CONCISE REVIEW Micro RNA Expression in Multipotent Mesenchymal Stromal Cells

Abstract

Multipotent mesenchymal stromal cells (MSC) isolated from various adult tissue sources have the capacity to self-renew and to differentiate into multiple lineages. Both of these processes are tightly regulated by genetic and epigenetic mechanisms. Emerging evidence indicates that the class of single-stranded non-coding RNAs known as “microRNAs” also plays a critical role in this process. First described in nematodes and plants, microRNAs have been shown to modulate major regulatory mechanisms in eukaryotic cells involved in a broad array of cellular functions. Studies with various types of embryonic as well as adult stem cells indicate an intricate network of microRNAs regulating key transcription factors and other genes which in turn determine cell fate. In addition, expression of unique microRNAs in specific cell types serves as a useful diagnostic marker to define a particular cell type. MicroRNAs are also found to be regulated by extracellular signaling pathways that are important for differentiation into specific tissues, suggesting that they play a role in specifying tissue identity. In this review we describe the importance of microRNAs in stem cells focusing on our current understanding of microRNAs in MSC and their derivatives.

Introduction

Global gene expression analysis of stem cells is one method to produce a clear understanding of the variations in biological conditions across different types of stem cells. A similar approach has begun to help us identify and understand the epigenetic signatures of stem cells. Several microarray-based techniques for the systematic analysis of epigenetic changes have become available allowing broad-scale analysis on a new level of transcriptional and translational regulation, particularly for studying the regulation of microRNAs and their predicted mRNA targets. Understanding the importance of microRNAs in MSC differentiation will require knowledge of the mechanisms explaining mRNA targeting, the mechanisms underlying microRNA regulation, and an accurate assessment of microRNA-mRNA targeting.

The discoveries of artificial silencer RNAs and genomic-encoded microRNAs have led to an explosion in the study of small non-coding RNAs^1-7. Various types of small, non-coding RNAs exist as modulators of gene expression, affecting transcription rate, mRNA stability, and mRNA translation into functional proteins. What was once believed to be a specialized function occurring only in plants^{8, 9} or worms^{1, 10} has now been shown to control regulatory pathways in several mammalian biological process such as cell growth and apoptosis^11-14, viral infection^15-19, neural function^20-23 and stem cell function^{21, 24-37}. We will outline the issues related to microRNA expression profiling in mesenchymal stem cells as an example of how these studies are useful for classifying differentiating cells and, potentially, understanding molecular networks active during differentiation.

MSC

The plastic-adherent, fibroblast-like cells isolated from bone marrow with osteogenic potential was first described by Friedenstein³⁸ and subsequently described by several studies^38-44. Since then the ability of these cells to differentiate into multiple connective tissue cell types has been reported^43-47 with the term mesenchymal stem cell first proposed by Caplan⁴⁸. MSC has been isolated from multiple adult tissue sources such as cord blood, placenta, adipose and dermal tissues, synovial fluid, deciduous teeth, and amniotic fluid^49-54. This broad distribution of sources combined with their ability to differentiate into multiple mesenchymal phenotypes such as bone, cartilage, tendon, and adipose tissue has lead them to be evaluated as potential therapeutic candidates for several disease and degenerative applications^55-61. The varied source and methodology used for MSC isolation has also largely led to ambiguities resulting in the lack of an universally accepted criteria to define these cells. To address this issue, the International Society of Cellular Therapy has recently designated these cells as “Multipotent mesenchymal stromal cells” (MSC)⁶² based on the adherence to plastic, expression of specific surface antigen and their differentiation potential.

Despite this level of interest, a clear understanding of the factors involved in their growth regulation remains rudimentary. Several reports have described that only one third of the expanded MSC colonies truly retain tri-lineage differentiation potential, that is, the ability to differentiate into adipocytes, osteocytes and chondrocytes, with the remaining cells in MSC cultures thought to be either having only bi- or mono-lineage or to be committed progenitor cells^{46, 63}. This could largely be due to clonal expansion which in general could alter the phenotype of most stem cell types. Global gene expression analysis has revealed that MSC differentiation into specific mature cell types is a temporally controlled and regulated process involving the activities of various transcription factors, growth facts and signaling pathways^{64, 65}. However, microRNAs would be expected to regulate mRNA translation and/or stability, so perhaps the regulation of microRNA expression patterns represents a novel regulatory network in MSC.

What are microRNAs?

