FormalPara Key Points

Prognostic impact of laboratory diagnostics may be enhanced by the use of microRNA.

Promising microRNA panels have been identified for differential diagnostics of hepatitis B virus-associated chronic disease and liver cancer, and for enterovirus-associated cardiomyopathy.

The potential of viral microRNAs should be thoroughly explored in any viral disease where the detection of nucleic acids is considered clinically relevant.

1 Outline

In this narrative review some viral and human microRNAs (miRNA) of diagnostic or prognostic potential in human diseases associated with viral infections will be discussed. This review is not exhaustive; instead, examples will be presented where the use of viral or human miRNAs may contribute to better laboratory diagnostics of severe viral diseases. The emphasis will be on miRNAs whose potential has not only been suggested in experimental systems but has also been studied in representative sets of human samples. Technical approaches and pitfalls, as well as therapeutic use of miRNAs will also be discussed.

2 MicroRNAs and Their Function

MicroRNAs are small, 18–25 nucleotide (nt) long single-stranded noncoding regulatory RNA molecules, which regulate the expression of many, if not the majority of genes. It has been estimated that the expression of at least 60 % of human genes is regulated by microRNAs [1]. In humans and other mammals, microRNA encoding genes are found both in intragenic regions and in introns of messenger RNAs (mRNAs) or within non-coding RNA (ncRNA) genes [2, 3]. All human chromosomes are known to harbor miRNA sequences, the X chromosome being the richest and the Y chromosome the poorest in miRNA genes [4]. More than one-third of human miRNAs appear in clusters, i.e. they are non-randomly located in the same DNA strand relatively close to each other and transcribed together as polycistronic transcripts [5, 6].

The canonical pathway of mammalian miRNA biogenesis is presented in Fig. 1 (modified from [7]). Primary miRNA (pri-miRNA) is transcribed from the genome by RNA polymerase II. Cleavage by Drosha, a class III RNase, produces precursor miRNA (pre-miRNA) hairpin, which is then exported from the nucleus to the cytoplasm. Pre-miRNA is further cleaved by the Dicer RNaseIII to produce 18–25 nt miRNA duplex consisting of 5′ and 3′ arms. The single-stranded functional or guide miRNA usually originates from one of these arms to produce either 5p or 3p miRNA stemming from the 5′ arm or the 3′ arm of the original hairpin, respectively. Both arms may also produce functional miRNA, in which case both 5p and 3p miRNAs are produced from the same locus. For this reason, the original nomenclature using star (*) miRNA for the nonfunctional passenger strand is now being replaced by the 5p/3p nomenclature. The functional mature miRNA is then loaded to RNA-induced silencing or RISC complex where it binds to the complementary 3′ untranslated region (UTR) of target mRNA molecules to regulate gene expression. miRNAs may also bind to the 5′ UTR or open reading frame of mRNA, or to DNA directly. Binding of miRNA to mRNA leads to downregulation of gene function by destabilization and degradation of mRNA, or by translational arrest.

Fig. 1
figure 1

Modified from [7]

Canonical pathway of mammalian microRNA biogenesis. Primary miRNA (pri-miRNA) is transcribed from the genome by RNA polymerase II (Pol II) in the nucleus. Cleavage by Drosha, a class III RNase, produces precursor miRNA (pre-miRNA) hairpin, which is exported to the cytoplasm. Pre-miRNA is cleaved by the Dicer RNaseIII to produce 18–25 nt miRNA duplex. The single-stranded functional mature miRNA is then loaded to RNA-induced silencing or RISC complex where it binds to the complementary 3′ untranslated region of target mRNA molecules to regulate gene expression, and the nonfunctional strand is degraded.

Full size image

MicroRNAs cause gene silencing by binding to their mRNA targets. Full complementarity is not required; a minimal requirement is complementarity of the seed sequence at nt 2–8 of mature miRNA. Many viral or plant miRNAs bind to perfectly complementary sequences in the target mRNA, which is uncommon for human or animal miRNAs. It seems that binding of a miRNA to a fully complementary sequence directs cleavage of the target mRNA [8], while partial complementarity causes repression of protein translation [9–11]. Because partial complementarity is sufficient for regulation, an individual miRNA may have hundreds of targets. However, perfect complementarity of many viral miRNAs to their target may narrow down the number of targets and increase the specificity of regulation [12].

miRNAs regulate key cellular processes such as development, cell proliferation, cell differentiation, hematopoiesis, immune responses, stress response, apoptosis and cell death, stress response, chemokine and cytokine production, neurodegeneration, and oncogenesis. miRNAs may indeed be implicated in most if not all chronic human diseases.

