Serum Exosomal miRNAs for Grading Hepatic Fibrosis Due to Schistosomiasis

Chronic infection with Schistosoma japonicum or Schistosoma mansoni results in hepatic fibrosis of the human host. The staging of fibrosis is crucial for prognosis and to determine the need for treatment of patients with schistosomiasis. This study aimed to determine whether there is a correlation between the levels of serum exosomal micro-ribonucleic acids (miRNAs) (exomiRs) and fibrosis progression in schistosomiasis. Reference gene (RG) validation was initially carried out for the analysis of serum exomiRs expression in staging liver fibrosis caused by schistosome infection. The expression levels of liver fibrosis-associated exomiRs in serum were determined in a murine schistosomiasis model and in a cohort of Filipino schistosomiasis japonica patients (n = 104) with different liver fibrosis grades. Of twelve RG candidates validated, miR-103a-3p and miR-425-5p were determined to be the most stable genes in the murine schistosomiasis model and subjects from the schistosomiasis-endemic area, respectively. The temporal expression profiles of nine fibrosis-associated serum exomiRs, as well as their correlations with the liver pathologies, were determined in C57BL/6 mice during S. japonicum infection. The serum levels of three exomiRs (miR-92a-3p, miR-146a-5p and miR-532-5p) were able to distinguish subjects with fibrosis grades I-III from those with no fibrosis, but only the serum level of exosomal miR-146a-5p showed potential for distinguishing patients with mild (grades 0–I) versus severe fibrosis (grades II–III). The current data imply that serum exomiRs can be a supplementary tool for grading liver fibrosis in hepatosplenic schistosomiasis with moderate accuracy.


Correlations of the Serum Levels of Eleven exomiRs with the Grades of Hepatic Fibrosis in a Cohort of Subjects from a Schistosomiasis Endemic Area
Using the most stable exomiR (miR-425-5p) as reference, we analyzed the correlations of the serum levels of the other eleven RGs with hepatic fibrosis grades. A positive correlation was confirmed between the serum levels of the two exomiRs (miR-103a-3p and miR-181a-5p) with the grades of liver fibrosis within the 42 clinical samples tested for the identification of RGs ( Figure 5).

Correlations of the Serum Levels of Eleven exomiRs with the Grades of Hepatic Fibrosis in a Cohort of Subjects from a Schistosomiasis Endemic Area
Using the most stable exomiR (miR-425-5p) as reference, we analyzed the correlations of the serum levels of the other eleven RGs with hepatic fibrosis grades. A positive correlation was confirmed between the serum levels of the two exomiRs (miR-103a-3p and miR-181a-5p) with the grades of liver fibrosis within the 42 clinical samples tested for the identification of RGs ( Figure 5). These two exomiRs along with the initially selected nine exomiRs and were further validated with a cohort of schistosomiasis japonica patients (n = 104) recruited from an endemic area of the Philippines ( Table 4). The potential associations between the levels of the nine exomiR signatures in serum and the grades of liver fibrosis were then assessed. The serum levels of two exomiRs miR-146a-5p and miR-532-5p were inversely correlated with liver fibrosis grades (r = −0.2524 and −0.2365, These two exomiRs along with the initially selected nine exomiRs and were further validated with a cohort of schistosomiasis japonica patients (n = 104) recruited from an endemic area of the Philippines ( Table 4). The potential associations between the levels of the nine exomiR signatures in serum and the grades of liver fibrosis were then assessed. The serum levels of two exomiRs miR-146a-5p and miR-532-5p were inversely correlated with liver fibrosis grades (r = −0.2524 and −0.2365, and p = 0.0097 and 0.0156, respectively), whereas the others did not significantly correlate with hepatic fibrosis grading ( Figure 6).

Serum-Derived exomiRs Can Discriminate Different Liver Fibrosis Grades
The levels of three exomiRs, miR-92a-3p, miR-146a-5p, and miR-532-5p, in serum exhibited a significant difference between subjects with fibrosis grades I-III compared with those without fibrosis (p = 0.0311, 0.0325, and 0.0020, respectively) (Figure 7a), while the levels of the other eight exomiRs did not show any significant difference between the two groups. The receiver operating characteristic

