Increased number of circulating exosomes and their microRNA cargos are potential novel biomarkers in alcoholic hepatitis

Confirmed cases of alcoholic hepatitis (n = 14) and healthy individuals (n = 10) were enrolled in the study. The diagnosis of alcoholic hepatitis was established by expert clinicians based on medical history, physical examination and laboratory data according to the guidelines of the American College of Gastroenterology [21]. In cases of uncertain diagnosis, liver biopsy was obtained. Patients were included from different stages of alcoholic hepatitis. Healthy individuals were defined as being free of any systemic and non-systemic diseases based on patients’ history and routine laboratory findings identified by primary care physician. These individuals provided history of “only social” alcohol consumption. Both controls and patients were enrolled consecutively to study to avoid selection bias. The study protocol was approved by Institutional Review Board for the Protection of Human Subjects in research at the University of Massachusetts Medical School (Worcester, MA, USA). Written informed consents were obtained from all subjects.

6 to 8-week-old female C57BL/6 mice were purchased from Jackson Laboratory. The animal study protocol was approved and conducted according to the regulations of the Institutional Animal Care and Use Committee (IACUC) of the University of Massachusetts Medical School (Worcester, MA, USA). In the alcohol experimental group (n = 12), mice received 5% (v/v) ethanol (36% ethanol-derived calories) containing Lieber–DeCarli diet (EtOH) for 4 weeks. The matched pair-fed animals received (n = 12) an isocaloric alcohol-free diet (pair-fed) to ensure equal nutrient and calorie intake between groups. Mice were euthanized at the end of experimental time course and serum was separated from the whole blood and the livers were collected. Serum ALT was measured using a kinetic method (D-Tek LLC).

For histopathological analysis, sections of formalin-fixed livers were stained with hematoxylin–eosin to study for steatosis, necrosis and inflammation. Flowcytometry was performed to study the immune cell population in the liver of the pair fed and ethanol fed mice as previously described [22].

For exosome isolation from sera of mice and plasma of human subjects, exosomes were isolated from 150 µl of sera or plasma following the previously established protocol [17]. Briefly, the serum or plasma were centrifuged at 1,500×g for 5 min to remove cells and 10,000×g for 20 min to remove residual cellular debris. Samples were serially filtered through 0.8, 0.44 μm (Millipore, Billerica, MA, USA). The filtered supernatant was used to precipitate exosomes with Exoquick-TC™ (according to the manufacturer’s guidelines). Exosome pellet was re-suspended in PBS. For isolating smaller subpopulation of exosomes (diameter <200 nm) we added an additional step before re-suspension in PBS and filtered the samples with 0.2 µm filters.

The concentration and size range of circulating EVs in human subjects and mice were identified by a NanoSight NS300 system (NanoSight, Amesbury, UK) supplied with a fast video capture and Nanoparticle Tracking Analysis (NTA) software. Before performing the experiments, the instrument was calibrated with 100 nm polystyrene beads (Thermo Scientific, Fremont, CA, USA). The samples were captured for 60 s at room temperature. NTA software processed the video captures and measured the concentration of the particles (particles/ml) and size distribution (in nanometer). Each specimen was measured three times.

Isolated EVs from plasma and serum were re-suspended in PBS and placed on a formvar-coated copper grid and then allowed to settle for 45 min. Sequential PBS washing of the grid was done by positioning droplets of PBS on the top and applying absorbing paper in between. The samples were fixed by positioning the grid on the top of 2% paraformaldehyde placed on the parafilm for 15 min. After fixation and three washing steps, samples were contrasted by adding drops of 2% uranyl acetate for 15 min. A drop of 0.13% methyl cellulose and 0.4% uranyl acetate was placed on the parafilm grid was incubated at the top for 10 min. The grid was visualized by a Philips CM10 transmission electron microscope and images were recorded using a Gatan CCD digital camera.

After isolating exosomes from mice sera, presence of CD63 and GRP78 were determined with western blot, using our laboratory’s established protocol. Briefly, exosomes were lysed in RIPA buffer and run on a 15% polyacrylamide gel. Proteins were transferred to nitrocellulose membrane overnight then blocked for 3 h in blocking buffer (1× TBST with 5% w/v nonfat dry milk). Primary antibody against CD63 and GRP78 (Abcam, Cambridge, MA, USA) was used overnight at 4°C at 1:2,000 dilution in the blocking buffer. For detection, secondary goat anti-mouse HRP-linked antibody (Santa Cruz Biotechnology, Dallas, TX, USA) was used for 1 h at a dilution rate of 1:5,000 in blocking. The immunoreactive bands were visualized by a Clarity™ Western ECL substrate kit (BioRad) according to the manufacturer’s protocol and LAS-4000IR Ver.2.02 (Fujifilm, Valhalla, NY, USA). Mouse primary hepatocytes were used as negative control.