MicroRNAs are single-stranded RNAs of 19−23 nucleotides that derive from a ∼70 nucleotide precursor and are found in a wide variety of organisms, from plants to insects to humans. Estimates suggest there are about 120 microRNA genes in each invertebrate species, and at least 250 genes in mammals, with some groups predicting from 1,000 per genome^66-68 to as many as 10,000⁶⁹. After transcription, microRNA precursors are cleaved in the nucleus by Drosha⁷⁰, exported to cytoplasm by exportin^{71, 72}, and inserted into a RISC (RNA induced silencing complex) complex after additional cleavage by Dicer⁷³. Evidence for the requirement of the processing of microRNA in stem cell function and differentiation comes from studies of the Drosha complex partners Loquacious (homolog of human TRBP), which is required for germ-line stem cell maintenance⁷⁴, and DGCR8⁷⁵, which is required for embryonic stem cell self-renewal ³⁷. Similarly, Dicer knockouts exhibit defects in stem cell differentiation²⁵. Clearly, microRNAs underlie key differentiation mechanisms.

Relatively little is known of the regulation of microRNA expression, although a few microRNA transcriptional control sequences have been described^{14, 76-78}. Interestingly, microRNAs are often clustered in the genome and two or more closely-related microRNA precursors have been detected in polycistronic precursors^{35, 79-81}. Other microRNAs are encoded within introns of other genes^{82, 83} and some microRNAs are edited^{84, 85}, producing a dizzying mix of co-expression, predicted transcriptional control and post-transcriptional modifications. There is much to learn about the regulatory mechanisms controlling microRNA expression.

microRNA regulation mechanisms & target prediction

MicroRNAs are known to negatively regulate gene expression. This could be achieved by direct mRNA cleavage^86-90; mRNA decay by deadenylation^{91, 92} or via translational repression⁹³. Complexes containing microRNAs and those of the RISC complex involved in RNAi are similar. RNA silencing, using a perfectly complementary siRNA, normally cleaves an mRNA target. In animals, microRNAs usually have imperfect complementarity to a 3’ UTR element in their mRNA targets and thus are primarily believed to attenuate translation of the target mRNA. Many studies suggest a continuum of RISC behaviors using similar components, since endogenous microRNAs with perfect complementarity can cleave mRNAs⁹⁴ and exogenously introduced siRNAs can attenuate translation of mRNAs having imperfect complementarity^{95, 96}.

Few target genes that are regulated by microRNAs are currently documented, but the ones identified have important roles in apoptosis, homeobox regulation, development^{30, 97-101}, cell growth and apoptosis^{13, 14}, viral infection^{15, 19} and human cancer^{81, 102-104}. This type of gene control represents a new regulatory mechanism, and is predicted to affect many crucial cellular processes including developmental programs.

An important challenge for incorporating microRNA regulation into gene expression mechanisms is the difficulty in predicting or verifying mRNA targets of specific microRNAs. Efforts to map microRNA binding sites specifically in the transcriptome in animal cells have been challenging, relying primarily on computational predictions. However, it has been suggested that microRNAs regulate gene expression of more than 30% of protein coding genes in humans¹⁰⁵. One clue to decoding this mystery is the conservation of microRNAs and mRNA target sequences across species. For example, more than a third of the C. elegans microRNAs have easily recognizable homologs in humans¹⁰⁶. Sequence conservation argues for conserved function throughout evolution. A number of prominent prediction algorithms have been developed^{66, 69, 107-110}, in most cases based on this observed conservation of complementary “seed” regions conserved across species. A recent review considers these methods and proposes an approach for applying these algorithms¹¹¹. A few reports extend these predictions to collect supporting evidence for specific biochemical targeting interactions¹¹², microRNA-mRNA presence in polysomes¹¹³, or RNA-protein complex composition^{114, 115}. One direct method for validating targets uses a luciferase reporter plasmid with a replaceable 3'UTR¹³. By cloning a 3'UTR from the suspected target mRNA into this reporter and co-transfecting it with microRNA precursors, inhibition of luciferase production can be detected. In general, many of the computationally-predicted targeting interactions are not consistent between algorithms and relatively little direct experimental evidence substantiates these predictions. In our experience, computed target lists have a high rate of false positives (unpublished results).

