Simultaneous determination of plasma indoles by HPLC with fluorescence detection: Application in polycystic ovary syndrome patients with/without depression

Abstract Microbiota-derived indoles were closely associated with psychological disorders like depression. We aimed to develop an HPLC method with fluorescence detection (FLD) for simultaneous determination of plasma indoles, including indoxyl sulfate (3-INDS), indole-3-acetic acid (IAA), indole-3-propionate (IPA), indole (IND) and 3-methylindole (3-MI). Diethyl ether and ethyl acetate were selected as the mixed extractant for the sample preparation. The separation was carried out on a Shim-Pack VP-ODS column (150 × 4.6 mm, 4.6 µm) with the mobile phase composed of 10 mmol/L sodium dihydrogen phosphate/methanol (40:60, v/v). The excitation and emission wavelengths were set at 280 and 355 nm, respectively. The linearities in plasma were obtained in the range of 1.56–400.0 μmol/L for 3-INDS, 0.312–10.0 μmol/L for IAA, 0.125–6.00 μmol/L for IPA, 6.25–400.0 nmol/L for IND, and 1.56–400.0 nmol/L for 3-MI, respectively. The coefficients of variation (CVs) for each analyte of the intra-day and inter-day precisions were within 4.0 and 5.2%, respectively. The recoveries were in the range of 90.1–109.3%. This sensitive and accurate method was applied to analyze plasma indoles for the diagnostic of depression in the patients with polycystic ovary syndrome (PCOS). Graphical Abstract


Introduction
Tryptophan is an essential amino acid in the human body and its metabolism follows three major pathways, the gut microbiota pathway, the serotonin pathway, and the kynurenine pathway. [1] In the gut microbiota pathway, tryptophan is transformed into indole (IND) and its derivatives by the gut microbiota. These microbiota-derived indoles may lead to human diseases, such as bacterial infections, intestinal inflammation, cancers, diabetes, as well as neurological diseases. [2] Studies addressed the impact of serotonin and kynurenine pathways on psychological disorders. [3,4] However, few studies focused on the microbiota pathway. Polycystic ovary syndrome (PCOS) is one of the most common endocrinopathies in reproductive-aged women, with the worldwide prevalence ranging from 4 to 21%. [5] Patients with PCOS are at an increased risk for metabolic diseases, and suffering from psychological disorders like depression as well. [6,7] PCOS patients with depressed mood experience a low quality of life. And screening for depression in PCOS to perform early intervention may be warranted. [8,9] It has been proved that the presence or absence of conventional intestinal microbiota influenced the development of behavior, and was accompanied by neurochemical changes in the brain. [10] Increasing evidence has demonstrated a strong correlation between gut microbiome composition and the development of psychological disorders like depression, indicating an important role of microbial metabolites. Therefore, researches are expected to investigate the relationship between microbiota-derived indoles and depression.
Simultaneous quantification of IND and its derivatives is important to illustrate the metabolic characteristics of indoles in vivo. Several methods have been applied to determine plasma indoles. Using the sample pre-treatment of protein precipitation, Chen's group [11] detected indole-3-acetic acid (IAA) and indole-3-propionate (IPA) by UPLC-MS/MS, while Zeng's group [12] detected IPA and indoxy sulfate (3-INDS) by HPLC-MS/MS. Our group [13] previously detected IAA, IPA, IND, and 3-methylindole (3-MI) by HPLC-FLD with the sample pre-treatment using diethyl ether as the extractant. However, indoxy sulfate (3-INDS) with strong polarity could not be detected successfully. The chemical structures of the indoles were presented in Figure 1. Using the sample preparation by the mixed extractant of diethyl ether and ethyl acetate, 3-INDS with strong polarity and the above four indoles could be quantified simultaneously by HPLC-FLD in this study. This method is simple, economic, and sensitive for the simultaneous detection of five plasma indoles with different polarities. Furthermore, the developed method was applied to detect the concentrations of plasma indoles in PCOS patients with or without depression. The potential role of microbiotaderived indoles for depression in patients with PCOS was explored preliminarily.  [14] which met at least two of the following features: (1) oligo-or anovulation, (2) clinical and/or biochemical hyperandrogenism, and (3) a polycystic appearance of the ovaries upon ultrasonography. Individuals who had hyperprolactinemia, thyroid disease, Cushing's syndrome, diabetes mellitus, or any other chronic comorbidity were excluded. Venous blood samples were collected in K 2 EDTA tube after an overnight fast and centrifuged at 2000 Â g for 5 min. The plasma samples were stored at À20 C before analysis. Participants were asked to fill in the self-rating depression scale (SDS). [15] According to the test results, participants with scores !53 were classified as the depression group, and score <53 were classified as the non-depression group. The study was approved by the ethical committee of the First Affiliated Hospital of Chongqing Medical University. Informed consent was obtained from all participants. Preparation of standard solutions Stock solutions (35.00 lmol/L for 1-naphthol, 100.00 lmol/L for IAA, 60.00 lmol/L for IPA, 4.00 lmol/L for IND, and 4.00 lmol/L for 3-MI, respectively) were prepared in methanol. The stock solution of 4.00 mmol/L 3-INDS was prepared in 50% methanol, and all of the stock solutions were stored at À20 C. These stock solutions were further diluted with 50% methanol to obtain final concentrations of 70.00 nmol/L 1-naphthol, 15