In addition, the Firefly miRNA multiplex assay (Firefly™ bioworks, Cambridge, MA, USA) was used to profile EV-associated miRNA isolated from mice sera for a targeted list of 68 miRNAs. Total RNA (12 μl) was assayed by the Firefly standard protocol and analyzed with a Millipore Guava 8HT flow cytometer.

We used TaqMan^® miRNA Assays (Applied Biosystems, Catalog number: 4366597) for confirmation of microarray and evaluating the expression level of miRNA-122 (Applied Biosystems, Catalog number: A25576), miRNA-30a (Applied Biosystems, Catalog number: 4427975) and miRNA-192 (Applied Biosystems, Catalog number: A25576). Reverse transcription was performed using a TaqMan stem loop primer, TaqMan primers, 10 ng RNA, and miRNA reverse transcription kit in an Eppendorf Realplex Mastercycler (Eppendorf, Westbury, NY). The cycle was as follows: 30 min, 16°C; 30 min, 42°C; 5 min 85°C. The cDNA was mixed with TaqMan ^® Universal PCR Master and quantitative real-time PCR was performed using a Bio-Rad CFX96 iCycler device. Synthetic C. elegans-miRNA-39 was spiked as external control during the total RNA isolation from human plasma samples. Consistent with the array findings, miRNA-15b was used for normalization of mice sera in confirmatory qPCR analysis. The amplification efficiency of qPCR was verified in our lab and it was very close to 100%. Samples were processed in duplicate and the relative expression level of specific miRNA calculated by 2^−ΔΔCt method. NormFinder was used to identify the suitable endogenous normalizer [23].

Data are expressed as mean ± standard error (SE). For pair-wise comparisons, non-parametric Mann–Whitney test or parametric t-test were used based on underlying distribution. The ability to discriminate between the alcohol fed mice/alcoholic hepatitis patients and control groups was evaluated by the receiver operating characteristic curve, and the area under the curve (AUC) was calculated as a measure of test accuracy. For group-wise comparison, Kruskal–Wallis non parametric test or ANOVA parametric test were used and a p ≤ 0.05 was considered as statistically significant. Correlation of ALT with exosome numbers was identified by Spearman correlation test. For statistical analysis and heat map data visualization, GraphPad Prism v.4.03 (GraphPad Software Inc.) and Firefly™ Analysis Workbench software were used, respectively.

The total number of EVs was significantly increased in mice sera after chronic alcohol feeding compared to pair-fed controls (p < 0.05), measured by NanoSight (Fig. 1a). Most of the EVs had a smaller diameter and were in the size of exosomes subfraction of EVs compared to the larger size microvesicles sub-fraction. Serum ALT levels were significantly increased after 4 weeks of alcohol feeding indicting alcohol-induced liver damage, as seen before in chronically alcohol-fed mice compared to the paired-fed controls (Fig. 1b) [24]. EVs isolated from alcohol-fed mice demonstrated the previously described morphology of EVs on electron micrograph [25], and expressed the exosomal marker CD63 (Fig. 1c, d). Loading of the same amounts of input proteins, exosomes isolated from sera of alcohol-fed mice and pair-fed mice showed similar amount of CD63 while primary hepatocytes showed less expression of CD63. GRP78, the intracellular endoplasmic reticulum marker, was present in primary hepatocytes while it was absent in the exosomes (Fig. 1d). Size distribution of the EVs in mice sera in both alcohol-fed mice and pair-fed mice indicated an abundance of smaller vesicles in the range of 50–150 nm supporting predominance of exosomes (Fig. 1e). Exosomes were isolated based on combination of serial filtration and precipitation reagent (ExoQuick). Quantitative measurement of particles with Nanoparticle Tracking Analysis (NTA) showed no significant difference in the distribution of EVs in crude serum and exosomes after isolation (data not shown). We used a 4-week Lieber–Decarli diet model for alcohol induced liver injury in mice which is an established model resulting in mild alcoholic steatohepatitis [26]. The histological and molecular changes confirmed liver injury, steatosis and inflammation in the Lieber Decarli model compared to pair-fed mice are shown in Additional file 1: Figure S1. Salient features included steatosis, macrophage/KC infiltration and activation, neutrophils activation and increased pro-inflammatory cytokines in the livers of alcohol-fed mice (Additional file 1: Figure S1).

Exosome number (particles/ml) showed a strong and statistically significant correlation with ALT levels (IU/L) (Spearman’s Rho, r = 0.80) (p < 0.05) (Fig. 2a). Interestingly, a great portion of total serum RNA were packed within the serum-exosomes compared to the exosome depleted serum (Fig. 2b).

