The miR-15/107 group of microRNA genes: evolutionary biology, cellular functions, and roles in human diseases

Abstract

The miR-15/107 group of microRNA (miRNA) genes is increasingly appreciated to serve key functions in humans. These miRNAs regulate gene expression involved in cell division, metabolism, stress response, and angiogenesis in vertebrate species. The miR-15/107 group has also been implicated in human cancers, cardiovascular disease, and neurodegenerative diseases including Alzheimer’s disease. Here, we provide an overview of (1) the evolution of miR-15/107 group member genes, (2) the expression levels of the miRNAs in mammalian tissues, (3) evidence for overlapping gene regulatory functions by the different miRNAs, (4) the normal biochemical pathways regulated by miR-15/107 group miRNAs, and (5) the roles played by these miRNAs in human diseases. Membership in this group is defined on the basis of sequence similarity near the mature miRNAs’ 5′ end: all include the sequence AGCAGC. Phylogeny of this group of miRNAs is incomplete so a definitive taxonomic classification (for example, designation as a “superfamily”) is currently not possible. While all vertebrates studied to date express miR-15a, -15b, -16, -103, and -107, mammals alone are known to express miR-195, -424, -497, -503, and -646. Multiple different miRNAs in the miR-15/107 group are expressed at moderate-to-high levels in human tissues. We present data on the expression of all known miR-15/107 group members in human cerebral cortical gray and white matter using new miRNA profiling microarrays. There is extensive overlap in the mRNAs targeted by miR-15/107 group members. We show new data from cultured H4 cancer cells that demonstrate similarities in mRNAs targeted by miR-16 and miR-103, and also support the importance of the mature miRNAs’ 5′ seed region in mRNA target recognition. In conclusion, the miR-15/107 group of miRNA genes is a fascinating topic of study for evolutionary biologists, miRNA biochemists, and clinically-oriented translational researchers alike.

Introduction

Here we describe a group of microRNA (miRNA) genes with strong influences on human biology in health and disease: the miR-15/107 gene group. Basic information about miRNA genomics, biogenesis, processing, and mRNA target recognition will not be described here but many other sources are available^{1; 2; 3; 4; 5; 6; 7; 8}. New data will be presented that focuses on specific biological aspects of miR-15/107 gene group members including the expression of individual miRNAs in human brain and their functional overlap in cultured cells. To provide introductory information about these homologous genes, we will first review prior studies related to miRNA evolutionary biology.

Heterogeneity and evolution of the miR-15/107 group of miRNA genes

The 5′ end-portion of miRNAs is particularly important determinant of miRNA function ^{9; 10; 11; 12}, and a group of vertebrate miRNAs shows 5′ end sequence similarity and also many common targets. There are 12 annotated human genes in the miRBase miRNA Registry¹³ with the sequence AGCAGC starting at either the first or second nucleotide (nt) from the 5′ end of the mature (~22nt single-stranded) miRNA, a motif which we refer to as AGCx2. The human AGCx2 miRNAs are shown in Figure 1, along with a diagram that demonstrates how the mature miRNAs cluster hierarchically according to sequence similarity. Half of the AGCx2 human genes (hsa-miR-15a¹⁴, hsa-miR-16-1¹⁴, hsa-miR-16-2¹⁵, hsa-miR-103-1¹⁵, hsa-miR-103-2¹⁵, and hsa-miR-107¹⁵) were originally identified based on studies in cultured HeLa cervical cancer cells¹⁶. The remaining six AGCx2 miRNA genes were identified in other human-derived cancer cell lines (hsa-miR-424¹⁷, and hsa-miR-646¹⁸) or from non-neoplastic tissues (hsa-miR-15b¹⁹, hsa-miR-195²⁰, hsa-miR-497²¹, and hsa-miR-503^{21; 22}).

Figure 1.

Members of the miR-15/107 miRNA gene group that are known to be expressed in humans. The list of genes (top) shows the 5′ AGCAGC sequence highlighted in black. The 5′ portion of mature miRNAs is known to confer specificity in terms of regulated mRNA targets. Note that the AGCAGC sequence starts at the first nucleotide (nt) of miR-103 and miR-107 but the second nt for the other genes in the group (*). Also note, most of the miRNAs highly-expressed in humans incorporate a 7-nt common sequence AGCAGCA (#). The sequences of miR-15/107 group members were compared using a web-based alignment tool (GeneBee; http://www.genebee.msu.su/services/phtree_reduced.html using the “Cluster” algorithm) that provide an unbiased graphic illustration of their sequence similarity (bottom).

Reconstructing the functional evolution of medically-relevant genes can help to predict the clinical consequences of the sequence variation we observe. However, we are unable to clearly delineate a taxonomy for the AGCx2 miRNA genes. A gene family is a group of genes with a common phylogenetic origin and possible functional homology^{23; 24}. As a practical matter, the reconstruction of gene phylogenies is nearly always based on incomplete information. The miRBase miRNA Registry²⁵ provides the most definitive miRNA annotation, yet this miRNA list is a work in progress with additional miRNAs still unidentified. Compounding the problem of missing data, there is no consensus about how miRNA gene families (as well as sub-families and superfamilies) should be defined. Current miRNA classifications recognize distinct miR-15, miR-16, miR-103, and miR-107 families ^{26; 27; 28; 29; 30; 31}.

