Mol. Cells 2017; 40(7): 466-475
Published online July 6, 2017
https://doi.org/10.14348/molcells.2017.2298
© The Korean Society for Molecular and Cellular Biology
Correspondence to : *Correspondence: young.goo@northwestern.edu (YAG); bear7905@naver.com (SYS)
Dietary supplements have exhibited myriads of positive health effects on human health conditions and with the advent of new technological advances, including in the fields of proteomics, genomics, and metabolomics, biological and pharmacological activities of dietary supplements are being evaluated for their ameliorative effects in human ailments. Recent interests in understanding and discovering the molecular targets of phytochemical-gene-protein-metabolite dynamics resulted in discovery of a few protein signature candidates that could potentially be used to assess the effects of dietary supplements on human health. Persimmon (
Keywords biomarker candidates, diabetes, mass spectrometry, persimmon leaf extract, proteomics, saliva, serum, urine
Diabetes mellitus, the most prevalent form of diabetes, is a metabolic disease due to impaired insulin secretion and/or action (Das and Elbein, 2006), which causes hyperglycemia (Gerich, 1998). In 2010, an estimated 285 million people worldwide were affected by diabetes, and this figure is predicted to reach 439 million by 2030 (Shaw et al., 2010). Over the next decade, the diabetes related long-term complications, morbidity, and mortality are expected to be more prevalent due to increasing rates of childhood and adulthood obesity (CDC, Atlanta, USA). Therefore, not only effective approaches to control blood glucose levels are required but also the novel, multi-directional approaches that help in monitoring the metabolic and proteomic changes associated with the diabetes are needed. Although, along with the pharmacotherapy, life style interventions including diet and exercise can help patients with diabetic complications (Yamagishi and Imaizumi, 2005), novel protein biomarkers that can be used in assessing proteome changes, as a function of anti-diabetic therapeutics treatment will be very useful in evaluating the treatment success as well as patient compliance. The most common oral medicines to control sugar levels in diabetic patients, Thiazolidinediones and Metformin® , often causes either weight gain and/or organ toxicity (Derosa and Maffioli, 2011; Mitri and Hamdy, 2009). Hence alternative strategies to manage glucose levels in diabetics are required, including development and evaluation of traditional/folk medicines, food supplements, along with the standard drug-therapy.
Persimmon (
Dietary supplements have exhibited myriads of positive health effects in different human ailments and over time, new technological advances, including in the fields of proteomics, genomics, and metabolomics, contributed immensely in understanding biological and pharmacological activities of dietary supplements (Astle et al., 2007), with particular emphasis on discovering molecular targets of phytochemicalgene-protein-metabolite dynamics (Milner, 2002). It is reported that the individual responses to dietary supplements are variable/pleiotropic and the food supplements may influence multiple biological pathways (Astle et al., 2007). The proteome, unlikely to relatively static genome, changes dynamically in response to thousands of inter-, intra- and extracellular signals and modulations, thereby reflects the health and/or pathogenesis conditions more accurately. For the last two decades, mass spectrometry (MS) based proteomics, particularly quantitative proteomics, has contributed significantly to the discovery of several biomarkers associated with different human pathophysiology; including irritable bowel syndrome, inflammation, renal failure, interstitial cystitis, Parkinson disease, cancer, diabetes etc (Acosta-Martin et al., 2011; Goo and Goodlett, 2010; Goo et al., 2010; 2012; Herrero et al., 2012; Kim et al., 2010; Lin et al., 2012; Matheson et al., 2010; Pan et al., 2012; Sanna et al., 2015; Zhang et al., 2005). Interestingly, quantitative proteomics based biomarker discovery has been successfully accomplished in all kinds of biological samples; including serum, saliva, urine, and tissues (Goo and Goodlett, 2010). Several biomarkers discovered using proteomics have been graduated into validated assays and are being applied routinely within the clinic, thereby leaving enormous impact on public health. Potentially, discovery of biomarker candidate(s) that could reliably reflect the recent and/or habitual consumption of dietary supplements would greatly help in assessing therapeutic and prophylactic effects of food supplements on human health conditions. In the present MS based proteomics study, we aim to find molecular markers that can be used to evaluate efficacy of PLE dietary supplements on diabetes amelioration, based upon the proteomic profile changes in saliva, serum, and urine samples obtained from the prediabetic patients under PLE treatment.
PLE was obtained from Wanju (Korea) by Dongsangmyeon Saramdeul Inc., as described previously (Bae et al., 2015). PLE was standardized to contain 7.5 mg total Quercetin 3-O-2″galloylglucoside (C24H24O19) and Kaempferol 3-O-2″galloylglucoside (C24H24O18) per 1 g of extract. It was administered as a tablet (500 mg/tab and 2 g/day) composed of 62.6% PLE and 37.4% diluting agent. The placebo tablets were matched with regard to flavor, appearance, and dosage. All subjects were instructed to take 4 tab/day (2 tablets each after breakfast and dinner). PLE and placebo tablets were packaged indistinguishably and labeled with the subject’s number.
The study subjects were recruited from the Clinical Trial Center for Functional Foods at Chonbuk National University Hospital, Jeonju, Republic of Korea during 2014. Subjects of both sexes 20–75 years of age with prediabetes [fasting plasma glucose (FPG) 100–140 mg/dl or impaired glucose tolerance (IGT) 140–250 mg/dl] who had not been diagnosed with any disease and met the inclusion criteria were recruited for this study (Table 1 and 2). All participants gave written informed consent prior to beginning the study. The protocol was approved by the Functional Foods Institutional Review Board (FFIRB) of Chonbuk National University Hospital (FFIRB number: 2013-02-016).
Participants with prediabetic condition were subjected to a randomized crossover study where they were treated with persimmon leaf extract (PLE, n = 5) or placebo (n = 5) twice a day for eight weeks, followed by a four-week wash out period before inverting the treatment regimen (Fig. 1). Saliva, serum, and urine samples from the study subjects were collected for proteomic analyses.