Sample preparation
Liquid-liquid extraction (LLE): Ten microliters of 10.00 mg/ L 1-naphthol as the internal standard (IS) was added to 300 mL of plasma. Then the sample was treated with 540 lL of diethyl ether and 270 lL of ethyl acetate. And then, it was shaken at the rate of 150 rpm for 15 min and centrifuged at 13,300 g for 10 min at 25 C. The supernatant was transferred into another centrifuge tube and dried in a vacuum concentrator. The residue was dissolved in 90 mL of a mixture of 10 mmol/L sodium dihydrogen phosphate/methanol (40:60, v/v). The sample was further centrifuged at 13,300 g for 10 min at 25 C, and ultimately 50 lL of the supernatant was injected for HPLC analysis.
Protein precipitation (PPT): Ten microliters of 10.00 mg/ L IS was added into 300 mL of plasma. Then the sample was treated with 900 lL of acetonitrile. And the mixture was vortexed for 30 s, and then centrifuged at 13,300 g for 10 min at 25 C. The supernatant was performed as described above.

HPLC conditions
The samples were analyzed on Agilent 1100 HPLC equipped with a G1321A fluorescence detector. The analysis was carried out at 35 C with a Shim-Pack VP-ODS column (150 Â 4.6 mm i.d., 4.6 mm, Shimadzu) equipped with a C18 security guard column (3 Â 4 mm i.d., 5 mm, Phenomenex). Separation was performed with 10 mmol/L sodium dihydrogen phosphate/methanol (40:60, v/v) at a flow rate of 0.80 mL/min. The excitation and emission wavelengths of fluorescence detection were set at 280 and 355 nm, respectively.

Statistical analysis
Mann-Whitney U test was used to evaluate the difference in indoles between groups. Receiver operating characteristic (ROC) curve analysis was used to assess the diagnostic efficiency of IND and the cut-off value was defined as the highest sum of sensitivity and specificity calculated based on the Youden index. Binary logistic regression analysis was performed to evaluate risk factors of depression in PCOS. p < 0.05 was considered as with significant difference. Statistical analysis was carried out by SPSS 25.0 software.

