J. Gen. Appl. Microbiol., 67, 100–105 (2021) doi 10.2323/jgam.2020.08.002 2021 Applied Microbiology, Molecular and Cellular Biosciences Research Foundation

Full Paper

Detection of 2-hydroxy-fatty acids and 2-hydroxy-fatty acid-containing ceramides in a gliding marine bacterium

(Received May 15, 2020; Accepted August 4, 2020; J-STAGE Advance publication date: March 13, 2021) Kazuyoshi Kawahara,1,∗ Hirokazu Iida,1 and Akira Yokota2 1 Department of Biosciences, College of Science and Engineering, Kanto Gakuin University, Yokohama, Kanagawa 236-8501, Japan 2 Department of Microbial Resources, Tohoku University Graduate School of Agricultural Science, Sendai, Miyagi 980-0845, Japan

lipopolysaccharide (Alexander and Rietschel, 2001) in The cellular fatty acid composition of Aureispira addition to non-polar fatty acids mainly present in marina IAM 15389T (JCM 23197T), a gliding bac- glycerophospholipids in the membrane. An hydroxy-fatty terium isolated from the coastline of Thailand, was acid profile is, therefore, an important feature for the tax- re-examined by using a standard MIDI method onomy of Gram-negative (Oyaizu and Komagata, based on alkaline hydrolysis, and two other meth- 1983), and should be carefully examined when we analyze ods. The direct transesterification using 5% HCl/ the fatty acid composition of newly isolated bacteria. methanol or 4 M HCl hydrolysis followed by me- Aureispira marina is a gliding Gram-negative bacterium thyl esterification revealed that 2-hydroxy-15-me- with helical filamentous morphology isolated from ma- thyl-hexadecanoic acid (2-OH-iso-C17:0) and 2-hy- rine plant debris from the southern coastline of Thailand droxy-15-methyl-hexadecenoic acid (2-OH-iso- (Hosoya et al., 2006). This bacterium was reported to have C17:1), which were not reported in a previous pa- a unique fatty acid composition. It contained arachidonic per, were found to be major cellular fatty acids of acid (C20:4) as the most abundant fatty acid and saturated this bacterium, and the amount of 2-OH-iso-C17:1 fatty acids with normal and branched hydrocarbon chains. was even higher than that of arachidonic acid As for hydroxy-fatty acids, 3-hydroxy-hexadecanoic acid (C20:4), a characteristic polyunsaturated fatty acid (3-OH-C16:0), 3-hydroxy-15-methyl-hexadecanoic acid (3- present in this bacterium. These 2-hydroxy-fatty OH-iso-C17:0), and 3-hydroxy-octadecanoic acid (3-OH- acids were contained in two cellular lipids that were C18:0) were detected with small amounts only, compared relatively stable against alkaline hydrolysis. One with a taxonomically related bacterial , Saprospira of them was analyzed by mass spectrometry, 1H- grandis, and 2-hydroxy-fatty acid was reported to be ab- nuclear magnetic resonance, and other chemical sent. The similar fatty acid composition was reported on methods, and identified as a ceramide composed another species of the same genus, Aureispira maritima of 2-hydroxy-fatty acid and sphingosine of 19 car- (Hosoya et al., 2007). These data of unusual hydroxy-fatty bons with three double bonds. A minor ceramide acid composition as a Gram-negative bacterium have containing 18 carbon sphingosine with three dou- prompted us to investigate more about the fatty acids of ble bonds was also detected. A. marina using several analytical methods. The present study revealed that the bacterium contains 2-hydroxy-fatty Key Words: Aureispira marina; ceramide; 2-hy- acids as major fatty acids and a novel ceramide contain- droxy-fatty acid; mass spectrometry; sphingosine ing these fatty acids.

