Expression of human miR-200b-3p and -200c-3p in cytomegalovirus-infected tissues

Human cytomegalovirus (HCMV) infection can cause inflammatory tissue-invasive end-organ diseases upon lytic replication. In humans, mature miR-200b-3p and -200c-3p suppress the synthesis of HCMV immediate early 2 (IE2) protein by binding to the 3′-UTR of the mRNA encoded by the unique long (UL) 122-123 region in human foreskin fibroblasts and pre-transplant peripheral blood mononuclear cells stimulated with HCMV. The present study aimed to quantitate the expression of Homo sapiens (hsa)-miR-200b-3p and 200c-3p in HCMV-infected tissues. We collected 240 HCMV-infected and 154 HCMV-non-infected, formalin-fixed, paraffin-embedded tissue samples of the gastrointestinal (GI) tract and bronchi/lungs. MiRNAs, HCMV, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were quantitated by quantitative reverse transcription-PCR (qRT-PCR) and quantitative PCR (qPCR) on the basis of standard curves generated using miRNA mimics, the HCMV strain from National Institute for Biological Standards and Control (NIBSC) 09/162, and GAPDH control. To avoid the effect of cell counts on the qRT-PCR and qPCR results, the data were normalized to GAPDH levels. HCMV-infected tissues had significantly lower levels of 200b-3p/GAPDH (3.03 ± 1.50 compared with 3.98 ± 1.08 log10 copies/μl, P<0.001) and 200c-3p/GAPDH (4.67 ± 1.84 compared with 6.35 ± 1.47 log10 copies/μl, P<0.001) than normal tissues. The values for 200b-3p/GAPDH (r = −0.51, P<0.001) and 200c-3p/GAPDH (r = −0.54, P<0.001) were significantly inversely correlated with HCMV load. Low tissue levels of 200b-3p and 200c-3p in humans are associated with cytopathic inflammation due to HCMV infection.


Introduction
Human cytomegalovirus (HCMV) replication can cause serious harm in solid organ transplantation (SOT) and hematopoietic stem cell transplantation (HSCT) recipients [1][2][3]. In particular, post-transplant HCMV viremia has various indirect immunomodulatory effects, such as chronic allograft dysfunction and failure, in SOT recipients [1,2]. HCMV reactivation has been associated with increased mortality and prolonged treatment of mechanical ventilation in critically ill intensive care unit patients without prior objective immunocompromised conditions [4][5][6][7]. This worsening of outcomes by HCMV infection has led to the implementation of regular HCMV monitoring as well as well-organized and patient-tailored preventive strategies with risk stratification in routine clinical practice of SOT and HSCT recipients [1][2][3]. In addition, clinical trials have been conducted for evaluating the effects of HCMV prevention on outcomes in non-immunocompromised critically ill patients with acute respiratory distress syndrome [8].
Numerous basic and clinical studies fully support the need for continuous suppression and elimination of HCMV replication, a need that is currently unmet [9]. In vitro experiments on cultured cells have shown that epigenetic regulation through chromatin structural changes in the major immediate-early promoter (MIEP) and miRNA originating from the human host and/or HCMV play important roles in balancing latency and cytopathic replication [10][11][12][13][14]. Therefore, complete blockage of the production of immediate-early (IE) protein 2 (pp86), which is essential for the lytic virion structure and is encoded by the HCMV UL122-123 region, could be an effective approach to maintaining latency [12][13][14][15].
Seed sequences of human (Homo sapiens (Hsa)) miRNAs bind to the 3 -UTR of certain mRNAs and generally inhibit protein synthesis [16]. In vitro studies revealed that hsa-mature miRNA (miR)-200b-3p, -200c-3p, and -429, belonging to the miRNA-200 family, bind the 3 -UTR of HCMV UL122-123, in accordance with in silico predictions, and decrease IE2 synthesis in HCMV-infected fibroblasts [12,17]. Moreover, higher levels of hsp-miR-200b-3p and -200c-3p in pre-transplant blood with in vitro HCMV stimulation were associated with lower post-transplant HCMV replication rates in SOT recipients [17]. These results suggest that hsa-miR-200b-3p, -200c-3p, and -429 might serve as biomarkers to predict HCMV infection and/or disease and as therapeutic targets to control HCMV replication. The present study aimed to confirm the association between the levels of hsa-miR-200b-3p or -200c-3p or -429 and HCMV load in a clinical setting, with a focus on tissue-invasive disease, using formalin-fixed paraffin-embedded (FFPE) tissues.