The known phylogenetic distribution of AGCx2 miRNAs suggests that they are unique to deuterostomes (a superphylum of animals that includes chordates but not insects or worms), and the number of AGCx2 genes has increased dramatically within the mammals^{32; 33; 34}. No such miRNA has been described in a plant or non-chordate animal (Figure 2). Seven of the 12 AGCx2 miRNAs expressed in humans are widespread within the vertebrata, having been identified in other primates, rodents, birds, amphibians, and bony fishes. The other five are apparently mammalian specific, and miR-646 appears in humans and chimpanzees but has not been found in Rhesus monkeys. MiR-322 has only been identified in mammals other than humans, so will not be discussed here. Note that all twelve of the human AGCx2 miRNA genes are associated with protein-coding genes, or transcribed pseudogenes, compared to ~50% of human miRNA genes that reside in intergenic regions^{35; 36}. As can happen with paralogs that arise via gene duplication, some genes are paired genomically in tandem: miR-15 paralogs with miR-16 paralogs, miR-424 with miR-503, and miR-497 with miR-195; these syntenies are evolutionarily conserved.

Figure 2.

Human genomic context including chromosome (Chr) and evolutionary conservation for the miR-15/107 gene group between human and Pan troglodytes (Chimpanzee; Chp), Macacca mulatta (Rhesus; Rhes), Mus musculus (Mouse; Mse), Xenopus tropicalis (Frog), Gallus gallus (Chicken; Chick), Danio rerio (Zebrafish; Fish), Ciona intestinalis (Sea squirt; Ssq), Drosophila melanogaster (Fruitfly; Fly), and Caenorhabditis elegans (Worm; Wrm). MiR-15/107 gene group members are only expressed in chordates, and tend to be located within or near a protein-coding gene or transcribed pseudogene. Mir-103 and miR-107 are entirely conserved in vertebrates whereas miR-15 and miR-16 show more variability. A number of miR-15/107 gene group members are mammalian specific and miR-646 is only known to be shared between humans and chimpanzees. Source: miRBase miRNA registry ¹³.

According to a recent phylogenetic analysis, miR-103 and miR-107 originated in the common ancestor of deuterostomes; Peterson and co-workers define a miR-103 superfamily that includes these two genes along with miR-2013³⁰. MiR-103 and miR-107 paralogs have been stable across vertebrate genomes, residing almost always in pantothenate kinase (PANK) gene introns. One exception: in the frog, Xenopus tropicalis, miR-103 is in a PANK gene intron but miR-107 is located intergenically. PANK genes encode polypeptides that are important in metabolism-related cell functions³⁷ (see below). Intronic miRNAs tend to be co-expressed with their protein-coding ‘mother’ genes although this is not always the case^{38; 39; 40}.

A miR-15 superfamily, comprising miR-15a, miR-15b, miR-16, miR-195, miR-322, miR-424, miR-457, and miR-497 are hypothesized to have evolved in the common ancestor of vertebrates ^{32; 33; 34}. There are two miR-15/16 clusters in mammals. Many vertebrate species have three or more paralogs of miR-15 and miR-16 as shown in Figure 2. The location of miR-15a/16 in the DLEU2 (“deleted in leukemia 2”, a non-coding transcript that affects cell proliferation^{41; 42}) gene seems to have evolved more recently. It is tempting to speculate that the physical location of miR-15/16 (located within chromatin/DNA and cell proliferation-related genes SMC4 and DLEU2) and miR-103/107 (located within cellular metabolism-related PANK genes) along with their known overlap in targets (see below), provide clues that these genes serve functional roles linked to cross-talk between the key cellular functions of cell division and metabolism.

Cephalochordates are generally regarded as the sister group of vertebrates ⁶, and the model cephalochordate, Branchiostoma floridae, expresses four miRNAs with 5′-end AGCAGC motifs^{6; 43} (Figure 3). The B. floridae mir-2013 is most similar to human miR-107, with 17/23 (74%) of the mature miRNAs’ nts matching exactly, including the 5′-end eight nts. A comparison between the different pre-miRNAs in Danio rerio and B. floridae show that the zebrafish miR-103 and miR-107 cluster in an intermediate position between the B.floridae pre-miRNAs and the zebrafish miR-15/16 pre-miRNAs ⁴⁴.

Figure 3.

The miR-15/107 gene group is chordate-specific but exact phylogenetic relationships are elusive A. Table compares some features of Drosophila melanogaster (Fruitfly; D.mel), the amphioxus cephalochordate Branchiostoma floridae (Lancelet; B.fl), Danio rerio (Zebrafish; D.rer), and Homo sapiens (H.sap). Note B. floridae, a relatively close evolutionary relative of vertebrates, does express four miRNAs with 5′AGCAGC sequence. B. A phylogenetic tree was created to evaluate the similarities in the sequences of the ~70 nucleotide (nt) pre-miRNA sequences of B. floridae and D.rerio, the former a relatively late-evolving pre-vertebrate species and the latter a relatively early-evolving vertebrate species. This phylogenetic tree shows that the miR-103 and miR-107 genes have sequences with intermediate similarity between the B. floridae pre-miRNAs and the other members of the gene group expressed in D.rerio. C. In addition to the apparent homology of the pre-miRNAs, the sequence of the mature miR-107 has the closest similarity to mature B. floridae miRNAs. Note that between vertebrates’ miR-107 and B. floridae bfl-mir-2013 17/23 (74%) of the mature miRNAs’ nts matching exactly.