Materials and Methods Introduction Strain and growth condition. A. marina IAM 15389T (JCM 23197T) (type strain of the species) was used in this Gram-negative bacteria usually contain 3-hydroxy-fatty study. This strain was grown in a Sap2 liquid medium acids and, in some cases, 2-hydroxy-fatty acids in their

*Corresponding author: Kazuyoshi Kawahara, Department of Biosciences, College of Science and Engineering, Kanto Gakuin University, 1-50-1, Mutsuura-higashi, Kanazawa-ku, Yokohama, Kanagawa 236-8501, Japan. E-mail: [email protected] None of the authors of this manuscript has any financial or personal relationship with other people or organizations that could inappropriately influence their work. 2-OH-fatty acids in Aureispira marina 101

(Hosoya et al., 2006) containing 0.5 × artificial sea water Table 1. Fatty acid contents in dry cells of A. marina IAM 15389T determined by three different methods. (1.5% NaCl, 0.035% KCl, 0.54% MgCl2·6H2O, 0.27% MgSO4·7H2O, and 0.05% CaCl2·2H2O), 0.1% tryptone Fatty acid Fatty acid content (nmol/mg dry cells)* (BD, Sparks, MD, USA), and 0.1% yeast extract (BD), Method I Method II Method III adjusted to pH 7.0, at 30°C for 24 h with reciprocal shak- iso-C 13 14 13 ing (80 cycles/min). 15:0 iso-C16:0 14 11 11 C 29 29 30 Analytical methods of cellular fatty acid. The bacterial 16:0 iso-C 25 26 27 cells were killed by heating at 105 C for 10 min, collected 17:0 ° C 111 by centrifugation, and lyophilized to prepare dry cells. Dry 17:0 3-OH-C16:0 213 cells (5 mg) were used for fatty acid analysis by three Unknown-1 27 96 92 methods. Unknown-2 23 41 39 3-OH-iso-C 325 Method I (MIDI method) (Glucksman et al., 2000): Dry 17:0 C 65 78 69 cells were hydrolyzed (100°C, 30 min) in one ml of alka- 20:4 line solution (15% NaOH in 50% methanol), two ml of *Mean values of data from triplicated experiments. acid solution (mixture of 13 ml 6 M HCl and 11 ml metha- nol) was then added, and fatty acids were esterified by heating (80°C, 10 min). Spectrometric analysis. The molecular weight of lipids Method II (Komagata and Suzuki, 1987): Dry cells were was analyzed by Matrix Assisted Laser Desorption/Ioni- mixed with one ml of 5% HCl/methanol (Kokusan Chemi- zation Time of Flight Mass Spectrometry (MALDI-TOF cal) and heated (100°C, 3 h) for direct transesterification. MS) (AXIMA-CHR plus, Shimadzu) with 2,5- Method III (Kawahara et al., 1991): Dry cells were dihydroxybenzoic acid (Fujifilm Wako Pure Chemical) as hydrolyzed with one ml of 4 M HCl, and liberated fatty a matrix material. Positive ions were detected with acids were extracted, and esterified with 5% HCl/metha- reflectron mode. Samples were solubilized in CDCl / nol by heating (85°C, 2 h). 3 CD OD (2:1, v/v), and 1H-Nuclear Magnetic Resonance Methyl ester samples were analyzed by gas-liquid chro- 3 (NMR) was measured using the equipment JNM-ECX 500 matography (GLC) equipped with a capillary column (500 MHz, JEOL RESONANCE) at 45°C. CBP1 (Shimadzu) for determination, and by gas chroma- tography/mass spectrometry (GC-MS) (QP2010Plus, Results Shimadzu) for identification. Electron impact ionization (EI) was used for GC-MS. Pentadecanoic acid (C ) was 15:0 Cellular fatty acid composition used as an internal standard for quantitative determina- Using dry cells of A. marina IAM 15389T, cellular fatty tion. The reduction of peaks of hydroxy-fatty acids be- acids were analyzed by three methods: alkaline hydroly- cause of tailing was corrected by using the peak area ratio sis and methyl esterification (Method I), direct of authentic C , 3-OH-C , and 2-hydroxy-hexadeca- 15:0 16:0 transesterification by HCl/methanol (Method II), and noic acid (2-OH-C ). 