Reconstruction of metabolic module with improved promoter strength increases the productivity of 2-phenylethanol in Saccharomyces cerevisiae

Background 2-phenylethanol (2-PE) is an important aromatic compound with a lovely rose-like scent. Saccharomyces cerevisiae is a desirable microbe for 2-PE production but its natural yield is not high, and one or two crucial genes’ over-expression in S. cerevisiae did not improve 2-PE greatly. Results A new metabolic module was established here, in which, permease Gap1p for l-phenylalanine transportation, catalytic enzymes Aro8p, Aro10p and Adh2p in Ehrlich pathway respectively responsible for transamination, decarboxylation and reduction were assembled, besides, glutamate dehydrogenase Gdh2p was harbored for re-supplying another substrate 2-oxoglutarate, relieving product glutamate repression and regenerating cofactor NADH. Due to different promoter strengths, GAP1, ARO8, ARO9, ARO10, ADH2 and GDH2 in the new modularized YS58(G1-A8-A10-A2)-GDH strain enhanced 11.6-, 15.4-, 3.6-, 17.7-, 12.4- and 7.5-folds respectively, and crucial enzyme activities of aromatic aminotransferases and phenylpyruvate decarboxylase were 4.8- and 7-folds respectively higher than that of the control. Conclusions Under the optimum medium and cell density, YS58(G1-A8-A10-A2)-GDH presented efficient 2-PE synthesis ability with ~ 6.3 g L−1 of 2-PE titer in 5-L fermenter reaching 95% of conversation ratio. Under fed-batch fermentation, 2-PE productivity at 24 h increased 29% than that of single-batch fermentation. Metabolic modularization with promoter strategy provides a new prospective for efficient 2-PE production.


Background
2-Phenylethanol (2-PE) is a kind of aromatic alcohol with a rose-like fragrance. It naturally exists in the essential oils of many plants and flowers, and is also a natural product of some fermented foods [1,2]. 2-PE has a wide range of applications in the perfumery, cosmetics and fermented food industries for its lovely scent [3], and also can be used as a substrate for the synthesis of other flavors or pharmaceutical compounds [4]. Compared to the traditional methods such as chemical synthesis and natural extraction, production of 2-PE by microbial fermentation is a breakthrough which is efficient and economic [5]. As a significant eukaryotic model organism with high tolerance towards many stress factors, Saccharomyces cerevisiae is a desirable strain for 2-PE production [4,6].
Two principal pathways exist in S. cerevisiae for 2-PE synthesis. Shikimate pathway is a long de novo synthesis pathway with multiple-branches and a variety of feedback inhibitions, whose 2-PE production is very low. When l-phenylalanine (l-Phe) is used as the sole nitrogen source in the medium, 2-PE can be synthesized via Ehrlich pathway (Fig. 1a), in which aromatic aminotransferases (Aro8p and Aro9p), phenylpyruvate decarboxylase (Aro10p) and alcohol dehydrogenases mainly participate [4,7]. S. cerevisiae can adjust the physiological metabolism according to the content and quality of nitrogen source [8]. Gap1p, a general amino acids permease [9], can be induced to express and transport l-Phe into yeast cells [9,10]. In Ehrlich pathway, transamination reaction catalyzed by Aro8p and Aro9p [11,12] and decarboxylation mainly catalyzed by Aro10p [13] are very crucial for 2-PE synthesis. Aro8p expression is regulated by general control of amino acid biosynthesis, and Aro9p and Aro10p are induced to express by high concentrations of aromatic amino acids [14][15][16][17]. Enhancing Aro8p, Aro9p or Aro10p expression could promote 2-PE production [10,17,18]. Some regulatory factors for Ehrlich pathway have been partly elucidated in recent years. Aro80p, a zinc finger transcriptional activator in the Zn2Cys6 family, can accept the inducing signals from aromatic amino acids and bind to the repeat sequences on promoters of ARO9 and ARO10 to activate them expressing [11,14]. Expression of ARO9 and ARO10 can also be activated by Gat1p and Gln3p [10,19]. Gln3p and Gat1p are GATA-family zinc finger transcriptional activators regulated by global nitrogen quality control known as nitrogen catabolite repression (NCR) [19][20][21][22][23]. In a good nitrogen condition, Gln3p and Gat1p are sequestered in the cytosol. When the nitrogen source is poor, Gln3p and Gat1p translocate to the nucleus, activating the NCR-sensitive genes such as ARO9 and ARO10 as well as GAP1 with GATAsequences motif resulting in 2-PE synthesis increased [8,10,24]. Wuster and Babu [25] ever predicted that transcription factors Cat8p and Mig1p were related to Modified cassettes with different promoter strengths. c Diagram of new expression module some genes' expression responsible for aromatic alcohol synthesis. Cat8p, a zinc cluster transcriptional activator, can bind to the carbon source-responsive element (CSRE) motifs to make key gluconeogenic enzymes derepressed and activated [26][27][28][29][30]. Mig1p is a transcription factor involved in glucose repression [31,32]. In recent research, Cat8p and Mig1p were proved to be regulatory factors of 2-PE synthesis. Cat8p over-expression or Mig1p deletion could positively regulate transaminase and phenylpyruvate decarboxylase activities leading 2-PE synthesis enhanced [18].