Next, liver miRNA microarray was performed on exosome-associated miRNA, using The Firefly™ microRNA Assay which includes miRNA signatures of 68 miRNAs associated with liver inflammatory pathways. First to ensure reproducibility and reliability, spiked positive control values were compared with historical value identified for the same amount of miRNAs. The spiked positive controls showed excellent correlation with historical values (p = 0.98) (Fig. 2c). This was a quality control step to evaluate the ability of the assay to distinguish different value of spiked positive control. Our results indicated that assay readouts were sensitive to the amount of miRNA and increase linearly when the amount of spiked in control increased. Of the 68 miRNAs indicated in the inflammatory panel, 63 miRNAs were detectable in our samples (Fig. 2d).

Lack of suitable endogenous control normalizer has been one of the main limitations of using miRNA and EVs associated miRNA diagnostics. Regarding the suitable endogenous control normalizer, miRNA-15b, miRNA-20a and miRNA-15b expression levels showed the lowest intragroup and intergroup variability, with the stability value of 0.189 and 0.232 respectively, according to NormFinder algorithm (Table 1). In RT-qPCR analysis, miRNA-15b consistently showed minimal variability in expression levels among different group of experiments.

Exosomes were isolated from the sera of 12 chronic alcohol-fed mice and 12 pair-fed mice. To obtain enough exosome-associated total RNA for running microarrays and subsequent qPCR validation, we pooled two samples randomly for a total of six samples in each control and alcohol-fed group. In two samples from the alcohol-fed pair-fed groups, respectively, we isolated the smaller subpopulation of exosomes with the diameter of less than 200 nm, to compare the differential signature between the whole population of exosomes and smaller subpopulation of exosomes. miRNA analysis of microarray data showed no significant difference in miRNA signature between total exosomes and smaller subpopulation of exosomes (with diameter less than 200 nm) in mice sera (Fig. 3). This is in accordance with our previous findings regarding the presence of homogenous exosomes in mice sera [27].

The heat map and the list of deregulated miRNAs in EVs isolated from alcohol-fed and pair-fed mice are shown in Fig. 3 and Table 2 respectively. Confirmatory qPCR using miRNA-15b as endogenous normalizer and identical amount of exosomal RNA input, showed that miRNA-192, miRNA-122, and miRNA-30a levels in EVs isolated from sera of alcohol-fed mice were significantly increased in alcohol-fed mice compared to the control pair-fed mice (Fig. 4a). In addition, miRNA-744 (1.49-fold), miRNA-1246 (2.40-fold), miRNA-30b (1.50-fold) demonstrated a significant increases in alcohol-fed mice compared to the pair-fed mice (p < 0.05).

Consistent with the highly significant increase of miRNA-122, miRNA-30a, and miRNA-192 (Fig. 4a–c), those miRNAs showed promising diagnostic values. To test the diagnostic value of the identified miRNAs, we performed Receiver–Operator Characteristic (ROC) curves analysis, and we found that miR-192 has the highest area under the curve (AUC = 0.96; p < 0.001). miRNA-122 and miRNA-30a showed an AUC of 0.92 (p < 0.001) and 0.85 (p < 0.05), respectively (Table 3; Fig. 4d–f).

We evaluated the number and profile of EVs in patients’ plasma (n = 14) and found a significant increase in the number of EVs in the plasma of patients with alcoholic hepatitis, compared to healthy controls (Fig. 5a). The mean age for patients with alcoholic hepatitis and control subjects were 46.28 (SD 6.43) and 28.9 (SD 3.03), respectively. The baseline characteristics of patients are shown in the Additional file 1: Table S1. Consistent with the animal study, most of the circulating EVs in plasma of alcoholic hepatitis patients were in the range of exosomes (less than 150 nm) (Fig. 5a). The level of miRNA-30a and miRNA-192 showed elevated levels in exosomes in patients with alcoholic hepatitis compared to healthy controls (Fig. 5b, c). miRNA-122 showed elevated levels in exosomes isolated from patients with alcoholic hepatitis, but it did not reach statistical significance (Fig. 5d). Based on ROC curve analysis, exosome-associated miRNA-192 can act as an accurate diagnostic test for discrimination of liver disease in the population (AUC = 0.95; p < 0.001) (Fig. 5e, f; Table 3). Using target prediction algorithms, we identified the predicted target protein networks of human miRNA-192 and miRNA-30a (Fig. 6a, b). The first thirty ranked target genes of miRNA-192 and miRNA-30a, as well as gene description are shown in Additional file 1: Tables 2 and 3.