Urochordates are generally regarded as the sister group to vertebrates plus cephalochordates. The sea squirt Ciona intestinalis appears to have lost (or never expressed) miR-103 and miR-107 genes. C. intestinalis has only been shown to express a single AGCx2 miRNA, which resembles miR-15 and resides in an intron of an SMC4 (“structural maintenance of chromosomes 4”) ortholog. The genomic location in C. intestinalis SMC4 is significant because as stated above, a miR-15 ortholog also is expressed in a SMC4 intron in vertebrates. SMC4 is an essential, evolutionarily-conserved ATPase that participates in DNA/chromatin dynamics^{45; 46}. More distantly-related species such as Drosophila and Caenorhabditis express no AGCx2 miRNAs.

We conclude that the origins of the miR-15 and miR-107 genes remain an unsolved puzzle for now. We cannot reject definitively the possibility that the miR-15 superfamily members had an independent origin in an early vertebrate ancestor, and they have converged in some respects with a separately-evolved miR-107 superfamily. However, it also is possible that the miR-15 superfamily and the miR-107 superfamily evolved from a common ancestor possessing a 5′-end AGCAGC sequence. If so, these miRNAs may exhibit functions that have been retained from their common ancestor and it would be appropriate to recognize the “miR-15/107 gene superfamily.” The existence of functional overlap between human miR-15/-16 and miR-103/-107 genes lends credibility to this hypothesis (see below). In either case, insights into the biology of the ancestral member of this group of genes could be gleaned from functional analyses of miRNAs in non-vertebrate deuterostomes. Other insights can be gained by studying the function and expression of these intriguing miRNAs in mammals.

Expression of the miR-15/107 group in mammalian tissues

MiRNA profiling experiments have shown that miR-15/107 group miRNAs are expressed at moderate-to-high levels in many mammalian tissues. Here we are distilling a large, sometimes contradictory literature and readers are encouraged to refer to the primary sources for more detailed information. There are technical challenges in assessing these genes’ expression levels, and a wide range of methods were used in preparing and profiling the miRNAs in the various studies. Further, in many miRNA profiling platforms there is cross-reactions in terms of closely-paralogous mature miRNAs such as miR-103 and miR-107.

Perhaps due to the abovementioned technical factors, there has been inconsistency among prior studies in terms of miR-15/107 gene group member expression profiling in mammals. Figure 4 shows results from a subset of prior studies that evaluated miRNA expression in multiple mammalian tissues including heart, skeletal muscle, brain, lung, liver, kidney, spleen, and placenta^{39; 47; 48; 49; 50; 51; 52}. There may be similar expression profiles among some group members across human tissues (miR-15, miR-16, and miR-195) ³⁹, but this is not always seen. Some but not all studies have indicated that miR-15/miR-16 expression is relatively high in hematopoietic cells including T lymphocytes^{47; 48; 51; 53; 54; 55; 56}. Published data is also discrepant as to whether miR-103/7 levels are highest in brain^{39; 48; 49; 52; 56; 57} and whether miR-195 is expressed at highest levels in lung^{47; 56}. Nor is there a clear-cut consensus for the miR-15/107 group member(s) that tend to have highest expression. The complexity of this situation is illustrated by Ason et al⁵⁸ who used in situ hybridization to study embryos from different species, finding that miR-107 is essentially brain-specific in the Japanese killifish, highly expressed in almost all tissues in chicken, but neither pattern is observed in mouse. Bargaje et al⁵¹ identified a subset of 18 miRNAs that are “constitutively” expressed in all mammalian tissues. From the miR-15/107 group, only miR-103 and miR-107 were on this list, which may indicate that the other miRNAs have more specialized functions. Distilling all these observations, the miR-15/107 gene group of miRNAs can be described as ubiquitous in that as far as we know there has not been a human cell type described that lacks expression of at least one miR-15/107 gene group member.

Figure 4.

Expression profiling of selected miR-15/107 gene group members from prior studies in human and mouse tissues. Profiling was performed from heart (Ht), skeletal muscle (M), brain (Br), liver (Li), lung (Lu), kidney (Ki), spleen (Sp), and placenta (Pl). The amount of miRNAs expressed is displayed in a semi-quantitative way from negligible/nondetected to very high expression. White squares indicate that the miRNAs were not evaluated in that tissue in the cited study. A variety of different profiling platforms is represented: microarray (MA), northern blots (NB), PCR, and a meta-analysis of other profiling studies (Meta). The semi-quantitative scoring reflects a subjective interpretation from the primary sources cited in the right-hand column. The results could not be mapped in a purely quantitative manner due to the nature of the data (for example, NBs yield no quantitative readout), known differences in dynamic ranges, and differential specificities to individual miRNAs. Note that there is no obvious tissue-specific pattern of expression that has been found across the different studies. However, the multiple miR-15/107 gene group members are consistently found to be expressed in multiple mammalian tissues.

In order to include some of the more recently annotated miR-15/107 group genes in our discussion, the results of recent expression profiling of miR-15/107 group members in human brain samples is displayed in Figure 5 (superior and middle temporal gyri, N=4 each from nondemented individuals older than age 80). These were evaluated using a locked nucleic acid (LNA)-based microarray and here raw data numbers from the microarrays are described. The results were different for gray matter and white matter -- the latter contains more oligodendrocytes and fewer neurons. Note that miR-16 is the highest-expressing miRNA according to this platform and is expressed at highest levels in white matter. Using a different microarray platform we found that miR-107 was the highest expressing group member in similarly processed brain tissue ⁵⁹; this discrepancy is not surprising. Researchers from many laboratories have now found that miRNA profiling results obtained using different platforms can vary dramatically for individual miRNAs despite good overall inter-platform correlation^{60; 61; 62; 63; 64}. Notwithstanding these technical issues, an interesting correlation is observed: the miRNAs apparently expressed at highest levels in human tissues (miR-15a, miR-15b, miR-16, miR-103, miR-107, and miR-195) include a 7-nt common sequence: AGCAGCA (see Figure 1).

