A study of microRNAs in silico and in vivo: diagnostic and therapeutic applications in cancer

Authors

Errata

This article is corrected by:

  1. Errata: Corrigendum Volume 276, Issue 23, 7201, Article first published online: 6 November 2009

S. A. Waldman, 132 South 10th Street, 1170 Main, Philadelphia, PA 19107, USA
Fax: +1 215 955 5681
Tel: +1 215 955 6086
E-mail: scott.waldman@jefferson.edu

Abstract

There is emerging evidence of the production in human tumors of abnormal levels of microRNAs (miRNAs), which have been assigned oncogenic and/or tumor-suppressor functions. While some miRNAs commonly exhibit altered amounts across tumors, more often, different tumor types produce unique patterns of miRNAs, related to their tissue of origin. The role of miRNAs in tumorigenesis underscores their value as mechanism-based therapeutic targets in cancer. Similarly, unique patterns of altered levels of miRNA production provide fingerprints that may serve as molecular biomarkers for tumor diagnosis, classification, prognosis of disease-specific outcomes and prediction of therapeutic responses.

Abbreviations
CLL

chronic lymphocytic leukemia

miRNA

microRNA

PTEN

phosphatase and tensin homolog

Cancer is a leading cause of mortality in the USA, with ∼ 25% of deaths attributable to neoplasia [1,2]. Worldwide, cancer-related global mortality follows only cardiovascular and infectious diseases [3]. In this context of expanded incidence and growing prevalence, clinical oncology is poised for unprecedented innovation. Through harnessing discoveries in disease pathobiology, increasingly propelled by the development of high-throughput technologies including genomics, proteomics and metabolomics, modern cancer biology offers previously unavailable diagnostic and therapeutic paradigms tailored to meet the needs of individuals and populations [4]. Transforming clinical management is predicated on translation of the new science into application of advanced markers and targets for personalized cancer prediction, prevention, diagnosis and treatment [4–6].

Indeed, the epigenetic, genetic and postgenetic circuits restricting cell destiny are becoming increasingly decoded, and their dysfunction is being linked to lineage-dependence underlying tumorigenesis [2,7]. Critical in cell-fate specification is the post-transcriptional regulation of gene expression by microRNAs (miRNAs) (Fig. 1) [8], which arise as transcripts from cognate genes in noncoding regions of chromosomes. miRNAs undergo nuclear and cytoplasmic processing [8,9], producing the targeting core of a multimeric complex by hybridizing with mRNA molecules resulting in their sequestration or degradation, thereby defining the genes available for lineage commitment [10,11]. This is the most recent addition to the hierarchical spectrum of molecular mechanisms defining nuclear–cytoplasmic information exchange [12] and forms the interface among transcriptional, translational and post-translational regulation [13] . Significantly, miRNAs represent a regulatory, rather than a structural, mechanism that co-ordinates normal gene expression and whose dysregulation underlies neoplastic transformation [8,10,11].

Figure 1.

 miRNA generation and gene regulation [9]. Mature miRNAs of about 22 nucleotides originate from primary miRNA (pri-miRNA) transcripts. Nuclear pri-miRNAs of hundreds to thousands of base pairs are converted into stem–loop precursors (pre-miRNA), of about 70 nucleotides, by Drosha, an RNase III endonuclease, and by Pasha, a homologue of the human DiGeorge syndrome critical region gene 8 (DGCR8). Precursor miRNAs (pre-miRNAs) undergo cytoplasmic translocation, which is mediated by exportin 5 in conjunction with Ran-GTP, and are subsequently processed into RNA duplexes of about 22 nucleotides by Dicer, an RNase III enzyme, and Loqacious (Loqs), a double-stranded RNA-binding-domain protein that is a homologue of the HIV transactivating response RNA-binding protein (TRBP). The functional strand of the miRNA duplex guides the RNA-induced silencing complex (RISC) to the mRNA target for translational repression or degradation. Figure reproduced from a previous publication [9].

miRNAs and cancer

The essential nature of this novel mechanism indelibly patterning gene expression in cell-lineage specification [8], in the context of the established model of cancer as a genetic disease in which pathobiology recapitulates cell and tissue ontogeny [14,15], naturally implicates miRNAs in neoplastic transformation. In fact, an altered level of miRNA production is a defining trait of tumorigenesis [16,17]. While the production of some miRNAs is universally altered in tumors, more often unique patterns of miRNA production reflect the lineage-dependence of tumors, relating to their tissues of origin [16–22]. Similarly, fundamental processes underlying tumorigenesis, including genomic instability, epigenetic dysregulation and alterations in the expression, or function, of regulatory proteins, directly alter the complement of miRNAs produced by cancer cells [8]. Additionally, miRNAs regulate key components integral to tumor initiation and progression, including tumor suppressors and oncogenes [8,17,23]. Furthermore, miRNA signatures are a more informative source for classification of tumor taxonomy than genomic profiling [16]. Moreover, miRNAs can serve as unique targets for diagnostic imaging in vivo for taxonomic classification of tumors [24]. The emerging role of miRNAs in neoplasia highlights their potential value as mechanism-based therapeutic targets and biomarkers for diagnosis, prognosis of disease-specific outcomes and prediction of therapeutic responses [25]. While there are numerous detailed reviews in this field, the purpose of this minireview was to provide, in overview, a summary of the potential application of miRNAs as diagnostic and therapeutic targets in cancer.

