An anticonvulsive drug, valproic acid (valproate), has effects on the biosynthesis of fatty acids and polyketides in microorganisms

Valproic acid or valproate (VPA) is an anticonvulsive drug used for treatments of epilepsy, bipolar disorder, and migraine headaches. VPA is also an epigenetic modulator, inhibiting histone deacetylase, and it has been subjected to clinical study for cancer treatment. During the investigation of VPA on a metabolite profile in a fungus, we found that VPA has significant effects on the production of some fatty acids. Further exploration of VPA on fatty acid profiles of microorganisms, fungi, yeast, and bacteria, as well as representative gut microbiome, revealed that VPA could enhance or reduce the production of some fatty acids. VPA was found to induce the production of trans-9-elaidic acid, a fatty acid that was previously reported to have cellular effects in human macrophages. VPA could also inhibit the production of some polyketides produced by a model fungus. The present work suggests that the induction or inhibition of fatty acid biosynthesis by VPA (100 µM) in gut microbiome could give effects to patients treated with VPA because high doses of VPA oral administration (up to 600 mg to 900 mg) are used by patients; the concentration of VPA in the human gut may reach a concentration of 100 µM, which may give effects to gut microorganisms.

and HIV infection 21 . VPA also inhibits MAP kinase signaling and cell cycle progression in the yeast model 22 and sensitizes hepatocellular carcinoma cells to proton therapy through the suppression of NRF2 activation 23 . VPA in combination with other anticancer drugs has been subjected to Phase II clinical study for cancer therapy 24 . Gut microbiome plays important role for human health and diseases, and it has gradually received attention over the past 15 years 25,26 . Many studies have revealed the functional interactions between the host (human) and gut microbiome, as well as the functions of gut microbiome in the healthy state and in certain disease states, e.g., diabetes, obesity, cancer, and liver diseases 25 . Drug-microbiome interactions have recently received attention from the scientific community. Previous works demonstrated the drug interaction with human gut microbiome, for example, atypical antipsychotic drug interacting with the gut microbiome in a bipolar disease cohort 27 . Herein we report the effects of an anticonvulsive drug VPA on the production of fatty acids and polyketides in microorganisms. Preliminary study on the effects of VPA on representative gut yeast and fungal strains is also investigated.

Results
Effects of VPA on a fatty acid profile of microorganisms. It is known that epigenetic modulators, i.e., DNA methyltransferase and histone deacetylase inhibitors, could alter the biosynthesis of fungal metabolites, and thus changing natural product profiles with enhanced chemical diversity 28 . Many studies revealed the effectiveness of epigenetic modifiers for the production of new secondary metabolites in microorganisms, for example, proteasome-inhibitor 29,30 and histone deacetylase inhibitor; [31][32][33] this technique is collectively known as "One strain many compound" (OSMAC) approach. Previously, VPA, an inhibitor of histone deacetylase, was found to enhance tenfold-production of a fungal alkaloid 34 , fumiquinazoline C, and could change the profile of secondary metabolites 35 and enhance antimicrobial activity of fungal extracts 36,37 . Initially, we preliminarily  Table 2. Effect of VPA (100 µM) on fatty acid profile of fungi. a p ≤ 0.01 was selected as the minimum criterion for significance (for the same fatty acid and the same fungus with or without VPA). b p ≤ 0.05 was selected as the minimum criterion for significance (for the same fatty acid and the same fungus with or without VPA).