Figure 5.

Raw microarray data from mammalian brain experiments help illustrate particular aspects of miR-15/107 gene group member expression. A. Results of a new experiment using the LNA-microarray including the relatively recently characterized genes miR-497, miR-503, miR-424, and miR-646. These data include readings from four aged nondemented persons and the RNA was isolated from the human superior and middle temporal gyri (Brodmann Areas 21 and 22). White matter and gray matter were dissected and analyzed separately. Note that all the highly-expressed miRNAs in mammals include the sequence AGCAGCA at the 5′ end (see Figure 1). Also note that the gray matter and white matter have different expression for some miRNAs (especially miR-15a, miR-15b, miR-16 which are more highly expressed in white matter). B. Microarray results at different developmental time points from the microarray study by Miska et al ⁷⁰. Mouse brains were studied at embryonic day 12.5 and 17.5, postnatal days 4 and 18, and adults. Note that most members of the miR-15/107 gene group had highest expression at some pre-adult developmental stage. The data for miR-103 and miR-195 (as shown with *) are averages of two different results.

In addition to tissue- and cell-specific factors, developmental stages constitute another layer of gene expression complexity. Recall that animal miRNAs were discovered in the context of stage-specific expression during worm development^{65; 66; 67}. There are altered expressions of miR-15/107 group members in certain stages of mammalian development also. For example, miR-15 and miR-16 expression correlates to particular stages in erythropoiesis^{26; 68}. Assessing archival human brains that had been formalin-fixed and paraffin-embedded we found that miR-16 expression is highly expressed in early human brain development ⁶⁹. Figure 5B shows the results from Miska et al⁷⁰, which provides compelling evidence that multiple miR-15/107 group members have increased expression levels during early mouse brain development.

We conclude the following about the expression of miR-15/107 group members in mammalian tissues:

• Expression of multiple miR-15/107 group miRNAs are moderate-to-high across many tissues including cells derived from mesoderm, ectoderm, and endoderm;

• Gene expression among group members tends not to be purely specific to particular tissue or cell types;

• MiR-15/107 group gene expression is relatively high in some early developmental stages;

• In humans the highest-expressing miRNAs tend to be miR-15a, miR-15b, miR-16, miR-103, and miR-107 (all of which contain the sequence AGCAGCA), with the caveat that some miRNAs were not assessed in earlier studies. Thus, there are important similarities between the highly-expressed miR-15/107 group members.

MiR-15/107 group miRNA processing and mRNA targeting

Clues related to the evolution and gene expression of miR-15/107 group members support the idea of a homologous gene group, but what about the functions of the mature miRNAs? There is compelling evidence that different members of the miR-15/107 group regulate overlapping lists of mRNA targets. This is presumably because the AGCAGC(A) sequence near the miRNAs’ 5′ end controls the mRNA target specificity. Some studies have focused on the mRNA targeting overlap between the “miR-15/16 group” paralogs such as miR-15a, miR-15b, miR-16, miR-195, miR-424, and miR-503^{71; 72; 73}. Yet there is also functional overlap even in the more dissimilar group members, miR-103 and miR-107^{74; 75}. In a rigorous and systematic analysis, Linsley et al²⁷ performed mRNA-detecting microarray studies after miRNA transfections in cultured cancer cell lines. They concluded that “(miR-15a, miR-16, and miR-103) are members of a single miRNA group”. Linsley et al²⁷ also note that transfections with “miR-15a, miR-16, and miR-195 also gave nearly identical expression profiles, but miR-103 gave an overlapping but distinct profile compared to other members of this group, consistent with the 1-nt offset of the miR-103 seed sequence.” More specifically: “Only a subset of [consensus miR-16 responsive] transcripts was regulated by miR-103 and miR-107. These transcripts were not regulated in cells transfected with miR-16 duplexes having mismatches at positions 2 and 3 or 4 and 5, but their regulation was unaltered in cells transfected with duplexes with mismatches at positions 18 and 19 and 19 and 20.” ²⁷

We compared experimentally the effects of transfections using different miR-15/107 group members in cultured cancer cells (Figure 6). Studies were performed using H4 “glioneuronal” cells using methods that we recently described in detail^{76; 77}. For these experiments, an individual mRNA was considered to be a target of miRNA X if the level of that mRNA was significantly reduced following a miRNA transfection with miRNA X, relative to the level of that mRNA after transfection with a Negative control miRNA (sequence shown in Figure 6). Three transfections were performed for each miRNA. We found that mRNAs that are knocked down after transfection by miR-107, miR-103, and miR-16 tend to overlap substantially. Note that in this experiment the correlation was best (in terms of mRNAs knocked down) between close paralogs miR-103 and miR-107, but even between miR-103 and miR-195 the correlation was better than to controls. MiR-15b* was used as a control miRNA because its sequence (roughly antisense to miR-15b) has similar G/C content. Two additional non-physiological controls (miR-107MUT1 and miR-107MUT2; see Figure 6) were used that have been described before⁷⁵. Inclusion of these controls helped to strongly support the importance of the miRNA 5′ seed portion in terms of dictating mRNA targeting specificity, because the miR-107MUT2, lacking the 5′ seed sequence, does not impact the same mRNAs as miR-103 or miR-16. It should be emphasized that these data describe mRNAs downregulated in lysates following miRNA transfections, rather than co-immunoprecipitated definitive “targets”, so some of these downregulated mRNAs may represent downstream effects. However, these results underscore the appreciable overlap between different miR-15/107 group members in terms of their biochemical impact.