Saliva samples were prepared by using a modified method from
Urine samples were processed following the method outlined in a previous study (Goo et al., 2012). Briefly, ~10 ml of urine samples per subject were thawed at RT and then half of the anti-protease tablet (Roche) was added to avoid protein degradation. Samples were centrifuged at 2000 rpm for 10 min at 4°C and supernatant was collected. Using a Amicon Ultra-15 Centrifugal MWCO 3 kDa filter (EMD Millipore, USA), sample supernatant volume was reduced to ~0.5 – 1 ml. To accomplish TCA precipitation for protein purification from concentrated urine samples, samples and TCA were mixed at 1: 1 ratio using 20% TCA (ice-cold, made fresh) followed by vortex for 15 s. Sample was left on ice for additional 30 min, followed by centrifugation at 14,000 rpm for 15 min at 4°C. Supernatant was discarded and the pellet was washed twice with 100 μl of cold acetone, centrifuged, and the pellet was air-dried for few seconds, prior to re-suspending in 1 ml of 0.5 × PBS. BCA assay and gel analyses were done on processed urine samples, before exposing them to trypsin digestion and LC-MSMS analyses.
Undepleted serum samples (300 μg, ~3 μl) were used for the study without further sample treatment.
Following the BCA analysis of the samples 100 μg of protein per sample (300 μg for undepleted serum) was used for trypsin digestion.
Filter-aided sample preparation (FASP) method for protein digestion, alkylation, and desalting was adapted from Wiśniewski et al. (2009). Briefly, 100 μg of protein sample (300 μg for undepleted serum) was mixed with 250 μl of 7.5 mM Tris-(2-carboxyethyl)-phosphine hydrochloride (TECP), 8M Urea, and 100mM Ammonium bicarbonate pH 8.0 (Ambic) and loaded into a 3K MWCO filter, followed by incubation in a dry-bath at 37°C for 1 h. Post-reduction, sample was centrifuged at 14,000 rpm for 15 min and after adding 250 μl of 8M Urea, 100 mM Ambic, and 50 mM Iodoacetamide (IAM), the reaction was left for alkylation in dark for 1 h at room temperature. Post-alkylation, sample was centrifuged at 14,000 rpm for 15 min and 25 μl of 500 mM DL-Dithiothreitol (DTT) was added by pipetting to deactivate residual IAM and sample was centrifuged again. Processed sample was further washed four-times with 300 μl of 50 mM Ambic, each time discarding the filtrate, before collecting the clean sample for trypsinization. Trypsin (Promega, Mass spectrometry grade, USA) was added to the sample in 50 protein: 1 trypsin weight ratio and incubated for overnight at 37°C. Digested samples were diluted to a final concentration of 95% water: 5% acetonitrile: 0.1% trifluoroacetic acid and speedvac to dryness. Dried peptides were resuspended in 95% water: 5.0% acetonitrile: 0.1% formic acid to a concentration of 0.5 μg/μl protein. Samples were stored at -80°C till LC-MS/MS analyses.
Trypsin-digested samples were analyzed using MS/MS data-dependent acquisition by electrospray ionization in the positive ion mode on a tribrid quadrupole-orbitrap-ion trap mass spectrometer (Orbitrap Fusion™; Thermo Fisher, USA). The Fusion was coupled with a nanoflow LC system (NanoAcquity ™, Waters Corporation, USA) equipped with online helium-degasser. Prior to peptide separation on a homemade 75 μm i.d. ×180 mm long analytical column pulled using a Sutter Instruments P-2000 CO2 laser puller (Sutter Instrument Company, USA) and packed with 5 μm (100 Å pore, C18AQ: Michrom BioResources Inc., USA) particle, peptides were trapped on an in-house packed 100 μm i.d. ×20 mm long pre-column (5 μm, 200 Å pore, C18AQ; Michrom Bio-Resources). An estimated 1 μg of peptide mixture was injected per-LC-MS/MS run onto a pre-column at 4 μl/min flow rate using mobile phase-A (water, 0.1% formic acid) and mobile phase-B (acetonitrile, 0.1% formic acid) in a A:B ratio 95/5. Peptides were separated and eluted using an acetonitrile gradient of 5–35% (mobile phase-B) over 30 min for a total of 95 min duration. Data-dependent MS acquisition was accomplished using Xcalibur v3.0 (Thermo Fisher), followed by tandem mass spectrometry (MS/MS) data acquisition for the 20 most abundant precursor ions, excluding singly charged ions. To avoid/minimize any data redundancy, dynamic range was set for 60 s. All samples were run in triplicate and ion source conditions were optimized using the vendor-recommended tuning and calibration solutions, prior to any acquisition. Tandem MS data were queried for protein identification against the UniprotKB
Proteins that were identified with ≥ 2 peptides were subjected to a label-free quantitation in order to compare relative expression in control and PLE-treated samples. MaxQuant LFQ intensities were converted to Log2 values using Perseus software (version 1.5.0.9) and differences in relative expression of proteins [(Log2 (postPLE/prePLE) and Log2 (postPlacebo/prePlacebo) of ≥ +1 or ≤ -1 (± 2-folds)], principle component analyses, and cluster analyses were achieved in Perseus. The t-test was used to test statistical significance with differences considered as statistically significant for p-values < 0.05 between two treatment groups. The data filtering was accomplished using R v3.2.2 statistics software tool. Volcano plots for
Western blotting was used to validate the observed potential biomarker candidate(s) levels in pre-PLE and post-PLE treatment patient groups. 10 μg of protein extract was electrophoresed on 4–15% gradient gels (Mini-PROTEAN TGX™, BioRad, USA) and transferred to polyvinylidene fluoride (PVDF) membrane using the wet transfer method at 350 mA, for 1 h over ice. Membrane was washed in tris buffered saline w/0.1% tween 20 (TBST), blocked with 5% bovine serum albumin (BSA) in TBST for an hour and probed overnight at 4°C with primary antibodies for, mouse monoclonal Ezrin (Abcam, USA) or polyclonal Sheep human Uromodulin antibody (R&D Systems, USA), rabbit monoclonal GAPDH (Cell Signaling Technology-CST, USA). Membrane was washed in TBST and probed with secondary antibodies, anti-Mousehorse radish peroxidase (HRP) (1:5000, Cell Signaling Technology-CST) or anti-Rabbit-HRP (1:5000, BioLegend, USA) or anti-sheep-HRP (1:5000, Abcam). Membrane was developed by enhanced chemiluminescence (Amersham ECL® Select, GE Healthcare, USA).