investigated the effect of VPA on the metabolite production of the marine fungus Trichoderma reesei, which normally produces only mevalonolactone as a secondary metabolite ( 1 H and 13 C NMR spectra of mevalonolactone are in Supplementary Information). The aim of this work is to use VPA, an epigenetic modulator, to enhance the production of other metabolites, possibly derived from mevalonolactone through the mevalonate pathway, the common biosynthetic pathway of terpenes and steroids. Previous works used different concentrations of VPA, e.g., 50 µM, 60 µM, 100 µM, and 500 µM, for the studies of the influence of VPA on the metabolite profile [34][35][36][37] . In the present work, we initially fed VPA with the concentrations of 100 µM and 300 µM to a culture of the marine fungus Trichoderma reesei, however, we found that VPA with the concentration at 300 µM or higher than 300 µM could inhibit the growth of the fungus. Therefore, we performed the experiment at the concentration of 100 µM. The present work revealed that VPA did not induce the fungus Trichoderma reesei to produce new terpenes and steroids, but it showed the effects on fatty acid profiles as observed from 1 H NMR spectrum of a crude cell extract. Detailed analysis by gas chromatography (GC) revealed that VPA significantly induced the production of palmitic acid (C16:0) from 9.39% in a control (without VPA) to 19.89% (2.11 times increase) in the fungus treated with VPA, while it reduced the production of oleic acid (C18:1) from 71.51% in a control to 57.19% (1.25 times decrease) ( Table 1). The amounts of palmitoleic acid (C16:1), stearic acid (C18:0), linoleic acid (C18:2), and α-linolenic acid (C18:3) of a control were relatively the same as that in the VPA treated fungus. The fungus treated with VPA (49.99%) had the total fatty acid less than a control (65.26%) ( Table 1). As mentioned above, undesirable side effects related to fatty acid metabolism were observed in patients after treatment of VPA, for example, the influence on lipid metabolism 2 . VPA interfered with mitochondrial β-oxidation via fatty acid β-oxidation pathway, and thus causing serious inborn errors of metabolism after VPA treatment 4 . VPA induced hepatotoxicity and weight gain for patients because of the interference on a key enzyme  Table 3. Effect of VPA (100 µM) on fatty acid profile of yeast. a p ≤ 0.01 was selected as the minimum criterion for significance (for the same fatty acid and the same yeast with or without VPA). b p ≤ 0.05 was selected as the minimum criterion for significance (for the same fatty acid and the same yeast with or without VPA).  Table 4. Effect of VPA (100 µM) on fatty acid profile of bacteria. a p ≤ 0.01 was selected as the minimum criterion for significance.
www.nature.com/scientificreports www.nature.com/scientificreports/ for fatty acid β-oxidation 5 . The interferences of VPA on fatty acid metabolism in patients and our preliminary data of VPA on the fatty profile of the fungus Trichoderma reesei (Table 1) prompted us to investigate the effects of VPA on fatty acid profile in other microorganisms including representative gut microbiome.
Scientific RepoRtS | (2020) 10:9300 | https://doi.org/10.1038/s41598-020-66251-y www.nature.com/scientificreports www.nature.com/scientificreports/ fatty acid of X. globosa and C. militaris. Overall, these results indicated that VPA could stop or enhance the production of certain fatty acids in fungi, and it also had effects on the total fatty acid of some fungi.
Next, we tried to investigate the effects of VPA on the fatty acid profile of other microorganisms, e.g., yeast and bacteria (Tables 3 and 4). The yeasts, Saccharomyces cerevisiae TBRC1563, Candida utilis TBRC360, and Lachancea thermotolerans TBRC4347 were used as model microorganisms (Table 3); these strains are normally used in food and beverage production. The bacteria Pediococcus acidilactici TBRC7580, Bacillus amyloliquefaciens TBRC293, and Acetobacter cerevisiae TBRC6687, were used as model bacteria (Table 4). P. acidilactici and A. cerevisiae are normally used in fermented dairy products (e.g., yoghourt production) and meat (e.g., Thai fermented pork sausage or "Naem" in Thai), while B. amyloliquefaciens is a known source of α-amylase for the starch hydrolysis in food industry. As shown in Table 3, VPA completely stopped the production of α-linolenic acid (C18:3) in the yeasts, S. cerevisiae and C. utilis, and it also stopped the production of palmitoleic acid (C16:1) in C. utilis. Elevated levels of palmitic acid (C16:0) from 29.03% to 47.16% (increased