Different expression pattern of human cytomegalovirus-encoded microRNAs in circulation from virus latency to reactivation

Background Human cytomegalovirus (HCMV) is a beta-hersvirinae that has a high latent infection rate worldwide and can cause serious consequences in immunocompromised patients when reactivation; however, the mechanism of how HCMV convert from latent to reactivation has rarely been investigated. In the present study, we aimed to perform a comprehensive analysis of the HCMV-encoded microRNA (miRNA) profile in serum of patients upon HCMV reactivation from latency and to further evaluate its clinical significance for the disease monitoring and preventing usefulness. Methods Serum samples from 59 viremia patients and 60 age-gender matched controls were enrolled in this study for screening and validation of different expression of HCMV miRNAs. Serum concentrations of 22 known HCMV miRNAs were determined by a hydrolysis probe-based stem-loop quantitative reverse transcription polymerase chain reaction (RT-qPCR) assay. HCMV DNA was measured by quantitative real-time PCR (qPCR) with the whole blood sample. Serum HCMV IgG and IgM were assessed using enzyme linked immunosorbent assay (ELISA). Another 47 samples from 5 patients at different time points were collected to evaluate the monitoring effectiveness and disease prediction ability of differential expression HCMV-miRNAs during the antiviral treatment. Results The RT-qPCR analysis revealed that the serum levels of 16 of the 22 examined HCMV miRNAs were elevated in HCMV viremia patients compared with controls, and a profile of 8 HCMV miRNAs including hcmv-miR-US25-2-3p, hcmv-miR-US4-5p, hcmv-miR-US25-2-5p, hcmv-miR-US25-1-3p, hcmv-miR-US25-1, hcmv-miR-UL36, hcmv-miR-UL148D, hcmv-miR-US29-3p were markedly elevated (fold change > 2, P < 0.01). Receiver operating characteristic curve (ROC) analysis were performed on the selected HCMV-miRNAs in all of the patients and controls that enrolled in this study, and which ranged from 0.72 to 0.80 in the autoimmune patients. In addition, hcmv-miR-US25-1-3p levels were significantly correlated with HCMV DNA load (r = 0.349, P = 0.007), and were obviously higher in the reactivation set than the latency set in the autoimmune patients, which could be a predictor for the monitoring of the antiviral treatment. Conclusions HCMV miRNAs profile showed markedly shift-switch from latency to reactivation in circulation from HCMV infected patients and hcmv-miR-US25-1-3p may be served as a predictor for the switch upon reactivation from latency in patients suffered with autoimmune diseases.


Background
Human cytomegalovirus (HCMV) is a member of the herpesviridae family, beta-hersvirinae subfamily that latently infects approximately 70-100% of the population worldwide in their lifetime [1]. Initial infection and reactivation of HCMV usually does not result in morbidity in healthy individuals, whereas reactivation of HCMV from the latency in immunocompromised people, such as AIDS patients, solid organ transplant recipients and neonates can lead to severe morbidity and mortality [2][3][4]. However, the mechanisms involving in HCMV latency and reactivation remain poorly understood.
MicroRNAs (miRNAs) are a subset of non-coding RNA molecules (19)(20)(21)(22)(23) nucleotides in length) that mediate post-transcriptional gene silencing. HCMV encodes at least 26 mature miRNAs, which were derived from 15 stem-loop precursors and have been implicated in the regulation of viral replication, immune modulation, and immune evasion [5,6]. Differential HCMV encoded miRNAs expression was observed in the latency and activation infection by HCMV in vitro. For instance, hcmv-miR-UL148D facilities latent viral infection by modulating the IER5-CDC25B axis in host cells [7]. Hcmv-miR-UL112-1 can attenuate replication of HCMV and implicates in latency control of HCMV by targeting HCMV IE1, UL112/113, UL120/121 and UL144 [8,9]. However, all of the HCMV miRNAs that discovered currently were detected in infected fibroblast cells [10,11], which due to the lack of appropriate cell-lines or animal models for studying HCMV latency.
Circulating miRNAs could be novel biomarker for the diagnosis of virous diseases, including viral infection diseases [6,12,13]. In vivo evidence of the link between HCMV miRNAs and diseases processes is now emerging, with the description of hcmv-miR-UL112-3p as a biomarker of essential hypertension, diabetes mellitus and glioblastoma [14,15], and hcmv-miR-UL22A-5p as a biomarker in solid organ transplantation [16]. Our group also demonstrated a distinct expression pattern of HCMV-encoded miRNAs in oral lichen planus (OLP) [17]. Moreover, one recent study showed that there was a different hcmv-miRNAs pattern between latent and lytic in vitro [11]. Nevertheless, no report about the relationship between the HCMV DNA load and HCMV miRNA's expression levels in vivo. In addition, expression patterns of HCMV miRNAs and their roles in the transformation from latency to reactivation have not yet been examined in vivo.
Since understanding the HCMV miRNA expression pattern during latency phase, the reactivation phase and the shift expression between the above two phases will offer great benefit for HCMV associated diseases therapy, and may also provide clues for preventing reactivation of the virus from latency. In the present study, we assessed the in vivo expression pattern of HCMV miRNAs in the patients which with the detection of HCMV IgG seropositive, IgM seronegative, differ by the HCMV DNA level up and below for the 500 IU/mL (viremia or not which equivalent to reactivation or latency phase), and examined their potential as predictors of clinical and virological endpoints. We found a panel of HCMV-encoded miRNAs that showing different expression level between the latency and the reactivation, and some may be used as HCMV indicators, especially in the patients who suffered with autoimmune diseases and co-infected with the latency HCMV virus.

