Hepatocyte-derived microRNAs as serum biomarkers of hepatic injury and rejection after liver transplantation

Abstract

Recent animal and human studies have highlighted the potential of hepatocyte-derived microRNAs (HDmiRs) in serum as early, stable, sensitive, and specific biomarkers of liver injury. Their usefulness in human liver transplantation, however, has not been addressed. The aim of this study was to investigate serum HDmiRs as markers of hepatic injury and rejection in liver transplantation. Serum samples from healthy controls and liver transplant recipients (n = 107) and peritransplant liver allograft biopsy samples (n = 45) were analyzed via the real-time polymerase chain reaction quantification of HDmiRs (miR-122, miR-148a, and miR-194). The expression of miR-122 and miR-148a in liver tissue was significantly reduced with prolonged graft warm ischemia times. Conversely, the serum levels of these HDmiRs were elevated in patients with liver injury and positively correlated with aminotransferase levels. HDmiRs appear to be very sensitive because patients with normal aminotransferase values (<50 IU/L) had 6- to 17-fold higher HDmiR levels in comparison with healthy controls (P < 0.005). During an episode of acute rejection, serum HDmiRs were elevated up to 20-fold, and their levels appeared to rise earlier than aminotransferase levels. HDmiRs in serum were stable during repeated freezing and thawing. In conclusion, this study shows that liver injury is associated with the release of HDmiRs into the circulation. HDmiRs are promising candidates as early, stable, and sensitive biomarkers of rejection and hepatic injury after liver transplantation. Liver Transpl 18:290–297, 2012. © 2012 AASLD.

MicroRNAs (miRNAs), a class of small noncoding RNAs, are important regulators of gene expression, and they control many cellular processes through the posttranscriptional suppression of gene expression.1, 2 Altered tissue expression levels of miRNAs have lately been linked to various pathological conditions in humans, including malignant, infectious, metabolic, autoimmune, and cardiovascular diseases.3-9 These findings have led to increased interest in miRNAs as potential diagnostic markers and as targets for therapeutic interventions.

Hepatocytes express a distinct set of miRNAs, and miR-122 is the most abundant.10 miR-122 has been found to be an important regulator of cholesterol metabolism11 and iron homeostasis12 and a crucial host factor for hepatitis C virus infection and replication.13, 14 In addition to these important cellular functions, recent studies in rodents have demonstrated that miR-122 and other hepatocyte-abundant miRNAs are released from cells during drug-induced liver injury.15, 16 These hepatocyte-derived microRNA (HDmiRs) were detected in serum or plasma, and their levels increased with the drug dose and the duration of the drug exposure. HDmiRs were found to correlate with serum aminotransferases [aspartate aminotransferase (AST) and alanine aminotransferase (ALT)] and liver histology. Importantly, the rise in the serum miRNA levels of these animals appeared earlier than the rise in the aminotransferase levels. In addition to the diagnostic potential of miRNAs, experimental animal studies have shown that miRNAs are feasible targets for therapeutic interventions designed to minimize and even reverse severe tissue injury caused by ischemic insults.17 In humans, it has recently been shown that the HDmiR miR-122 can be detected in serum, and its level was found to be elevated in patients with hepatocyte injuries caused by viral, alcoholic, or chemical-related hepatotoxicity.18, 19 In these patients, the serum and plasma miR-122 levels also showed a close correlation with aminotransferases and liver histology. However, this has not been evaluated in the setting of liver transplantation.