by 1.62 times) and stearic acid (C18:0) from 6.60% to 10.91% (increased by 1.65 times) were observed in the VPA treated culture of C. utilis, while decreased level of linoleic acid (C18:2) from 15.86% to 7.72% (decreased by 2.05 times) was found in the VPA treated culture of C. utilis (Table 3). Moreover, the total fatty acid of the yeast, C. utilis, decreased from 24.35% to 15.62% (decreased by 1.55 times) was observed in the VPA treated culture of C. utilis (Table 3). In contrast, VPA did not have notable effects on the fatty acid profile and the total fatty acid of the yeast L. thermotolerans (Table 3). This result clearly showed that VPA had significant effects, both on the fatty acid profile and the total fatty acid, toward the yeast C. utilis. Interestingly, the yeast, C. utilis, was found as gut microbiome in pediatric patients with inflammatory bowel disease 38 . The effects of VPA on fatty acid profile of bacteria are shown in Table 4;  Table 6. Effect of VPA (100 µM) on fatty acid profile of representative gut yeast. a p ≤ 0.01 was selected as the minimum criterion for significance (for the same fatty acid and the same yeast with or without VPA). b p ≤ 0.05 was selected as the minimum criterion for significance (for the same fatty acid and the same yeast with or without VPA). www.nature.com/scientificreports www.nature.com/scientificreports/ VPA completely inhibited the production of lignoceric acid (C24:0) in the bacterium, P. acidilactici, and it also inhibited the production of oleic acid (C18:1) and arachidic acid (C20:0) in the bacterium, A. cerevisiae. Both P. acidilactici and A. cerevisiae are used in fermented dairy products and fermented meat (Thai sausage). The level of palmitoleic acid (C16:1) in the VPA treated culture of B. amyloliquefaciens was decreased from 9.72% to 4.41%, accounting for 2.2 times less than the control (Table 4).
Recently, gut microbiome has received attention worldwide 25,26 . Therefore, we investigated the effects of VPA drug on fatty acid profile of certain representative gut microbiome, e.g., fungi and yeast. Representative gut fungi are Penicillium shearii TBRC2865, Phialemonium sp. TBRC4709, Cladosporium sp. TBRC4134, and Aspergillus flavipes BCC28681; the genera Penicillium, Phialemonium, Cladosporium, and Aspergillus are the most prevalent in human gut [39][40][41] . VPA could enhance the production of linoleic acid (C18:2) from 12.46% (control) to 23.60% (increase 1.89 times) in the fungus P. shearii TBRC2865, however, it reduced the production of linoleic acid (C18:2) from 14.85% to 7.47% (1.98 times) in the fungus Phialemonium sp. (Table 5). VPA reduced the production of stearic acid (C18:0) from 17.62% (control) to 9.90% (1.77 times) in Cladosporium sp. TBRC4134, but it enhanced the production of stearic acid (C18:0) in A. flavipes BCC28681 from 8.70% (control) to 16.09% (1.85 times) ( Table 5). Other changes were the reduction of docosanoic acid (C22:0) in P. shearii TBRC2865 from 1.30% (control) to 0.67% (1.94 times); and increase of arachidic acid (C20:0) from 0.24% (control) to 1.77% (7.3 times), docosanoic acid (C22:0) from 0.24% (control) to 2.10% (8.75 times), and lignoceric acid (C24:0) from 0.29% (control) to 1.19% (4.1 times) in the fungus A. flavipes BCC28681 (Table 5). Three representative gut yeasts were Candida catenulata TBRC223, Candida butyri TBRC221 (syn. Candida aaseri), and Saccharomyces ludwigii TBRC2149; the genera Candida and Saccharomyces are commonly found as prevalent gut microbiome 42,43 , particularly the yeast Saccharomyces cerevisiae 44 . The yeast C. catenulata is one of the most prevalent species in gastrointestinal tract of turkeys 45 , while C. butyri is found as microbiota in green olive fermentations 46 . In the present work, the yeast S. cerevisiae TBRC1563 previously mentioned above would be one of the representative gut yeasts (Table 3). We found that VPA completely inhibited the biosynthesis of α-linolenic acid (C18:3) in the yeast S. cerevisiae TBRC1563 ( Table 3). The yeast C. utilis was found as gut microbiome in patients with inflammatory bowel disease 38 ; VPA completely stopped the production of palmitoleic acid (C16:1) and α-linolenic acid (C18:3) in C. utilis ( Table 3). As shown in Table 6, VPA completely inhibited the biosynthesis of palmitoleic acid (C16:1) and α-linolenic acid (C18:3) in the yeast Candida catenulata TBRC223, while it induced the production of palmitoleic acid (C16:1) in C. butyri TBRC221 and trans-oleic acid or trans-9-elaidic acid (trans-C18:1) in Saccharomyces ludwigii