The goal of this study was to find protein marker(s) that can be used to evaluate efficacy of PLE dietary supplements on diabetes amelioration, based upon the proteomic profile changes in different body fluids obtained from the prediabetic patients following controlled PLE intake. PLE-treatment showed an improvement in the blood glucose level of enrolled participants as determined by the PPG levels, in comparison to placebo treated group (Table 1). Five subjects with prediabetic condition (Table 2) were participated in 20 week long study where the different body fluid samples, saliva, urine, and serum, were collected in four time points at week 0, 8, 12, and 20 as described in the Fig. 1 for proteomics study. Data obtained from the different time points were subjected for various statistical analyses. Initial t-test showed no significant difference in protein expression levels post PLE treatment. Principal component analysis (PCA) revealed that better correlation within individual subject, but not by the PLE treatment, suggesting the ameliorative effects caused by the PLE might be too minimal to overcome individual’s genetic barrier. The same observation was made from all different sample types including saliva, serum, and urine (Fig. 2).
This initial observation prompted an alternative analytical approach to look for changes induced by the PLE treatment as shown in Fig. 3. First, protein lists for proteomic changes between two major groups; PLE treatment (PostPLE) versus Pre PLE treatment (PrePLE) and PostPlacebo versus PrePlacebo were populated based on the criteria that any protein that show Log2-fold change of ≥ +1 or ≤ −1 (2-fold up and down regulation, respectively) should be listed. Second, common proteins present in the placebo group (PostPlacebo versus PrePlacebo) were removed from the treatment group (PostPLE versus PrePLE), in order to get unique proteins list potentially due to PLE treatment but not due to genetic or environmental effects. Finally, the proteins that showed consistent up/down-regulation in at least three out of five patients group were analyzed for their relevance to diabetes and termed as potential protein marker(s) of PLE treatment. The candidate protein list, that met above described criteria, and their expression level are shown in Table 3 and Fig. 4.
While we were able to detect proteins that undergo changes by the PLE treatment in saliva and urine samples, we could not find proteins that showed statistically significant differences in expression from serum samples except for only small fold changes (data not shown). Perhaps the subtle proteomic changes in low-abundance proteins were not detected by mass spectrometry, due to presence of high-abundance proteins (Merrell et al., 2004).
Proteomics results identified salivary and urinary proteins through comparative proteomics analysis between pre and post PLE treatments. Literature search showed no previous implications of SPARC-like protein 1 (SPARC-L 1), leukocyte-associated immunoglobulin-like receptor-1 (LAIR-1), immunoglobulin kappa variable 4 (IGKV4; V4-1), and complement C7 with diabetes (Table 3). Of these, a renal glycoprotein uromdulin, also known as Tamm-Horsfall protein (Lo et al., 2014), showed down-regulation in prediabetic subjects after the PLE treatment whereas, Ezrin (Wasik et al., 2014), which is involved in diabetes and obesity, was found to be upregulated due to PLE-treatment, suggesting ameliorative effects of PLE. Physiological functions of uromodulin are still elusive, despite the fact that it is involved in the pathophysiology of various diseases including diabetes, renal inflammation, acute and chronic kidney diseases. Conventionally, uromodulin was regarded as an instigator in kidney injury; however recent studies suggested it has a protective role in acute kidney inflammation (El-Achkar and Wu, 2012). Furthermore, in chronic kidney disease, uromodulin excretion is increased and has been proposed that uromodulin exacerbates kidney injury. Previously, decreased levels of uromodulin have been correlated with severity of chronic kidney diseases, diabetes, and diabetes-associated albuminuria (Chakraborty et al., 2004; Lapolla et al., 2009; Lo et al., 2014; Lynn and Marshall, 1984; Roscioni et al., 2013). Though, diabetic patients and patients with early kidney disease have shown increased uromodulin secretion, correlating with the fact that diabetic renal systems might exhibit hyper-filtration (Torffvit and Agardh, 1994). However, microalbuminuria often does not correlate well with uromodulin excretion (Pfleiderer et al., 1993). While baseline uromodulin measurements potentially are useful in predicting disease progression, longitudinal measurement will provide more accurate picture (El-Achkar and Wu, 2012). Urinary proteomics has shown that with down-regulation of uromodulin, albuminuria increases which suggests progressive dysfunction of urinary system, particularly nephrons (Roscioni et al., 2013). Interestingly, Elevated urinary uromodulin levels preceded the development of chronic kidney disease (CKD) and showed association with a common polymer phism in uromodulin gene (Köttgen et al., 2010). In present study, uromodulin expression was found to be down regulated in one patient, while the other two patients showed very subtle changes as found by the Western blot (WB) validation (Fig. 6). Unfortunately we were unable to see strong quantitative correlation for uromodulin in urine samples; in addition GAPDH was used as a loading control for saliva but not detected in urine WB assay. In future studies, we will use transferrin or beta tubulin as a loading control for urine samples, which are better control for urine samples (Beretov et al., 2015). Nonetheless, we were able to detect Uromodulin protein in urine samples by WB assay.
Ezrin is down regulated in glomeruli from slightly diabetic and insulin-resistant obese Zucker rats or in podocytes of type 2 diabetes patients, suggesting ezrin down-regulation may add to the diabetes-mediated kidney diseases (Wasik et al., 2014). Earlier reports suggest that ezrin-podocalyxin-NHERF2 interaction may modulate podocyte functions (Takeda et al, 2001) and a decrease of podocytes causes diabetes related nephropathy in patients (Huang, 2011). WB validation (Fig. 6) confirmed that the PLE treatment resulted in increased expression of ezrin protein levels from the two of three subjects tested. MS efficiently detected Ezrin (P15311; Mol. weight [kDa] 69.4) across the samples with sequence coverage of 39.4%, detecting 23 total peptides with 13 being unique peptides, covering sequence length of 586, intensity 2.9E9, and total MS/MS counts of 339. All in all, we were able to specifically detect and quantify Ezrin levels in salivary samples (Fig. 5); a representative MS/MS spectrum for a unique peptide “SQEQLAAELAEYTAK” (MH+ 1651.8174; m/z 551.2773 for 3+) for Ezrin protein (comet search accuracy = 0.999) is shown in Fig. 7.