Results and discussion
Sample preparation optimization PPT has been used for the preparation of plasma samples for the quantification of indoles by LC-MS/MS. [11,12] In this work, PPT was investigated for the analysis of indoles in human plasma by HPLC-FLD method. However, analytes cannot be separated, as shown in Supplemental Figure 1c. Due to the presence of the matrix interference effect, PPT was not applicable for the sample preparation in HPLC-FLD.
A plasma sample prepared by LLE with diethyl ether can be analyzed by HPLC-FLD for the quantitation of plasma IND and 3-MI. [13,16,17] Five indoles were found separated by this extraction, as shown in Supplemental Figure 1a. However, the peak areas (A) of 3-INDS and IAA were too small to be accurately quantified. Hence, the mixed extractant composed of diethyl ether and ethyl acetate was adopted to improve the extraction efficiency. The Furthermore, the volume ratio of the two extractants was investigated. With the increased dose proportion of diethyl ether, A IAA and A IPA decreased. However, A 3-INDS and A 3-MI increased and reached a maximum when the volume ratio of diethyl ether to ethyl acetate was 2:1 and then declined, as shown in Figure 2. As A IPA was much larger than A 3-INDS and A 3-MI , we focused more on the extraction efficiency of substances with smaller peak areas. The volume ratio of diethyl ether to ethyl acetate was optimized to 2:1 for all the subsequent experiments.
Besides, the extraction time and the shaken speed were selected as 15 min and 150 rpm (in Supplemental Figure 2), respectively. In addition, a satisfactory peak shape could be obtained when the residue was dissolved in a mixture of 10 mmol/L sodium dihydrogen phosphate/methanol (40:60, v/v).

Method development and optimization
Plasma indoles were detected simultaneously by the HPLC method with fluorescence detection. Indoles could be assayed by fluorescence detection because of the native fluorescence produced by the indole ring. The emission and excitation wavelengths were set at 280 nm and 355 nm, respectively. The selection of IS was referred to in our previous research. [16] The mobile phase was composed of sodium dihydrogen phosphate and methanol. The percentage of methanol in the mobile phase, the temperature of the column, and the rate of flow were investigated, as shown in Supplemental Figure 3. With the proportion of methanol increased, except for IS/3-MI, the resolution between the adjacent substances decreased significantly. When the volume ratio of methanol was 60%, a satisfactory resolution could be achieved. Finally, the experimental conditions including the proportion of methanol in the mobile phase, the column temperature, and the flow rate were selected as 60%, 35 C, and 0.80 mL/min, respectively. Under the optimized liquid chromatographic conditions, each analyte achieved baseline separation within 10.0 min with a resolution >2.0. The chromatograms of the plasma sample were shown in Figure 3.  Table 1.

Precision and recovery
Precision was estimated by analyzing pooled plasma samples of low and high concentration levels. The intra-day reproducibility of the assay was obtained by analyzing the pooled samples for five times in a day and the inter-day reproducibility was obtained by screening the pooled samples for 5 consecutive days. The analytical coefficients of variation (CVs) of the intra-day and inter-day precisions were within 4.0 and 5.2%, respectively. The results were summarized in Supplemental Table 1. Recovery was investigated by spiking pooled plasma samples with low, medium, and high concentrations of analytes and measured in three replicates. The recoveries of plasma analytes were in the range of 90.1-109.3% with CVs ranging from 1.6 to 8.2%, as shown in Supplemental Table 2.

Interference test
The interference test was performed according to the Clinical and Laboratory Standards Institute (CLSI) of America. The concentrations of indoles in plasma before and after adding standards of interferences were defined as X C and X T , respectively. The interference value (expressed as X C À X T ) less than 1.96 S showed no significant interferences, which was expressed by N. While the interference value more than 1.96 S indicated significant interferences, which was expressed by I. Since hemolysis and jaundice are common interference phenomena in clinical plasma sample tests, hemoglobin and bilirubin are supposed to be interferents in plasma to evaluate the anti-interference ability of the method. No significant interferences were observed when the concentrations of hemoglobin and bilirubin were lower than 0.52 g/L and 30.48 lmol/L, respectively. The results were shown in Supplemental Tables 3 and 4. It indicated that the developed method was with acceptable anti-interference abilities.

