Formation of gold nanoparticles by glycolipids of Lactobacillus casei

Gold nanoparticles have particular properties distinct from those of bulk gold crystals, and such nanoparticles are used in various applications in optics, catalysis, and drug delivery. Many reports on microbial synthesis of gold nanoparticles have appeared. However, the molecular details (reduction and dispersion) of such synthesis remain unclear. In the present study, we studied gold nanoparticle synthesis by Lactobacillus casei. A comparison of L. casei components before and after addition of an auric acid solution showed that the level of unsaturated lipids decreased significantly after addition. NMR and mass spectrum analysis showed that the levels of diglycosyldiacylglycerol (DGDG) and triglycosyldiacylglycerol (TGDG) bearing unsaturated fatty acids were much reduced after formation of gold nanoparticles. DGDG purified from L. casei induced the synthesis of gold nanoparticles in vitro. These results suggested that glycolipids, such as DGDG, play important roles in reducing Au(III) to Au(0) and in ensuring that the nanoparticles synthesized remain small in size. Our work will lead to the development of novel, efficient methods by which gold nanoparticles may be produced by, and accumulated within, microorganisms.


Results
Reaction of auric acid solution with L. casei cells. Concentrated L. casei cells (2.0 g/L) were washed with distilled water to remove the culture medium and incubated in an auric acid solution (0.5 mM K[AuCl 4 ]) for 24 h. The colors of the control auric acid solution and L. casei suspension were transparently yellow and turbidly white, respectively, at this time (Fig. 1a,b). The suspension containing both L. casei cells and auric acid was violet in color after 24 h (Fig. 1c). The suspensions were next subjected to ultrasonication and a UV/VIS spectroscope was used to measure absorbance at wavelengths from 400 to 700 nm. The violet suspension had an absorption maximum at 534 nm (Fig. 1c). However, not all suspensions of L. casei cells mixed with auric acid were violet in color (Fig. S1) Identification of gold nanoparticles. To explore why the violet color developed, we subjected both Au(− ) and Au(+ ) L. casei samples to both transmission and scanning electron microscopy (TEM and SEM, respectively). The Au(− ) L. casei cells were baculiform in shape (1.9 μ m in length, 0.4 μ m in width) (Fig. 2a). However, many nanoparticles (7-56 nm in diameter) were evident in Au(+ ) cells (Fig. 2b). The sizes of synthesized particles were measured by ImageJ. The size value of the highest frequency was about 30 nm. The size distribution graph was described in Fig. S2. Similarly, SEM of Au(− ) L. casei showed only cells that were baculiform in shape (Fig. 2c), whereas the L. casei Au(+ ) cells had many bright surface-attached nanoparticles (Fig. 2d). Energy dispersion spectroscopy (EDS) showed that the nanoparticles were composed of Au (Fig. 2e). X-ray diffraction (XRD) measurements on dried L. casei Au(+ ) cells revealed several peaks corresponding to gold crystals with a face-centered cubic structure (Fig. 2f). These results indicated that L. casei cells mixed with an auric acid solution produced gold nanoparticles. As in liquid suspension, the gold nanoparticles absorbed light of wavelengths between 500-600 nm via surface plasmon resonance (SPR), the violet color of the mixture of L. casei with auric acid was attributable to formation of such nanoparticles.