Due to limited number of subjects available, statistically, the study has some limitations, however, this is the first proteomics study that explores the effects of PLE dietary supplement on prediabetic patients from the three different body fluids. Given we don’t have access to the PK/PD data, it is hard to understand the PLE bioavailability. Infact, Naoxinqing, a commercially available PLE tablet showed poor dissolution in water, and Li et al. (2011) showed that the self-nanoemulsifying drug delivery system for PLE improved its bioavailability. Future studies using nanoemulsifying drug delivery systems might help with increased bioavailability, thereby further assisting in PLE-mediated disease amelioration. Nonetheless, a recent discovery based proteomics study reported that SPARC-like protein 1 precursor (SPARCL1) levels were elevated in non-proliferative diabetic retinopathy subjects (Chee et al. 2016). In our study, the PLE-treated patients showed down-regulation of SPARCL1 (Table 3). Interestingly, Gao et al. (2008) characterized the vitreous proteome of nondiabetic individuals (NDM), diabetic with proliferative retinopathy (PDR) or without retinopathy (noDR) and reported elevated levels of the proteins involved in complement system; in comparison to NDM, noDR as well as PDR showed elevated levels of Complement proteins including C 7. Interaction between the coagulation and complement system is well known as they are critically involved in the inflammatory response and can undergo massive activation after injury causing vascular lesions (Amara et al., 2008). In present study, we have found that PLE-treatment group showed down-regulation of Complement C7 (Table 3), suggesting dietary food supplement PLE exhibited ameliorative effects on prediabetic conditions. Further targeted proteomic studies are warranted to comprehensively elucidate the correlation of salivary proteome to the disease progression and/or amelioration, when treated with food supplements or small-molecule drugs.
Diabetes requires non-invasive, easy methodology for life-long glucose monitoring as well as for assessing treatment efficacy associated with the pharmacotherapy. Salivary proteins can be used for evaluating treatment effectiveness as well as improving patient compliance. MS-based proteomic studies were accomplished in order to discover proteomics molecular signatures to understand therapeutic potentials of PLE in diabetes amelioration. Subtle protein changes were observed after PLE treatment from all three-sample types including saliva, urine and serum and a stronger clustering was observed by individuals but not by treatment type. This indicates genetic integrity exceeded the changes may have induced by PLE treatment. In our study design, the intake of PLE was for twice a day for eight weeks. Unlike a drug treatment, food supplements might take longer period to show significant proteomic changes across patient groups. Saliva sample collection is simple, noninvasive, and allows multiple time-points collection of sample with comparative ease thereby aiding in diagnosis, understanding and monitoring disease progression, or treatment responses. Recently, saliva has emerged as an alternative body fluid for clinical diagnostics (Hu et al., 2007), other than urine, which, however, requires comparatively time-consuming sample preparation steps and usually has sample degradation issues. Salivary protein markers, ezrin or SPARCL1 responded well to the PLE-treatment as determined by proteomic analyses; downstream validation for Ezrin by WB validation and previous research supports (Amara et al., 2008; Gao et al., 2008) that PLE treatment implicates possible restoring of the protein function and/or expression for diabetes amelioration. Potentially an antibody-based method can be developed to monitor proteomic changes in salivary proteins using potential biomarker candidate ezrin or SPARCL1. However, a longitudinal study with a bigger patient cohort and longer duration of PLE intake would be needed to follow-up study to assess the specificity and sensitivity of salivary markers. Nonetheless, protein candidate signature(s) from the present study could serve as a starting point for the future quantitative mass spectrometry-based peptidomic analyses, such as multiple- or parallel- reaction monitoring (MRM or PRM) (Shi et al., 2016), in order to accomplish longitudinal data for PLE-treated subjects.
Postprandial glucose obtained prior to and after treatment with PLE or placebo
PLE group (n = 34) | Placebo group (n = 34) | ||||||
---|---|---|---|---|---|---|---|
0 w | 8 w | 0 w | 8 w | ||||
PPG120 min (mg/dl) | 171.36 ± 38.38 | 160.91 ± 35.62 | 0.133 | 164.73 ± 32.95 | 172.03 ± 33.61 | 0.126 | 0.029* |
Values are presented as mean ± S.D.
2)Linear mixed model,
*
Demographic characteristics of the study participants
Subjects (n = 5) | |
---|---|
Age (year) | 58.00 ± 8.43 |
Sex (M/F) | 2/3 |
Height (cm) | 158.60 ± 6.02 |
Weight (kg) | 62.80 ± 7.12 |
BMI (kg/m2) | 24.90 ± 1.17 |
FPG (mg/dl) | 103.80 ± 12.17 |
2 h-PPG (mg/dl) | 158.60 ± 14.22 |
Proteins showed at least 2-fold changes among study subjects by the PLE treatment.
Protein Name | Patient#1 | Patient#2 | Patient#3 | Patient#4 | Patient#5 |
---|---|---|---|---|---|
Saliva proteomics analyses | |||||
SPARC-like protein 1 | −1.24 | ND | ND | −1.66 | −1.13 |
Ezrin | 1.35 | 1.10 | 1.13 | ND | ND |
Urine proteomics analyses | |||||
LAIR-1 | −2.22 | ND | ND | −1.23 | −2.08 |
IGKV4 | −3.19 | ND | −2.20 | −1.11 | ND |
IGKV4-1 | 1.29 | ND | 1.30 | ND | 3.97 |
Complement C7 | −1.43 | ND | −1.45 | ND | −3.81 |
Uromodulin | ND | ND | −1.14 | −1.08 | −1.27 |
The value represents log2-fold changes (ND = not detected)
Mol. Cells 2017; 40(7): 466-475
Published online July 31, 2017 https://doi.org/10.14348/molcells.2017.2298
Copyright © The Korean Society for Molecular and Cellular Biology.
