCO, freeze dry system).

| Pressurized liquid extraction
Extractions of R. marinus samples were performed using an accelerated solvent extractor (ASE 350, Dionex, Sunnyvale, CA), equipped with a solvent controller. To avoid any possible oxidation effects and to remove the dissolved oxygen, ethanol was sonicated for 45 min prior to use. Extractions were performed at 100°C for three extraction cycles (2 min per extraction cycle) based on an experimental method used for the extraction of carotenoids from microalgae (Plaza et al., 2012) with some modifications. Each extraction started with heating the extraction cell for 5 min. Extractions were performed in 10-11 ml extraction cells, containing 0.2-0.6 g of freeze-dried sample. The extraction procedure was as follows: (1) the extraction cell was loaded into the oven; (2) the cell was filled with solvent up to a pressure of 1500 psi; (3) heat-up time was applied, (4) a static extraction was performed with all system valves closed; (5) the cell was rinsed (with 60% cell volume using extraction solvent); (6) solvent was purged from the cell with N 2 gas and (7) depressurization took place. Between extractions, the complete system was rinsed to prevent extract carry-over.
The extracts obtained were dried with a Reacti-Vap TM Evaporator (Thermo Fisher, Germering, Germany) until total sample dryness. The dry extracts were redissolved in ethanol at 10 mg/ml concentration and filtered through 0.2 μm PTFE filters (VWR International, West Chester, PA) and were ready for UHPSFC separation without further clean-up. The extracts obtained were stored at −80°C.

| Analysis of carotenoids by UHPSFC-DAD-QTOF/MS
The analysis of carotenoids was performed using a Waters Acquity Ultra Performance Convergence Chromatography (UPC 2 ) system (Waters, Milford, MA) coupled to a photodiode array detector (PDA) and a quadrupole and orthogonal acceleration time-of-flight tandem mass spectrometer (XEVO-G2 QTOF) with electrospray ionization (ESI) (Waters, MS Technologies, Manchester, UK). The system was controlled by Waters ® Empower TM Chromatography software; while MassLynx TM (V 4.1, SCN 779, Waters Corp., Manchester, UK) was used for MS data acquisition and treatment. Separation was carried out with CSH fluoro-phenyl (100 mm × 3 mm, 1.8 μm) from Waters (Milford, MA). The mobile phase consisted of (A) CO 2 and (B) methanol in a gradient elution analysis programmed as follows: 0 min, 5% (B); 0-1 min, 5% (B); 1-3 min, 5-15% (B); 3-7 min, 15% (B); 7-8 min, 15-5/% (B); and 8-10 min, 5% (B). The flow rate was 1.5 ml/min; injection volume, 5 μL; column temperature, 30°C, and backpressure, 160 bar. Ammonium formate (0.2% ) in methanol was used as a makeup solvent, 0.5 ml/min. UV-vis spectra were recorded in the range of 200-500 nm. The ESI interface was operated in positive mode, and full-scan HPLC-qTOF-MS spectra were obtained by scanning from m/z 50 to m/z 1,000. The mass spectrometer was calibrated using a solution of sodium formate. Data were collected in continuum mode and mass was corrected during acquisition using an external reference (Lock-Spray TM ) comprising of 10 μL/min solution of leucineenkephalin (2 ng/μl) via a lock-spray interface. The capillary and cone voltage were set at 3 kV and 40 V, respectively. Nitrogen was used as both cone gas (50 L/hr) and desolvation gas (800 L/h). The source and desolvation temperature were set at 150 and 300°C, respectively.
Simultaneous acquisition of exact mass at high and low collision energy, MS E (where E represents collision energy), was used to obtain full-scan accurate mass fragment, precursor ion, and neutral loss information. The collision energy in function 1 (low energy) was off while in function 2 (high energy) and the collision energy ranged between 15 V to 60 V. peaks were from a single carotenoid in peak 3, a few drops of NaOH (0.5 mmol/L) was added to the make-up solvent to increase Na + adducts.