Identification of key organic molecules in L. casei for the synthesis of gold nanoparticles. After ultrasonication, the supernatants of L. casei Au(− ) and Au(+ ) cells were turbidly white with an extreme meniscus and transparently violet without an extreme meniscus, respectively (Fig. S3). We assumed that these differences in physical properties were probably attributable to differences in lipid composition. L. casei Au(− ) and Au(+ ) cells were suspended in chloroform/methanol solution (2:1, v/v) to extract lipids, and these extracts were subjected to [ 1 H]-NMR analysis. Comparison of the [ 1 H]-NMR spectra showed that the heights of the 1.94-and 5.35-ppm peaks were significantly lower in the Au(+ ) than the Au(− ) extract (Fig. 3a,b), suggesting that particular lipids may have been associated with gold nanoparticle synthesis. We next subjected both lipid extracts to two-dimensional thin-layer chromatography (TLC). A comparison of the spot patterns showed that two distinct spots (1 and 2, circled in Fig. 3c,d) in the Au(− ) extract were fuzzy, smeared, and tailed in the Au(+ ) extract. To identify the lipids of spots 1 and 2, the silica gel regions bearing these spots were scraped off the TLC plate, the lipids extracted was dissolved into a chloroform/methanol solution (2:1, v/v), and subjected to MS analyses. MALDI-TOF MS showed that spot 1 contained four major peaks at 939.6, 953.6, 967.6, and 981.6 (m/z), and certain isotope peaks (Fig. 4a). These peaks were also evident in the Au(− ) L. casei extract (Fig. 4b). However, especially in the region around 930-990 (m/z), the peak profiles differed dramatically (Fig. S4). The intensities of all such peaks (m/z) were reduced in the Au(+ ) extract. Four major peaks [at 941.6, 955.6, 969.3, and 983.6 (m/z)] were detected (Fig. 4c). In the Au(+ ) extract, the heights of these four peaks were twice those of the Au(− ) extract. MS analyses (ESI-TOF-MS spectroscopy and MALDI-TOF/TOF MS spectroscopy) revealed that the four major peaks at 939.6, 953.6, 967.6, and 981.6 (m/z) corresponded to diglycosyldiacylglycerol (DGDG) with two unsaturated fatty acids, the carbon chains of which differed in length (Fig. S5, Fig. 4d, and Table 1). Two larger masses (around m/z 969.3) detected in Au(+ ) L. casei also corresponded to DGDG, but with one saturated and one unsaturated fatty acid ( Fig. S5 and Table 1).
NMR spectrum analyses revealed the structures of the DGDG saccharides. H1-6 and H1′ -6′ indicate the protons of galactose and glucose, respectively. C1-6 and C1′ -6′ indicate the carbons of galactose and glucose, respectively. The DQF-COSY spectrum revealed anomeric protons at 4.96 ppm (Fig. S6). The 4.96 ppm signal had two correlations, suggesting that two anomeric protons from two monosaccharides overlapped. The signals at 5.2 ppm are from two protons of carbons bound to the glycerol ester. The signal at 5.3 ppm is from a proton of a carbon bound to the glycerol ether. DQF-COSY signal assignment also revealed the signals from H1-5 of galactose and H1′ -H4′ of glucose. The HSQC spectrum revealed all carbons of galactose and glucose (Fig. S7). The HMBC spectrum revealed a correlation between C1 and H2′ , suggesting that carbon 1 of galactose was bound to carbon 2′ of glucose (Fig. S8). We also noted a correlation between C1′ and the glycerol protons, suggesting that carbon 1′ of glucose was bound to glycerol (Fig. S9). These results suggested that DGDG (Fig. S9). On the other hand, MALDI-TOF MS analysis showed that the extract of spot 2 contained four major peaks at 1,101.2, 1,115.2, 1,129.2, and 1,143.2 (m/z), and certain isotope peaks (Fig. S10). The intensities of these four peaks were reduced  Table 1).
Synthesis of gold nanoparticles using purified DGDG. DGDG extracted from the silica gel around spot 1 was solubilized in ethanol, mixed with auric acid, and incubated at 37 °C for 24 h. At this time, the color of   (Fig. 4e, Fig. S12). The absorption maximum varied by the DGDG concentration. TEM showed that gold nanoparticles were synthesized (Fig. 4f). The diameter of the gold nanoparticles formed was slightly greater than that of nanoparticles synthesized by L. casei cells. The size value of the highest frequency was about 60 nm (Fig. S12b). On the other hand, the commercially available purified DGDG (caDGDG: 1,2-diacyl-3-O-[(α -D-galactosyl(1-6)-β -D-galactosyl]glycerol with two chains of fatty acids (C16:1 and C20:5))) was also used for the formation of gold nanoparticles. The color of the solution (5 mg/L, 10 mg/L and 100 mg/L caDGDG) also became violet. The absorption spectra of caDGDG were very similar to those of purified DGDG from TLC spot. TEM observation showed that gold nanoparticles were synthesized (Fig. S13). The size value of the highest frequency was about 60 nm (Fig. S13c). These results suggested that the glycolipid such as DGDG is one of the key factors for the formation of gold nanoparticles in L. casei.