FFPE tissue collection
We collected 240 whole-block HCMV-infected FFPE tissues between January 2007 and August 2016 in Shinchon and Gangnam Severance Hospital (Seoul, Korea) on the basis of information retrospectively extracted from electronic medical records. HCMV-infected tissues were strictly defined as those distinctly positive in immunohistochemical staining (IHS) using mouse anti-HCMV monoclonal antibody (mAb) (Clone DDG9 and CCH2; Agilent Dako Technologies, Seoul, South Korea) along with cytopathic histopathological findings of inflammation with/without necrosis as well as ulcerative/erosive lesions. After the exhaustive review of pathologic reports, we excluded FFPE tissues with following features: (i) suspected viral infection as indicated by the presence of giant cells with macronuclei/smudged chromatin/inclusion bodies, but IHS for HCMV was negative or not performed; (ii) co-infections with other organisms; (iii) patient was ≤18 years of age; and (iv) pathologic findings such as malignancies and acute/chronic inflammation caused by factors other than HCMV. As controls, 154 blocks of HCMV-non-infected age-and sex-matched FFPE tissues were obtained, between 2011 and 2016, from the resection margin of solid cancers and without any abnormal histological findings were included. The present study was approved by the institutional review board with waiver of written informed consent.

DNA and RNA preparation
Total RNA was isolated from 20-μm-thick FFPE tissue sections using a miRNeasy R FFPE Kit per the manufacturer's instructions (Qiagen, Seoul, Korea). Genomic DNA (gDNA) was extracted with a QIAamp R DNA FFPE Tissue kit per the manufacturer's instructions (Qiagen). All RNA and gDNA samples had A 260 /A 280 values of 1.8-2.0 as measured in a NanoDrop TM 2000 spectrophotometer (Thermo Fisher Scientific, Seoul, Korea). All samples were stored at −80 • C until use.
To exclude the effects of severity of necrosis on HCMV quantity [14,23], we measured the expression of the housekeeping protein glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in 100 ng/μl of gDNA through qPCR using TaqMan TM human GAPDH endogenous control (6FAM TM /MGB probe, non-primer limited) according to the manufacturer's instructions (Applied Biosystems/Thermo Fisher Scientific). qPCRs were run in a LightCycler R 480 platform (Roche Diagnostics, Seoul, Korea). Each sample was tested in triplicate. Reliability analysis using inter-class correlation of C t values for each qRT-PCR or qPCR showed nearly perfect reproducibility as 0.996 for hsa-miR-200b-3p, 0.994 for -200c-3p, 0.987 for HCMV, and 0.959 for GAPDH.
The quantity of miRNAs or GAPDH was expressed as log 10 copies/μl of input RNA or log 10 copies/μl of input DNA. The HCMV load was expressed as log 10 IU/μl of input DNA. The limit of detection of all qRT-PCRs was defined as 2 log 10 copies/μl. All data under this threshold were considered as undetectable for categorical and zero for continuous variable analysis.

Statistical analyses
Data were expressed as mean + − S.D. or median (interquartile range) or number (percent). Statistical analyses were performed using SPSS V23.0 (SPSS, Chicago, IL, U.S.A.) and GraphPad Prism V6 (GraphPad Software, La Jolla, CA). Differences in categorical and continuous variables amongst two groups were analyzed by chi-square and independent t test or Mann-Whitney U test, respectively. We used one-way ANOVA test and post-hoc Bonferroni's multiple comparison tests to compare three groups. Correlations between two continuous variables were expressed as Pearson's r value and 95% confidence interval (CI). To adjust the type I error by multiple independent comparisons, we considered the significant two-tailed P-value as less than 0.002 (0.05/25 of independent comparisons in the present study). Table 1 summarizes the clinical characteristics of patients and FFPE tissues. All tissues originated from HCMV seropositive subjects. Mean age (P=0.188) and sex (P=0.872) were not different between HCMV-infected and non-infected tissues. The majority of tissues originated from the gastrointestinal (GI) tract, including the esophagus, stomach, colon, and rectum, and there was no significant difference in organ distribution between the two groups (P=0.918).  Figure 1B). The HCMV load in the 240 HCMV-infected FFPE tissues was 6.15 + − 1.19 (range: 4.11-9.69) log 10 IU/ml, whereas that in all normal tissues was less than 0.7 log 10 IU/ml of the lower limit of quantitation.