TBRC2149. Normally, trans-9-elaidic acid is present in yeast and is degraded by peroxisomal multifunctional enzymes 47 . It was found that trans-9-elaidic acid is less toxic than its cis isomer, oleic acid, and that they exhibited different effects in gene expression regulation and handling of excess fatty acids in yeast 48 . Moreover, trans-9-elaidic acid could inhibit β-oxidation in human peripheral blood macrophages and increase intracellular www.nature.com/scientificreports www.nature.com/scientificreports/ Zn 2+ in human macrophages 49,50 . Interestingly, the present work revealed that VPA could induce the production of trans-9-elaidic acid in the representative gut yeast. VPA reduced the production of oleic acid (C18:1) from 48.75% (control) to 23.63% (2.06 times) and linoleic acid (C18:2) from 5.61% (control) to 0.35% (16.02 times) in S. ludwigii TBRC2149 (Table 6). VPA markedly enhanced the production of the total fatty acid in the yeast C. butyri TBRC221 from 7.39% (control) to 14.18% (1.91 times). Although the present work has not investigated the effects of VPA on fatty acid profile of the representative gut bacteria, it is more likely that VPA may give effects on the biosynthesis of fatty acid of gut bacteria. The genus Pediococcus is normally found as intestinal flora of humans and animals 51 , the bacterium Pediococcus acidilactici TBRC7580 mentioned in Table 4 may be used as the representative gut bacterium. Normally, P. acidilactici is used as probiotic, and oral feeding study revealed that this bacterium can survive in gastrointestinal tract of volunteers about two weeks after feeding 52 . The present work showed that VPA completely inhibited the biosynthesis of lignoceric acid (C24:0) in P. acidilactici (Table 4).
It is known that the enzymes responsible for the biosynthesis of fatty acids and polyketide natural products share a great deal of similarities 53 . Since the above data demonstrated that VPA has effects on the production of fatty acids, we envisage that VPA may have effects on the biosynthesis of polyketide natural products because fatty acid synthases and polyketide synthases have similar catalytic elements, for example, the use of common precursors and catalytic roles. Therefore, we investigated the effects of VPA on the production of polyketide natural products using the endophytic fungus Dothideomycete sp., which is a known source of polyketides in our laboratory [54][55][56] . The fungus Dothideomycete sp. was previously found to produce a tricyclic polyketide, and other polyketides such as azaphilone, hybrid azaphilone-pyrone, calbistrin, and isochromanone, and it produced a large amount of austdiol (1) as the major azaphilone [54][55][56] . However, in the present study, in addition to austdiol (1), the fungus Dothideomycete sp. produced other types of polyketide, i.e., known isobenzofuranone polyketides 2, 3, and 5, and a polyketide 4 (Fig. 1). Compounds 2 and 3 were previously obtained as a mixture (1:1 ratio), which could not be separated by C18 reversed phase HPLC 57 . In the present work, compounds 2 and 3 were obtained after repeated HPLC separation; structures of both compounds were elucidated by analysis of 1D and 2D NMR spectra, as well as by data comparison with those published 57 . The absolute configuration of isobenzofuranones (e.g., 2, 3, and 5) was well established by the modified Mosher's method and CD spectra after derivatization 58 , as well as by X-ray analysis for quadricinctone A (5) 59 . The isomer with 3 R and 8 S (e.g., 2 and 5) had negative values, while that with 3 R and 8 R (e.g., 3) had positive values 58,59 . Therefore, a polyketide 2 had 3 R and 8 S configuration because of negative optical rotation, [α] 27 The endophytic fungus Dothideomycete sp. was cultivated in 100 μM of VPA, and the comparison of the metabolite profile by HPLC analysis between the VPA treated culture and the control (without an addition of VPA) was investigated (Fig. 2). As shown in Fig. 2A, HPLC chromatogram of the control fungal culture showed peaks at retention time (t R ) of 6.5 min for austdiol (1), at t R of 7.4 min for compounds 2 and 3, at t R of 8.0 min for compound 4, and 10.2 min for quadricinctone A (5). HPLC chromatogram of VPA treated culture of the fungus Dothideomycete sp. is shown in Fig. 2B, showing a marked reduction (>90%) of the polyketide austdiol (1) and ca 50% reduction of quadricinctone A (5). However, the amounts of polyketides 2, 3, and 4 were not affected by VPA. This experiment demonstrated that VPA could affect the production of certain fungal polyketides.