Liver transplantation has developed from a risky, experimental procedure to a lifesaving and effective treatment of end-stage liver failure. However, despite this success, transplant recipients can suffer from serious side effects of long-term immunosuppression and remain at risk for de novo malignancies,20 and they can lose their allografts because of rejection, recurrent disease, or biliary complications.21, 22 The potential benefits from tapering immunosuppressive medications in patients to reduce toxicity are countered by the potential risk of losing the graft to immune-mediated rejection. Therefore, there is an urgent need for better biomarkers that can provide earlier and more sensitive signs of rejection or liver graft dysfunction in a noninvasive fashion. Because of their cell type–specific distribution, biological stability, and detection sensitivity, HDmiRs could be promising candidates for this. Indeed, several recent studies in the setting of kidney transplantation have highlighted the potential of messenger RNA and miRNA as biomarkers for assessing the renal allograft status.23-26 Current protein-based markers for liver injury, AST and ALT, are also expressed outside the liver in muscle tissue, and they can cause false elevations during muscle injury.27 Therefore, an assessment of the liver allograft status often still requires tissue biopsy for more definite proof of hepatic injury. Particularly after liver transplantation, trough-cut biopsy is a relatively perilous procedure associated with pain, bleeding, and infections.28-31 Alternatively, more sensitive, specific, and noninvasive methods for monitoring graft injury are needed to minimize the need for liver biopsy and to allow safer weaning from immunosuppressive medication.

The aim of the current study was to investigate the utility of serum HDmiRs as markers of hepatic injury and acute rejection after liver transplantation. We found that the expression of miR-122 and miR-148a in liver tissue was significantly diminished with prolonged graft warm ischemia times and, conversely, was elevated in serum during ischemia/reperfusion injury and acute rejection. HDmiRs were found to be promising candidates as biomarkers for assessing the allograft status after liver transplantation.

Abbreviations:

ALT, alanine aminotransferase; AST, aspartate aminotransferase; cDNA, complementary DNA; HDmiR, hepatocyte-derived microRNA; miRNA, microRNA; PCR, polymerase chain reaction; RT-PCR, real-time polymerase chain reaction.

PATIENTS AND METHODS

Patient Samples

All liver transplants were performed at Erasmus MC (Rotterdam, the Netherlands). Liver graft biopsy samples (n = 45) were obtained during transplantation 60 minutes after portal reperfusion and were directly snap-frozen for storage. Serum samples were taken from 12 healthy controls and 43 recipients at different times after liver transplantation; 13 patients with histologically proven acute rejection were included. All blood samples were collected with a standard protocol, and serum samples were processed within 2 hours and were quickly stored at −80°C. Serum samples with signs of red blood cell lysis were not used. Patient demographics and clinical variables were extracted from a prospectively filled database and are summarized in Table 1. The intrinsic stability of HDmiRs in serum was determined by the subjection of 4 individual serum samples from liver transplant recipients to 5 freezing and thawing cycles (−80 and +20°C). The Erasmus MC medical ethics council approved the use of the human samples, and all patients provided informed consent for the use of materials for medical research.

Table 1. Characteristics of the Patients and the Healthy Controls

Characteristic	Healthy Controls	After Transplantation
		Nonrejectors	Rejectors
* Not determined.
Serum samples (n)	12	33	62
Mean AST (IU/L)	—*	369 ± 83	135 ± 37
Mean ALT (IU/L)	—*	386 ± 60	144 ± 29
Subjects/patients (n)	12	30	13
Age (years)	42 ± 3	45 ± 3	39 ± 6
Sex: male/female (n/n)	7/5	18/12	6/7
Underlying disease (n)	—
Viral	—	14	6
Cholestatic	—	5	3
Alcoholic	—	4	2
Other	—	7	2

Serum levels of AST and ALT < 50 IU/L were considered normal. Acute cellular rejection was defined by the presence of all 3 of the following criteria: a transient rise in AST and ALT levels above the upper limit of normal, a rejection activity index of 6 or more (according to the results of consequent needle biopsy during a histological examination), and a decrease in the aminotransferase levels upon treatment with methylprednisolone.32

RNA Isolation

Total RNA was extracted from approximately 10 mg of liver tissue with the miRNeasy mini kit according to the manufacturer's instructions (Qiagen, Hilden, Germany). A modified protocol was used to isolate total RNA from serum. For this, 1.5 mL of the QIAzol lysis reagent was added to 200 μL of serum, and extensively mixed by vortexing. Chloroform (300 μL) was added, and after centrifugation (15 minutes at a relative centrifugal force of 16.000), 800 μL of an aqueous RNA-containing layer was obtained; this was further processed according to the manufacturer's protocol (Qiagen). RNA that was extracted from liver tissue was quantified with a NanoDrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA) and normalized to a concentration of 50 ng/7.5 μL. RNA that was extracted from serum could not be quantified because of its low concentration and was normalized only for the initial serum input.