Serum-Derived exomiRs Can Discriminate Different Liver Fibrosis Grades
The levels of three exomiRs, miR-92a-3p, miR-146a-5p, and miR-532-5p, in serum exhibited a significant difference between subjects with fibrosis grades I-III compared with those without fibrosis (p = 0.0311, 0.0325, and 0.0020, respectively) (Figure 7a), while the levels of the other eight exomiRs did not show any significant difference between the two groups. The receiver operating characteristic (ROC) analysis for discriminating liver fibrosis grades I-III from grade 0 showed that the area under the curve (AUC) levels of the three exomiRs, miR-92a-3p, miR-146a-5p, and miR-532-5p, were 0.6326, 0.6315, and 0.6884, respectively (p = 0.0315, 0.0329, and 0.0023, respectively) (Figure 7b). The best diagnostic performance was obtained with a combination of exosomal miR-146a-5p and miR-532-5p, which showed an AUC value of 0.6962 (p = 0.0015) (Figure 7b). For discriminating liver fibrosis grades II-III from 0-I, we observed the serum level of exosomal miR-146a-5p or a combination of exosomal miR-146a-5p and miR-532-5p was significantly higher in individuals with mild fibrosis (grades 0-I) compared with those with severe disease (grades II-III) (p = 0.0227 and 0.0364, respectively) (Figure 8a), with an AUC of 0.6305 and 0.6200, respectively (p = 0.0231 and 0.0367, respectively) (Figure 8b). No exomiRs tested were able to discriminate subjects with liver fibrosis grades 0-II from those with grade III (data not shown).  (grades I-III) or without (grade 0) liver fibrosis. Boxes represent the interquartile range of the data with lines across the boxes indicating the median values. The hash marks above and below the boxes indicate the 90th and 10th percentiles, respectively. p values were calculated using a Mann-Whitney U test (* = p < 0.05, ** = p < 0.01). (b) The area under the curve (AUC) values were calculated using receiver operating characteristic (ROC) analysis to access capabilities of individual exomiR candidates or in combination in serum for discriminating subjects with stage I-III to those without liver fibrosis (grade 0). The AUC values were calculated using ROC analysis to access capabilities of exomiR candidates individually or in combination in serum for discriminating subjects with severe fibrosis grades (II-III) from those without significant fibrosis (grades 0-I). The AUC values were calculated using ROC analysis to access capabilities of exomiR candidates individually or in combination in serum for discriminating subjects with severe fibrosis grades (II-III) from those without significant fibrosis (grades 0-I).