Discussion
Although many previous studies have shown that L. casei cells produce gold nanoparticles, the molecular mechanism remains unknown. In the present study, we sought to identify organic molecules involved in gold nanoparticle synthesis.
In the previous report, gold nanoparticles formed in the culture medium 24 . However, we found that such synthesis was unreliable, because the culture medium contains many components either inhibiting or promoting nanoparticle synthesis, and such synthesis varied by the growth stage of L. casei. To identify the key organic molecules involved, we found it necessary to remove the culture medium. Therefore, we washed the cells with distilled water and then mixed the washed cells with auric acid. We found that the ratio of L. casei cell numbers to auric acid concentration was important in terms of nanoparticle synthesis. Excess auric acid inhibited L. casei metabolism, and an excessive number of L. casei cells resulted in the production of large amounts of reducing agents, which may induce polymerization of certain cellular components, in turn inhibiting gold nanoparticle synthesis.
DGDG and TGDG were involved in gold nanoparticle synthesis by L. casei. These glycolipids are probably components of the membrane lipid bilayer, replacing phospholipids; glycolipid synthesis requires less energy than phospholipid synthesis. Therefore, DGDG and TGDG of L. casei are probably present in the cell membrane [25][26][27] . TEM and SEM showed that gold nanoparticles formed on the cell surface, suggesting that the reduction reaction [Au(III) to Au(0)] occurred on the cell membrane containing DGDG and TGDG. After nanoparticle synthesis, most DGDG and TGDG with two unsaturated fatty acids disappeared; small amounts of DGDG and TGDG containing single unsaturated fatty acids remained. One previous report also found that unsaturated fatty acids present in vegetable oil reduced metal ions and bound to the surface of nanoparticles 28 . Therefore, the unsaturated bonds in the fatty acids of DGDG and TGDG may be associated with reduction of Au(III).
Kumar et al. 28 , suggested that the hydrogen in the allylic position of an unsaturated fatty acid can be desorbed by light or an oxidizing agent, creating a fatty acid radical 28 . The electron of the hydrogen radical may be transferred to Au(III) in a reducing reaction. Au(III) then becomes the unstable Au(II); two Au(II) ions become an Au(III) and an Au(I) ion via a disproportionation reaction. Au(I) then receives an electron from a hydrogen radical to become Au(0). On the other hand, autoxidation creates lipid peroxides in DGDG and TGDG; these are ultimately decomposed in a radical chain reaction. According to this mechanism, the speculative autoxidation process for the formation of gold nanoparticles was described in Fig. S14. The reaction between glycolipids containing unsaturated fatty acids (GLU) and Au(III) is shown below: degraded products Au(I) GLU Au(0) GLU Au(0) gold nanoparticles In the present study, incubation of DGDG with auric acid in vitro resulted in production of gold nanoparticles, suggesting that DGDG functioned not only as a reducing agent, but also as a dispersing agent. The carboxylic groups of citric acid bind to the surface of Au(0) to create nanoparticles 11 . The carboxylic groups of the DGDG fatty acids may interact with the Au(0) surface to stabilize nanoparticles. However, the gold nanoparticles synthesized from DGDG were slightly larger than those synthesized by L. casei cells, suggesting that any dispersing m/z capacity of DGDG may be limited; other (as yet unknown) dispersing agents within L. casei cells may also play roles in inhibiting Au(0) aggregation and crystal growth. Further work is thus needed to reveal exactly how L. casei forms gold nanoparticles. Identification of the key organic molecules may allow modification of L. casei genes, permitting such recombinants to produce gold nanoparticles more effectively. In the future, L. casei strains modified along these lines may find applications in metal recycling and phytoremediation.