Figure 6.

There is considerable overlap in the mRNAs targeted by various miR-15/107 gene group members in cultured H4 glioneuronal cancer cells. Transfections with “RIP-Chip” experimental design were performed as previously described in detail ⁷⁵; ⁷⁷. Briefly, H4 cells (American Type Culture Collection, Manassas, VA), were cultured under the vendor’s recommended conditions and plated at a density of 2.5×10⁶/10cm plate a day before transfections. Cells were transfected with 25 nM of exogenous RNA duplexes (Ambion, Austin, TX) with sequences shown on bottom, using RNAiMAX (Invitrogen, Carlsbad, CA) according to manufacturer’s instructions. Transfections with “miR-107MUT” artificial miRNAs was recently described⁷⁵. Cells were harvested 48 hr after transfection, the RNA was isolated, converted to cDNA and profiled using Affymetrix 1.0 ST Gene microarrays according to manufacturer’s instructions. A total of 21,898 annotated genes were profiled. Genes were evaluated in which mRNA levels were lower after miR-103 transfection (N=1484 genes; blue bars) or after miR-16 transfections (N= 945 genes; red bars) relative to the “Neg control” miRNA transfections (N=3 biological replicates for each transfection). Comparisons were performed using unpaired Student’s t-test with a P<0.01 cutoff. Note that a majority of genes knocked down by miR-103 were also knocked down by close paralog mir-107 and the miR-107MUT2 reagents, and also miR-16 and miR-195. Similarly, genes knocked down by miR-16 tended to also knocked down after miR-195 and somewhat less so with miR-103, miR-107, and miR-107MUT2 but not controls. These results strongly support the importance of the 5′ AGCAGC sequences in determining which mRNAs are targeted by genes in the miR-15/107 gene group.

Before discussing biological pathways and individual targets affected by miR-15/107 group genes, we note that miRNAs work through multiple distinct mechanisms^{78; 79; 80; 81; 82}. There are special considerations related to mRNA targeting by genes in the miR-15/107 group. For example, miR-107 may act preferentially by systematically targeting sequences in mRNAs’ open reading frame, as opposed to their 3′ untranslated region¹². It has been established that miRNAs may target the protein-coding sequence of target mRNAs⁸³, with ~25% of miRNA targets in mouse brain mRNA corresponding to protein-coding sequences according to high-throughput sequencing of RNA isolated by crosslinking immunoprecipitation (HITS-CLIP) data⁸⁴. However, miR-107 is the first example of an individual miRNA that may recognize protein-coding portions of target mRNAs systematically. The phenomenon (a focus of ongoing work in the Nelson laboratory) was discovered using RIP-Chip technique in H4 cells and will require confirmatory studies in other cells, tissues, and organisms.

Beyond mRNA targeting, there are additional levels of complexity that distinguish the miR-15/107 gene group. MiR-107 causes feedback inhibition of the entire miRNA processing machinery by repressing expression of Argonaute proteins with profound global changes in miRNA processing ⁷⁷. Martello et al reported that miR-107 also targets Dicer, the exonuclease that processes precursor miRNAs⁸⁵. The interplay between miR-107 and the miRNA-processing machinery is additionally complicated, because the miR-107 gene contains a specific sequence motif that controls how the precursor miRNA molecule is processed by Dicer, and thus how the mature miRNA is expressed. This novel mechanism, with special relevance to miR-107 and only a few other miRNAs, is associated with the uridylyl transferase protein “TUTase 4” (TUT4)⁸⁶. It is currently unknown whether these phenomena relate to the abovementioned tendency of miR-107 to target mRNAs’ protein-coding sequences. We are also only beginning to understand the complex mechanisms regulating transcription of genes in the miR-15/107 group^{42; 87; 88; 89}. In sum, the many unanswered questions pertaining to miR-15/107 group members’ biological chemistry are an important consideration as we begin to assess the kaleidoscopic extant data about the hypothesized functions of miR-15/107 gene group members.

The biochemical impact of a well-expressed mature miRNA is thought to relate to the “sum of its targets”--the aggregate mRNAs targeted weighted by the strength of each interaction. Computational predictive methods may provide information but these methods entail some shortcomings; computational algorithms differ amongst themselves and each is anchored in prior knowledge^{90; 91}. For a bench researcher wishing to prove that a particular miRNA targets a specific mRNA, the experimental strategies usually involve a combination of the following:

• miRNA transfection or over-expression→ down-regulation of target (mRNA and/or protein);

• In vivo or cell culture miRNA knockout/knockdown→ up-regulation of target (mRNA and/or protein);

• reporter assays with subcloned 3′UTR sequences help assess sequence specificity and confirm direct interaction;

• and target(s) co-purify with anti-Argonaute co-immunoprecipitation with or without RNA cross-linking.