3 Viral MicroRNA

Many viruses are known to encode their own miRNAs [12]. Both coding and noncoding viral sequences harbor miRNA genes, which is an efficient way to package regulatory capacity into viral genomes often having small size and limited coding capacity. Viruses can thus efficiently extend their possibilities to regulate the expression of their own genes as well as host genes by encoding miRNAs. Viral miRNAs are generated by canonical (Fig. 1) or noncanonical, Drosha-independent pathways [13]. Viral miRNAs can efficiently manipulate host gene expression, because they can regulate a number of mRNA targets, and they are less likely to be recognized by the host immune system than viral proteins [14]. Known functions of viral miRNAs include regulation of viral persistence or latency and reactivation, proliferation and long-term survival of host cells despite the presence of virus, and host immune evasion contributing to establishment of infection. In addition to host gene expression, viral miRNAs may also regulate the expression of viral protein coding genes [14]. Many viral miRNAs function by immune modulation and immune evasion by direct targeting of pro-apoptotic genes and controlling viral latency, which support the establishment of viral persistence [15]. Altogether, miRNAs seem to represent yet one entity of interactions between viruses and their hosts.

miRNAs encoded by herpesviruses are the most numerous and best studied. Among them, alphaherpesviruses herpes simplex (HSV) 1 and 2, cytomegalovirus (CMV) betaherpesvirus, as well as Epstein–Barr virus (EBV) and Kaposi’s sarcoma-associated herpesvirus (KSHV, or human herpesvirus 8 or HHV-8) gammaherpesviruses encode a number of miRNAs, whereas Varicella Zoster virus (VZV) alphaherpesvirus encodes none [16, 17]. Among DNA tumor viruses, human polyomaviruses and adenoviruses encode miRNAs [17]. Further, we have validated human papillomaviruses (HPV) encoded miRNAs [18] (Virtanen et al., unpublished data). Putative miRNA has also been identified for hepatitis B virus (HBV) [19], and among the rare miRNA encoding RNA viruses, for human immunodeficiency virus (HIV) [20, 21] and the Ebola filovirus [22]. At present the diagnostic potential of any viral miRNA has not been fully established yet. Some considerations will, however, be presented in later sections of this article for selected virus groups.

4 Human MicroRNAs in Viral Diseases

Interplay between viruses and their hosts by mechanisms involving miRNA is yet another example of their long coexistence and coevolution. Indeed, human miRNAs have been suggested to have their evolutionary origin as an innate immune defense mechanism against viral infections [23]. Human miRNA expression is substantially dysregulated in many if not all viral infections and in chronic viral diseases [24]. Computer prediction of complementary sequences suggested that human miRNAs may have potential antiviral effects against more than 200 human viruses, and this was narrowed down by further extraction to 62 viruses from six families, mostly single-stranded RNA viruses [25]. Further, host miRNAs may influence virus replication and gene expression, as well as latency or tumorigenesis, which can be achieved by expression of cellular functional orthologs of viral miRNAs as has been shown for some herpesviruses [26]. This raises the interesting question whether host miRNAs evolved to combat viral infections or whether viruses, which evolve faster, evolved to mimic specific host miRNAs [26].

Future exploration of miRNA functions will presumably illuminate the basis for tissue specificity of viruses. For example, efficient hepatitis C virus (HCV) replication depends on the liver-specific hsa-miR-122 or miR-122 (from now on the species abbreviation hsa- for Homo sapiens will be omitted) in hepatocytes (see Sects. 9 and 15) [27]. Identification of miRNA markers of individual immunological status or host immune responses to pathogens would open a whole new field of applications to miRNA-based diagnostics and therapies. This is exemplified by miR-155, which is a key modulator of T-cell responses and seems to play a central role in the immune response to many viral and bacterial infections [28–30].

Many human miRNAs have been identified which may serve as diagnostic or prognostic markers or be useful in monitoring disease course or treatment success, although few miRNAs have been established in diagnostic practice at present. In the following sections the use of both viral and cellular miRNA markers in the diagnostics of viral diseases will be discussed.