Synthesis of gold nanoparticles by L. casei. L. casei (strain JCM1134) was purchased from RIKEN
Microbe Division. L. casei was cultured in the MRS liquid medium (1% casein peptone, tryptic digest/1% meat extract/0.5% yeast extract/2% glucose/0.1% tween80/0.2% K 2 HPO 4 /0.5% CH 3 COONa/0.2% diammonium citrate/0.02% MgSO 4 · 7H 2 O/0.005% MnSO 4 · 5H 2 O) at 37 °C for 8 h of preculture and for 14 h of main culture with shaking (300 rpm). The correlation between OD 600 and dry bacterial cell weight was calculated to estimate the amounts of bacterial cells in the solution (Fig. S15). were coated with carbon before observations. X-ray microanalysis was performed using energy-dispersive spectroscopy (EDS) with an ultra-thin window detector (HORIBA) equipped to the SEM. The particle sizes observed by TEM were measured by ImageJ. The bigger precipitates more than 100 nm were removed from the counts.

X-ray diffraction (XRD) measurements. The dried powders of L. casei cells incubated in auric acid
solution were filled up in a nonreflecting silicon sample holder for XRD. XRD patterns were collected using a RINT-Ultima + diffractometer (RIGAKU) with CuKα radiation emitted at 40 kV and 30 mA.
Extraction of lipid from L. casei. The extraction of lipid was performed according to the Bligh & Dyer method 29 . Chloroform/methanol (2:1) solution was added to the precipitation of bacterial cells with or without the treatments of auric acid solution. After suspension and centrifugation, chloroform layer was extracted to remove the water layer and precipitation using the 22G syringe (TERUMO). For the NMR analyses, the deuterated chloroform/methanol instead of normal chloroform/methanol was used to extract the lipid from L. casei.

MS analysis.
Mass spectra were measured on a matrix-assisted laser-desorption ionization-time-of-flight mass spectrometer (ultraflex MALDI-TOF/TOF, Bruker). The sample solution was mixed with 500 mM 2,5-dihydroxybenzoic acid in chloroform/methanol (1:1) solution as the matrix and dried on the plate. High resolution mass spectra were measured on a time-of-flight mass spectrometer JMS-T100LC AccuTOF (JEOL) equipped with an electrospray ionization source in the positive ion mode. The extracted samples were dissolved in methanol and applied to the mass spectrometer. The settings of the instrument were calibrated using sodium trifluoroacetate. The ion-source temperature was 250 °C. The mass analyzer was scanned from m/z 300 to 1500 for the full scan analysis.
Extraction of diglycosyldiacylglycerol (DGDG). The spot of TLC analysis containing DGDG was scraped from the surface of thin chromatography layer. DGDG was extracted from the powder of silica gel using chloroform/methanol solution (2:1). The extract was evaporated to concentrate DGDG. The purified DGDG powder was dissolved in ethanol.
Synthesis of gold nanoparticles using purified DGDG. DGDG extraction from the TLC plate dissolved in 40 μ l ethanol was applied to 960 μ l of auric acid solution (final concentration of K[AuCl 4 ]: 250 μ M). The mixture solution of DGDG and auric acid was incubated at 37 °C for 24 h. 40 μ l of ethanol without DGDG was mixed with 960 μ l of auric acid solution (final concentration of K[AuCl 4 ]: 250 μ M) as a negative control experiment. UV/VIS spectra of the supernatant were measured and the synthesized gold nanoparticles were observed by TEM. On the other hand, we used the commercially available purified DGDG (caDGDG) for the same experiment described above. caDGDG is 1,2-diacyl-3-O-[(α -D-galactosyl(1-6)-β -D-galactosyl]glycerol with two chains of fatty acids (C16:1 and C20:5) (Carbosynth).