miRNAs as mechanism-based therapeutic targets in cancer

The case for miRNAs as tumor suppressors and oncogenes reflects their loss or gain, respectively, as a function of neoplastic transformation, their dysregulation in different tumors, the relevance of their mRNA targets to mechanisms underlying tumorigenesis and their ability to alter tumorigenesis directly in model cells and organisms (Fig. 2; Table 1) [8,26,27]. Typically, miRNAs that serve as oncogenes are present at high levels, which inhibit the transcription of genes encoding tumor suppressors. Conversely, tumor-suppressor miRNAs are present at low levels, resulting in the overexpression of transcripts encoded by oncogenes.

Figure 2.

 miRNA oncogenes and tumor suppressors [26]. (A) Normally, miRNA binding to target mRNA represses gene expression by blocking protein translation or inducing mRNA degradation, contributing to homeostasis of growth, proliferation, differentiation and apoptosis. (B) Reduced miRNA levels, reflecting defects at any stage of miRNA biogenesis (indicated by question marks), produce inappropriate expression of target oncoproteins (purple squares). The resulting defects in homeostasis increase proliferation, invasiveness or angiogenesis, or decrease the levels of apoptosis or differentiation, potentiating tumor formation. (C) Conversely, overexpression of an oncogenic miRNA eliminates the expression of tumor-suppressor genes (pink), leading to cancer progression. Increased levels of mature miRNA could reflect amplification of the miRNA gene, a constitutively active promoter, increased efficiency in miRNA processing or increased stability of the miRNA (indicated by question marks). ORF, open reading frame. Figure reproduced from a previous publication [26].

Table 1.   miRNAs in tumorigenesis. CLL, chronic lymphocytic leukemia; B-CLL, B cell CLL.
miRNAGene locusTumor typesGene targetsReferences
Suppressors
 mir-15a, 16-113q14CLL, prostate, mantle cell lymphoma, multiple myelomaBCL-2[28–32]
 let-7Eight clustersLung, gastricRAS[22,23,26,34,35]
Oncogenes
 mir-17 cluster13q31-32B-CLL, follicular lymphoma, mantle cell lymphoma, cutaneous B cell lymphoma, colon, lung, breast, pancreas, prostatePTEN
TGF-β RII
E2F1
[17,36–38,40–43]
 mir-2117q23.2Breast, colon, lung, prostate, gastric, endocrine pancreas, glioblastomas, leiomyomasPTEN
BCL-2
Tropomyosin I
[17,44–50,54]

miRNA tumor suppressors

The best characterized tumor-suppressor miRNAs are miR-15a and miR-16-1. B-cell chronic lymphocytic leukemia (CLL) is the most common adult leukemia in developed countries and is universally associated with the loss of chromosomal region 13q14 [8,27,28]. Within this deletion is a region of ∼ 30 kb in which miR-15a and miR-16-1 reside, which are lost in ∼ 70% of patients with CLL [29]. Similarly, the loss of chromosomal region 13q14, including miR-15a and miR-16-1, occurs in prostate cancer, mantle cell lymphoma and multiple myeloma [29,30]. Tumor suppression by miR-15a and miR-16-1, in part, reflects inhibition of the expression of the anti-apoptotic oncogenic protein Bcl-2, which is characteristically overexpressed in CLL, promoting the survival of leukemia cells [31]. Indeed, there is a reciprocal relationship between the expression of miR-15a and miR-16-1 and of Bcl-2, and the heterologous production of these miRNAs suppresses Bcl-2 levels [32]. Suppression is specifically mediated by complementary binding sites for those miRNAs in the 3′-UTR of the Bcl-2 transcript [32]. Furthermore, heterologous expression of miR-15a and miR-16-1 produces apoptosis in leukemia cell lines [32]. Moreover, mouse models of spontaneous CLL possess a mutation in the 3′-UTR of miR-16-1 that is identical to mutations in patients with CLL and associated with decreased production of that miRNA [33]. Heterologous expression of miR-16-1 in CLL cells derived from those mice alters the cell cycle, proliferation and apoptosis of these tumor cells [33].