5 Extracellular MicroRNAs: Circulating and Exosomal MicroRNAs

miRNAs can be detected in diseased tissue and site of infection, but also in various body fluids such as plasma, serum, urine, saliva, or cerebrospinal fluid. Circulating miRNAs play an important role in host-pathogen interactions and display host responses to pathogen. Circulating miRNAs in the blood may originate from peripheral blood mononuclear cells (PBMC) or may be released from diseased tissue by cytolysis, tissue injury, or within apoptotic bodies [31]. miRNAs can also be actively secreted from cells in exosomes or other microvesicles, or as RNA-protein complexes. The origin and contents of exosomal miRNAs may be different in health and disease. Exosomes can be taken up by neighboring or distant cells and the content miRNAs may modulate gene expression in recipient cells, and viral miRNAs can thus exert regulation of gene expression in cells that are not infected by the virus. In clinical practice, circulating miRNAs may provide novel biomarkers to narrow down the need of invasive sampling or costly imaging examinations to smaller patient groups. However, miRNA levels in body fluids may vary and may not reflect the levels in diseased tissues, and thus for many solid tumors miRNA or any other biomarker may have to be analyzed directly from tissue.

6 MicroRNA as a Diagnostic Tool

New or accessory methods are required in clinical practice if new pathogens or new pathogen strains emerge, novel information about known pathogens accumulate, or new disease associations are established. New antivirals are continuously being launched, some of them with novel mechanisms of action, which requires new methods to monitor disease course and success of therapy. Detection of miRNA may be one of the next new fields where a biological or technical discovery finds clinical use in laboratory medicine.

MicroRNAs are promising diagnostic markers for several reasons (modified from [32]), but their use also requires careful considerations. (1) miRNAs are stable in various body fluids including the RNase rich environment of blood, at broad pH and temperature range, and in varying laboratory and transport conditions. (2) Established quantitative PCR (qPCR) methods exist for miRNA detection, but these methods, as well as protocols to normalize and quantify miRNAs, need standardization. (3) miRNA expression profiles are specific to tissue and disease, and therefore sample material and controls have to be carefully selected. (4) Levels of circulating miRNAs in various body fluids may reflect the changes in diseases tissues, but may also differ due to different sources of circulating miRNAs. The use of body fluids would be preferable to invasive samples whenever possible.

Expression of a miRNA can be regulated by many factors, and one miRNA may have even hundreds of targets, implicating that the biomarker potential of a miRNA has to be carefully assessed to establish clinical relevance. Any individual miRNA may be a marker of several disease conditions, and the same miRNA may be upregulated in one condition and downregulated in another—even in the same individual, which presents a challenge to result interpretation. In a diagnostic context detecting presence or upregulation of a biomarker is conceptually more feasible than detecting downregulation or absence.

In order to identify miRNAs of biomarker potential, pilot profiling of miRNA expression using miRNA microarray or miRNA sequencing (miRNA-seq) in a small number of relevant samples from sick (disease) and healthy (control) individuals should be carried out. From the beginning the choice of relevant sample matrix considering the putative final diagnostic test is important. Typically, a large number of miRNAs are found regulated in any disease condition, and altogether downregulation is much more frequent than upregulation. After microarray or miRNA-seq profiling a feasible set of miRNAs are selected based on robust regulation and clinical relevance of predicted or known target mRNAs. Selected miRNAs are then clinically validated by quantitative PCR in a carefully selected and characterized sample material representing both (different stages of) disease and healthy condition and relevant sample matrix.

A number of diagnostic challenges can be identified where the possibility to apply miRNA should be explored. These include early diagnosis, triage or differential diagnostics to narrow down patients at high risk, monitoring of disease progression and treatment follow-up, risk of reactivation of chronic or latent viral infection, risk of recurrence or relapse, and prognosis of patients in cases where there is typically a long latency period between infection and disease. Further, miRNAs can be used both as drug targets and as drug molecules.

Detection of viral nucleic acids may be a sign of infection, a sign of contaminating virus without infection, or a marker of long persistence of nucleic acid in the organism even after clearance of infection. Future research will establish whether the same applies to miRNA.

7 Technical Issues

Amounts of individual miRNAs vary between sample matrices, and measured expression levels from different matrices are not comparable. Carefully selected positive and negative controls need to be included. Any miRNA has biomarker potential only if its levels can be assumed to remain stable in healthy individuals.