16:0 strong acid hydrolysis and methyl esterification (Method Thin-layer chromatography and column chromatogra- III). The mean values of fatty acid contents (nmol/mg dry phy. Cellular lipids were extracted with chloroform/ cells) of triplicated experiments are shown in Table 1. As methanol (2:1, v/v), and hydrolyzed with 0.1 M NaOH described in a previous paper (Hosoya et al., 2006), the (100°C, 30 min). Extracted lipids and lipid hydrolysate bacterium contained 13-methyl-tetradecanoic acid (iso- were analyzed by thin-layer chromatography (TLC) us- C15:0), 14-methyl-pentadecanoic acid (iso-C16:0), palmitic ing silica gel 60 aluminium plates (Merck, Darmstadt, acid (hexadecanoic acid) (C16:0), 15-methyl-hexadecanoic Germany). Lipids were developed with chloroform/metha- acid (iso-C17:0) as major fatty acids as well as C20:4, and nol/acetic acid/water (25:15:4:2, v/v/v/v), and visualized the amounts of these fatty acids were determined to be by spraying 10% H2SO4/ethanol and heating. Lipids were similar by all three methods. Only small amounts of 3- separated by silica gel column chromatography with the hydroxy fatty acids, 3-OH-C16:0 and 3-OH-iso-C17:0, were elution solvent, chloroform, chloroform/methanol (4:1, v/ identified, and their amounts were determined to be rela- v), chloroform/methanol (2:1, v/v), and chloroform/metha- tively higher by Method III. In addition to those fatty ac- nol (1:1, v/v) in a stepwise manner. ids, two large peaks with slightly different retention times (difference of about 0.06 min) (described as Unknown 1 Chemical reactions. Lipids solubilized in 2 ml of petro- and Unknown 2 in Table 1) were detected by Methods II leum ether were hydrogenated in a closed glass vial with and III. These peaks were also detected, but much smaller 5 mg platinum black. The atmosphere was exchanged with in the chromatogram of Method I, suggesting that Method hydrogen gas, and stirred for 40 min at room temperature. I was not suitable for determining these peaks. The Un- Methanolysis was performed by heating (100°C, 2.5 h) known 1 and Unknown 2 peaks gave EI-mass spectra with 1.0 M HCl/methanol prepared by dilution from 5% shown in Figs. 1A and 1B, respectively. The spectrum of HCl/methanol. Hydroxyl and amino groups of lipids were Fig. 1B showed mass fragment peaks m/z 90 and 103, peracetylated with one ml of acetic anhydride/pyridine which were characteristic peaks of 2-hydroxy-fatty acid (1:1, v/v) by heating (100°C, 30 min) in a closed vial. methyl ester. If the peak of m/z 300 was assumed to be a Acetylated derivatives were analyzed by GLC and GC- molecular peak (M), the fragment peak m/z 241 could be MS. a M-59 peak, which was also a characteristic fragment of 102 KAWAHARA, IIDA, and YOKOTA