Reverse Transcription and Real-Time Polymerase Chain Reaction (RT-PCR)

The TaqMan miRNA reverse-transcription kit (Applied Biosystems, Carlsbad, CA) was used to prepare complementary DNA (cDNA) for multiple miRNAs in 1 reaction with a modified protocol. Every multiplex cDNA reaction consisted of 0.4 μL of a deoxyribonucleotide triphosphate mix, 1.35 μL of the MultiScribe reverse-transcription enzyme, 2.0 μL of a 10× reverse-transcription buffer, 0.25 μL of a ribonuclease inhibitor, 1.0 μL of each reverse-transcription primer, and 7.5 μL of diluted template RNA. The total reaction volume was adjusted to 20 μL with nuclease-free water. On the basis of the literature, 15 miRNAs were initially tested: miR-30a, miR-30c, miR-30e, miR122, miR-133a, miR-148a, miR-191, miR-192, miR-194, miR-198, miR-200c, miR-222, miR-296, miR-710, and miR-711.15, 33-36 Three highly expressed hepatocyte-rich miRNAs—miR-122, miR-148a, and miR-194—were selected and further used. For the analysis of liver biopsy samples, additional cDNA was prepared for a small nuclear RNA, RNU43, which served as a reference gene for the normalization of the RNA input. For serum samples, 2 additional non–liver-abundant miRNAs, miR-133a (muscle-abundant) and miR-191 (blood-abundant), served as controls. All cDNA reactions were performed according to the manufacturer's instructions. Polymerase chain reaction (PCR) reactions were carried out in duplicate and according to the manufacturer's instructions. Each reaction included 10 μL of the TaqMan universal PCR master mix, 0.5 μL of an miRNA-specific PCR primer (Applied Biosystems), and 5.0 μL of previously diluted (1:10) cDNA. The final volume of every PCR reaction was adjusted to 20 μL with nuclease-free water.

Statistical Analyses

Statistics for correlations were generated with Spearman's rank correlation test. Comparative statistics between groups were tested with the Mann-Whitney U test and the Wilcoxon matched pairs test with GraphPad Prism software (GraphPad Software, Inc., San Diego, CA). P values < 0.05 were considered significant.

RESULTS

Reduced Hepatic miRNA Levels in Liver Grafts with Long Warm Ischemia Times

To investigate changes in intrahepatic miRNA expression in response to ischemia/reperfusion injury, we analyzed 45 biopsy samples that were taken from liver grafts 1 hour after reperfusion. The mean cold ischemia time was 484 ± 25 minutes, and the mean warm ischemia time was 35 ± 2 minutes. As shown in Fig. 1A, there was a significant positive correlation between the levels of hepatocyte-abundant miRNAs. The levels of miR-122 strongly correlated with the levels of miR-148a and miR-194 (r ≥ 0.85, P < 0.001) but were approximately 20 times higher. As shown in Fig. 1B, the levels of miR-122 and miR-148a (but not miR-194) in these liver graft biopsy samples showed a significant reverse correlation with the length of the warm ischemia time (miR-122, r = −0.307, P = 0.038; miR-148a, r = −0.404, P = 0.005). No significant correlation between the miRNA levels and the cold ischemia time was observed (data not shown). These findings suggest that graft injuries associated with longer warm ischemia times reduced the levels of specific hepatocyte-abundant miRNAs, possibly through the release of miRNAs from injured cells.