| Trolox equivalent antioxidant capacity (TEAC) assay
The TEAC assay, based on a previously described method (Re et al., 1999) with some modifications, was used to measure the antioxidant capacity of the extracts. ABTS radical cation (ABTS ·+ ) was produced by reacting 7 mmol/L ABTS with 2.45 mmol/L potassium persulfate and allowing the mixture to stand in the dark at room temperature for 12-16 hr before use. The aqueous ABTS ·+ solution was diluted with ethanol to an absorbance of 0.70 (±0.02) at 734 nm. One milliliter of ABTS· + radical solution was added to 10 µL of each sample concentration. After 50 min at room temperature, 300 μl of the mixture was transferred into a well of the microplate, and the absorbance was measured at 734 nm on a microplate spectrophotometer reader (Multiskan GO, Thermo Fisher, Germering, Germany). Trolox was used as a reference standard and results were expressed as TEAC values (mmol Trolox/g dry bacterial extract). These values were obtained from at least four different concentrations of each extract tested in the assay giving a linear response between 20% and 80% of the initial absorbance. All analyses were done at least in triplicate for each extract.

| DPPH radical scavenging assay
The antioxidant capacity of all the obtained extracts was measured using the DPPH (2,2-diphenyl-1-picrylhydrazyl) radical scavenging assay based on the protocol by Brand-Williams et al. (1995). Briefly, a solution was prepared by dissolving 23.5 mg of DPPH in 100 ml of methanol. This stock solution was further diluted with methanol 1:10.
Both solutions were stored at 4°C until use. Four different concentrations of extracts were tested. Twenty-five microliter of these solutions were added to 975 μl of DPPH diluted solution to complete the final reaction medium (1 ml). After 4 hr at room temperature, 300 μl of the mixture was transferred into a microplate well and the absorbance was measured at 516 nm in a microplate spectrophotometer reader (Multiskan GO, Thermo Fisher). DPPH-methanol solution was used as a reference sample. The DPPH concentration remaining in the reaction medium was calculated from a calibration curve. The percentage of remaining DPPH against the extract concentration was then plotted to obtain the amount of antioxidant necessary to decrease the initial DPPH concentration by 50% or EC 50 . Therefore, the lower the EC 50 , the higher the antioxidant activity. For rational reasons of clarity, the antioxidant capacity was determined as the inverse value of the efficient concentration EC 50 (mg/ml), representing a comparable term for the effectiveness of antioxidant and radical scavenging capacity (1/EC 50 ). The larger the antioxidant capacity, the more efficient an antioxidant. Measurements were done at least in triplicate for each extract.

| RESULTS AND DISCUSSION
The R. marinus strains DSM 4252 T , DSM 4253, and PRI 493 appeared as red colonies on medium 162, while strain SB-71 (ΔtrpBΔpurAcrtBI'::trpB) produced white colonies. Low cell densities of R. marinus resulted in a long lag phase and a procedure for maintaining the exponential growth during transfers was therefore developed.
In short, colonies were transferred into 5 mL liquid LB medium for precultivation and diluted 10-fold into baffled shake flasks after 24 hr.
The cells were then grown for 8 hr before inoculating the bioreactor, which were harvested at the end of the exponential phase.
The carotenoids of R. marinus were extracted using pressurized liquid extraction (PLE). The cell extract were red for all strains except for strain SB-71 (ΔtrpBΔpurAcrtBI'::trpB), while the cell pellets were colorless (see supplementary data). Extraction conditions (solvent, temperature and time) were selected based on previous results obtained with the microalgae Chlorella vulgaris (Plaza et al., 2012). PLE uses solvents at elevated temperatures subjected to high enough pressures to keep the solvents in a liquid state during the extraction process. The extraction process is faster and more efficient at higher temperature due to faster diffusion rates (Björklund, Nilsson, & Bøwadt, 2000) and the high pressure that penetrates the sample matrix (Mustafa & Turner, 2011). In order to decrease the risk of carotenoid breakdown due to high temperature, a lower extraction temperature was used in comparison to literature and more importantly, a very short extraction time of 6 min was applied (Cha et al., 2010). Conventional carotenoid extraction procedures use organic solvents, harmful both for the environment and for human health, which has made PLE increasingly popular, especially for analytical purposes (Mustafa & Turner, 2011). Several studies have successfully used PLE for the extraction of carotenoids (Mustafa & Turner, 2011).