Participants and study design
A total of 3986 subjects with high risk HCMV infection [2] mainly from the Departments of respiratory (27.2%), hematology (17.6%), immunology (13.6%), infection disease (12.9%), gastroenterology (8.8%), intensive care unit (3.0%) and others (16.9%) were recruited in this study. All the patients were hospitalized in Nanjing Drum Tower Hospital between January 2016 and March 2017. After HCMV DNA load and HCMV serological examination, a training cohort that containing 23 patients with HCMV viremia (HCMV DNA > 500 IU/mL), HCMV IgG seropositive and HCMV IgM seronegative as the case set (defined as reactivation infectious), and another 24 patients with HCMV DNA levels less than 500 IU/ mL, HCMV IgG seropositive and HCMV IgM seronegative as the control set (defined as latency infectious) was used for screening the differential expression pattern of HCMV miRNAs. A validation cohort that containing 36 patients for the case set and 36 patients for the control set with the same above criterial was used to confirm the results of the training cohort. An additional independent cohort including 47 samples from 5 patients with leukemia (2 severe aplastic anemia patients, 1 myelodysplastic syndromes patient, 1 acute myeloid leukemia M2a Conclusions: HCMV miRNAs profile showed markedly shift-switch from latency to reactivation in circulation from HCMV infected patients and hcmv-miR-US25-1-3p may be served as a predictor for the switch upon reactivation from latency in patients suffered with autoimmune diseases. Keywords: Human cytomegalovirus, Serum microRNA, HCMV DNA load, Latency, Reactivation, Switch patient and 1 acute lymphoblastic leukemia patient) who underwent bone marrow transplantation (the samples were collected at different time points during the antiviral therapy with ganciclovir) were also collected. The overall study design is shown in Fig. 1. All clinical data and blood samples were obtained from participants who had given written informed consent, according to protocols approved by the Ethics Committee of Nanjing Drum Tower Hospital. For all the patients, the age, gender, diagnosis, white blood cell count, C reactive protein (CRP), HCV, EBV and HIV status were recorded and used for the study (Table 1).

HCMV DNA load
The HCMV DNA level in peripheral blood leukocytes were determined by quantitative real-time PCR (qPCR). In brief, DNA was extracted from 200 µL peripheral blood leukocytes using QIAamp DNA Mini kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. Two microliters of DNA were tested with TaqMan PCR assays using diagnostic kit for quantification of human cytomegalovirus DNA (DAAN GENE, Guangzhou, China) on a Roche LightCycler ® 96 PCR System (Roche diagnostics, Mannheim, Germany) according to the manufacturer's protocol. Ten-fold diluted recombinant plasmid that contained the HCMV target sequence was used to construct calibration curve. The absolute HCMV DNA level of each sample were calculated through the calibrator. Results were expressed as IU per 1 mL blood [18].

Anti-HCMV IgG and IgM antibodies determination
Serum anti-HCMV IgG and IgM were tested using a commercially available ELISA kit (MEDSON, NJ, USA) according to the manufacturer's instructions. In brief, serum (1:100) was added to the 96-well plate which containing HCMV antigen and incubated at 37 ℃ for 1 h, the mixture was then washed for four times and incubated with a horse-radish peroxidase conjugated at 37 °C for 1 h. After four times washing, reactivity was determined using o-phenylenediamine and the reaction was blocked with 2.5 M sulfuric acid. For the IgG-ELISA, a calibration curve, calibrated against the 1st WHO international standard, was used to quantitatively determine IgG antibody concentrations in each sample. For the IgM-ELISA, the test results were calculated using the optical density (OD) value at 450 nm, and the cut-off value for positivity was OD > 0.25.