Discussion
Liver fibrosis in schistosomiasis occurs during the development of a series of complex hepatopathologies, involving immune inflammation, granuloma formation, and liver damage [4], a scenario that may be more complicated than the etiology of other chronic liver diseases, such as those due to viral infections, alcoholic hepatitis, or non-alcoholic steatohepatitis. The current study aimed to determine whether the levels of exosome-derived miRNAs in serum are able to grade the intensities of liver fibrosis due to schistosomiasis. Identification of suitable RGs is a prerequisite for the quantification of miRNAs using qRT-PCR [27,28]. In this study, different exomiRs were identified as the most stable genes in a murine schistosomiasis model and within clinical samples. Across the infection course in C57BL/6 mice, exosomal miR-103a-3p was identified as the as the most stable RG, while exosomal miR-425-5p was determined as the most stable RG in patients infected with S. japonicum. Previous work recommended miR-103a-3p as a suitable RG for analysis of plasma miRNAs in a murine model with acetaminophen-induced hepatotoxicity [29]. Furthermore, miR-103a-3p was determined as the second stably expressed reference gene by the geNorm and NormFinder algorithms for analyzing serum exosomal miRNA expression in patients with hepatitis B or hepatocellular carcinoma [30]. On the other hand, miR-425-5p was identified as the most stable expressed miRNA in plasma of patients with vulvar intraepithelial neoplasia lesions and vulvar carcinoma [31] and in human plasma exosomes [21]. In the murine schistosomiasis model, the experimental time frame employed (4-11 weeks p.i.) represents an acute toxemic phase, which is characterized by complicated hepatic pathologies, including necrosis, inflammation and granuloma formation within a short period [32]. We noted in this study that the levels of several miRNA signatures in the circulatory system were not only associated with the severity of liver fibrosis but also with the extent of liver damage and the degree of granulomatous involvement (Table 3), features also evident in a previous study we undertook by directly measuring serum miRNAs in two murine models of schistosomiasis [9]. In contrast, schistosome infection in the human host tends to be chronic [9]. Thus, the animal model does not precisely represent the pathological status of human schistosomiasis, a feature that further highlights the importance of identification of suitable normalizers for the analysis of miRNA expression in extracellular vesicles in different disease models/conditions. Comparative miRNAome analysis revealed different miRNA expression profiles in bovine sera and serum-derived exosomes [33]. Although it has been suggested that RNA packaging into extracellular vesicles/exosomes is a selective process, the precise loading mechanism for this remains elusive [34,35]. It is not surprising to observe the inconsistent performance of specific miRNA signatures derived from serum and serum exosomes as biomarkers for a particular disease. For example, Lambrecht et al. reported that levels of vesicle-associated miRNA-192, -200b-3p, -92a-3p, and -150-5p in plasma showed diagnostic power in discriminating early stage liver fibrosis, but this was not the case when these markers were directly measured in plasma [8]. In contrast, Matsuura et al. [7] showed that the plasma levels of let-7 family members significantly declined over the progression of fibrosis in patients with chronic hepatitis C, with an AUC value between 0.734 and 0.790; however, the detection power was reduced when targeting the exosome-derived let-7 family members. Similarly, we showed previously that the serum levels of miR-150-5p, let-7a-5p, and let-7d-5p in patients were able to discriminate mild from severe hepatic fibrosis caused by schistosome infection, with an AUC value of 0.6838, 0.6598, and 0.6270, respectively [9]. And, here, we found that the levels of these three miRNAs in serum exosomes were not capable of discriminating liver fibrosis grades 0 versus I-III, nor grades 0-1 versus II-III or grades 0-II versus III in the human subjects tested. Nevertheless, when compared with our previous study [9], similar trends were evident in the expression levels of several miRNAs (i.e., miR-150-5p, miR-192-5p, miR-200b-3p, let-7a-5p, let-7d-5p, and miR-146a-5p) in serum or in serum exosomes in C57BL/6 mice during S. japonicum infection.
It has been recently documented that miR-146a-5p participates in the biogenesis of fibrosis in a variety of tissues [36][37][38][39][40]. In the pathogenesis of nonalcoholic fibrosing steatohepatitis, it has been shown that miR-146a-5p was significantly down-regulated in activated hepatic stellate cells (HSCs) and overexpression of miR-146a-5p suppressed HSC activation, as well as extracellular matrix deposition [41]. During the development of liver fibrosis in a murine model of schistosomiasis, He et al. found that by targeting Signal Transducer And Activator Of Transcription 1 (STAT1), miR-146 suppressed the interferon (IFN)-γ-induced differentiation of macrophages to M1 cells [38]. In relation to the application of extracellular miR-146a-5p in staging liver fibrosis, Appourchaux et al. found that the serum level of miR-146a-5p was higher in chronic hepatitis C patients with advanced fibrosis and cirrhosis (F3-F4) than those with mild and moderate fibrosis (F1-F2); in contrast, no difference in the serum level of miR-146a-5p was observed between chronic hepatitis B patients with F3-F4 cirrhosis and those with F1-F2 [42]. In the current study, consistent with our previous study of direct measurement of serum miR-146a-5p [9], down-regulation of exosomal miR-146a-5p in the sera of C57BL/6 mice was observed over the course of S. japonicum infection, although no direct correlation between the serum level of exosomal miR-146a-5p and HP or HF was found. Furthermore, a negative correlation between the serum level of this miRNA and the grades of liver fibrosis was observed in the human cohort we investigated, which enables the discrimination of individuals with moderate to severe (grade II-III) liver fibrosis from those with mild (grade 0-I) disease. Nevertheless, there is still a concern that using a single miRNA/exomiR marker in the circulatory system for predicting a particular disease may incur a potential problem of non-specificity. For example, it has been shown recently that the level of serum exosomal miR-146a-5p declined significantly in systemic lupus erythematosus patients compared with healthy controls [43]. Accordingly, the use of a combination of multiple miRNA/exomiR signatures may improve diagnostic specificity.
Recent reports indicate that miR-532 family members are dysregulated in various tumors and are associated with carcinogenesis, thus having been considered as potential novel therapeutic interventions for the treatment of specific tumors [44][45][46]. Furthermore, Chen et al. found that serum EV miR-532-5p was significantly higher in control subjects versus F3/F4 fibrosis patients, using U6 as a reference gene [24]. Here, we showed that exosomal miR-532-5p inversely correlated with liver fibrosis grades in the human schistosomiasis cohort and was able to discriminate hepatic fibrosis I-III from those without fibrosis (AUC: 0.6884). However, in C57BL/6 during S. japonicum infection, a strong positive correlation was observed between the levels of serum exosomal miR-532-5p with the grades of liver fibrosis, which further echoed the different pathological severities in the murine schistosomiasis model, which usually represents a high dosage infection, compared with schistosome-infected patients.
Moderate diagnostic values were obtained in the grading of schistosomal liver fibrosis by detection of an individual exomiR or in combination. For discriminating individuals without liver fibrosis from those with grades I-III, the best diagnostic performance was obtained with the combination of serum exosomal miR-146a-5p and miR-532-5p (AUC: 0.6962, p = 0.0015) (Figure 7b). For discriminating mild from severe liver fibrosis, the diagnostic performance of exosomal miR-146a-5p in serum was not superior but comparable with that of miRNAs directly isolated from serum, i.e., AUCs of 0.6575 and 0.6305 were obtained for serum miR-146a-5p and serum exosomal miR-146a-5p, respectively. The following factors may contribute to this phenomenon: (i) It has been shown that re-infection with S. japonicum and co-parasitism with intestinal helminths and protozoa were common in the cohort subjects investigated [47,48], a feature which potentially might alter the host immune responses and further dysregulate the serum levels of the targeted exomiRs; (ii) as the study cohort individuals were from a rural area with rudimentary water sources and poor sanitation and hygiene [47], acute respiratory infections, diarrheal diseases, and other communicable diseases are likely to have had a considerable influence on the expression of the exomiRs investigated; (iii) only a limited number of exomiRs were tested in the current study, and it would be informative to identify more potential exomiRs specific for schistosomiasis fibrosis by profiling; (iv) importantly, not only the liver, but also other organs, i.e., the intestine, spleen and other ectopic sites (such as kidney and brain), are likely affected during hepatosplenic schistosomiasis [49]. Since serum exosomes can be released from all affected organs, the pathological alternation in multiple organs further likely changes the levels of exomiRs tested. Thus, assessment of miRNAs from sub-population exosomes, such as liver-derived [50] or, more specifically, HSCs-derived exosomes [51], may represent a promising pipeline to improve diagnostic power, as has been suggested for improving the performance of tumor diagnosis by tagging tumor-derived exomiRs in the blood circulation [52]. In this regard, membrane markers for liver-specific or HSC-specific exosomes need to be identified prior to applying this approach in the discovery of biomarkers for liver fibrosis or other liver diseases [53,54].