Each of these methods entails both benefits and difficulties. In terms of potential pitfalls, the physiological relevance and specificity of cell culture miRNA perturbation techniques bears critical scrutiny due to the intrinsic artificiality of the experiments, and reporter assays tend to depend on mechanisms that are not identical to the target mRNAs’ biochemical contexts^{91; 92; 93; 94}. Co-immunoprecipitation methods entail other assumptions (such as the stability of the miRNA:mRNA particles) and may only capture a subset of miRNA:target interactions⁹¹. High-throughput methods such as HITS-CLIP and PAR-CLIP^{84; 95; 96} may be more indicative of the wide web of miRNA effects than lower-throughput tools because the -CLIP methods identify miRNA recognition sites transcriptome-wide. However, every current technique for miRNA target identification is at least somewhat biased by current dogma and there are probably many interactions yet to be identified. In sum, each and every experimental method – as with each computational method – includes potential limitations. These caveats are not mentioned out of pessimism but as a reminder that the field of miRNA biology is still very much in flux with many exciting discoveries ahead. In that spirit, we provide updates of various “works in progress” pertaining to miR-15/107 group members.

Biological functions of the miR-15/107 group

A. Cell division

Cellular pathways related to mitosis in vertebrate species are partly regulated by miRNAs in the miR-15/107 gene group. Most studies have focused on the impact of miR-15 and miR-16 paralogs on cell division, but other miR-15/107 group members have also been studied in this context. Linsley et al²⁷ described this group of miRNAs as the “miR-16 group” and noted that miR-15a and miR-16 (and to a lesser extent miR-103) target a highly disproportionate number of cell cycle genes and transfection with these miRNAs leads to G0/G1 arrest. Specifically, for miR-103, miR-15a, and miR-16, the percentage of targets that were involved in cell cycle regulation were 12%, 14%, and 22% respectively as determined using microarray data after miRNA transfection with correlation to GO process terms, and these percentages were much higher than would have been expected by chance²⁷. Other labs have confirmed that miR-15/107 group members (again usually focusing on miR-15 and miR-16) target cell cycle-related genes (and see below) ^{27; 28; 42; 97; 98; 99}. MiR-107 transfection also leads to decreased rates of cell division with cell cycle arrest ¹⁰⁰. The biological significance of miR-15/107 gene group-mediated cell cycle arrest is still not known but is an area of active enquiry partly because of the obvious implications in cancer biology as discussed below.

B. Cellular metabolism and stress

Numerous lines of evidence indicate that miR-15/107 group genes participate in pathways involved in cellular metabolism. Tang et al¹⁰¹ studied miRNA expression changes in response to glucose, using a cDNA array to test the effects of 20 mM pyruvate, 1 or 25 mM glucose in the cell medium of pancreatic β-cell line MIN6. A handful of miRNAs were up-regulated following increased extracellular glucose, including miR-107. We subsequently confirmed this effect of glucose on miR-107 expression using a nearly identical experimental design but with cultured H4 cells⁷⁵. MiR-107 is also up-regulated in cells exposed to folate deficiency¹⁰², and dietary intake of various lipids altered specifically miR-107 expression in the colonic epithelium and cancerous tissue from mice¹⁰³. Further, mice forced to perform endurance exercise had over 50% increased miR-107 expression in quadriceps femoris muscle tissue relative to a sedentary control group (this was the largest increase of any miRNA tested)¹⁰⁴.

There are other hints that the miR-15/107 group genes are involved in metabolic pathways. In vertebrate species, miR-103 and miR-107 paralogs reside in introns of genes that also encode PANK. PANK enzymes catalyze the universal rate-determining step in the formation of Coenzyme A (CoA). CoA participates in many metabolic reactions^{105; 106}. MiR-103/107 paralogs and PANK genes may be coordinately transcribed gene products with synergistic effects in regulating cellular Acetyl-CoA levels³⁷. Bioinformatics predict that miR-103/107 paralogs exhibit an exceptional tendency to target mRNAs in a pattern suggestive of synergy with PANK function. Like PANKs, miR-103/7 may function to increase acetyl-CoA levels available for the Krebs Cycle, and miR-103/7 may also inhibit the synthesis and metabolism of cellular lipids³⁷. Although the research to date cannot be fully harmonized into a unitary framework, miR-15/107 group members clearly are both responsive to, and impactful on, metabolic pathways in human cells.

It may be artificial to consider biological mechanisms separately from each other because in a cellular context these pathways interact dynamically. For example, perturbations in cellular metabolism can cause cell stress and vice versa. Multiple experimental stress paradigms are associated with changed expression of miR-15/107 group members. In cultured cells, miR-16 expression increases after ultraviolet light damage and then seems to participate in regulating the DNA-damage response and cell division machinery¹⁰⁷. It has also been shown that exposure to mice of hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX), a common environmental toxicological contaminant, led to decreased miR-15 and miR-16 expression¹⁰⁸. Members of the miR-15/107 group have not been shown to be aberrantly regulated in heat stress¹⁰⁹ which may indicate a more focused cellular role. MiR-107 expression is down-regulated following rodent traumatic brain injury (TBI)¹¹⁰. We have found that TBI-related miR-107 down-regulation may trigger increased GRN expression in mouse hippocampi after TBI⁷⁵, which may be an adaptive response in that context. The miR-15/107 group may also help regulate cells’ response to ischemic stress: miR-107 and miR-424 are downregulated in hypoxia^{111; 112}, and miR-15a and miR-497 may help cells respond to oxygen deprivation in vivo and cell culture models^{113; 114}. It remains to be seen how these cell stress pathways relate to the other physiological pathways miR-15/107 group members participate in.