Discussion
It is known that one of the side effects of VPA for patients is on lipid metabolism 2 ; VPA interferes β-oxidation pathway of fatty acids, and thus causing toxicity. The negative impact of VPA treatment on serious inborn errors of metabolism is well documented 4 . VPA induced hepatotoxicity and weight gain for patients because it interferes carnitine palmitoyl-transferase I, a key enzyme in fatty acid β-oxidation 5 . The present work revealed that an anticonvulsive drug, VPA, has effects on the fatty acid profile of fungi, bacteria, and yeast, suggesting that this drug affects the biosynthesis of fatty acids in microorganisms. Normally, oral administration of VPA at doses of 10 to 15 mg/kg/day (i.e., 600 mg to 900 mg for the patient with 60 kg weight) is used for the treatment of epilepsy. However, doses of VPA for the prophylaxis of migraine headaches are 250-500 mg/day, while the treatment of manic episodes associated with bipolar disorder uses the dose up to 750 mg/day. In the present work, we found that VPA with the concentration 100 µM has effects on the biosynthesis of fatty acids in certain representative gut microbiome. It is possible that when patients taking high doses of VPA, i.e., 600 mg to 900 mg, the amount of VPA in patient gastrointestinal tract may reach at the concentration of 100 µM, which may affect the fatty acid biosynthesis in gut microbiome of patients. VPA induced abnormal autism-like behaviors in mice 8 , and many studies showed the possible risk of VPA on autism spectrum disorders [9][10][11][12][13][14] . The present work demonstrated that the biosynthesis of certain fatty acids was completely inhibited by VPA, while that of some fatty acids was induced by VPA. The present work showed that VPA could induce the production of trans-9-elaidic acid in S. ludwigii TBRC2149 (Table 6); this is worth mentioning because trans-9-elaidic acid was previously found to inhibit β-oxidation in human peripheral blood macrophages, and it could increase intracellular Zn 2+ in human macrophages 49,50 . Therefore, the induction or inhibition of certain fatty acids by VPA drug may have direct effects to patients treated with VPA. Fatty acid metabolism has a critical role in human since it sustains balanced homeostasis and the negative perturbations that would lead to disease development 61 . The present work also demonstrated that VPA has effects on the biosynthesis of certain polyketide natural products produced by the fungus Dothideomycete sp.; this is because there are a great deal of similarities of the catalytic elements of fatty acid synthases and polyketide synthases. This knowledge may be applied for natural product research, aiming to diversify the polyketide structures.

Scientific RepoRtS |
(2020) 10:9300 | https://doi.org/10.1038/s41598-020-66251-y www.nature.com/scientificreports www.nature.com/scientificreports/ In spite of the fact that VPA provides the effects on the biosynthesis of fatty acids in microorganisms such as fungi, yeast, and bacteria, the limitation of the present work is that representative gut microbiome used in this work does not well cover gut microbe community, particularly gut bacteria. We had selected certain genus of gut bacteria from the culture collection of Thailand Bioresource Research Center, however, most of them are anaerobic strains, growing under the conditions without oxygen. We tried to cultivate the selected gut bacteria representative using a homemade plastic incubator sealed with rubber, but failed to obtain cells for lipid extraction. In the present work, we could obtain cells from only one anaerobic bacterium, Pediococcus acidilactici, for lipid extraction; VPA was found to inhibit the production of lignoceric acid (C24:0) in the bacterium, P. acidilactici (Table 4). Recent study revealed that VPA and some psychotropic drugs gave effects on gut bacteria and short-chain fatty acids in rats, and VPA could decrease levels of propionate and butyrate, but enhancing the levels of isovalerate 62 . Many psychotropic drugs have substantial effects of gut microbiome 63 . The study in an animal model or a cohort study in patients that monitors the levels of fatty acids in feces and the composition of gut microbiota between the group treated with VPA and those without VPA will provide information regarding the precise effects of VPA on changes of fatty acids in gut microbiome.