Figure 1. Decreased levels of hepatic miRNAs in liver grafts with extended warm ischemia times. Liver graft tissue biopsy samples (n = 45) were analyzed for hepatocyte-abundant miRNAs (miR-122, miR-148a, and miR-194) with quantitative RT-PCR. The miRNA levels were normalized to a small nuclear RNA (RNU43), which served as a reference gene. (A) The relative expression levels of miR-122 correlated significantly with the miR-148a and miR-194 levels in the liver grafts (r ≥ 0.85, P < 0.001). The miR-122 levels were approximately 20 times higher than the miR-148a and miR-194 levels. (B) Decreased levels of miR-122 and miR-148a in the liver graft biopsy samples correlated significantly with the length of the warm ischemia time to which the graft had been exposed during liver transplantation (P < 0.05).

Serum HDmiRs are Associated with Peritransplant Ischemic Liver Injury

Serum samples from healthy individuals and liver allograft recipients within 2 weeks of transplantation were analyzed for the presence of HDmiRs. All 3 HDmiRs (miR-122, miR-148a, and miR-194) and both control miRNAs (miR-133a and miR-191) were detectable in serum samples from healthy individuals and patients. As shown in Fig. 2, the levels of HDmiRs were significantly elevated in the patients after liver transplantation versus the healthy controls. In serum samples with high AST and ALT levels (>50 IU/L), the levels of miR-122 were elevated 124- and 102-fold, respectively, in comparison with the average levels of the healthy controls (P < 0.001). In comparison with the healthy controls, the levels of miR-148a and miR-194 were 30- and 40-fold higher, respectively, in the high-aminotransferase groups (P < 0.001). The levels of HDmiRs were significantly higher in the high-AST and high-ALT groups versus the low-AST and ALT-groups (P < 0.005, Fig. 2), except for miR-194 in the high-ALT group, which was elevated only 2-fold (not statistically significant). The levels of the control miRNAs, miR-133a and miR-191, were not significantly different between any of the groups (Fig. 2). The HDmiRs appeared to be sensitive because the patients with normal aminotransferase values had significantly elevated levels of miR-122, miR-148a, and miR-194 in comparison with the healthy controls (11-, 7-, and 9-fold higher in the low-AST group and 8-, 6-, and 17-fold higher in the low-ALT group, P < 0.005). As shown in Fig. 3, a positive correlation was observed between serum HDmiR levels and aminotransferases in the patients. The correlation with AST and ALT resulted in r coefficients of 0.80 and 0.77, respectively, for miR-122, whereas for miR-148a, r was 0.60 for both AST and ALT (P < 0.001). No significant correlations were found for miR-194 (r < 0.30, P > 0.05).

Figure 2. HDmiR levels are elevated in serum during peritransplant ischemic liver injury. The HDmiRs miR-122, miR-148a, and miR-194 were quantified with RT-PCR in 92 serum samples obtained from liver transplant recipients (n = 40) and healthy controls (n = 12). In comparison with the levels in the healthy controls, the miR-122, miR-148a, and miR-194 levels were significantly elevated in the serum samples from the patients with low AST and ALT levels (11, 7, and 9 times and 8, 6, and 17 times, respectively). The levels were further elevated in the serum samples from the patients with aminotransferase levels above the clinical diagnostic threshold of 50 IU/L. In comparison with the low-AST and low-ALT groups, the miR-122, miR-148a, and miR-194 levels were 11, 5, and 5 times higher in the high-AST group and 13, 5, and 2 times higher in the high-ALT group. The levels of the control miRNAs (miR-133a and miR-191) were not significantly elevated in any of the serum samples versus the healthy controls. *P < 0.005.

Figure 3. Serum HDmiR levels in liver transplant recipients correlate with AST and ALT levels. The HDmiRs miR-122 and miR-148a were quantified with RT-PCR in 80 serum samples obtained from liver transplant recipients. The serum levels of miR-122 and miR-148a correlated significantly with the levels of AST and ALT in the same samples.

Additional experiments for testing the stability of HDmiRs in serum showed that the levels of miR-122, miR-148a, and miR-194 in serum were not significantly affected after 5 cycles of freezing (−80°C) and thawing to room temperature (with respect to the untreated baseline values, the mean values were 120% ± 11%, 100% ± 6%, and 99% ± 19%, respectively).