Animal Infection, Serum Collection, Histological Assessment, and Biochemical Analyses
Eight-week-old female C57BL/6 mice were divided into 6 groups and were percutaneously infected with 16 ± 2 of S. japonicum cercariae. Subsequently, mice were sacrificed at 0 (uninfected), 4, 6, 7, 9, and 11 weeks post-infection, and~1 mL blood was collected from each mouse at each time point by cardiac puncture. The median lobe was collected from each mouse liver and subjected to histological assessment after formalin fixation [32]. Slides were scanned using the Aperio Slide Scanner (Aperio Technologies, Vista, CA, USA). Liver hydroxyproline (HP) content was assessed by a hydroxyproline colorimetric assay kit (Bioo Scientific, Austin, TX, USA). Serum alanine transaminase (ALT) and aspartate transaminase (AST) levels were tested using ALT and AST color endpoint assay kits (Bioo Scientific), respectively.

Study Cohort
The human cohort was recruited from 18 schistosomiasis-endemic but non-malaria-endemic barangays in the municipalities of Laoang and Palapag, Northern Samar, the Philippines in 2012. For more than over the last three decades, the area has had active schistosomiasis control programs, including a mass drug administration (MDA) program which commenced in 2008. Free annual treatment (Praziquantel 40 mg/kg in a single dose) is delivered to all individuals aged between 5-65 years in accordance with the Department of Health Administrative Order 2007-0015, but the compliance rate varies [55]. It is worthwhile to note that the S. japonicum reinfection rate in the area is also high [56]. Detailed information of this study cohort is described elsewhere as part of a hepatic morbidity study [25,47]. In brief, Kato-Katz thick smear stool examination was performed on each individual in the cohort using two stool specimens provided over the course of a week to determine infection status and intensity. Six Kato-Katz thick smears were prepared on microscope slides using standard 50 mg templates according to established protocols and examined under a light microscope by experienced technicians. All subjects in the cohort were assessed for liver fibrosis severity by ultrasound scan using a portable gray-scale ultrasonogram (SONOACE X1; Madison Co., Ltd., Seoul, South Korea) [25]. The parenchymal pattern of hepatic fibrosis was determined according to practical guidelines recommended by the WHO for ultrasonographic examination for schistosomiasis-related morbidity [57]. In the current study, 104 individuals (all negative for hepatitis B virus/hepatitis C virus infection) from the cohort were enrolled. For serum collection,~10 mL blood was drawn from each subject enrolled; serum was subsequently separated by centrifugation, and stored at 2-8 • C. The serum samples were then transported to the Research Institute for Tropical Medicine (RITM), Manila, at a temperature of 4 • C and stored at −80 • C. Subsequently, sample aliquots were transported on dry ice to QIMRB, Australia.