C. Angiogenesis

An exciting recent discovery is the connection between miR-107 and angiogenesis. Angiogenesis, the dynamic process of growing new small blood vessels from existing vessels, is directly relevant to development, neoplasia, neurobiology, and various healthy responses to stress. Yamakuchi et al⁸⁸ reported that miR-107 is regulated transcriptionally by P53 protein and in turn miR-107 inhibits HIF-1beta (also known as aryl hydrocarbon receptor nuclear translocator, ARNT). Through this pathway, miR-107 regulates colon cancer tumor growth through a mechanism related to angiogenesis. This study provides insights into upstream regulation of miR-107: agents were identified that up-regulate miR-107 expression through specific p53-responsive promoter regions. Thus, a small non-coding RNA neither necessary nor sufficient to subserve any cellular function is hypothesized to represent a key regulatory intermediary connecting p53 expression, hypoxia, angiogenesis, and the clinical-biological behavior of colon cancer. These changes may be synergistic with other miR-15/107 group members. For example, it has been shown that miR-16 can regulate VEGF strongly in angiogenesis¹¹⁵. As in other important biological processes, it can be assumed that miRNAs in addition to those of the miR-15/107 group play roles in angiogenesis. Specifically, miR-17-92 cluster, miR-378, miR-143, miR-145, miR-93a, miR-27b, miR-130a, let-7f, miR-221, and miR-222 have been shown to participate in blood vessel development in normal and/or tumor tissues^{116; 117}.

In summary, the ubiquitously expressed miR-15/107 gene group may help orchestrate gene expression adaptively in boundary areas between important human cellular pathways (e.g., cross-talk between hypoxia with angiogenesis, and linking cell proliferation with metabolic needs). The known functions of this miRNA gene group dovetail intriguingly with the hypothesis that miRNAs provide cells with a “safeguard against turmoil”¹¹⁸ by regulating the stability and translation of mRNA. However, any adaptive mechanism can go awry. Thus it comes as no surprise that dysregulation of miR-15/107 group gene expression may contribute to, or even cause, human diseases.

Human diseases

A. Neoplasia

An important role for genes in the miR-15/107 gene group has been described in human cancers. Multiple miRNAs from this group have been implicated, spanning a broad range of tumor types. Table 1 summarizes miR-15/107 group gene expression changes that have been found in human neoplastic conditions. In seminal studies, Calin et al^{119; 120; 121} established that the miR-15a and miR-16 tandem gene locus was the important genomic susceptibility factor within the minimal deletion region that determines susceptibility to chronic lymphocytic leukemia (CLL), as had been hinted at by Lagos-Quintana et al²⁰. Subsequent studies have produced a compelling corpus of experimental evidence to suggest that miR-15a/miR-16 constitute key tumor suppressors whose deletion contribute to cancer. Genes with mRNAs targeted by miR-15a and miR-16, and thought to have clinico-biological impact for CLL patients, include BCL-2, CCND3, CCNE1, CDK6, CAPRIN1, and HMGA1^{27; 28; 42; 97; 98; 99}. This is an active area of ongoing research and recent reviews are available^{122; 123; 124}.

Table 1.

MiR-15/107 gene group may be potential cancer prognostic biomarkers, either up-regulated (“oncomiRs”) or down-regulated (tumor suppressors) in tumor cells

miRNA expression is a possible prognostic biomarker		Reference(s)
miR-16, miR-195	Esophageal carcinoma	159
miR-15a, miR-16	Myelodysplastic syndrome	160
miR-16	Acute lymphoblastic leukemia	161
miR-103/7	Esophageal carcinoma	162
miR-195	Adrenocortical carcinoma	163
miR-15b	Hepatocellular carcinoma	164
miRNA expression increased in tumor vs. normal tissues
miR-16	Head and neck carcinoma	165
miR-16	Renal cell carcinoma	166
miR-15b, miR-16	Cervical carcinoma	167
miR-503	Retinoblastoma	168
miR-424	Colonic carcinoma	169
miR-15b	Colonic carcinoma	170
miR-15b	Malignant melanoma	171
miRNA decreased in tumor vs. normal tissues
miR-15a, miR-16	Prostatic carcinoma	172; 173; 174
miR-15a, miR-16	Chronic lymphocytic leukemia	120; 175
miR-15a, miR-16	Non-small cell lung carcinoma	97
miR-15a, miR-16	Pituitary adenomas	176; 177
miR-15a, miR-16	Ovarian carcinoma	178; 179; 180
miR-107	Colonic carcinoma	103
miR-107	Head and neck carcinoma	181
miR-107	Acute promyelocytic leukemia	182
miR-107	Pancreatic carcinoma	89; 183
miR-107	Tongue squamous cell carcinoma	184
miR-195	Hepatocellular carcinoma	185
miR-195	Bladder carcinoma	186
miR-195	Gastric carcinoma	187
miR-195	Chronic lymphocytic leukemia	188
mir-195, miR-397	Primary peritoneal carcinoma	189
miR-497	Male breast carcinoma	190
miR-503	Parathyroid carcinoma	191

The roles of miRNAs in cancer can provide new insights into disease mechanisms and the rich complexity of mammalian gene regulation. Lee et al reported that promoter methylation may induce down-regulation of miR-107 in pancreatic carcinomas⁸⁹. We also note that granulin (GRN), which is regulated by miR-15/107 gene group members, is an active mitogen and growth factor relevant to many cancers ^{125; 126; 127; 128; 129; 130}. It remains to be seen if the impact of miR-107 in pancreatic and other cancers is mediated partly through GRN. Unlike the case for CLL, pancreatic carcinoma, and other cancers, miRNAs in the miR-15/107 group are up-regulated in some other cancers such as cervical carcinoma (Table 1). This is compatible with the hypothesis that miRNAs can act as either tumor suppressors or oncomirs^{131; 132; 133}. Providing yet another layer of complexity in cancers as in other conditions, the degree of functional redundancy among the different members of the miR-15/107 group is an important, unresolved, and deceptively complex issue.