Mohd M. Khan1,6, Bao Quoc Tran1,7, Yoon-Jin Jang2, Soo-Hyun Park3, William E. Fondrie4, Khadiza Chowdhury4, Sung Hwan Yoon1, David R. Goodlett1, Soo-Wan Chae2,3, Han-Jung Chae2, Seung-Young Seo5,*, and Young Ah Goo1,8,*
1Department of Pharmaceutical Sciences, School of Pharmacy, University of Maryland, Baltimore, MD 21201, USA, 2Department of Pharmacology, Chonbuk National University Medical School, Jeonju 54907, Korea, 3Clinical Trial Center for Functional Foods, Chonbuk National University Hospital, Jeonju 54907, Korea, 4University of Maryland School of Medicine, Baltimore, MD 21201, USA, 5Department of Internal Medicine, Research Institute of Clinical Medicine, Chonbuk National University Medical School and Hospital, Jeonju 54907, Korea, 6Present address: University of Maryland School of Medicine, Baltimore, MD 21201, USA
Correspondence to:*Correspondence: young.goo@northwestern.edu (YAG); bear7905@naver.com (SYS)
Dietary supplements have exhibited myriads of positive health effects on human health conditions and with the advent of new technological advances, including in the fields of proteomics, genomics, and metabolomics, biological and pharmacological activities of dietary supplements are being evaluated for their ameliorative effects in human ailments. Recent interests in understanding and discovering the molecular targets of phytochemical-gene-protein-metabolite dynamics resulted in discovery of a few protein signature candidates that could potentially be used to assess the effects of dietary supplements on human health. Persimmon (
Keywords: biomarker candidates, diabetes, mass spectrometry, persimmon leaf extract, proteomics, saliva, serum, urine
Diabetes mellitus, the most prevalent form of diabetes, is a metabolic disease due to impaired insulin secretion and/or action (Das and Elbein, 2006), which causes hyperglycemia (Gerich, 1998). In 2010, an estimated 285 million people worldwide were affected by diabetes, and this figure is predicted to reach 439 million by 2030 (Shaw et al., 2010). Over the next decade, the diabetes related long-term complications, morbidity, and mortality are expected to be more prevalent due to increasing rates of childhood and adulthood obesity (CDC, Atlanta, USA). Therefore, not only effective approaches to control blood glucose levels are required but also the novel, multi-directional approaches that help in monitoring the metabolic and proteomic changes associated with the diabetes are needed. Although, along with the pharmacotherapy, life style interventions including diet and exercise can help patients with diabetic complications (Yamagishi and Imaizumi, 2005), novel protein biomarkers that can be used in assessing proteome changes, as a function of anti-diabetic therapeutics treatment will be very useful in evaluating the treatment success as well as patient compliance. The most common oral medicines to control sugar levels in diabetic patients, Thiazolidinediones and Metformin® , often causes either weight gain and/or organ toxicity (Derosa and Maffioli, 2011; Mitri and Hamdy, 2009). Hence alternative strategies to manage glucose levels in diabetics are required, including development and evaluation of traditional/folk medicines, food supplements, along with the standard drug-therapy.
Persimmon (
Dietary supplements have exhibited myriads of positive health effects in different human ailments and over time, new technological advances, including in the fields of proteomics, genomics, and metabolomics, contributed immensely in understanding biological and pharmacological activities of dietary supplements (Astle et al., 2007), with particular emphasis on discovering molecular targets of phytochemicalgene-protein-metabolite dynamics (Milner, 2002). It is reported that the individual responses to dietary supplements are variable/pleiotropic and the food supplements may influence multiple biological pathways (Astle et al., 2007). The proteome, unlikely to relatively static genome, changes dynamically in response to thousands of inter-, intra- and extracellular signals and modulations, thereby reflects the health and/or pathogenesis conditions more accurately. For the last two decades, mass spectrometry (MS) based proteomics, particularly quantitative proteomics, has contributed significantly to the discovery of several biomarkers associated with different human pathophysiology; including irritable bowel syndrome, inflammation, renal failure, interstitial cystitis, Parkinson disease, cancer, diabetes etc (Acosta-Martin et al., 2011; Goo and Goodlett, 2010; Goo et al., 2010; 2012; Herrero et al., 2012; Kim et al., 2010; Lin et al., 2012; Matheson et al., 2010; Pan et al., 2012; Sanna et al., 2015; Zhang et al., 2005). Interestingly, quantitative proteomics based biomarker discovery has been successfully accomplished in all kinds of biological samples; including serum, saliva, urine, and tissues (Goo and Goodlett, 2010). Several biomarkers discovered using proteomics have been graduated into validated assays and are being applied routinely within the clinic, thereby leaving enormous impact on public health. Potentially, discovery of biomarker candidate(s) that could reliably reflect the recent and/or habitual consumption of dietary supplements would greatly help in assessing therapeutic and prophylactic effects of food supplements on human health conditions. In the present MS based proteomics study, we aim to find molecular markers that can be used to evaluate efficacy of PLE dietary supplements on diabetes amelioration, based upon the proteomic profile changes in saliva, serum, and urine samples obtained from the prediabetic patients under PLE treatment.
PLE was obtained from Wanju (Korea) by Dongsangmyeon Saramdeul Inc., as described previously (Bae et al., 2015). PLE was standardized to contain 7.5 mg total Quercetin 3-O-2″galloylglucoside (C24H24O19) and Kaempferol 3-O-2″galloylglucoside (C24H24O18) per 1 g of extract. It was administered as a tablet (500 mg/tab and 2 g/day) composed of 62.6% PLE and 37.4% diluting agent. The placebo tablets were matched with regard to flavor, appearance, and dosage. All subjects were instructed to take 4 tab/day (2 tablets each after breakfast and dinner). PLE and placebo tablets were packaged indistinguishably and labeled with the subject’s number.
The study subjects were recruited from the Clinical Trial Center for Functional Foods at Chonbuk National University Hospital, Jeonju, Republic of Korea during 2014. Subjects of both sexes 20–75 years of age with prediabetes [fasting plasma glucose (FPG) 100–140 mg/dl or impaired glucose tolerance (IGT) 140–250 mg/dl] who had not been diagnosed with any disease and met the inclusion criteria were recruited for this study (Table 1 and 2). All participants gave written informed consent prior to beginning the study. The protocol was approved by the Functional Foods Institutional Review Board (FFIRB) of Chonbuk National University Hospital (FFIRB number: 2013-02-016).
Participants with prediabetic condition were subjected to a randomized crossover study where they were treated with persimmon leaf extract (PLE, n = 5) or placebo (n = 5) twice a day for eight weeks, followed by a four-week wash out period before inverting the treatment regimen (Fig. 1). Saliva, serum, and urine samples from the study subjects were collected for proteomic analyses.