The miRNA, let-7, a phylogenetically conserved gene product that regulates the transition of cells from proliferation to differentiation in invertebrates [34], also serves as a tumor suppressor [27]. There are 12 let-7 homologs in humans, forming eight distinct clusters of which four are localized to chromosomal regions lost in many malignancies [35]. In that context, the down-regulation of let-7 family members in lung cancer is associated with poor prognosis [22]. A role for these miRNAs in growth regulation and in the expression of the tumorigenic phenotype is highlighted by the ability of heterologous let-7 expression in lung cancer cells in vitro to inhibit colony formation [36]. Key downstream targets for let-7 include the human Ras family of proteins, oncogenes that are commonly mutated in many human tumors [23]. Indeed, KRas and NRas expression in human cells is regulated by let-7 family members [27]. Moreover, loss of let-7 expression in human tumors correlates with the overexpression of Ras proteins [23].

miRNA oncogenes

The miR-17 cluster comprises a group of six miRNAs (miR-17-5p, miR-18a, miR-19a, miR-20a, miR-19b-1 and miR-92) at 13q31–32, a chromosomal region amplified in large B-cell lymphoma, follicular lymphoma, mantle cell lymphoma and primary cutaneous B-cell lymphoma [37]. Consistent with their functions as oncogenes, overexpression of this miRNA cluster is associated with amplification of the 13q31–32 genomic region in lymphoma cells in vitro [37,38]. These miRNAs are overexpressed in many types of tumors, including lymphoma, colon, lung, breast, pancreas and prostate [17,38,39]. Interestingly, expression of the miR-17 cluster is induced by c-Myc, an oncogene overexpressed in many tumors. Heterologous expression of c-Myc up-regulates expression of the miR-17 cluster, mediated by direct binding of that transcription factor to the chromosomal region harboring those miRNAs [40]. In turn, the miR-17 cluster appears to regulate several downstream oncogene targets. Thus, miR-19a and miR-19b may regulate phosphatase and tensin homolog (PTEN), a tumor suppressor with a broad mechanistic role in human tumorigenesis, through interactions with complementary sites in the 3′-UTR of this transcript [41]. Similarly, miR-20a may reduce the expression of transforming growth factor-β receptor II, a tumor suppressor frequently mutated in cancer cells and which regulates the cell cycle, imposing growth inhibition [17]. The best-characterized target of the miR-17 cluster is the E2F1 transcription factor whose expression is regulated by miR-17–5p and miR-20a [42]. In turn, E2F1 regulates cell cycle progression by inducing genes mediating DNA replication and cell cycle control [43]. Beyond the regulation of key targets contributing to transformation, the miR-17 cluster directly induces the tumorigenic phenotype. Heterologous expression of the miR-17 cluster increased proliferation in lung cancer cells in vitro [39]. Moreover, components of this cluster accelerate the process of lymphomagenesis in mice [44].

The miRNA miR-21 is overexpressed in many solid tumors, including breast, colon, lung, prostate and stomach, and in endocrine pancreas tumors, glioblastomas and uterine leiomyomas [17,45–47]. This miRNA is encoded at chromosome 17q23.2, a genetic locus that is frequently amplified in many tumors. The tumorigenic effects of miR-21 are mediated, in part, by targeting a number of mediators in critical cell-survival pathways. Thus, in glioblastoma cells in vitro, miR-21 modulates the expression of the common tumor suppressor PTEN, a central regulator of cell growth, proliferation and survival, which is mediated by the phosphatidylinositol3-kinase/Akt pathway [48]. Also, miR-21 regulates breast cancer cell growth by reciprocally regulating apoptosis and proliferation, in part reflecting regulation of the anti-apoptotic protein, Bcl-2 [49]. Moreover, miR-21 controls expression of the tumor suppressor tropomyosin 1, whose overexpression in breast cancer cells suppresses anchorage-independent growth [50]. Beyond signaling analyses, elimination of miR-21 expression in glioblastoma cells induces caspase-dependent apoptosis, underscoring the importance of this miRNA in mediating the survival phenotype [51]. Similarly, antisense oligonucleotides to miR-21 suppress the growth of breast cancer cells in vitro and in xenografts in mice [48].