Fig. 1. EI-mass spectra of unknown peaks of cellular fatty acids detected by GLC. Cellular fatty acids were analyzed by Method III, and two unknown peaks detected by GLC was analyzed by GC-MS. A. Unknown 1; B. Unknown 2.

2-hydroxy-fatty acid methyl ester derived from C1-C2 acid-fatty acid complex such as ornithine lipid (Ratledge cleavage. These data indicated that the Unknown 2 was 2- and Wilkinson, 1988), and, in those lipids, fatty acids are hydroxy-heptadecanoic acid (2-OH-C17:0) methyl ester. bound to amino groups of other components through al- Additionally, the retention time of this peak was found to kaline-stable amide linkage, cellular lipids of A. marina be shorter than that of the normal (non-branched) chain were hydrolyzed with 0.1 M NaOH at 100°C for 30 min, 2-OH-C17:0 methyl ester, which could be deduced from and the remaining lipids were analyzed by TLC. As shown the retention time data of authentic non-branched 2-hy- in Fig. 2, two alkaline-stable lipids were detected at Rf droxy-tetradecanoic acid (2-OH-C14:0) and 2-OH-C16:0. 0.7 and 0.5 (Lane 2). These lipids were purified by silica These data strongly suggested that the fatty acid was 2- gel column chromatography (Fig. 2, Lanes 3 and 4), and hydroxy-15-methyl-hexadecanoic acid (2-OH-iso-C17:0). fatty acids in these purified fractions were analyzed. The The EI-mass spectrum of Unknown 1 (Fig. 1A) was simi- more hydrophobic lipid (Rf 0.7) contained about the same lar to Fig. 1B, but it lacked m/z 90 and 103, and m/z 239 amounts of 2-OH-iso-C17:0 and 2-OH-iso-C17:1 in addition was present instead of m/z 241. Some of the other frag- to other cellular fatty acids liberated by alkaline hydroly- ments were two mass units smaller than those of Fig. 1B. sis and contaminated to this preparation, which were re- The fatty acid preparation was then hydrogenated to ex- moved by further purification using preparative TLC. The amine the difference of the two mass units, and as a re- lipid shown in Lane 4 (Rf 0.5) contained mainly 2-OH- sult, the peak of Unknown 1 disappeared, and only the iso-C17:1 with a trace amount of 2-OH-iso-C17:0. peak of Unknown 2 (2-OH-iso-C17:0) remained. These re- sults clearly indicated that Unknown 1 was 2-hydroxy- MALDI-TOF MS of 2-hydroxy-fatty acid containing 15-methyl-hexadecenoic acid (2-OH-iso-C17:1). As shown lipids in Table 1, these two 2-hydroxy-fatty acids were major 2-hydroxy-fatty acid containing lipids were analyzed by cellular fatty acids of A. marina. MALDI-TOF MS. Several peaks were detected in the spec- trum of the lipid of Rf 0.7, probably because of impurity, Purification of 2-hydroxy-fatty acid containing lipids but the highest peak was m/z 598.5 (Fig. 3), and the two As most common bacterial lipids that contain hydroxy- mass units larger peak, m/z 600.6 was also present. The fatty acids are lipopolysaccharide, ceramide, or an amino lipid (Rf 0.5) gave two major peaks of m/z 704.6 and m/z 2-OH-fatty acids in Aureispira marina 103

602.6 (data not shown), but this lipid was not investigated gested the presence of several double bonds in this lipid. further in this study. These signals disappeared after the lipid was hydrogen- The lipid (RF 0.7) was analyzed also by 1H-NMR. The ated. This hydrogenated lipid was again analyzed by strong signals around 1.2 ppm in the spectrum indicated MALDI-TOF MS, and the spectrum shown in Fig. 4 was the presence of hydrocarbon (-CH2-) chain(s), and many obtained. There were four main peaks in the spectrum, complex signals observed between 5.0 and 6.0 ppm sug- and the difference of these peaks were about 22 or 14 mass units, suggesting that there were two similar lipids (des- ignated M1 and M2) in this lipid preparation with the dif- + ference of one CH2, and both of them gave [M+H] and [M+Na]+ peaks as indicated in Fig. 4. The differences of molecular weight between the peaks in Figs. 3 and 4 were six or eight mass units for both M1 and M2, if we assumed that the peaks in Fig. 3 were [M+Na]+, indicating that three double bonds were present in the lipid components bound to 2-hydroxy-fatty acids. The lipids contained 2-OH-iso-C17:0 as a sole fatty acid after hydrogenation. Therefore, one of the candidates for the additional lipid component was dihydrosphingosine. Dihydrosphingosine of C18 carbon chain and that of C19 carbon chain could form ceramide molecules with 2-OH- iso-C17:0 which is consistent with the molecular weights calculated from the spectrum in Fig. 4. The presence of dihydrosphingosine was therefore examined in the follow- ing experiments.