Saliva samples were prepared by using a modified method from
Urine samples were processed following the method outlined in a previous study (Goo et al., 2012). Briefly, ~10 ml of urine samples per subject were thawed at RT and then half of the anti-protease tablet (Roche) was added to avoid protein degradation. Samples were centrifuged at 2000 rpm for 10 min at 4°C and supernatant was collected. Using a Amicon Ultra-15 Centrifugal MWCO 3 kDa filter (EMD Millipore, USA), sample supernatant volume was reduced to ~0.5 – 1 ml. To accomplish TCA precipitation for protein purification from concentrated urine samples, samples and TCA were mixed at 1: 1 ratio using 20% TCA (ice-cold, made fresh) followed by vortex for 15 s. Sample was left on ice for additional 30 min, followed by centrifugation at 14,000 rpm for 15 min at 4°C. Supernatant was discarded and the pellet was washed twice with 100 μl of cold acetone, centrifuged, and the pellet was air-dried for few seconds, prior to re-suspending in 1 ml of 0.5 × PBS. BCA assay and gel analyses were done on processed urine samples, before exposing them to trypsin digestion and LC-MSMS analyses.
Undepleted serum samples (300 μg, ~3 μl) were used for the study without further sample treatment.
Following the BCA analysis of the samples 100 μg of protein per sample (300 μg for undepleted serum) was used for trypsin digestion.
Filter-aided sample preparation (FASP) method for protein digestion, alkylation, and desalting was adapted from Wiśniewski et al. (2009). Briefly, 100 μg of protein sample (300 μg for undepleted serum) was mixed with 250 μl of 7.5 mM Tris-(2-carboxyethyl)-phosphine hydrochloride (TECP), 8M Urea, and 100mM Ammonium bicarbonate pH 8.0 (Ambic) and loaded into a 3K MWCO filter, followed by incubation in a dry-bath at 37°C for 1 h. Post-reduction, sample was centrifuged at 14,000 rpm for 15 min and after adding 250 μl of 8M Urea, 100 mM Ambic, and 50 mM Iodoacetamide (IAM), the reaction was left for alkylation in dark for 1 h at room temperature. Post-alkylation, sample was centrifuged at 14,000 rpm for 15 min and 25 μl of 500 mM DL-Dithiothreitol (DTT) was added by pipetting to deactivate residual IAM and sample was centrifuged again. Processed sample was further washed four-times with 300 μl of 50 mM Ambic, each time discarding the filtrate, before collecting the clean sample for trypsinization. Trypsin (Promega, Mass spectrometry grade, USA) was added to the sample in 50 protein: 1 trypsin weight ratio and incubated for overnight at 37°C. Digested samples were diluted to a final concentration of 95% water: 5% acetonitrile: 0.1% trifluoroacetic acid and speedvac to dryness. Dried peptides were resuspended in 95% water: 5.0% acetonitrile: 0.1% formic acid to a concentration of 0.5 μg/μl protein. Samples were stored at -80°C till LC-MS/MS analyses.
Trypsin-digested samples were analyzed using MS/MS data-dependent acquisition by electrospray ionization in the positive ion mode on a tribrid quadrupole-orbitrap-ion trap mass spectrometer (Orbitrap Fusion™; Thermo Fisher, USA). The Fusion was coupled with a nanoflow LC system (NanoAcquity ™, Waters Corporation, USA) equipped with online helium-degasser. Prior to peptide separation on a homemade 75 μm i.d. ×180 mm long analytical column pulled using a Sutter Instruments P-2000 CO2 laser puller (Sutter Instrument Company, USA) and packed with 5 μm (100 Å pore, C18AQ: Michrom BioResources Inc., USA) particle, peptides were trapped on an in-house packed 100 μm i.d. ×20 mm long pre-column (5 μm, 200 Å pore, C18AQ; Michrom Bio-Resources). An estimated 1 μg of peptide mixture was injected per-LC-MS/MS run onto a pre-column at 4 μl/min flow rate using mobile phase-A (water, 0.1% formic acid) and mobile phase-B (acetonitrile, 0.1% formic acid) in a A:B ratio 95/5. Peptides were separated and eluted using an acetonitrile gradient of 5–35% (mobile phase-B) over 30 min for a total of 95 min duration. Data-dependent MS acquisition was accomplished using Xcalibur v3.0 (Thermo Fisher), followed by tandem mass spectrometry (MS/MS) data acquisition for the 20 most abundant precursor ions, excluding singly charged ions. To avoid/minimize any data redundancy, dynamic range was set for 60 s. All samples were run in triplicate and ion source conditions were optimized using the vendor-recommended tuning and calibration solutions, prior to any acquisition. Tandem MS data were queried for protein identification against the UniprotKB
Proteins that were identified with ≥ 2 peptides were subjected to a label-free quantitation in order to compare relative expression in control and PLE-treated samples. MaxQuant LFQ intensities were converted to Log2 values using Perseus software (version 1.5.0.9) and differences in relative expression of proteins [(Log2 (postPLE/prePLE) and Log2 (postPlacebo/prePlacebo) of ≥ +1 or ≤ -1 (± 2-folds)], principle component analyses, and cluster analyses were achieved in Perseus. The t-test was used to test statistical significance with differences considered as statistically significant for p-values < 0.05 between two treatment groups. The data filtering was accomplished using R v3.2.2 statistics software tool. Volcano plots for
Western blotting was used to validate the observed potential biomarker candidate(s) levels in pre-PLE and post-PLE treatment patient groups. 10 μg of protein extract was electrophoresed on 4–15% gradient gels (Mini-PROTEAN TGX™, BioRad, USA) and transferred to polyvinylidene fluoride (PVDF) membrane using the wet transfer method at 350 mA, for 1 h over ice. Membrane was washed in tris buffered saline w/0.1% tween 20 (TBST), blocked with 5% bovine serum albumin (BSA) in TBST for an hour and probed overnight at 4°C with primary antibodies for, mouse monoclonal Ezrin (Abcam, USA) or polyclonal Sheep human Uromodulin antibody (R&D Systems, USA), rabbit monoclonal GAPDH (Cell Signaling Technology-CST, USA). Membrane was washed in TBST and probed with secondary antibodies, anti-Mousehorse radish peroxidase (HRP) (1:5000, Cell Signaling Technology-CST) or anti-Rabbit-HRP (1:5000, BioLegend, USA) or anti-sheep-HRP (1:5000, Abcam). Membrane was developed by enhanced chemiluminescence (Amersham ECL® Select, GE Healthcare, USA).