microRNA analysis methods

Detection of microRNAs in tissues is possible with a variety of techniques. Several microarrays have been described^{102, 103, 116-127} and many commercial sources of microRNA arrays can be found (Table 1) Most of these are based on the publicly-available list of microRNAs found in the miRBase database maintained at the Sanger Institute (http://microrna.sanger.ac.uk)¹²⁸. With most of the array-based methods, it is difficult to claim resolution of specific microRNAs within 1 nt of the probe sequence since the melting temperatures are quite low compared with the longer probes often used for mRNA detection. However, higher specificity can be achieved using direct labeling of microRNAs to obtain RNA:DNA hybridization (Ncode™) or by locked nucleic acid (LNA) oligo probes (mirCURY™). Validation of microarray results originally depended on Northern blots but can also use quantitative real-time PCR (qPCR). In general, qPCR techniques yield the greatest dynamic range, improved specificity, and high sensitivity. In one case, qPCR allowed detection of microRNAs from single cultured neurons or even laser captured somatodendritic compartments¹²⁹. Other detection techniques include ELISA-like or bead-anchored hybridizations using probes and labeling similar to array methods to perform high-throughput analyses¹³⁰. A promising technology for detection and quantification of both known and novel microRNAs is MPSS or massively parallel signature sequencing. For example, Lu and colleagues¹³¹ sequenced 721,044 17-nt cDNAs prepared from Arabidopsis inflorescence, of which 67,528 were unique. Seventy-seven percent of these matched genome sequences, exceeding by more than 10-fold the previously identified Arabidopsis microRNA library. However, later studies described the importance of filtering these results for genomic hairpin structures and other types of small RNA, reducing the number of predicted or novel structures^{132, 133}. MPSS revealed novel microRNAs in Marek's disease virus that contribute to pathogenesis¹³⁴. MPSS also revealed 447 new microRNAs in chimpanzee and human brains¹³⁵. Furthermore, MPSS of mRNAs found ∼25% uncharacterized or novel genes expressed in hESC⁴². As MPSS becomes more widely available and the costs per sample come down it is possible that direct tag sequencing may become an important technique in the study of microRNAs.

Table 1.

Commercial sources of microRNA microarrays

Product Name	Source	References
Human miRNA Microarray	Agilent	155
mirVana™	Ambion/Applied Biosystems	93, 156
Species Specific MicroRNA Arrays	CombiMatrix
MirCURY™	Exiqon, Inc.	157, 158
GenoExplorer™	GenoSensor Corp.	159-161
NCode™	Invitrogen, Inc.	117, 144
Human microRNA Microarray	LC Sciences	30, 162

microRNAs in embryonic and adult stem cells

Several studies have identified populations of microRNAs that are associated with specific stem cell types or those that are regulated during stem cell differentiation (Table 2). MicroRNA has been shown to mediate regulation of stem cell division ²⁷, as well as differentiation in adipocyte¹¹⁶, cardiac¹³⁶, neural^{24, 137} and hematopoietic lineages^{88, 138}. For example, the involvement of miR181 in differentiation of the hematopoietic lineage is well established⁸⁸. miR181a is expressed at relatively low levels in Lin⁻ bone marrow progenitor cells compared to B-lymphocytes and is expressed in high levels in Thymus and primary lymphoid organs. Ectopic expression of miR181a in hematopoietic stem cells leads to increased fraction of B-cell lineage both in vitro and in vivo suggesting it to be a positive regulator for B-cell differentiation⁸⁸. microRNAs are also found to be highly expressed in the adult brain suggesting their involvement in neuronal function and plasticity¹³⁹. miR124 and miR128, both highly expressed in adult brain, are also expressed in neurons, while miR23 is restricted to astrocytes. The induction of microRNA expression during neural differentiation of embryonic stem cells further suggest their importance in neural development²⁹.

Table 2.

Studies of MicroRNA in Stem Cells

Cell type		Process	Reference
Embryonic Stem Cells	All	Review: miRNA in Stem cells	163 164
	Murine	Dicer in miRNA biogenesis in conditional dicer-deficient ESC	26
		DGCR8 in miRNA biogenesis and stem cell self-renewal	37
		ESC	35
		ESC and during differentiation	32
		ESC in response to DNA damage	165
		ESC-derived neurogenesis	166
	Human	ESC	141, 143
		expression pattern to characterize hESCs	144, 146
		ESC and during Differentiation	145
		regulation during Stem Cell Division	167
		miR125b in Differentiation	168
Neural cells		Review-miRNA in Neural cells
		expression during neural differentiation	22
		regulation of Lin-28 during neural differentiation	137
		miR124 in developing neural tube	20
		Neural cell specification	29; 169
Hematopoeitic cells		Review: miRNA in Hematology	170
		Murine and human erythroid differentiation	171
		in CD34+ cells	33
		in hematopoeitic lineage differentiation	88: 172
		in cord blood CD34+ erythorpoiesis	173
		miRs 17−5p, 20a and 106a in monocytopoiesis	174
		miR221 and 222 inhibit normal erythropoeisis	175
		miR181a in T cell selection	176
		Timing of miR-150 in mature B and T cells	177

The miR-15a-miR16 cluster resides in a region of human chromosome 13 thought to harbor a tumor suppressor gene^{102, 140} and also to target Bcl2, potentially modulating apoptosis⁹⁸. MicroRNAs have also been found to control G1/S checkpoint bypass in Drosophila stem cells²⁷, demonstrating their intimate role in stem cell division.