Fig. 2. TLC profile of cellular lipids and alkaline stable lipids extracted Identification of dihydrosphingosine T from A. marina IAM 15389 . The hydrogenated lipid preparation was decomposed by Cellular lipids were extracted and treated by 0.1 M NaOH to prepare methanolysis, and the lipid fraction that showed ninhy- alkaline stable lipids, and analyzed by TLC. Lanes 1, cellular lipids; 2, drin-positive reaction was purified by preparative TLC, alkaline stable lipids; 3, purified alkaline stable lipid of Rf 0.7; 4, puri- fied alkaline stable lipid of Rf 0.5. and peracetylated. The GC-MS analysis of peracetylated

Fig. 3. MALDI-TOF mass spectrum of an alkaline stable lipid prepared from A. marina IAM 15389T. Alkaline stable lipids were purified, and a lipid showing Rf 0.7 by TLC was analyzed by MALDI-TOF MS with a positive mode. 104 KAWAHARA, IIDA, and YOKOTA

Fig. 4. MALDI-TOF mass spectrum of a hydrogenated alkaline stable lipid. The lipid analyzed in Fig. 3 (Rf 0.7 by TLC) was hydrogenated with hydrogen gas and catalyst, and analyzed again by MALDI-TOF MS with a positive mode.

Fig. 5. EI-mass spectra of derivatives in peracetylated sphingosine preparation. The hydrogenated lipid analyzed in Fig. 4 was decomposed by methanolysis, and the liberated sphingosines were peracetylated. Minor and major peaks detected by GLC were analyzed by GC-MS. A. a minor peak; B. a major peak. 2-OH-fatty acids in Aureispira marina 105 lipid fraction gave the EI-mass spectra shown in Figs. 5A prominent features of cellular lipids of this bacterium is and 5B, which corresponded to the minor and the major the high content of C20:4, and this polyunsaturated fatty peaks of the GLC chromatogram. The retention time of acid was shown to be a component of several the minor peak and the spectrum shown in Fig. 5A with a glycerophospholipids and to be located probably at the C2 characteristic fragment of m/z 144 were identical to those position of glycerol (Sakoda, K. and Kawahara, K., of the derivative of authentic C18-erythro- unpubl.). Both the ceramide containing the polyunsatu- dihydrosphigosine, indicating that the minor peak derived rated sphingosine found in this study and C20:4-contain- from the component of M1 in Fig. 4 was peracetylated C18- ing glycerophospholipids should influence the fluidity of dihydrosphingosine. The major peak showed a retention the membrane, and could be closely related with the glid- time 0.43 min later than the minor peak, and the mass spec- ing feature of this bacterium. Our data also showed that trum (Fig. 5B) was similar to that in Fig. 5A. The frag- the ratio of 2-OH-C17:1 to 2-OH-C17:0 increased when the ment peaks (m/z 262 and 308) were 14 mass units larger bacterium was cultured at a lower temperature (20°C) than the corresponding peaks in Fig. 5A (m/z 248 and 294). (Yoshihama, H. and Kawahara, K., unpubl.), which also From the difference in retention time and the fragment suggests the relation between lipid unsaturation and mem- peaks, although not identified clearly, the derivative of brane fluidity. the major peak was identified as peracetylated C19- As shown in Fig. 2 (Lane 4), there was another 2-hy- dihydrosphingosine. From these results, hydrogenated droxy-fatty acid containing lipid besides the ceramide in lipids detected in Fig. 4 were proved to be ceramide mol- this bacterium. This lipid contains mainly 2-OH-C17:1, and ecules composed of 2-OH-iso-C17:0 and C18- less hydrophobic than the ceramide. Although it gives clear dihydrosphingosine (M1) or C19-dihydrosphingosine (M2), peaks by MALDI-TOF MS analysis, we do not have and the original lipids present in A. marina were the cor- enough data at present to speculate the structure of this responding ceramides with three double bonds in the sphin- lipid. Further study on this lipid is highly required to clarify gosine portion, and one more double bond if the fatty acid the whole lipid profile of this unique marine bacterium. component was 2-OH-iso-C17:1. Acknowledgments Discussion The authors greatly thank Kazuya Sakoda, Hiroyuki