The methods for improving 2-PE yield of yeast mainly include traditional breeding methods, fermentation conditions optimization and in situ 2-PE removal [33][34][35][36][37][38][39]. With the elucidation of regulatory mechanism of 2-PE biosynthesis, metabolic engineering strains were constructed, whose 2-PE synthetic ability were improved in different degrees [10,17,18]. Currently, the engineering strains were mainly operated by over-expressing one or two genes of transaminases, phenylpyruvate decarboxylase, alcohol dehydrogenases, transcription factors or permease [10,17,18,40,41], however, these genetically modification did not improve 2-PE yield greatly, probably due to the limitation of substrate transportation or unfluent reactions in Ehrlich pathway or some repression.
Up to now, no globally enhanced synthetic net including substrate supplement and all reaction steps has been considered and investigated. In this study, a new metabolic module was established, in which, Gap1p for l-Phe transportation and three enzymes respectively responsible for transamination, decarboxylation and reduction were assembled, besides, glutamate dehydrogenase was also harbored aiming for relieving the repression of glutamate and re-supplying another substrate 2-oxoglutarate, in the meantime, cofactor NADH was regenerated. Promoter strategy was adopted in this work. The genes in the new module cassette were expressed under different promoter strengths. Based on these modifications, the new modularized S. cerevisiae presented fluent and high efficient 2-PE synthesis ability.

New metabolic module design with different promoter strengths
Natural 2-PE synthesized in S. cerevisiae is low, overexpression of one or two crucial genes up-regulated 2-PE synthesis via Ehrlich pathway but did not improve 2-PE yield greatly [10,17,18,40], therefore, new strategy should be considered. Up to now, most of the regulatory transcription factors such as Aro80p, Gat1p, Gln3p and Cat8p were reported to promote 2-PE synthesis by targeting at genes of amino acid permease, crucial enzyme aromatic aminotransferase and phenylpyruvate decarboxylase [10,18,24], hence, it is appropriate and necessary in the new design to include relative elements responsible for substrate transportation and crucial reactions. Module balance should also be considered, thus, non-rate-limiting alcohol dehydrogenase was contained. The by-product glutamate produced from the first transamination reaction is a kind of good nitrogen for yeast probably repressing Ehrlich pathway to synthesize 2-PE, in order to relieve the repression and make the whole pathway fluently, non-pathway-specific glutamate dehydrogenase Gdh2p was further designed to the metabolic module. Harboring Gdh2p would lead cofactor NADH regeneration making the third reduction in Ehrlich pathway more fluent (Fig. 1a, c). The mechanism by which 2-PE is exported from the cells into the culture medium remains unknown [7]. It is conceivable that export occurs by simple passive diffusion across the lipid bilayer [42], hence, no relative gene for exportation was designed in our module.