For most miRNA assays, total RNA is used as a template without the need to enrich for small RNAs. Individual miRNAs may be better extracted with a specific extraction method, but selection of the absolutely best extraction method for each miRNA may have to be compromised for fluent workflow. The amount of input material, weight or volume, is important for choice of extraction method. Standard sample volume rather than amount of RNA is feasible for body fluid samples in a diagnostic setting. Protein content of the sample may be an issue; for example, plasma is rich in protein but poor in miRNA, which may be challenging for efficient miRNA extraction. Established RNA extraction methods include phenol-based (e.g. Trizol, Thermo Fisher Scientific, Wilmington, DE), column-based (e.g. miRCURY, Exiqon A/S, Vedbæk, Denmark) or combination techniques (mirVana kits, Ambion, Life Technologies, Austin, TX; miRNeasy, Qiagen, Gaithersburg, MD) [33]. These methods yield RNA preparations of different degrees of purity, which may greatly affect the choice of downstream detection/quantification method and should be considered case by case. miRCURY kits yield highly pure miRNA, of better quality than Trizol or miRNeasy, and give good results in miRNA PCR assays if not too much starting material is loaded to saturate the column, or not too little to get low amounts of small RNAs [33, 34]. miRNeasy also yields good enrichment of high-quality small RNAs if starting material is not too ample, and may thus be optimal for biological fluids. Extraction method should be selected according to sample type and expected quantity of cells in the sample, and according to final use of the RNA preparation. We have successfully used mirVana for miRNA profiling in cell lines [18, 35] (Virtanen et al., unpublished data), mirVana Paris and miRCURY for body fluids and swabs [36] (Virtanen et al., unpublished data), and RecoverAll (Ambion) for formalin-fixed paraffin-embedded (FFPE) samples [18] (Virtanen et al., unpublished data). Although miRNAs are considered stable within natural matrices, their stability for quantitative diagnostic purposes in extracted RNA preparations has not yet been established.

Cellular samples offer the possibility to use endogenous control miRNAs to control the performance of sample extraction, reverse transcription (RT) and qPCR steps. One frequently used endogenous control for human miRNA assays is RNU6B small nuclear RNA, which is ubiquitously expressed at equal levels in many tissues and cells. Spike-in controls are useful to control for RT and qPCR, and for normalization. However, the stability of synthetic spike-in miRNA oligonucleotides is not as good as miRNAs in natural sample matrices, and thus spike-in controls may have to be added after sample extraction to avoid degradation in sample lysis buffer. A widely used spike-in control, also included in several commercial assays, is Caenorhabditis elegans cel-39-3p miRNA, which does not share homology with any known human miRNA [37]. The appropriate amount of spike-in control miRNAs is within the femtomole (10−15) to attomole (10−18) range.

Two types of qRT-PCR assays are broadly used in miRNA detection. TaqMan stem-loop primer based assays (Applied Biosystems, Thermo Scientific, Foster City, CA) are well established and have high sensitivity, and a large number of assays are readily available. Assays can be tailored for individual use, but the assay parameters may not be compatible with all putative miRNA sequences [18]. miRCURY LNA-enhanced primer-based assay design (Exiqon) may have more flexibility and is claimed by the manufacturer to be more specific, although in some cases at the cost of reduced sensitivity as observed in our own experiments (Virtanen et al., unpublished data). Running samples in triplicates at minimum is recommended, but there is no fixed recommendation as to how much variation between replicates should be tolerated.

A method of normalization, whether for solid tissue or body fluids, has to be selected to quantify the results [38, 39]. For body fluids which contain little or no cells a spike-in control is required. No established reference miRNAs exist for the analysis of circulating miRNAs for normalization. miR-16 and miR-93 are examples of reference miRNAs which have been used for serum samples from gastric cancer patients [40], but the appropriate reference miRNA may need to be selected according to disease and sample matrix.

Similar to mRNA expression levels, miRNA expression levels are quantified using 2ΔΔCt calculations based on normalized cycle of threshold (Ct) values in the qPCR assay. For each “sick” or “healthy” sample, each miRNA Ct is first normalized by subtracting the Ct for reference miRNA, for example cel-39-3p, to produce ΔCt (logarithmic parameters). Then ΔCt of healthy control is subtracted from ΔCt of sick individual to obtain ΔΔCt or normalized fold change of miRNA in sick versus healthy condition.

$$\begin{aligned} \Delta {\text{Ct}}_{\text{sick}} &= {\text{ Ct }}\left( {{\text{Target}} - {\text{miR}}_{\text{sick}} } \right) \, {-}{\text{ Ct }}\left( {{\text{Reference}} - {\text{miR}}_{\text{sick}} } \right)\\ \Delta {\text{Ct}}_{\text{healthy}} & = {\text{ Ct }}\left( {{\text{Target}} - {\text{miR}}_{\text{healthy}} } \right) \, {-}{\text{ Ct }}\left( {{\text{Reference}} - {\text{miR}}_{\text{healthy}} } \right)\\ \Delta \Delta {\text{Ct }} &= \, \Delta {\text{Ct}}_{\text{sick}} {-} \, \Delta {\text{Ct}}_{\text{healthy}} \end{aligned}$$