A unique set of microRNAs is reported to be expressed in mouse embryonic stem cells (ESC) that changes with differentiation during differentiation via embryoid body (EB) formation^{25, 26, 35, 141, 142}. Northern blot and cloning analysis identified several microRNAs in human ESC, several of which were identical to microRNAs previously reported in mouse ESCs¹⁴³. We recently reported that several microRNAs are highly expressed across multiple hESC lines¹⁴⁴. Furthermore, as these cells differentiate, the microRNA profiles change significantly. Combining the profiling of microRNAs and mRNAs, correlation of gene and microRNA expression predicts regulatory interaction of microRNA and mRNAs involved in both maintenance of pluripotency as well as differentiation¹⁴⁵. Based on this, microRNA expression pattern in human embryonic stem cells is included in the panel of tests suggested for qualification of human embryonic stem cells¹⁴⁶.

Consistent with the observation for individual microRNAs, mouse ES knockout cells lacking Dicer²⁵, required for cleavage of microRNA precursors, or DGCR8³⁷, an RNA-binding protein essential for processing microRNAs, exhibits a failure to undergo differentiation and completely prevented mouse ESC from differentiating into embryoid bodies, suggesting that microRNAs are required to inhibit ES self-renewal. Furthermore, knockout mice lacking Argonaute2, the catalytic component of the RISC complex, exhibited severe defects in neural development, including the failure to close neural tube¹⁴⁷. Collectively, these studies implicate the importance of microRNAs as key regulators during the process of stem cell maintenance and differentiation.

A unique set of microRNAs are expressed in embryonic stem cells (Table 3). Using Northern blot analysis, cloning, or array methods, a set of microRNAs specific to pluripotent cell types such as mouse embryonic stem (ES) cells³⁵, mouse or human embryonic carcinoma cells²², or human ESC^{32, 143-145}, has been identified. While these studies could not yet ascribe regulatory functions to these microRNA populations, their unique presence in stem cells and their disappearance during differentiation suggest roles in maintaining “stemness” by suppressing pluripotency and/or restricting cell differentiation. In one example, miR-134 was found to be enriched after retinoic acid treatment of mouse ES cells, and miR-134 reduces expression of Nanog and LRH1³⁰, enhancing differentiation to specific lineages.

Table 3.

MicroRNAs associated with embryonic stem cells

miRNA	Species	Function	Reference
miR291,292, 293,294,295	Mouse	Present in ESC, low in EB, absent in MEF/NIH3T3	35
miR29a, 29b, 193, 199		Present in MEF/NIH3T3, absent in ESC,EB
miR296		ESC specific
miR21, miR22		Spontaneously diff cells
miR15a, 16, 19b, 92, 93, 96, 130, 130b		General physiological aspects
miR106a, 127, 130b, 15b, 16, 18, 19b	Human	Present in single hESC cells	141
miR20, 21, 290, 291, 292, 293, 294, 295,
miR92, 93
miR302b, 302c, 302a, 302a, 367	Human	Present in Both hESC and EC	143
miR200c, 368, 154, 371, 372, 373		ESC only
miR200c, 371, 372, 302a, 302d	Human	Present in hES, low in EB, absent in adult brain	145
miR373, 302c, 21, 222, 296, 494, 367		Higher in ESC compared to EB or adult brain
miR17M, 92, 93		Higher in EB comapred to ESC or adult brain
Let7a, miR29a, 143		Higher in adult tissue followed by ESC and EB
miR134, 18a, 18b, 16		Unchanged between hESC & EB
miR183,290,291−3p, 291−5p, 292−5p, 293, 294, 295, 302a, 302b, 302c, 302d, 367, 292−3p	Mouse	Stem cell specific	32