Elevated Serum HDmiRs during Acute Rejection

Serum HDmiRs were analyzed in liver transplant recipients experiencing an episode of acute rejection. As shown in Fig. 4A, the serum HDmiR miR-122 level was significantly elevated during rejection. On average, a 9-fold increase was observed at the time of rejection in comparison with the levels observed 6 months after the resolution of rejection (P < 0.005). For 5 patients, a longitudinal series of serum samples taken before, during, and after acute rejection was analyzed. One representative patient is shown in Fig. 4B. The serum levels of miR-122 and miR-148a showed kinetics similar to those of AST and ALT and increased up to 20-fold during acute rejection. The levels of the control miRNAs, miR-133a and miR-191, did not increase during acute rejection (Fig. 4B). Although the miR-122 levels showed similar kinetics, they appeared to rise and drop 1 or 2 days earlier than the aminotransferase levels (Fig. 4B). As shown in Fig. 4C, a similar trend was observed in the pooled data of 5 patients. When acute rejection was diagnosed and treatment was started (0 hours), miR-122 was already elevated to its maximum level. The levels of miR-122 dropped quickly after the start of an intravenous methylprednisolone treatment, whereas the levels of AST and ALT continued to rise even after the start of the treatment and took longer to normalize.

Figure 4. Changes in serum HDmiR levels during acute rejection. Serum samples from 13 liver transplant recipients experiencing 1 or more episodes of biopsy-proven acute rejection were analyzed. (A) Serum miR-122 levels were significantly elevated during acute rejection (approximately 9-fold) in comparison with the levels in the same recipients 6 months after the rejection was resolved (n = 13, *P < 0.005). (B) A longitudinal series of serum samples that were taken at daily intervals from 5 of these patients was analyzed. Representative results for 1 patient are shown. The miR-122 and miR-148a serum levels increased up to 20-fold during acute rejection (middle panel) and showed kinetics similar to those of AST and ALT (top panel). The peak HDmiR levels appeared to precede the peak aminotransferase levels (indicated by the dashed line) and quickly normalized after the treatment with intravenous methylprednisolone was begun (see the arrow on the x axis). The levels of the control miRNAs (miR-133a and miR-191) did not increase during acute rejection (bottom panel). (C) The serum aminotransferase and miR-122 levels of the 5 patients at the time of their histological diagnosis and the start of their methylprednisolone treatment (0 hours) and up to 96 hours before and after the start of their treatment are shown. miR-122 reached a maximum level at the start of the treatment and quickly decreased after the treatment, whereas the aminotransferase levels still continued to rise 24 hours later.

DISCUSSION

Small noncoding RNAs and particularly miRNAs have emerged as important genetic regulators of cellular processes, including tissue injury and repair responses.17 Recent studies in small animal models as well as humans have demonstrated that HDmiRs are highly stable and sensitive serum biomarkers of liver injury.15, 16, 18, 19 In both humans and rodents, HDmiR levels in serum appear to increase earlier and more rapidly than AST and ALT levels. In particular, miR-122 was significantly elevated even in subjects with aminotransferases below the threshold of 50 IU/L.15, 16, 18, 19 Here we provide evidence that the concept of miRNAs as biomarkers of hepatic injury is also feasible in the setting of liver transplantation. Serum levels of HDmiRs were elevated in patients with liver injury after liver transplantation (Fig. 2) and during acute rejection (Fig. 4). Conversely, hepatic miRNA levels in liver graft biopsy samples exhibited diminished expression with prolonged warm ischemia times (Fig. 1). During acute rejection, serum HDmiRs showed similar kinetics; however, miRNA levels increased and decreased earlier than aminotransferase levels (Fig. 4B and C). As in previous studies,15, 18 miRNAs showed higher sensitivity than aminotransferases, and miRNA stability, which was proposed in earlier studies,6, 9, 37-40 was confirmed.