The goal of this study was to find protein marker(s) that can be used to evaluate efficacy of PLE dietary supplements on diabetes amelioration, based upon the proteomic profile changes in different body fluids obtained from the prediabetic patients following controlled PLE intake. PLE-treatment showed an improvement in the blood glucose level of enrolled participants as determined by the PPG levels, in comparison to placebo treated group (Table 1). Five subjects with prediabetic condition (Table 2) were participated in 20 week long study where the different body fluid samples, saliva, urine, and serum, were collected in four time points at week 0, 8, 12, and 20 as described in the Fig. 1 for proteomics study. Data obtained from the different time points were subjected for various statistical analyses. Initial t-test showed no significant difference in protein expression levels post PLE treatment. Principal component analysis (PCA) revealed that better correlation within individual subject, but not by the PLE treatment, suggesting the ameliorative effects caused by the PLE might be too minimal to overcome individual’s genetic barrier. The same observation was made from all different sample types including saliva, serum, and urine (Fig. 2).
This initial observation prompted an alternative analytical approach to look for changes induced by the PLE treatment as shown in Fig. 3. First, protein lists for proteomic changes between two major groups; PLE treatment (PostPLE) versus Pre PLE treatment (PrePLE) and PostPlacebo versus PrePlacebo were populated based on the criteria that any protein that show Log2-fold change of ≥ +1 or ≤ −1 (2-fold up and down regulation, respectively) should be listed. Second, common proteins present in the placebo group (PostPlacebo versus PrePlacebo) were removed from the treatment group (PostPLE versus PrePLE), in order to get unique proteins list potentially due to PLE treatment but not due to genetic or environmental effects. Finally, the proteins that showed consistent up/down-regulation in at least three out of five patients group were analyzed for their relevance to diabetes and termed as potential protein marker(s) of PLE treatment. The candidate protein list, that met above described criteria, and their expression level are shown in Table 3 and Fig. 4.
While we were able to detect proteins that undergo changes by the PLE treatment in saliva and urine samples, we could not find proteins that showed statistically significant differences in expression from serum samples except for only small fold changes (data not shown). Perhaps the subtle proteomic changes in low-abundance proteins were not detected by mass spectrometry, due to presence of high-abundance proteins (Merrell et al., 2004).
Proteomics results identified salivary and urinary proteins through comparative proteomics analysis between pre and post PLE treatments. Literature search showed no previous implications of SPARC-like protein 1 (SPARC-L 1), leukocyte-associated immunoglobulin-like receptor-1 (LAIR-1), immunoglobulin kappa variable 4 (IGKV4; V4-1), and complement C7 with diabetes (Table 3). Of these, a renal glycoprotein uromdulin, also known as Tamm-Horsfall protein (Lo et al., 2014), showed down-regulation in prediabetic subjects after the PLE treatment whereas, Ezrin (Wasik et al., 2014), which is involved in diabetes and obesity, was found to be upregulated due to PLE-treatment, suggesting ameliorative effects of PLE. Physiological functions of uromodulin are still elusive, despite the fact that it is involved in the pathophysiology of various diseases including diabetes, renal inflammation, acute and chronic kidney diseases. Conventionally, uromodulin was regarded as an instigator in kidney injury; however recent studies suggested it has a protective role in acute kidney inflammation (El-Achkar and Wu, 2012). Furthermore, in chronic kidney disease, uromodulin excretion is increased and has been proposed that uromodulin exacerbates kidney injury. Previously, decreased levels of uromodulin have been correlated with severity of chronic kidney diseases, diabetes, and diabetes-associated albuminuria (Chakraborty et al., 2004; Lapolla et al., 2009; Lo et al., 2014; Lynn and Marshall, 1984; Roscioni et al., 2013). Though, diabetic patients and patients with early kidney disease have shown increased uromodulin secretion, correlating with the fact that diabetic renal systems might exhibit hyper-filtration (Torffvit and Agardh, 1994). However, microalbuminuria often does not correlate well with uromodulin excretion (Pfleiderer et al., 1993). While baseline uromodulin measurements potentially are useful in predicting disease progression, longitudinal measurement will provide more accurate picture (El-Achkar and Wu, 2012). Urinary proteomics has shown that with down-regulation of uromodulin, albuminuria increases which suggests progressive dysfunction of urinary system, particularly nephrons (Roscioni et al., 2013). Interestingly, Elevated urinary uromodulin levels preceded the development of chronic kidney disease (CKD) and showed association with a common polymer phism in uromodulin gene (Köttgen et al., 2010). In present study, uromodulin expression was found to be down regulated in one patient, while the other two patients showed very subtle changes as found by the Western blot (WB) validation (Fig. 6). Unfortunately we were unable to see strong quantitative correlation for uromodulin in urine samples; in addition GAPDH was used as a loading control for saliva but not detected in urine WB assay. In future studies, we will use transferrin or beta tubulin as a loading control for urine samples, which are better control for urine samples (Beretov et al., 2015). Nonetheless, we were able to detect Uromodulin protein in urine samples by WB assay.
Ezrin is down regulated in glomeruli from slightly diabetic and insulin-resistant obese Zucker rats or in podocytes of type 2 diabetes patients, suggesting ezrin down-regulation may add to the diabetes-mediated kidney diseases (Wasik et al., 2014). Earlier reports suggest that ezrin-podocalyxin-NHERF2 interaction may modulate podocyte functions (Takeda et al, 2001) and a decrease of podocytes causes diabetes related nephropathy in patients (Huang, 2011). WB validation (Fig. 6) confirmed that the PLE treatment resulted in increased expression of ezrin protein levels from the two of three subjects tested. MS efficiently detected Ezrin (P15311; Mol. weight [kDa] 69.4) across the samples with sequence coverage of 39.4%, detecting 23 total peptides with 13 being unique peptides, covering sequence length of 586, intensity 2.9E9, and total MS/MS counts of 339. All in all, we were able to specifically detect and quantify Ezrin levels in salivary samples (Fig. 5); a representative MS/MS spectrum for a unique peptide “SQEQLAAELAEYTAK” (MH+ 1651.8174; m/z 551.2773 for 3+) for Ezrin protein (comet search accuracy = 0.999) is shown in Fig. 7.