| Analysis of carotenoids by UHPSFC-DAD-QTOF/MS
In order to obtain a suitable separation between the lipids and ca- The chromatogram produced several peaks for R. marinus strains DSM 4252 T , DSM 4253, and PRI 493 with retention times of 2.52, 2.96, 3.33, and 3.59 min, respectively ( Figure 1). As expected, no absorption could be detected with the DAD for the R. marinus SB-71 knock-out strain, indicating the absence of carotenoids. The co-eluting carotenoid peaks were determined to be chain length variations of the carotenoid acyl group, which was verified by MS analysis. The C 11 -C 17 fatty acid pattern (Figure 2) could be detected in all of the chromatogram peaks and in R. marinus strains DSM 4252 T , DSM 4253, and PRI 493. It has previously been reported that R. marinus DSM 4253 produces iso-C 15 , anteiso-C 15 , iso-C 17 , and anteiso-C 17 fatty acids (Lutnaes et al., 2004) but that the composition greatly varies depending on the growth conditions including medium composition (Bjornsdottir et al., 2006). The carotenoids with iso13:0 fatty acid group consistently had the highest peak intensity in the MS, while the C 15 fatty acid variant had the second highest intensity. Peak intensity of similar compounds with the same ionization characteristics can be assumed to have a proportional correlation with concentration, which would indicate a higher concentration of the carotenoid with the iso13:0 fatty acid constituent (Tang & Kebarle, 1993). The pre-  Table 2 fatty acid also reported by Lutnaes et al., (2004). Sodium was added in the mass spectrometry make-up solution to exclude that the double ions peaks were two different compounds. The sodium addition  (Rivera et al., 2014).
These fragments could be detected in peaks 3 and 4. However, such fragments could not be detected for the carotenoids in peak 1 and 2, which verifies that the mass difference is located on the β-ionone ring.
A mass difference comparable to the hydroxylated and nonhydroxylated salinixanthin was found between peak 1 and 2. Hence, the ion F I G U R E 2 Spectrum of salinixanthin with variable fatty acid lengths (C 11 -C 17 ) in R. marinus strain DSM 4252 T in peak 4 of the chromatogram T A B L E 1 TEAC and DPPH antioxidant capacity values obtained for the studied cell extracts after the extraction by PLE. All measurements were done at least in triplicate   (Guinand, Vacheron, & Michel, 1970) and nonhydroxylated in Chloroflexus aurantiacus (Takaichi, Tsuji, Matsuura, & Shimada, 1995) and in species of the Myxococcus genus (Dembitsky, 2005). The reason why the previous report on carotenoids in R. marinus strain DSM 4253 (Lutnaes et al., 2004) did not detect the carotenoids found in this study could be due to their loss in the purification steps that were performed in that study.

| Bioinformatics analysis of putative carotenoid biosynthetic pathway genes
A gene homology search for known carotenoid biosynthetic genes was performed for the R. marinus DSM 4252 T genome (Bjornsdottir et al., 2006). In this search, six genes encoding homologues to enzymes in carotenoid biosynthesis pathways from other species were found (CrtB, CrtI, CrtY, CrtO, CruD, and CruC), and allowed identification of a gene cluster, with two adjacent operons, one small and one larger ( Figure 4). They appear to share a regulatory region, while being transcribed in opposite directions. The smaller operon consists of four genes and the larger of seven genes. A putative biosynthetic pathway for carotenoid production in R. marinus was suggested based on the results ( Figure 5).
In the suggested carotenoid biosynthesis pathway ( Figure 5 & Schmidt-Dannert, 2010;Rählert, Fraser, & Sandmann, 2009), but corresponding genes were not found in the R. marinus genome. Enzymes with the above functions, although with different sequence conservation, may however be encoded by the yet unidentified genes in the cluster (Figure 4).
A glycosyltransferase, (CruC) adds a unit of glucose to the carotenoids and an acetyltransferase (CruD) finally links a fatty acid to the glucose ( Figure 5). A gene encoding a CruD homologue is found in the smaller operon, whereas a gene located far from the gene cluster showed relatively high homology to the gene encoding CruC in S. ruber (Query cover 90%, Ident 57%). In order to verify the predicted genes in the two operons, knock-out mutations of the separate genes are vital for the validation of the suggested carotenoid biosynthetic pathway in R. marinus.
The biological function of the carotenoids produced by R. marinus is still under debate. It has previously been argued that carotenoids stabilize the cell membranes in thermophilic conditions but identical carotenoids have been found in the mesophilic S. ruber (Lutnaes et al., 2004). In S. ruber, salinixanthin has been shown to be a functional group in xanthorhodopsin, a protein/carotenoid complex that makes up a light-driven proton pump. (Balashov et al., 2005). However, homology searches between rhodopsin genes in different bacteroidetes species and the R. marinus genome fail to detect any matches.