HDmiRs could provide a solution for the urgent need for better noninvasive biomarkers that could serve as earlier and more sensitive signs of rejection or liver graft dysfunction. Better markers would greatly help with the management of liver transplant recipients and could allow safer reductions of immunosuppressive medications to achieve a better balance between effects (the prevention of graft rejection) and side effects (toxicity, infection, and malignancy). Long-term complications of immunosuppressive drugs, such as nephrotoxicity and de novo cancers, are becoming bigger problems because of the long survival of liver transplant recipients.20 Currently, the potential benefits from tapering immunosuppressive medications in patients are countered by the potential risk of losing the graft to immune-mediated rejection. Serum ALT and AST are often insufficient for the early and definitive diagnosis of acute rejection, and liver biopsy is necessitated. Particularly in the setting of liver transplantation, liver biopsy poses a significant risk for complications such as pain, bleeding, and infections.28-31 The feasibility of the concept of a minimally invasive diagnosis of acute rejection based on the detection of messenger RNA has been demonstrated for kidney transplants.24, 25

Currently, little is known about the mechanism and biology of the release of hepatocyte-abundant miRNAs in response to liver injury. Ideally, an unbiased genome-wide approach would be preferred to study this release, but it is very challenging to perform gene array analysis with serum samples because of the low yields of RNA and the relatively high amounts required. In our initial analyses, we tested 15 different types of hepatocyte- and cholangiocyte-abundant miRNAs and control miRNAs that were selected from other studies.15, 33-36 These included miR-30a, miR-30c, miR-30e, miR122, miR-133a, miR-148a, miR-191, miR-192, miR-194, miR-198, miR-200c, miR-222, miR-296, miR-710, and miR-711, but only 3 HDmiRs were found to be significantly elevated during acute rejection. It is likely that many other miRNAs expressed in hepatocytes and other liver cells are released during hepatic injury, but only the most abundant and liver-specific miRNAs will be detectable in serum. Nevertheless, the hepatocyte-abundant miRNA miR-194, whose expression levels in liver tissue significantly correlated with miR-122 levels (Fig. 1A), did not correlate with aminotransferase levels (data not shown). This suggests that there may be sequence specificity or selectivity for the release of miRNAs rather than just a general leakage of all miRNAs from the injured cell. This hypothesis is supported by the observation that cellular miRNAs can be released from cells by the secretion of microvesicles, including exosomes, and only distinct sets of miRNAs are selectively packaged into microvesicles.40, 41

This specificity in release and the distinct repertoires of miRNAs expressed by various cell types in the liver may allow us in the future to distinguish between different causes and types of liver injuries (eg, cholangiocyte injury in bile ducts and endothelial cell injury in veins and arteries). Preliminary data from our research group indeed suggest that tissue levels of specific miRNAs expressed by biliary epithelial cells could be used to quantify biliary injury and predict the development of long-term biliary complications and graft loss after liver transplantation.42 In addition, miRNA-based diagnostics could facilitate allograft selection (particularly with marginal donors) and potentially enlarge the pool of grafts. For example, several experimental studies have demonstrated a role of hepatic miRNAs (including miR-122) in the regulation of cell proliferation during liver regeneration after partial hepatectomy.43-48 Although the exact biology is not clear, it is conceivable that the decrease in miR-122 expression during graft storage may be related to hepatic cell cycle progression in response to ischemic injury. It is tempting to speculate that the manipulation of miRNAs with antisense, anti-miRNA technology11 could allow therapeutic manipulation for the rescue of marginal grafts or the use of smaller split grafts through the minimization of injury and the stimulation of cell proliferation.17

In summary, we have demonstrated that the circulating HDmiRs miR-122, miR-148a, and miR-194 are stable and detectable during hepatic injury in patients after liver transplantation. The levels of 2 of these HDmiRs closely correlate with AST and ALT during posttransplant liver injury and acute rejection. These data support the potential of miRNA-based diagnostic tools for various types of liver injuries in liver transplant recipients.