Method application
The developed method was applied to detect the concentrations of plasma indoles in PCOS patients with or without depression. The general characteristics and SDS scores in  both groups were presented in Supplemental Table 5. The concentrations of plasma indoles from PCOS patients with or without depression were shown in Table 2. Significant differences in the concentrations of IND were observed between depression and non-depression groups in patients with PCOS. Besides, a forest plot of the binary logistic model was applied to estimate the odds ratio (OR) of indoles. The OR of IND was 1.144 (95%CI: 1.028-1.272, p ¼ 0.014), which indicated that IND was a risk factor for depression in patients with PCOS, as shown in Figure 4A. Moreover, ROC analysis was performed to evaluate the diagnostic efficiency of IND for depression in patients with PCOS, as shown in Figure 4B. The area under the curve (AUC) for IND was 0.873 (p < 0.001), and the cut-off value of IND was 53.64 nmol/L and the Youden index was 0.623. The sensitivity and the specificity were 90.9 and 71.4%, respectively. The results indicated that IND was a satisfactory diagnostic index for depression in patients with PCOS. Indoles could be produced by bacterial species. Since the link between the gut microbiota and the development of neurodegenerative disorders has been addressed, it is of great importance to explore the effect of indoles on human behaviors. [18] 3-INDS and IPA could modulate the activity of astrocytes and inflammation of the central nervous system via the aryl hydrocarbon receptor, which played an important role in depression and anxiety. [19,20] However, Edelenyi et al. [21] observed that insignificant associations were obtained between urinary 3-INDS level and recurrent depressive symptoms. Abildgaard et al. [22] found that behaviors related to anxiety and depression were not affected by IPA treatment in rats. Besides, IAA was illustrated to be correlated with anxiety and depression in a prospective pilot study. [23] In our work, no significant differences in IPA were found in PCOS with and without depression, which was consistent with the previous reports. Whereas, the association between IAA and depression was not found in our work. The inconsistency may be attributed to different scales and small sample sizes. Animal research demonstrated that anxiety or depressive disorders were more likely to occur in individuals with higher levels of IND production. [24] However, the association between indole and depression in humans has not been elucidated so far. In our study, IND was found to be a risk factor for depression in patients with PCOS. Because of the reflection of the dysbiosis of gut microbiota, the excessive production of IND might disturb the catabolism of tryptophan, further affect the serotonin and kynurenine pathways, and finally make the brain susceptible to depression. [25,26] The current understanding of depression is based on the monoamine hypothesis. However, approximately 30% of patients failed to respond to such antidepressant drugs. [27] Studies have also highlighted the role of inflammation in depression, and inflammatory cytokines are elevated in a third of subjects with depression in the absence of medical illness. [28,29] To date, no biomarkers are specific and sensitive enough to be used in clinics. In this work, IND was found to be a satisfactory diagnostic index for depression in patients with PCOS. Although the molecular mechanisms of the progression of depression were gradually explored, the pathophysiology of depression remained unclear. Microbiotaderived metabolites are released by the gut microbiota into the blood, which interacts with the host and further contributes to a variety of brain disorders. [30] So, IND was speculated to act in the same way. Further researches need to explore the potential mechanisms based on this association and develop new diagnostic and therapeutic strategies.

Conclusion
With the effective extraction of plasma indoles with different polarities by the mixed extractant of diethyl ether and ethyl acetate, an HPLC-FLD method was developed for the simultaneous determination of five indoles in plasma. This method is simple, and cost-effective, which provides a promising way to comprehensively analyze the metabolic characteristics of indoles in diseases. The method was then applied to detect plasma indoles in PCOS with or without depression. IND was found to be a risk factor and a satisfactory diagnostic index for depression in patients with PCOS, which provides insights into the pathophysiology of depression. Nevertheless, the existing findings should be replicated in larger populations.