It is challenging to define a healthy condition, and it might be particularly challenging for patient follow-up or monitoring. Also clinical interpretation of thresholds, significant fold changes or significant absolute Ct values have to be established using statistical testing like paired T test or similar. Cellular and viral miRNAs may both serve as quantitative biomarkers. Only viral miRNAs may be used as qualitative biomarkers. They may reflect either present or past disease, and in both cases they may have biomarker potential (see Sect. 13), whereas miRNA as a remnant of viral expression products of a cured infection has no use as a biomarker.

8 Herpesviruses

Herpesviruses encode the majority of known viral miRNAs, and many of them are associated with viral latency [41]. EBV encoded miR-BART2-5p, 13 and 15 miRNAs in plasma have been found to associate with high morbidity and mortality of chronic active EBV infection [42]. Strong upregulation of BART1-3p, 2-5p, 5, 6-5p, 6-3p, 7, 8, 9, 14, 17-5p, 18-5p and 19-3p miRNAs in serum, targeting genes involved in cancer metastasis such as E-cadherin, correlates with EBV copy numbers in nasopharyngeal carcinoma tissue [43]. This may also apply to other EBV-associated malignancies such as Burkitt’s lymphoma, Hodgkin’s lymphoma, or gastric carcinoma. EBV-encoded miRNAs are selectively enriched in exosomes from nasopharyngeal carcinoma cells [44], and exosomal EBV miRNAs from infected lymphocytes have been shown to exert their functions in noninfected cells [45]. Human miR-155 induction in B cells in EBV type III latency and high expression level in Hodgkin’s, primary mediastinal and diffuse large cell lymphomas suggests involvement and functional importance in malignancies associated with EBV [46].

Host cell encoded miR-132 is strongly upregulated in endothelial cells infected by KSHV. miR-132 may mechanistically link inhibition of antiviral innate immune responses and induction of abnormal endothelial cell proliferation, both essential phenomena in Kaposi’s sarcoma [47].

In addition to straightforward antigen or nucleic acid detection of VZV, biomarkers to differentiate infections prone to cause serious complications such as pneumonia, central nervous system (CNS) complications such as encephalitis, or secondary bacterial infections are needed. In a pilot study, a putative marker panel miR-197, miR-629, miR-363, miR-132, and miR-122 was identified, although no careful clinical validation was carried out as yet [48]. For the time being VZV has not been reported to encoded miRNAs of its own.

9 Hepatitis B and C Viruses

Putative miRNA biomarkers among viral diseases have been most thoroughly studied for chronic diseases and liver cancer associated with HBV and hepatitis C (HCV) viruses, where new biomarkers could greatly contribute to current diagnostic methods and prognosis of disease course [49]. One HBV-encoded miRNA has been validated but its function has not yet been established [19]. HCV-encoded miRNAs have not been reported.

Natural course of HBV infection is characterized by immune tolerant, immune active, and immune inactive stages. Conversion of active to inactive hepatitis coincides with hepatitis B envelope antigen (HBeAg) seroconversion to anti-HBe antibodies (HBeAb) followed by clinical remission and lifelong inactive stage. A panel of circulating plasma miRNAs with strong correlation to HBV DNA loads, surface antigen HBsAg, and the immune stage of pediatric HBV infection was identified. The levels of miR-99a-5p, miR-122-5p, miR-122-3p, and miR-125b-5p decreased significantly over time in immune tolerant and immune active children, reflecting poor immunological control [50, 51]. A serum MiR-B index based on the combined Cq values (a measure of amplification corresponding to Ct) of hepatocellular miR-122-5p, miR-99a-5p and miR-192-5p, with subtraction of the Cq values of three endogenous internal controls miR-126-3p, miR-335-5p and miR-320a, was created. MiR-B index was claimed to identify therapy-induced or spontaneous switch from active to inactive chronic HBV infection [52].