Serum RNA isolation and RT-qPCR assay
Total RNA was extracted from 100 μL of serum using a 1-step phenol/chloroform purification method and precipitated using isopropyl alcohol as previously described [17]. In brief, 100 μL of serum was mixed with 300 μL deionized water, 200 μL acid phenol, and 200 μL chloroform. The mixture was vortex-mixed vigorously and incubated at room temperature for 15 min. After phase separation, the aqueous layer was mixed with 1.5 volumes of isopropyl alcohol and 0.1 volumes of 3 mol/L sodium acetate (pH 5.3). The solution was stored at -20 °C for 1 h. The RNA pellet was obtained by centrifugation at 16,000g for 20 min at 4 °C. The resulting RNA pellet was washed once with 750 mL/L ethanol and dried for 10 min at room temperature. The pellet was dissolved in 20 μL of RNase-free water and stored at − 80 °C. To control the variability in RNA extraction and purification procedures, an exogenous plant small molecular RNA named MIR2911 (5′-GGC CGG GGG ACG GGC UGG GA-3′), was spiked into each sample with a final concentration of 10 6 fmol/L during RNA isolation as a synthetic external reference for the normalization of serum miRNAs [19]. Hydrolysis probe-based quantitative reverse transcription polymerase chain reaction (RT-qPCR) was carried out using a TaqMan miRNA PCR kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's instructions with a minor modification as previously described [19]. Briefly, 2 μL of total RNA was reverse transcribed to cDNA using AMV reverse transcriptase (TaKaRa, Dalian, China) and the stem-loop RT primer (Applied Biosystems, Foster City, CA, USA). Real-time PCR was performed using hydrolysis miRNA probes on a Roche LightCycler ® 96 PCR System (Roche diagnostics, Mannheim, Germany). All reactions, including no-template controls, were performed in triplicate. The Cq values were determined using the fixed threshold settings. Relative levels of HCMV miRNAs were then normalized to exogenous MIR2911 and were calculated using comparative Cq method (2 −ΔCq ).

Statistical analysis
Statistical analysis was performed with GraphPad Prism 6.0 software or SPSS statistical software (version 16.0). The miRNA concentrations were represented as means and standard errors (Mean ± SEM) and other clinical variables were showed as Mean ± SD or Median (interquartile range). The data analyses were performed using the non-parametric Mann-Whitney tests, Chi-square test and Pearson correlation analyses. Statistically significant was defined as a P < 0.05. For each miRNA, a receiver operating characteristic (ROC) curve was generated. The area under curve (AUC) values and 95% confidence interval (CI) were calculated to determine the specificity and sensitivity of diagnosis of HCMV reactivation.

HCMV DNA viral load and serological results
We examined the HCMV viral load in 5664 samples of peripheral blood leukocytes that collected from 3986 patients by quantitative real-time PCR. Among which 145 whole blood samples were defined as viremia with the HCMV DNA level upper than 500 IU/mL and others with HCMV DNA level lower than 500 IU/mL. Of the 145 whole blood samples, 59 non-repetitive patients with HCMV-IgG seropositive, HCMV-IgM seronegative were enrolled in this study. In the meanwhile, another 60 patients with HCMV-IgG seropositive, HCMV-IgM seronegative and HCMV DNA level below 500 IU/mL were selected as controls (Fig. 1). The mean levels of HCMV DNA viral load for the 59 viremia patients were 1930 (853, 6810) IU/mL, 1920 (1100, 3680) IU/mL in the training cohort and 1995 (795.5, 9995) IU/mL in the validation cohort, respectively (Table 1). For the concentrations of anti-HCMV IgG, there was no significant difference between the HCMV viremia patients (n = 59) and the controls (n = 60) (t = 0.7794, P = 0.4373). Similar results were also observed in the training cohort and the validation cohort (t = 0.5766, P = 0.5671 and t = 1.384, P = 0.1709), respectively.

Confirmation of the up-regulated HCMV-encoded miRNAs
Subsequently, the 8 up-regulated HCMV-encoded miR-NAs were confirmed in an additional cohort including 36 cases and 36 controls (refereed as validation cohort). The 8 miRNAs exhibited consistent alterations as the results from the training cohort (Fig. 2a-h). Moreover, when combined the results of the training set and validation set ( Fig. 2i-p), consistent with our expectations, the concentrations of all the eight hcmv-miRNAs were significantly increased in the viremia patients as compared with control group.
Receiver operating characteristic curve (ROC) analysis on the seven selected hcmv-miRNAs yielded areas under ROC curve (AUCs) ranged from 0.72 to 0.80 (Fig. 3h)

Independent cohorts of antiviral treatment
Five patients were received antiviral therapy, including 2 patients with severe aplastic anemia, one with myelodysplastic syndromes, one with acute myeloid leukemia M2a and one with acute lymphoblastic leukemia respectively, who were underwent bone marrow transplantation. Forty-seven serum samples were collected at different time points during the antiviral therapy, and hcmv-miR-US25-1-3p levels was significantly correlated with HCMV DNA levels during the antiviral therapy ( Fig. 6a-e).
The time of occurrence of viremia in the five patients was on the 33rd day (Fig. 6a), 65th day (Fig. 6b), 27th day (Fig. 6c), 42nd day (Fig. 6d), and 33rd day (Fig. 6e) after bone marrow transplantation, respectively. Of the five patients, the No. 3 patient developed viremia on 27th day (Fig. 6c) after allogeneic stem cell transplant, both HCMV DNA and hcmv-miR-US25-1-3p were at high levels at this time and decreased upon antiviral therapy. On 130th day after transplantation, the patient's hematopoietic disease relapsed and worsened again, and was transferred to the ICU for emergency treatment. In the late phase of monitoring, the viral DNA showed a certain degree