Promoter strength is significant for gene expression, different design among target genes according to their specificity and significance to the whole metabolism should be favorable for 2-PE synthesis. ADH1p, PGK1p and TEF1p are three constitutive strong promoters respectively from alcohol dehydrogenase I, 3-phosphoglycerate kinase and translational elongation factor EF-1 alpha. In our lab, constitutive and tunable promoter library with strength variation of 1-19 folds has been established using synthetic hybrid promoter approach, in which enhancer elements such as transcription factors Rap1p, Gcr1p and/or Sgp1p binding sites and upstream activation sequences (UAS) were included [43,44]. UAS TEF-RP-TEF1p, one member in our promoter library derived from TEF1p is stronger than TEF1p. Similarly, UAS TEF-RG-ADH1p derived from ADH1p is stronger than ADH1p. Permease Gap1p is essential for substrate transportation, constitutive promoter ADH1p was used in it. The stronger promoter PGK1p was designed to harbor in the upstream of aromatic aminotransferase Aro8p for catalyzing the first pathway-specific reaction in Ehrlich pathway. Phenylpyruvate decarboxylase (Aro10p) is the rate-limiting enzyme, the strongest UAS TEF-RP-TEF1p was used to regulate it (Fig. 1b). In the third step of reduction reaction, one of the alcohol dehydrogenases ADHII is a glucose-repressible enzyme, for speeding up reduction step, constitutive and stronger promoter UAS TEF-RG-ADH1p was chosen to put to the upstream of ADH2 (Fig. 1b). Glutamate dehydrogenase Gdh2p was not the pathway-specific enzyme for 2-PE metabolism, it was designed to use its own promoter. In the new module, there are gradually increased promoting strengths to regulate gene expressing responsible for relative enzymes from substrate transportation to most important decarboxylation reaction. l-Phe should be absorbed fast by yeast cells, transamination and decarboxylation ought to be boosted. Over-expressed Gdh2p is probably helpful for removing glutamate repression and for increasing substrate 2-oxoglutarate. NADH regeneration from reaction catalyzed by Gdh2p and enhanced ADHII would accelerate reduction reaction to synthesize 2-PE efficiently. The global modularized up-regulation of enzyme activities was supposed to improve 2-PE synthesis ability in a large degree (Fig. 1c).

Construction of expression cassettes, screening of engineering yeast strains and verification
Based on above design, a series of recombined plasmids were constructed. In the combined plasmid pGAP1-ARO8, expression cassettes ADH1p-GAP1 and PGK1p-ARO8 were harbored (Fig. 2). In plasmid Y-ARO10-ADH2, expression cassettes UAS -TEF-RP -TEF1p-ARO10 and UAS -TEF-RG -ADH1p-ADH2 were contained (Fig. 2). Plasmid pYCUP-GDH2 carried GDH2 expression cassette (Fig. 2). The sequencing results indicated that the coding sequences of the above genes were the same as model strain S288C in NCBI.
Analysis of enzyme activities indicated that the most crucial enzymes aromatic aminotransferases and phenylpyruvate decarboxylase in the module strain YS58(G1-A8-A10-A2)-GDH were respectively 4.8-and 7-folds higher than that of the control strain (Table 1), reflecting that increased RNA levels resulted in increased enzyme proteins' level. Since GAP1 over-expression could upregulate mRNA level of ARO9 [10], increased activities of aminotransferases in the module strain should result not only from over-expressed ARO8 but also from ARO9 up-regulated by GAP1 ( Fig. 3a and Table 1). Enhanced phenylpyruvate decarboxylase should be the reflection of ARO10 increase that was led by both the strongest promoter in the module and GAP1 regulation on original ARO10. Increase of the two crucial enzyme activities were accordance with that of their mRNA levels ( Fig. 3a and Table 1). Cell growths of control strains YS58-YEp, YS58(YEp-YKA) and YS58(YEp-YKA-YCUP) were similar, but all the engineering strains grew faster than control strains (Fig. 3b), which might result from the acceleration of utilization of l-Phe nitrogen source [18]. The results of 2-PE analysis in tube fermentation indicated that with the number of over-expressed genes being added, 2-PE titers increased gradually. Among these strains, module strain YS58(G1-A8-A10-A2)-GDH produced the highest 2-PE titer compared to other strains, 5.65-folds higher than that of the control strain YS58(YEp-YKA-YCUP) (Fig. 3b). Considering that the raised biomass also contributed to total 2-PE concentration (g L −1 ) increase, specific production of 2PE (g/L/OD 600 ) was also compared. The results showed that single gene expression such as ARO8, ARO10 or GAP1 up-regulated the yeast 2-PE (g/L/OD 600 ) level 10-12%, two (ARO8 + ARO10) or three (ARO8 + ARO10 + ADH2) genes' expression promoted yeast cells to synthesize 2-PE 17-19% higher than the host, four (GAP1 + ARO8 + ARO10 + ADH2) and five (GAP1 + ARO8 + ARO10 + ADH2 + GDH2) genes' over-expression respectively led to 28 and 32% increase of 2-PE production than the control strain (Fig. 3b). The designed module strain with five genes presented the highest 2-PE synthesis ability, it was chosen for further test.