microRNA expression in MSC and derivatives

Compared with these examples examining microRNAs in ESC, relatively few studies have identified microRNA expression patterns in differentiating MSC or specific pathways that could be targeted (Table 4). In one case, miR-103 and miR-107 were found to be encoded within an intron of the PANK gene in a pattern that is conserved among a large array of organisms¹⁴⁸. Both microRNAs (the high degree of sequence homology makes them virtually indistinguishable) are predicted to target a large number of genes involved in acetyl-CoA and lipid metabolism. Another study found that the modulation of a single microRNA could promote the formation of adipocytes from precursor cells¹¹⁶. This differential targeting was shown to be mediated by the translational regulation of ERK5, an intermediate in the LIF signaling pathway¹¹⁶. ERK5 is also known to mediate a neuronal survival response following retrograde signaling of neurotrophins^{149, 150}. A third example found that miR-140 was expressed during cartilage development and that miR-140 may act by inhibiting HDAC4, a likely co-repressor of Runx2¹⁵¹. These initial examples show that, similar to studies in ESC, microRNAs are likely to be mediators of key pathways during MSC differentiation.

Table 4.

microRNAs associated with MSCs

microRNA	Tissue/Pathway	Reference
miR-143	Adipocyte	116
miR-103, 107	Adipocyte	148
miR-140	Cartilage	151
miR-638, 663	Chondrocytes
miR-24, let-7a, let-7b, let-7c, miR-138, mir-320	Osteocyte/PDGF

We have recently begun mapping microRNA changes during MSC differentiation using microarrays and qPCR validation¹⁵². In general, we observe a great deal of variability between individual donors, making generalizations difficult. For example, MSC prepared from two different donors had a correlation coefficient of 0.912 over all detected mRNAs, and MSC from the same donor prepared from two subsequent passages had a correlation of 0.804. However, using pooled samples of MSC cultures, we identified 27 human microRNAs as regulated during MSC differentiation into adipocytes, osteocytes or chondrocytes. Expression patterns could be clustered to highlight specific microRNAs with restricted expression patterns, confirming previously-described results (e.g., miR-143 and miR-145 enriched in adipocytes) as well as identifying novel markers (e.g. miR-638 and miR-663 in chondrocytes). We found that several of the regulated microRNAs were attributable to changes upon osteocyte differentiation. Standard gene expression analysis found that osteocyte differentiation included pathway signatures for PDGF stimulation, as was first recognized by Ng et al.¹⁵³. To determine whether the class of osteocyte-specific microRNAs could by regulated by PDGF-stimulated transcriptional mechanisms, we bioinformatically associated PDGF-responsive transcription factor position weight matrices with genomic sequences upstream of regulated microRNAs. Our results predict a significant number of PDGF-responsive transcription factor sites are found upstream of regulated microRNAs, compared with expressed but unregulated microRNAs (Fig. 1). This prediction was confirmed by treating differentiating MSC with AG-370, a tyrphostin inhibitor of the PDGF signaling pathway. MicroRNAs ranking highly on a confidence list of predicted PDGF regulation targets were generally also inhibited by the addition of AG-370 during osteogenic differentiation (examples listed in Table 4). Taken together, these results identify an osteogenic class of microRNAs that are regulated by growth factor signaling during differentiation, demonstrating the regulation of specific microRNAs during MSC differentiation.

Figure 1.

Model of PDGF-regulated microRNAs modulating translation of targeted mRNAs.

Predicted microRNA-target interactions

The importance of microRNA regulation in MSC differentiation is demonstrated in the increasing evidence for specific microRNAs participating in known pathways through mRNA target interactions. There are fewer known examples for bone and connective tissue differentiation but we should expect to find target pathways as these microRNAs are described.

Recently, using a concept similar to the “biclustering” of gene expression data¹⁵⁴, we carried out correlation of an mRNA to each microRNA expression level in multiple hESC lines and their differentiated samples organized as separate groups¹⁴⁵. A similar approach was also employed to study the changes in miRNA expression during murine embryonic stem cell differentiation³². Such a cross-correlation clustering provides interpretable lists of microRNAs and associated mRNAs that may be hypothesized to interact through specific mechanisms providing impetus towards recognizing microRNAs involved in controlling stem cell fate.

Future directions

As the complexity of the regulatory network in differentiating stem cells begins to become known, identification of regulated microRNAs and their interactions with transcriptional networks and, ultimately, cellular regulatory systems, becomes valuable. The organization of the hierarchical order of all types of stem cells based on their pluripotency and linkage of their functional characteristics to genetic and epigenetic regulatory elements provides a novel means to understand and, potentially, manipulate cell fate. With the ultimate goal of using microRNAs to modulate differentiation of stem cells as potential therapeutic agents, we expect to produce a clearer understanding of microRNAs in specific cellular pathways such as development, proliferation, stem cell renewal, differentiation and in diseases such as cancer.