miRNAs as biomarkers in cancer

Their fundamental role in development and differentiation, and their pervasive corruption in lineage-dependent mechanisms underlying tumorigenesis, suggest that miRNAs may be a particularly rich source of diagnostic, prognostic and predictive information as biomarkers in cancer [8,26,52]. Differential production of miRNAs compared with their normal adjacent tissue counterparts is a characteristic of every type of tumor examined to date [8,52], a feature that could be particularly useful in diagnosing incident cancers in otherwise normal tissues. Indeed, this approach discriminates normal and neoplastic tissues in various cancer types, including CLL, breast cancer, glioblastoma, thyroid papillary carcinoma, hepatocellular carcinoma, lung cancer, colon cancer and endocrine pancreatic tumors [8,17–22,26,45,52–54]. Similarly, miRNA expression profiles provide a powerful source of molecular taxonomic information, with an accuracy for classifying tumors according to their developmental lineage and differentiation state that surpasses mRNA expression profiling [16,17]. These observations suggest the utility of miRNA expression profiling for identifying metastatic tumors of unknown origin, which represent ∼ 5% of all malignancies worldwide [16,17,52]. Also, differential miRNA expression patterns are associated with disease prognosis [8,52]. Specific patterns of miRNA expression identified patients with pancreatic cancer who survived for longer than 24 months, compared with those who survived for less than 24 months [53]. In addition, the expression of specific miRNAs predicted overall poor survival in patients with pancreatic cancer [53]. Similarly, overexpression of specific miRNAs was an independent prognostic variable associated with advanced disease stage and decreased survival in patients with colon cancer [54]. Beyond diagnosis and prognosis, miRNA expression patterns predict responses to therapy, and overexpression of oncogenic miRNAs was associated with improved survival following adjuvant chemotherapy in patients with colon cancer [54]. These observations highlight the potential of miRNAs as biomarkers for diagnosis, taxonomic classification, prognosis, risk stratification and prediction of therapeutic responses in patients with cancer.

Corruption of miRNA expression in cancer

The genetic basis of cancer, in part, reflects chromosomal re-arrangements encompassing translocations, deletions, amplifications and exogenous episomal integrations that alter gene expression. The essential role of miRNAs in tumorigenesis predicts coincidence between the location of their encoding genes and those cancer-associated chromosomal regions. Indeed, more than half of the miRNA genes are located in cancer-associated genomic regions in a wide array of tumors, including lung, breast, ovarian, colon, gastric, liver, leukemia and lymphoma [28,35]. Conversely, chromosomal regions harboring miRNAs are sites of frequent genomic alterations involved in cancer [28,55]. Additionally, the impact of chromosomal remodeling on gene copy number directly translates to altered miRNA expression [19,28,55]. Beyond structural re-organization, epigenetic remodeling of chromosomal regions harboring miRNA loci contributes to transformation, and tumor-suppressing miRNAs silenced by CpG island hypermethylation result in the dysregulation of essential proteins responsible for accelerating the cell cycle, including cyclin D and retinoblastoma [56,57]. Moreover, alterations in the machinery responsible for processing miRNA contributes to tumorigenesis, and impairment of Dicer enhances lung tumor development in experimental mouse models and is associated with poor prognosis in patients with lung cancer [58–60].

Therapeutic targeting of miRNAs

The causal role of miRNAs in molecular mechanisms underlying transformation, and the contribution of specific miRNA species to lineage-dependent tumorigenesis, suggest that these molecules could serve as therapeutic targets in the prevention and treatment of cancer [61]. In the context of established therapeutic paradigms in medical oncology, individualized therapy with miRNAs could re-establish the expression of silenced miRNA tumor suppressors, whereas antisense oligonucleotides could silence overexpressed oncogenic miRNAs [8,28,52,61]. Indeed, antisense oligonucleotides (with modified RNA backbone chemistry resistant to nuclease degradation) targeted to miRNA sequences irreversibly eliminate the overexpression of oncogenic miRNAs [61]. Similarly, locked nucleic acid analogs resist degradation and stabilize the miRNA target–antisense duplex required for silencing [62]. Moreover, single-stranded RNA molecules (termed antagomirs), complementary to oncogenic miRNAs, silence miRNA expression in mouse models in vivo [63]. The specificity of targeting inherent in nucleic acid base complementarity, coupled with their mechanistic role in neoplastic transformation, make miRNAs attractive therapeutic targets for future translation.

Summary

miRNAs represent one fundamental element of the integrated regulation of gene expression underlying nuclear–cytoplasmic communication. Disruption of these regulatory components in processes underlying tumor initiation and promotion contributes to the genetic basis of neoplasia. Beyond molecular mechanisms underlying pathophysiology that constitute therapeutic targets, unique patterns of miRNA expression characterizing lineage-dependent tumorigenesis offer unique opportunities to develop biomarkers for diagnostic, prognostic and predictive management of cancer. These novel discoveries are positioned to launch a transformative continuum, linking innovation to patient management. Advancement of these novel paradigm-shifting concepts into patient application will proceed through development and regulatory approval to establish the evidence basis for integration of miRNA-based diagnostics and therapeutics into clinical practice.

Acknowledgements

The authors are supported by grants from the NIH (SAW, AT), Targeted Diagnostic and Therapeutics, Inc. (SAW), and the Marriott Foundation (AT). SAW is the Samuel M. V. Hamilton Endowed Professor of Thomas Jefferson University. AT is the Marriott Family Professor of Cardiovascular Research at the Mayo Clinic. SAW is a paid consultant to Merck.

Ancillary