A panel of eight serum miRNAs was identified for differentiation or early diagnosis of HBV-related hepatocellular carcinoma (HCC) patients from healthy individuals (AUC or area under curve in ROC analysis = 0.893) and from cirrhosis patients (AUC = 0.879): miR-206, 141-3p, 433-3p, 1228-5p, 199a-5p, 122-5p, 192-5p, 26a-5p. The miRNA panel can be used in preventive screening and for timely start of therapy instead of liver biopsy, and it shows better accuracy than the widely used alpha-fetoprotein (AFP) [53]. Another panel of seven serum miRNAs was established with high accuracy in the diagnosis of early HBV-related HCC: upregulation of miR-192, miR-21, and miR-801, and downregulation of miR-122, miR-223, miR-26a and miR-27a. The panel showed high accuracy in differentiating HCC from healthy (AUC = 0.941), chronic hepatitis B (AUC = 0.842) and cirrhosis (AUC = 0.884) [54]. These panels have been validated in reasonably large patient materials and show true potential for clinical implementation (Table 1).

Table 1 Selected microRNA panels with potential in clinical laboratory diagnostics
Full size table

miR-122 is a liver-specific miRNA, which may constitute 70 % of the miRNA pool in the liver [55, 56], and it is required for efficient HCV replication [27, 57]. Interference of endogenous miR-122 function as a treatment for HCV has entered clinical trial as a first miRNA-based therapy in humans (see Sect. 15). Currently HCV RNA viral load is the gold standard in diagnosing HCV infection and in monitoring drug efficacy, although it is not an optimal correlate of disease severity or progression because it may remain stable for years. Studies on the use of serum or plasma miR-122 as a marker of HCV infection [58, 59], chronic hepatitis C patients [60] or treatment response [61] have as yet remained inconclusive. Upregulation of serum miR-320c has been established in HCV infection [62], as well as serum or plasma miR-20a and miR-92a in acute and chronic hepatitis C-infected patients, and a further increase of circulating miR-20a in progressing liver fibrosis [63]. Inverse correlation of circulating miR-92a and miR-122 with fibrosis stage, and a decrease in miR-92a levels in resolving HCV infection have also been shown [63, 64], suggesting that more research is needed to establish true significance.

Human miR-21 is one of the most frequently upregulated miRNAs in many cancers including HCC [65]. Serum miR-21 is a more sensitive biomarker than AFP, which is currently used for screening and early diagnosis of HCC and as a marker of recurrent HCC [66]. miR-155, miR-26a and miR-31 may function as prognostic markers for risk stratification of recurrence and hepatocarcinogenesis [67–69].

10 HIV and HTLV-1

T-cell counts and viral loads are established but suboptimal markers for disease progression in human immunodeficiency virus 1 (HIV-1) patients on highly active antiretroviral therapy (HAART). HIV-1 encodes a miRNA miR-N367, an ortholog of human miR-192, by which the virus regulates its own transcription and replication [20, 21], but no diagnostic potential has been established. Biomarker potential for disease progression and monitoring of resistance to therapy has been suggested for miR-150, which is downregulated in PBMC of HIV-infected individuals, returned to normal levels upon successful HAART, but downregulated again if resistance to treatment develops [70, 71].

Adult T-cell leukemia/lymphoma (ATL) is an aggressive tumor caused by human T-cell leukemia virus type 1 (HTLV-1) retrovirus. It is characterized by long clinical latency, very poor prognosis and resistance to chemotherapy, and monoclonal integration of HTLV-1 provirus. It was reported in a small group of patients using FFPE samples, that downregulation of miR-145 is significantly associated with poor overall survival of ATL patients and could thus be used as a prognostic marker [72].

11 Respiratory Viruses

Most respiratory viruses are RNA viruses and they are not expected to express miRNA. Adenoviruses are DNA viruses and they do encode miRNAs which are known to be vastly enriched in exosomes [17], but there are no data as yet concerning the role of miRNA in human adenovirus-associated diseases.

Some human miRNAs have been implicated in lethal influenza A cases [73, 74], providing important information on human response to viral infection. miR-21 and miR-223 were found considerably upregulated in lethal infections by 1918 pandemic H1N1 in mice or avian H5N1 in macaques, while modest or no upregulation was seen in infections by less pathogenic influenza A [73, 74]. MiR-1260, -26a, -335*, -576-3p, -628-3p and -664 were significantly regulated in blood and in throat swabs in H1N1 infection, and may also represent therapeutic targets [75].