Fermentation optimization of modularized engineering strain
With plasmids increase in yeast cells, the metabolic overload may increase and the medium used in previous study might be innutritious. Hence fermentation conditions optimization was done. The results of 2-PE titers at different concentrations of glucose and l-Phe indicated that less or excessive carbon source and nitrogen substrate were both unfavorable for 2-PE synthesis, the optimum concentration of glucose was about 40 g L −1 , and the concentration of l-Phe should be less than 10 g L −1 (Fig. 4a). Some ions are necessary for cellular metabolism, for example, magnesium is activator for many enzymes, phosphate ion affect redox reaction and other metabolic synthesis that is generally necessary elements for cells. According to the previous reports, magnesium and potassium were necessary ions [10,35,45,46], while zinc and sodium had no significant affection on 2-PE synthesis in S. cerevisiae [47]. In our previous study, the medium condition suitable for our other engineering strains for 2-PE synthesis was established [10,18]. In preliminary experiments of this study, the affection of NaCl on 2-PE synthesis was tested, differently, our results showed that NaCl suppressed 2-PE synthesis in our module strain, not necessary as for strain reported by Table 1 Activities of crucial enzymes in different yeast strains a YS58-YEp, strain with control vector; YS58(ARO8-58), YS58(ARO10-58) and YS58(YEpAP58), strains with ARO8, ARO10 and GAP1 expression cassettes respectively; YS58(ARO8 + ARO10), ARO8 and ARO10 co-expression strain; YS58(A8-A10-A2), ARO8 + ARO10 + ADH2 co-expression strain; YS58(G1-A8-A10-A2), strain with GAP1 + ARO8 + ARO10 + ADH2 co-expression; YS58(G1-A8-A10-A2)-GDH, strain with GAP1 + ARO8 + ARO10 + ADH2 + GDH2 co-expression cassettes ** Significance of difference P < 0.01
Besides carbon, l-Phe substrate and inorganic salts, it has been reported that ascorbic acid (Vitamin C, Vc) could improve the redox reaction and suppress the generation of by-products during biotransformation of natural 2-phenylethanol [48]. Considering that the increase of reducing power might promote the third step reaction in Ehrlich pathway, ascorbic acid was added into the fermentation medium in this work. The results showed that appropriate amount of ascorbic acid (0.15 g L −1 ) was indeed helpful to 2-PE synthesis (Fig. 4b).
Moreover, synthesis of aromatic alcohols are also controlled by cell density and morphology. In our work, when cell density was less than OD 600 30, 2-PE was subjected to positive feedback regulation (Fig. 4c), which was similar to that in Chen and Fink's report [49]. The optimum cell densities for efficient 2-PE synthesis were about OD 600 25-30 with OD 600 6-9 of inoculum (Fig. 4c). When cell density was too high, 2-PE yield decreased. Especially when cell density was more than OD 600 40, the strain had a greater drop in 2-PE yield which might be Under the optimum condition, modularized YS58(G1-A8-A10-A2)-GDH strain was fermented in flask and redox reaction was tested. Redox level reflects metabolic activity relative to cell growth, biosynthesis and so on [50,51]. NAD + , NADH and NADH/NAD + ratio are crucial indexes of redox reaction. The levels of NAD + and NADH content control metabolic flux in cells [52]. The ratio between NAD + and NADH regulate redox balance [50,51]. The results showed that both the content of NAD + and NADH in the module yeast strain with or without ascorbic acid addition decreased during the fermentation, however, adding ascorbic acid in the culture led to higher NADH/NAD + ratio during the period from 24 to 60 h of fermentation, reaching 1.7 folds higher than that without ascorbic acid at 42 h, which might enhance the yeast metabolic activity resulting in increased 2-PE titer (Fig. 5a).