Discussion
Our study revealed that the expression of hsa-miR-200b-3p and -200c-3p was decreased in HCMV-infected tissues. This finding is in-line with those in two previous studies that evaluated the levels of two miRNAs in pre-transplant peripheral blood mononuclear cells (PBMCs) from SOT recipients with in vitro HCMV stimulation, and performed co-transfection assays using a laboratory HCMV strain and two miRNA mimics in fibroblasts [12,17]. This is the first study using HCMV-infected FFPE tissues, which are more representative of the clinical situation because these tissues definitely indicate HCMV end-organ disease.
The suppressive role of hsa-miR-200b-3p and -200c-3p on HCMV lytic replication was supported by two stepwise objective bases: (i) accurate prediction through in silico analysis of miRNAs binding to the 3 -UTR in the HCMV UL122-123 rather than high-throughput screening methods, such as oligonucleotide microarrays, in which expression profiles could be biased due to sample characteristics, platforms, or statistical analyses [24,25], (ii) verification of the binding of identical seed sequences in hsa-miR-200b-3p and -200c-3p to the HCMV UL122-123 by luciferase reporter assay using wild-type and mutant recombinant vectors, the results of which precisely corresponded with in silico prediction [12,17].
The results of the different experiments in the present study were consistent, allowing us to derive a reasonable conclusion. In spite of the relatively small differences in hsa-miR-200b-3p or -200c-3p levels between HCMV-infected and non-infected upper GI tissues, the levels of both miRNAs were significantly decreased in HCMV-infected upper/lower GI tissues. These findings were consistent between the quantitative and categorical analyses. We identified a strong negative correlation between the two miRNAs and HCMV levels; however, we were not able to establish that this correlation is directly caused by a reduction in IE2 production. This inability could be attributed to the fact that the HCMV-infected tissues do not have measurable levels of IE2 owing to the presence of large numbers of complete virions. In fact, we could not detect the mRNA encoding IE2 by qRT-PCR (data not shown). Finally, normalization of the expression data to that of GAPDH did not significantly alter the findings [14,23]. Our findings suggest that hsa-miR-200b-3p and -200c-3p could play relevant roles in regulating the initiation of HCMV active replication and/or latent-to-lytic switch by blocking the synthesis of major IE2 protein. This IE2 suppression could result in a serial reduction in early and late proteins, including HCMV DNA polymerase or glycoproteins [26]. Finally, this process could inhibit the extracellular budding of complete HCMV virion with envelope. However, the cross-reaction between these hsa-miRNAs and miRNAs encoded by HCMV, as well as the effect of MIEP, a strong transcription factor for major IE proteins, should be explored [27][28][29][30].
The major limitation of the present study would be the quality of RNA extracted from the FFPE tissues. MiRNAs can be degraded in FFPE tissues, and this damage would affect qRT-PCR results [31,32]. We did not check RNA integrity number by capillary electrophoresis. However, in addition to good A 260 /A 280 values, the following features were considered while concluding that our RNA samples were of sufficient quality to yield satisfactory results: (i) triplicate qRT-PCRs and qPCRs for all samples showed nearly perfect inter-class correlation of C t values; (ii) very strong correlation between the levels of hsa-miR-200b-3p and -200c-3p, with r = 0.85; and (iii) 200c-3p levels were higher than 200b-3p levels, similar to the findings in human PBMCs [17]. Most tissue samples were of the GI tract and a small number of the lungs, because clinical specimens of HCMV-infected tissues such as liver, lung, and central nervous system, are hard to obtain. Despite these limitations, this first evaluation of hsa-miR-200b-3p and -200c-3p in end-organ FFPE tissues gives insights in HCMV infection in association with these miRNAs in the real clinical setting.
In conclusion, the levels of hsa-miR-200b-3p and -200c-3p are reduced in HCMV-infected tissues. These miRNAs likely play a biologically relevant role in the control of HCMV after transplantation.

Clinical perspectives
• HCMV infection can result in cytopathic tissue-invasive inflammatory diseases. In addition, active replication of HCMV is closely associated with mortality and several morbidities such as graft rejection/dysfunction/failure, graft-versus-host disease, increase in other opportunistic infections in recipients of SOT and HSCT.
• hsa-miR-200b-3p and -200c-3p suppressed the synthesis of IE2 by binding the 3 -UTR of mRNA encoded by the HCMV UL122-123 region in human foreskin fibroblasts (in vitro) and in pre-transplant PBMCs stimulated by HCMV (in SOT recipients).
• MiR-200b-3p and -200c-3p levels were lower in HCMV-infected tissues than in normal tissues. The levels of miR-200b-3p and -200c-3p were inversely correlated with HCMV load. These miRNAs likely play a biologically relevant role in the control of cytopathic inflammation or HCMV infection after transplantation.