12 Picornaviruses

Picornaviruses with single-stranded RNA genomes are not known to encode miRNAs. Severe conditions caused by picornaviruses include enterovirus-associated cardiomyopathy, which is mostly caused by Coxsackie virus B3 or other CVBs. Some affected individuals spontaneously clear the virus and recover, while in others the virus persists and the disease progresses to heart failure [76]. Predictive biomarkers have not been established in profiling of protein coding genes, but miRNA markers were identified in CVB3 cardiomyopathy patients when other possible causes of heart problems were carefully excluded [76]. A panel of eight miRNAs was established: miR-135b-5p, miR-155, miR-190a-5p, miR-422a, miR-489-3p, miR-590, miR-601, and miR-1290 were strongly induced in persistent CVB3 cardiomyopathy as compared to clearing disease or healthy hearts in endomyocardial biopsies taken at the baseline for setting the diagnosis. Induction of these miRNAs may reduce viral RNA recognition and innate immune response to virus, as suggested by their predicted target genes. The panel may represent a true predictive biomarker to assess the risk of virus persistence and progressive clinical deterioration (Table 1), as well as a putative drug target [76].

13 Polyomaviruses

Of the 13 human polyomaviruses known to date, BK polyomavirus (BKPyV or BKV), JC polyomavirus (JCPyV or JCV), and Merkel cell polyomavirus (MCPyV) encode 5p and 3p miRNAs, which are cleaved from one common precursor transcript [26]. The 3p miRNA of both BKPyV and JCPyV share identical sequence and thus its origin is not possible to specify. Polyomavirus miRNAs reduce host antiviral immune responses in an unusual way in cis to regulate the accumulation of viral gene products which would otherwise raise host immune responses, and some also in trans to downregulate the expression of host transcripts related to immune responses [77]. Polyomavirus-encoded miRNAs seem to contribute to the establishment of viral persistence in an exceptionally robust manner among miRNAs, partially due to perfect homology to their target sequences [24]. MCPyV miRNAs seem not to present suitable biomarkers, although they are crucial in the establishment of persistence [78, 79].

New tools are required for early diagnosis and monitoring of severe conditions associated with BKPyV among renal transplant recipients. A limited number of studies including our own pilot study [36] have explored polyomavirus-encoded miRNAs in human samples. Surprisingly frequent detection of BKPyV/JCPyV 3p miRNA suggests that it may serve as a marker of past exposure to one or both of these viruses, and may even be more sensitive than serology [36, 80]. Detection of bkv-miR-B1-5p is much less frequent and tends to associate with high viral loads in blood [36, 81, 82], which suggests that it could serve in monitoring of severe BKPyV disease. Human miR-155, which seems to be a key regulator of host CD8 + T-cell responses to viral infections, was found downregulated in BKPyV positive plasma samples in our setting of just a few samples [36], calling for further studies.

Progressive multifocal leukoencephalopathy (PML), a fatal brain disease, is caused by reactivation of latent JCPyV followed by lytic virus replication in the brain. A number of studies including our own recent study [83] suggest that PML is specifically caused by rearranged neurotropic viral strains, but past exposure to JCPyV is considered key for PML risk assessment at present. Very frequent detection of jcv-miR-J1-5p in plasma, urine and CSF samples from immunosuppressed and healthy individuals, both seropositive and seronegative, suggest biomarker potential for past infection [36, 84]. The presence of 5p miRNA might be a more sensitive marker for past exposure than serology based on VP1 ELISA [36, 84], and therefore the biomarker potential of jcv-miR-J1-5p should be thoroughly investigated.

14 Papillomaviruses

Human papillomaviruses (HPV) are double-stranded DNA viruses whose replication takes place largely in the nucleus, which would suggest that they could encode miRNAs of their own. In some previous work, no HPV-encoded miRNAs could be identified [85, 86, 87, 88], whereas in another study putative miRNAs were predicted [89]. We have identified several putative HPV-encoded miRNAs by massive parallel sequencing of small RNA libraries established from both human tissue samples and from HPV-containing cell lines, and performed their biological validation [18] (Virtanen et al., submitted). However, HPV-encoded miRNAs are expressed at too low levels for biomarker use.

High quality tests exist for HPV diagnostics but prognostic markers, preferably detectable in swabs to avoid the need for biopsy, would be valuable for the triage and prognosis among HPV-positive patients. For example, downregulation of miR-424/miR-375/miR-218 in exfoliated cells is associated with severe cervical disease [90, 91]. Downregulation of miR-143 in cervical cancer tissue is associated with tumor size, lymph node metastasis and HPV 16 infection [87, 92, 93]. miR-21 is upregulated in cervical cancer [87, 94–96] and the levels were even higher in radiation-resistant cancers [96]. Intriguingly, upregulated miR-21 levels could also be seen in serum of cancer patients, making this particular miRNA a tempting biomarker for cervical and other cancers.