Based on the above test, batch and fed-batch fermentation of modularized YS58(G1-A8-A10-A2)-GDH strain in 5 L fermenter were carried out for further comparison of 2-PE synthesis (Fig. 5b). Oxygen could inhibit the activity of alcohol acetyltransferase that catalyzes 2-phenylethanol to its ester 2-phenylethylacetate, hence, higher shaking speed was helpful to 2-PE accumulation avoid of its esterification [53]. In our study, when ventilatory capacity was about 5 L min −1 with 400-500 rpm shaking speed (~ 43-55% dissolved oxygen, DO) during the earlier 22 h and with 300 rpm shaking speed (~ 60% DO) during 22-60 h, 2-PE titers in 5 L of batch fermenter were about 6% higher than that in flask, indicating that raising aeration appropriately was helpful to 2-PE synthesis. Similar to flask fermentation, 2-PE titers during single-batch fermentation increased linearly before 96 h and reached a plateau at 96 h (Fig. 5b). During fed-batch fermentation, 2-PE productivity during 24-56 h was higher than that of single-batch fermentation, while 2-PE production after 72 h was similar to that of batch fermentation. The increased 29% of 2-PE titer and productivity at 24 h might mainly be due to higher biomass caused by glucose feeding (Fig. 5b). Feeding of glucose could speed up yeast cells growth, when the biomass reached the maximum, more feeding had no positive role on 2-PE yield. There were no significant differences among three feeding ways of l-Phe, initial l-Phe content (7 g L −1 ) with feeding, initial l-Phe content (9 g L −1 ) without or with feeding of l-Phe during the fermentation, therefore, enough content of l-Phe at beginning was appropriate. The differences of 2-PE production between batch and fed-batch were gradually minified with the biomass in single-batch fermentation increased, 2-PE titers were very close to each other during 72-120 h (Fig. 5b). When the fermentation was finished, 2-PE titer was ~ 6.3 g L −1 whose with conversation ratio 95% (Fig. 5b).
Although there was no difference of 2-PE production between single-batch and fed-batch after 72 h fermentation, the significantly enhanced 2-PE synthetic rate before 24 h was very favorable to fed-batch with product extraction. Since 2-PE titers at 12 and 24 h (3.79 and 4.99 g L −1 ) during fed-batch were very close to that at 24 and 48 h (3.88 and 4.88 g L −1 ) during single-batch, when in situ product adsorption was combined in later study, shortening half a day or 1 day was economic for the whole production craft.
Compared to other engineering strains, the module strain YS58(G1-A8-A10-A2)-GDH in our study presented efficient 2-PE synthesis ability. 2-PE titer in metabolic engineering S. cerevisiae with ARO9 + ARO10 + ARO80 co-expression strain and ALD3 deletion reported by Kim et al. reached 4.8 g L −1 at 195 h of fermentation [17], in our study, the module strain produced 4.99 g L −1 of 2-PE at 24 h under fed-batch condition, and produced 4.88 g L −1 of 2-PE at 48 h under single-batch condition (Fig. 5b), the reduced fermentation time and higher synthetic rate were advantageous for large-scale production later. Genetic engineered S. cerevisiae with ARO8 + ARO10 co-expression reported by Yin et al. produced 2.6 g L −1 of 2-PE at 60 h under fed-batch fermentation, higher 36.8% than that of control strain [41], in our study, 2-PE titer at 56 h under fedbatch condition was 5.65 g L −1 , 2.7 folds higher than that of control (Fig. 5b). 2-PE yield of our module strain at 56 h was 2.2 folds higher than that of engineered yeast of Yin et al. [41], indicating that the module design and promoter strategy in our study were helpful for efficient 2-PE bio-synthesis.

Conclusions
In this study, metabolic modularization and promoter strategy was feasible for 2-phenylethanol synthesis efficiently. Due to multi-proteins' over-expression, including permease for l-Phe transportation, catalytic enzymes transaminase, decarboxylase and dehydrogenase for 2-PE synthesis via Ehrlich pathway and glutamate dehydrogenase for re-supplying another substrate and relieving repression, the produced 2-PE in the module strain increased to ~ 6.3 g L −1 , whose higher 2-PE yield had potentiality for application. The enhanced productivity and shortened fermentation time by glucose feeding would be important for later two phases fermentation. Wang et al. Microb Cell Fact (2018) 17:60
Primer sequences used for plasmid construction are shown in Table 3.