Due to limited number of subjects available, statistically, the study has some limitations, however, this is the first proteomics study that explores the effects of PLE dietary supplement on prediabetic patients from the three different body fluids. Given we don’t have access to the PK/PD data, it is hard to understand the PLE bioavailability. Infact, Naoxinqing, a commercially available PLE tablet showed poor dissolution in water, and Li et al. (2011) showed that the self-nanoemulsifying drug delivery system for PLE improved its bioavailability. Future studies using nanoemulsifying drug delivery systems might help with increased bioavailability, thereby further assisting in PLE-mediated disease amelioration. Nonetheless, a recent discovery based proteomics study reported that SPARC-like protein 1 precursor (SPARCL1) levels were elevated in non-proliferative diabetic retinopathy subjects (Chee et al. 2016). In our study, the PLE-treated patients showed down-regulation of SPARCL1 (Table 3). Interestingly, Gao et al. (2008) characterized the vitreous proteome of nondiabetic individuals (NDM), diabetic with proliferative retinopathy (PDR) or without retinopathy (noDR) and reported elevated levels of the proteins involved in complement system; in comparison to NDM, noDR as well as PDR showed elevated levels of Complement proteins including C 7. Interaction between the coagulation and complement system is well known as they are critically involved in the inflammatory response and can undergo massive activation after injury causing vascular lesions (Amara et al., 2008). In present study, we have found that PLE-treatment group showed down-regulation of Complement C7 (Table 3), suggesting dietary food supplement PLE exhibited ameliorative effects on prediabetic conditions. Further targeted proteomic studies are warranted to comprehensively elucidate the correlation of salivary proteome to the disease progression and/or amelioration, when treated with food supplements or small-molecule drugs.
Diabetes requires non-invasive, easy methodology for life-long glucose monitoring as well as for assessing treatment efficacy associated with the pharmacotherapy. Salivary proteins can be used for evaluating treatment effectiveness as well as improving patient compliance. MS-based proteomic studies were accomplished in order to discover proteomics molecular signatures to understand therapeutic potentials of PLE in diabetes amelioration. Subtle protein changes were observed after PLE treatment from all three-sample types including saliva, urine and serum and a stronger clustering was observed by individuals but not by treatment type. This indicates genetic integrity exceeded the changes may have induced by PLE treatment. In our study design, the intake of PLE was for twice a day for eight weeks. Unlike a drug treatment, food supplements might take longer period to show significant proteomic changes across patient groups. Saliva sample collection is simple, noninvasive, and allows multiple time-points collection of sample with comparative ease thereby aiding in diagnosis, understanding and monitoring disease progression, or treatment responses. Recently, saliva has emerged as an alternative body fluid for clinical diagnostics (Hu et al., 2007), other than urine, which, however, requires comparatively time-consuming sample preparation steps and usually has sample degradation issues. Salivary protein markers, ezrin or SPARCL1 responded well to the PLE-treatment as determined by proteomic analyses; downstream validation for Ezrin by WB validation and previous research supports (Amara et al., 2008; Gao et al., 2008) that PLE treatment implicates possible restoring of the protein function and/or expression for diabetes amelioration. Potentially an antibody-based method can be developed to monitor proteomic changes in salivary proteins using potential biomarker candidate ezrin or SPARCL1. However, a longitudinal study with a bigger patient cohort and longer duration of PLE intake would be needed to follow-up study to assess the specificity and sensitivity of salivary markers. Nonetheless, protein candidate signature(s) from the present study could serve as a starting point for the future quantitative mass spectrometry-based peptidomic analyses, such as multiple- or parallel- reaction monitoring (MRM or PRM) (Shi et al., 2016), in order to accomplish longitudinal data for PLE-treated subjects.
. Postprandial glucose obtained prior to and after treatment with PLE or placebo.
PLE group (n = 34) | Placebo group (n = 34) | ||||||
---|---|---|---|---|---|---|---|
0 w | 8 w | 0 w | 8 w | ||||
PPG120 min (mg/dl) | 171.36 ± 38.38 | 160.91 ± 35.62 | 0.133 | 164.73 ± 32.95 | 172.03 ± 33.61 | 0.126 | 0.029* |
Values are presented as mean ± S.D..
2)Linear mixed model,
*
. Demographic characteristics of the study participants.
Subjects (n = 5) | |
---|---|
Age (year) | 58.00 ± 8.43 |
Sex (M/F) | 2/3 |
Height (cm) | 158.60 ± 6.02 |
Weight (kg) | 62.80 ± 7.12 |
BMI (kg/m2) | 24.90 ± 1.17 |
FPG (mg/dl) | 103.80 ± 12.17 |
2 h-PPG (mg/dl) | 158.60 ± 14.22 |
. Proteins showed at least 2-fold changes among study subjects by the PLE treatment..
Protein Name | Patient#1 | Patient#2 | Patient#3 | Patient#4 | Patient#5 |
---|---|---|---|---|---|
Saliva proteomics analyses | |||||
SPARC-like protein 1 | −1.24 | ND | ND | −1.66 | −1.13 |
Ezrin | 1.35 | 1.10 | 1.13 | ND | ND |
Urine proteomics analyses | |||||
LAIR-1 | −2.22 | ND | ND | −1.23 | −2.08 |
IGKV4 | −3.19 | ND | −2.20 | −1.11 | ND |
IGKV4-1 | 1.29 | ND | 1.30 | ND | 3.97 |
Complement C7 | −1.43 | ND | −1.45 | ND | −3.81 |
Uromodulin | ND | ND | −1.14 | −1.08 | −1.27 |
The value represents log2-fold changes (ND = not detected).
Hyeong Cheol Park*, Wonkyun Choi, Hee Jin Park,Mi Sun Cheong, Yoon Duck Koo, Gilok Shin, Woo Sik Chung, Woe-Yeon Kim, Min Gab Kim, Ray A. Bressan, Hans J. Bohnert, Sang Yeol Lee, and Dae-Jin Yun*
Mol. Cells 2011; 32(2): 143-151 https://doi.org/10.1007/s10059-011-2297-3Yong Jin Heo, Chanwoong Hwa, Gang-Hee Lee, Jae-Min Park, and Joon-Yong An
Mol. Cells 2021; 44(7): 433-443 https://doi.org/10.14348/molcells.2021.0042Jinyoung Kim, Yu-Mi Lim, and Myung-Shik Lee
Mol. Cells 2018; 41(1): 11-17 https://doi.org/10.14348/molcells.2018.2228