Exosome Isolation, RNA Extraction, Polyadenylation, and Reverse Transcription (RT)
Serum exosomes were isolated using ExoQuick reagent (SBI System Biosciences, Palo Alto, CA, USA), according to the manufacturer's protocol. Briefly, one hundred microliters of mouse or human serum was mixed with 31 µL of ExoQuick exosome precipitation solution. After incubation at 4 • C for 30min, the samples were then centrifuged at 1500× g for 30min. The supernatant was removed, and the exosome-rich pellet was re-suspended by adding 100 µL PBS. Subsequently, total RNA was extracted from the re-suspended samples using a Qiagen miRNeasy Mini Kit (Qiagen, Hilden, Germany), following the manufacturer's instruction. RNA concentration was quantified using a Qubit MicroRNA Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Polyadenylation and RT reactions were performed using the S-Poly(T) method in a one-step procedure by the combined use of two commercial kits: the Poly(A) Polymerase Tailing Kit (Lucigen, Middleton, WI, USA) and the TaqMan MicroRNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, MA, USA) [32,58]. One ng RNA (5 µL) was used in each RT reaction, which was carried out under the following condition: 37 • C for 30 min, 42 • C for 30 min, and 85 • C for 5 min. RT products were stored at −20 • C prior to subsequent analysis. The RT primers used in this study are listed in Supplementary Tables S1-S3.

QRT-PCR for miRNA Quantification
Quantification of miRNAs was performed by qRT-PCR with the Applied Biosystems Quantstudio 5 Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA) according essentially to protocols described previously [32]. The assays were performed with the following cycling conditions: 50 • C, 2 min, 95 • C 10 min, followed by 40 cycles: 95 • C for 15 s, and 60 • C for 1 min. The expression levels of miRNAs were calculated by the 2 −∆∆Ct method with normalization to the reference genes identified in this study for analysis [32].

Reference Genes Identification
To identify RGs used for the analysis of exomiR expression in serum for predicting the severity of liver fibrosis, we evaluated two different data sets: (i) the temporal expression data of candidate RGs in a murine model during S. japonicum infection; (ii) the expression data of candidate RGs in serum samples of 42 subjects from a schistosomiasis-endemic area with different liver fibrosis grades. The relative expression stability of the candidates was analyzed using four computational programs (NormFinder [59], GeNorm [60], BestKeeper [61], and the comparative Delta Ct method [62]), which are collectively available at a user-friendly web-based RefFinder tool [63] (https://www.heartcure. com.au/reffinder). A recommended comprehensive ranking of candidate RGs was then generated by RefFinder based on the results of the four algorithms.

Statistical Analysis
For analysis of the temporal levels of serum exomiRs in C57BL/6 mice during S. japonicum infection, one-way ANOVA was used followed by the Holm-Sidak multiple comparison. Pearson's correlation coefficient (r) was used for the determination of the correlations between continuous variables, while Spearman's rank correlation coefficient (rho) was employed for the assessment of the correlations between continuous and categorical variables. The Mann-Whitney U-test was used for analysis of the ability of the levels of serum exomiRs in discriminating different schistosomiasis liver fibrosis grades. Receiver operating characteristic (ROC) curve analyses were performed, and the area under the curve (AUC) was calculated to evaluate the potency of using the serum exomiRs as biomarkers for discriminating hepatic fibrosis caused by schistosomiasis. A p-value < 0.05 was considered statistically significant. Statistical analysis was performed with GraphPad Prism v. 6.00 for Windows.

Conclusions
In summary, optimal reference genes were determined for studying exomiRs in a schistosomiasis murine model and a clinical cohort from a schistosomiasis endemic area, which may be of value for future research focused on serum/plasma exomiRs in fibrosis. We also identified serum exomiRs that are capable of grading hepatic fibrosis due to schistosomiasis japonica. The diagnostic performance of these exomiRs is moderate and, consequently, additional optimization steps, such as evaluating miRNAs in liver-derived or HSCs-derived exosomes [50], will be required to improve the overall diagnostic power using this approach.
Supplementary Materials: Supplementary materials can be found at http://www.mdpi.com/1422-0067/21/10/ 3560/s1. Table S1: Primers and probe used for optimal reference gene identification; Table S2: Primers and probe used for quantification of serum exosomal miRNA candidates for grading the intensities of liver fibrosis in a murine schistosomiasis model; Table S3: Primers and probe used for quantification of serum exosomal miRNA candidates for grading liver fibrosis in a human cohort from a schistosomiasis endemic area.