The recombinants strains were preserved on selective medium, and were also stabilized by alternating cultivation in non-selective medium YPD for about 20 generations followed by shifting to the selective medium. Then they were verified by PCR. In addition, the strains after fermentation were also analyzed to convince that the plasmids were kept. The optimum fermentation medium used for 2-PE synthesis contained 40 g L −1 glucose, 5 g L −1 MgSO 4 ·7H 2 O, 5 g L −1 KH 2 PO 4 , 0.15 g L −1 ascorbic acid and 9 g L −1 l-phenylalanine (l-Phe) as a sole nitrogen source. Yeast strains were first grown in YPD medium with 2% inoculation, and cultivated at 30 °C (shaking speed 220 rpm) for 24 h, then inoculated the collected pellets to 40 mL fermentation medium in 250 mL shake flask or to 2.5 L fermentation medium in 5 L fermenter with initial OD 600 of ~ 6 or ~ 9, and cultivated at 30 °C in flask with shaking at 220 rpm or in batch/fed-batch fermenter with 300-500 rpm, 4-5 L min −1 of ventilatory capacity, 40-80% of dissolved oxygen (DO) and 0.05 MPa of tank pressure. Medium for fed-batch: 500 g L −1 glucose (Feed 1) and 30 g L −1 l-Phe (Feed 2) with the same concentration of MgSO 4 /KH 2 PO 4 /ascorbic acid as above. In tube tests, the yeast strains from YPD medium were inoculated into 5 mL of fermentation medium with initial OD 600 of ~ 1 or 1-30 according to experimental design, then shook at 220 rpm and 30 °C for 96 h. 2-PE titers and other indexes at different time points were analyzed and compared. Fold changes were calculated using the 2 −ΔΔCT method [58]. Enhanced folds of transcription levels were the ratios of gene/ACT1 values of transformants divided by that of control strain YS58-YEp.

Analysis of enzyme activities
Crude enzyme extraction: cell pellets from 2 mL yeast culture were first frozen in 1 mL of potassium phosphate (10 mM) with 2 mM EDTA (pH7.5) overnight, then re-suspended in 0.4 mL of potassium phosphate (100 mM) with 0.1 mM EDTA (pH6.5), after vortex oscillation with 50 mg of glass beads (diameter 0.5 mm) for 4 min, supernatant crude enzyme was obtained. The protein content was determined by Bradford method using Coomassie plus protein assay reagent with bovine serum albumin (BSA) as the standard protein [59]. Test for aromatic aminotransferases: The reaction mixtures (1 mL) contained potassium phosphate buffer (0.1 mM, pH 8.0), pyridoxalphosphate (0.1 mM), l-phenylalanine (1 mM), pyruvate (10 mM) as amino acceptor and cell free extract (0.2-0.6 mg protein). The reactions were started by the addition of l-phenylalanine. Two parallel reactions of aminotransferase activities were stopped after 2 and 10 min at 30 °C by the addition of 1 mL of 1 M NaOH on ice. Activity was measured by monitoring the increment of phenylpyruvate, the reaction product of the phenylalanine aminotransferase at 320 nm, ε = 1.7500 × 10 4 cm 2 mol −1 [12,60]. Enzyme activities for phenylpyruvate decarboxylase: The reaction mixtures (1 mL) contained potassium phosphate buffer (70 mM, pH 7.0), NAD + (2 mM), thiamine pyrophosphate (0.2 mM), yeast acetaldehyde dehydrogenase (0.35 U, dissolved in 1 mM dithiothreitol)), phenylpyruvic acid (2 mM) and cell free extract (~ 0.5 mg protein). The reactions were started by the addition of phenylpyruvic acid. Activity was measured by monitoring the reduction of NAD + at 340 nm, ε = 6.2 × 10 3 cm 2 mol −1 [61]. International unit (IU) was adopted for the above two enzyme activities: the amount of enzyme required to convert 1 μmol of substrate to product per minute under a designated condition. ). Optical density (OD 0 ) for time "zero" at 565 nm and OD 15 after 15 min of incubation at room temperature were detected. ∆OD for each sample by subtracting OD 0 from OD 15 was then used to calculate NAD(H) concentration [62,63]. l-Phe contents and 2-PE titers in the culture medium were detected using Agilent 1260 high performance liquid chromatography system equipped with C-18 column and DAD-detector at 254 and 260 nm respectively (Agilent Technologies). Methanol (50%) and water (50%) were pumped as mobile phase at 1.0 mL min −1 . All data are presented as the averages of the results of three independent experiments. Error bars show standard deviations.