Although this work employs the OSMAC approach using VPA as an epigenetic modulator to change the profile of secondary metabolites (natural products) in the fungus Trichoderma reesei, we propose that the effects of VPA toward the biosynthesis of fatty acids and polyketides may not through the epigenetic modulation. Previous works demonstrated that that VPA affected the fatty acid metabolism in patients 2 , and interfered with β-oxidation via fatty acid β-oxidation pathway 4 . Therefore, VPA may give direct effects toward fatty acids, i.e. the metabolism of fatty acids; however the epigenetic modulation on the biosynthesis of fatty acids could not be ruled out at this stage. Further study on the mechanistic insights into the cellular and molecular levels of VPA on changes of fatty acids and polyketides should be pursued.

Methods cultivation of microorganisms.
Methods for cultivation of microorganisms, fungi, yeast, and bacteria are in Supplementary Information. Individual microorganisms were cultivated without an addition of VPA (control) or in the presence of VPA (100 µM). Cells of filamentous fungi were separated from broth by filtration using a filter paper, while cells of yeast and bacteria were collected by centrifuge at 8000 rpm. Cells of microorganisms were dried by freeze drying, and lipid in dried cells was extracted by hexane. extraction of lipid and analysis of fatty acids. Dried cells of microorganisms were extracted twice by maceration in hexane overnight at room temperature. Crude fat extract was individually transesterified with 4% sulphuric acid in methanol 64 . A fat extract was dissolved in methanol containing 4% of sulphuric acid, and the mixture was heated at 90 °C for 1 h. Nonadecanoic acid (C19:0) was used as an internal standard. The esterified products were analyzed by a gas chromatography (GC) using a 30 m × 0.25 mm fused silica capillary column. The GC instrument was equipped with an automatic sampler and flame ionization detector (FID). The injector and detector temperatures were kept at 250 °C and 260 °C, respectively, and helium was used as a carrier gas at a linear velocity of 30 cm/s. The initial temperature for GC column at 200 °C was held for 10 min, and then increased at 20 °C/min to 230 °C, where it was held for 17 min. Individual fatty acid esters were identified based on the retention times relative to fatty acid methyl ester standards (Supelco 37 Component FAME Mix as the standard for methyl esters). isolation of mevalonolactone. A crude broth extract (198 mg) of the fungus Trichoderma reesei was separated by Sephadex LH-20 column chromatography (CC) (size 2 × 132 cm), eluted with methanol, yielding twelve fractions (F1-F12). A fraction F8 (79 mg) was further separated by Sephadex LH-20 CC (size 1.5 × 126 cm), eluted with methanol, to give nine fractions (F8 1 -F8 9 ). A fraction F8 1 was purified by C18 reversed phase HPLC, eluted with a solvent system of MeOH:H 2 O (60:40), to give mevalonolactone (11.7 mg). 1 H and 13 C NMR spectra of mevalonolactone are in Supplementary Information.
Structure elucidation of fungal metabolites by spectroscopic techniques. 1 H, 13 C, and 2D NMR spectroscopic data were obtained from on 400 MHz NMR spectrometer ( 1 H at 400 MHz, 13 C at 100 MHz), or 600 MHz NMR spectrometer ( 1 H at 600 MHz, 13 C at 150 MHz). Deuterated CDCl 3 was used as an NMR solvent. HRMS data were obtained from ESI-TOF mass spectrometer. Data for optical rotations for compounds 2 and 3 were obtained from a polarimeter.

Statistical analysis method.
Statistical analysis of all data, three replications per each condition of an individual microorganism, was performed using the IBM SPSS Statistics 22 software, Independent-Samples T test method. Differences of fatty acid content between each group (control without VPA or with 100 µM of VPA) were determined by two-tailed t test, and data was reported as mean±s.d. with the significance set at p < 0.01 or at p < 0.05.