Downregulation of miR-218 takes place in high-risk HPV-positive penile and oral cavity squamous cell carcinoma, and prognostic value has been suggested [97, 98].

Squamous cell cancers of the head and neck region (HNSCC) comprise two different disease entities of different patient groups: HPV-positive cancers with considerably better response to treatment and better prognosis, and HPV-negative cancers with bad prognosis, often among heavy smokers. Because HPV test sensitivities have been adjusted to detect cervical disease and their ability to detect HPV in HNSCC is unknown, sensitive markers for differential diagnostics are required. Overexpression of miR-363, resulting in decreased cell migration and invasion, has been established in HPV-positive HNSCC [90, 99]. Downregulation of miR-145 in HPV-positive as compared to HPV-negative HNSCC and association with aggressive progression and poor prognosis were shown, suggesting potential for prognostic biomarker; downregulation has also been shown in cervical cancer and precursor lesions [90, 92, 100–102]. Interestingly, significant association of miR-21 was shown with poor survival in HPV-negative but not in HPV-positive HNSCC [103].

15 MicroRNA-Based Drug Therapy

As master regulators of gene and protein expression, miRNAs hold great therapeutic promise in the control of viral infections and many human diseases. Blocking upregulated—or virus-induced—miRNA with antagomir, a synthetic oligonucleotide which inhibits binding of the natural miRNA to its target, is tempting and has proven technically feasible. Replacement of downregulated miRNA function using miRNA mimetic or mimic is more challenging and may lead to undesired and unspecific consequences. Other challenges are off-target effects of miRNA-based therapy due to multitude of miRNA targets, drug delivery to correct anatomical site, effective dose, and sustained effect of treatment. The clinical value of viral miRNA manipulation requires more exploration, although some studies report inhibition of virus replication in experimental systems lacking viral miRNA expression. The effects of antiviral drugs may also be mediated by regulation of cellular miRNA function. The field of miRNA-based therapies is the fastest developing field in genomic medicine.

Indeed, the first miRNA-targeting drug to enter human clinical trial is SPC3649 or miravirsen developed by Santaris Pharma A/S to treat chronic HCV infection. Miravirsen is a locked nucleic acid (LNA) modified oligonucleotide complementary to miR-122, a hepatocyte-specific miRNA required for efficient HCV replication in the liver. miR-122 binds to two sites in the 5′ UTR of HCV RNA leading to viral RNA stabilization and promotion of viral replication [24, 27], and its function is inhibited by miravirsen binding. Miravirsen is currently in Phase IIb clinical trial. It has shown remarkable safety, efficacy and tolerance, no sign of developing resistance, and continued efficacy in experimental animal models as compared to standard treatments [104, 105]. Further, miR-34 mimics (MRX34) to treat hepatocellular carcinoma and metastatic liver cancer are in clinical trial [106]. It is intriguing that miRNA has found its first drug applications among viral diseases in hepatocellular carcinoma.

16 Conclusions

MicroRNA has made its way from discovery up to clinical trials with unprecedented speed, and the diagnostic and particularly prognostic potential of miRNA markers should be thoroughly investigated. miRNA panels established for differentiation of HBV-associated hepatocellular carcinoma from other chronic HBV-associated diseases and from healthy individuals are examples of miRNA-based biomarkers closest to clinical application. Another promising application, importantly with predictive value, is a panel of human miRNAs for differential diagnostics of enterovirus-associated cardiomyopathy. miRNAs that are regulated in several diseases may prove feasible as biomarkers, but careful validation and interpretation of expression levels in different disease states and in different sample matrices is required. One example is miR-21, which is upregulated in liver cancer, cervical cancer and in many other cancers, and for example in lethal influenza A cases, and may even be detected in serum. The biomarker capacity of miR-155, a key player in host T-cell responses to many viral and bacterial pathogens, should be explored. Commonly regulated miRNAs may provide information on an underlying microbial infection or on the overall immune status of the host.

A novel biomarker or a new diagnostic method is valuable only if it improves patient management. For future clinical use of miRNA biomarkers proper standardization of laboratory methods is key. Also, specific algorithms of result interpretation have to be established for each laboratory approach, sample matrix, and disease. The assets of miRNA biomarkers that can be envisaged at present include the use of body fluids enabling more selective invasive sampling, earlier diagnosis, and improved precision in disease monitoring and prediction of disease course. The diagnostic possibilities provided by viral and cellular miRNAs should be exploited in their full capacity.