B. Neurodegenerative diseases

MiRNAs may have direct relevance to human neurodegenerative diseases including Alzheimer’s disease (AD), Parkinson’s disease, and frontotemporal dementias (FTD)^{134; 135; 136}. Members of the miR-15/107 group have been shown to be downregulated in AD by at least three different laboratories ^{59; 137; 138}. We studied the cerebral cortical gray matter of aged humans and found that miR-107 is downregulated very early in AD pathogenesis⁵⁹. This finding was recently validated using a different group of brain samples¹³⁹. MiR-107 targets the beta-amyloid cleavage enzyme 1 (BACE1) gene so the miRNA down-regulation may have important pathogenetic consequences^{59; 103}. Yet the physiological significance of the miR-107 downregulation in AD brains has not been systematically tested. Noren Hooten et al compared miRNA expression in young versus old individuals’ blood mononuclear cells by real-time RT-PCR analysis to test aging-related differences, and found that two of the nine miRNAs significantly decreased in aged individuals were miR-103 and miR-107¹⁴⁰. The link between decreased miR-103/107 expression and aging merits further study because age is the most significant risk factor for AD. As noted above, miR-107 and other group members strongly regulate the expression of the GRN gene⁷⁵. This targeting depends on elements in the open reading frame of GRN mRNA. GRN is a susceptibility gene for FTD^{141; 142} and we found that the regulation may also be relevant to neuroinflammation and the brain response to traumatic brain injury. Two laboratories independently reported that miR-15a is downregulated in AD although these studies employed different technical parameters^{137; 138}. Members of the miR-15/107 group are also predicted to target the neurodegeneration-related genes such as the β-amyloid precursor protein and α-synuclein^{136; 143}. In addition, miR-15/107 group members seem involved in the fine-tuning of tau phosphorylation (Hébert et al., unpublished data). The findings in AD may tie in with studies that have suggested that miR-15/107 group member miR-195 may regulate the important brain growth factor BDNF with relevance to schizophrenia^{72; 144; 145}. We note that miRNAs outside the miR-15/107 gene group, including miR-29a/b, miR-128, miR-146a, and others have also been shown to have possible relevance to AD pathogenesis^{134; 138; 146; 147; 148; 149; 150; 151; 152; 153; 154}. As with other conditions, it is not known how targeting by different miRNAs overlaps or interacts with the function of individual miR-15/107 group members in the context of neurodegenerative diseases.

In addition to the miRNA expression down-regulation in AD brains, there are other connections between the miR-15/107 gene group and neurodegenerative diseases. PANK2, which harbors an intronic miR-103 gene, is the susceptibility gene for a separate neurodegenerative disease, pantothenate kinase-associated neurodegeneration (PKAN)¹⁵⁵. PKAN is characterized neuropathologically by progressive brain atrophy with iron pigment deposition and axonal spheroids mostly in subcortical structures. Intriguingly, the brains of some PKAN patients harbor neurofibrillary tangles similar to those seen in AD patients’ brains^{156; 157}. We conclude that it is unclear whether miR-15/107 group members participate causally in neurodegenerative diseases; however, there are already numerous intriguing connections between the expression of the miR-15/107 gene group members and the manifestations of neurological diseases.

C. Heart disease

MiR-15/107 group members are among a subset of human-expressed miRNAs that have been implicated in cardiovascular disease. Most of this research has focused on miR-195. MiR-195 is up-regulated during cardiac hypertrophy and conversely over-expressing miR-195 in murine cardiomyocytes led to fatal dilated cardiomyopathy¹⁵⁸. An recent review on the roles of miRNAs in cardiovascular disease is available³⁰.

Summary and Conclusions

Research on the miR-15/107 gene group is less than a decade old but there is compelling evidence that these genes play essential roles in human biology (Figure 7). MiR-15, miR-16, miR-103, and miR-107 are stably situated in the genomes of vertebrates; however, the evolutionary biology of these miRNAs is still incompletely understood. Moderate-to-high expression of multiple miR-15/107 gene group members has been observed in most human tissues, with increased expression in particular tissues and at earlier stages of development. The miR-15/107 group genes that are relatively highly expressed in humans include the sequence AGCAGCA at or near the 5′ seed region of the mature miRNAs, and this sequence appears to be a strong determinant of mRNA target selection. Functionally, miRNAs in this gene group serve fundamental roles in human tissues. Some of these biological functions occur via novel mechanisms including systematic targeting of mRNA target protein coding sequences. Laboratory data dovetail with evidence from evolutionary biology suggesting that miR-15/107 group members could provide some sort of link between the mechanisms of cell division and metabolism. MiR-15, miR-107, and their paralogs are dysregulated in some human diseases and provide potential therapeutic targets. Although there has been substantial recent scientific progress, much more work remains to be done in studying these genetic regulators whose complexity and impact are enhanced by functional synergy between different miRNA genes.

Figure 7.

Summary cartoon regarding the regulation and function of the miR-15/107 gene group. Upstream positive regulation has been documented for glucose (through unknown transcription factors) and P53 with the caveat that each miRNA is transcriptionally activated via differing mechanisms. Suppression of miR-15/107 gene group member expression can occur in cell stress and trauma, and secondary to promoter methylation or MYC. The normal functions of miR-15/107 gene group members seem to include roles in cell division, angiogenesis, and metabolic pathways. Dysfunction in miR-107 expression may contribute to neoplasia, neurodegeneration, cardiovascular dysfunction, and other diseases.