Yoshihama, Nanako Oi, and other students of Kanto Gakuin University, for their For the analysis of cellular fatty acid composition, the contributions to this study. MIDI method (Method I in this study) is used in most of the recent studies. However, 3-hydroxy-fatty acids in the References lipopolysaccharide of Gram-negative bacteria (Zähringer et al., 1999) or 2-hydroxy-fatty acids in bacterial Alexander, C. and Rietschel, E. Th. (2001) Bacterial lipopolysaccharides glycosphingolipids (Kawahara et al., 1999) cannot be prop- and innate immunity. J. Endotox. Res., 7, 167–202. Glucksman, A. M., Skipper, H. D., Brigmon, R. L., and Domingo, J. W. erly determined because the amide-linkage between amino (2000) Use of the MIDI-FAME technique to characterize groups in sugar or sphingosine moieties and caroboxyl groundwater communities. J. Appl. Microbiol., 88, 711–719. groups of fatty acids is rather stable against alkaline hy- Hosoya, S., Arunpairojana, V., Suwannachart, C., Kanjana-Opas, A., drolysis or saponification, which is used in the MIDI and Yokota, A. (2006) Aureispira marina gen. nov., sp. nov., a glid- method. In the present study, we used two other methods ing, arachidonic acid-containing bacterium isolated from the south- ern coastline of Thailand. Int. J. System. Evol. Microbiol., 56, 2931– that are based on transesterification in acidic conditions 2935. or acid hydrolysis. By using these methods, we could iden- Hosoya, S., Arunpairojana, V., Suwannachart, C., Kanjana-Opas, A., and Yokota, A. (2007) Aureispira maritima sp. nov., isolated from tify two 2-hydroxy-fatty acids (2-OH-iso-C17:1 and 2-OH- marine barnacle debris. Int. J. System. Evol. Microbiol., 57, 1948– iso-C17:0) that were not reported in a previous paper (Hosoya et al., 2006), and could show that 2-OH-iso-C 1951. 17:1 Kawahara, K., Seydel, U., Matsuura, M., Danbara, H., Rietschel, E. Th. was the most abundant fatty acid in A. marina. However, et al. (1991) Chemical structure of glycosphingolipids isolated from it should be noted that these 2-hydroxy-fatty acids were Sphingomonas paucimobilis. FEBS Lett., 292, 107–110. detected in small amounts even by the MIDI method in Kawahara, K., Kuraishi, H., and Zähringer, U. (1999) Chemical struc- this study. The difference of the results between these two ture and function of glycosphingolipids of Sphingomonas spp and studies might be explained by a slight difference of heat- their distribution among members of the α-4 subclass of Proteobacteria. J. Ind. Microbiol. Biotechnol., 23, 408–413. ing conditions derived from laboratory equipment used in Komagata, K. and Suzuki, K. (1987) Lipid and cell wall analysis in each study. bacterial systematics. Meth. Microbiol., 19, 161–207. Oyaizu, H. and Komagata, K. (1983) Grouping of Pseudomonas spe- The position of the double bond in 2-OH-C17:1 is not yet determined, but when we compare the EI-MS spectra (Figs. cies on the basis of cellular fatty acid composition and the quinone 1A and 1B), an intensity difference of m/z 81 and 83 is system with special reference to the existence of 3-hydroxy fatty acids. J. Gen. Appl. Microbiol., 29, 17–40. noticed. If the fragment m/z 83 derives from the cleavage Ratledge, C. and Wilkinson, S. G. (1988) Fatty acids, related and de- between the 6th and 7th carbon from the methyl terminal, rived lipids. In Microbial Lipids Volume 1, ed. by Ratledge, C. and Wilkinson, S. G., Academic Press, London, pp. 23–53. the double bond position may be between C11 and C12 (n- 6). The position of double bonds in the fatty acid, and also Zähringer, U., Lindner, B., and Rietschel, E. Th. (1999) Chemical struc- in sphingosines, needs to be clarified and is under investi- ture of lipid A: recent advances in structural analysis of biologi- cally active molecules. In Endotoxin in Health and Disease, ed. by gation in our laboratory. Brade, H., Opal, S. M., Vogel, S. N., and Morrison, D. C., Marcel The present study demonstrated that A. marina contains Dekker, New York, pp. 93–114. large amounts of ceramides as cellular lipids. One of