Introduction

Cancer is a class of diseases characterized by uncontrolled cell proliferation. There are more than hundred types of cancers which can be classified on the basis of the organ it affects. One of the biggest strength in the cancer’s armor is that even after a substantial number of cancer cells have been killed by therapies, it has the ability to relapse and become even more lethal than it was in the first instance. Hence, it is imperative to know which cancer cells and how many of them must be eliminated to have a permanent cure for cancer. Two models have been described for the origin of cancer cells. One of the models referred to as clonal evolution model states that any of the tumor cells could initiate the tumor formation. It describes cancer as a disease of the genome which occurs through stepwise accumulation of mutations. According to this model, a cell becomes tumorogenic and malignant by acquiring six cumulative alterations which include (1) unlimited proliferation potential, (2) independence from growth signals, (3) evasion of anti-growth and (4) apoptotic signals, (5) the ability to metastasize, and (6) induce angiogenesis [1]. The second model, referred to as the cancer stem cell theory states that cancer has a hierarchical cell structure in which only a subpopulation, termed as cancer stem cell (CSC) is competent to initiate tumor formation. CSCs have the ability to grow indefinitely and give rise to all the differentiated cells of the tumor. They are also responsible for metastasis [2, 3] and the relapse of the cancer owing to their chemoresistant properties [4]. If the CSC hypothesis is correct, then elimination of CSCs would give us that perfect cure [5].

miRNAs are small single-stranded RNA molecules which bind to 3′UTR of the mRNA and inhibits gene expression by mRNA degradation or translational repression. They have been implicated in a variety of biological processes including cell cycle, proliferation, differentiation, metabolism, and apoptosis [6–10]. miRNA deregulation has been established as the causal factor in different cancers (Table 1). miRNAs can function as both oncogenes [11, 12] and tumor suppressor genes in the pathology of cancer [13–15]. miRNAs are also essential for the maintenance of stem cells. The mutation in Dicer1, an enzyme essential for biogenesis of miRNA, results in depletion of stem cells in mouse embryo [16] and the re-expression of Dicer1 in mouse embryos could restore stem cell phenotype [17]. Many miRNAs which regulate the self-renewal and differentiation potential of normal stem cells have also been elucidated, for example miR-124 (required for neuronal differentiation) [18], miR-24 (promotes skeletal differentiation) [19], and miR-21 (suppresses the self-renewal potential of mouse embryonic stem cells) [20]. The aberrant expression of miRNAs has also been observed in CSCs and if CSC hypothesis is true, then the aberrant expression profile of miRNAs may be the cause of initiation, development or/and maintenance of the tumor. This aspect, however, warrants further investigation.

Table 1 Validated causal miRNAs in cancer
Full size table

Cancer stem cells and related controversies

The term cancer stem cell was adopted at AACR workshop [21] in 2006 and it was used to define the population of cancer cells which have the ability to self-renew and differentiate just like the stem cells. The name, cancer stem cell, was coined to represent its properties of self-renewal and multipotency. CSCs can self-renew but it’s debatable whether CSCs can differentiate into multiple types of cells. Also, this nomenclature inadvertently implies that these cells have originated from normal stem cells. The stem cells may be a source of origin of CSCs but they are certainly not the only source (Fig. 1a). Because of these reasons, many researchers prefer to call them tumor initiating cells.

Fig. 1
figure 1

Cancer stem cells: origin and isolation (a). Stem cells are a small subpopulation present in almost all human tissues. Under normal physiological conditions, these stem cells, in a series of sequential stages, differentiate into cells with a specialized function. The differentiated cell also has a rare chance of reverting back to the stem cell in certain biological conditions (for instance, induction by TGFβ). The normal stem cells may undergo mutations in its DNA owing to genetic, epigenetic, or microenvironmental factors to generate CSCs which have the property of self-renewal just like the stem cells, but instead of differentiating into a normal cell, they differentiate into a cancer cell. The cancer cell may also develop from a mutation in the normal differentiated cell. Even the cancer cells can convert to CSCs via EMT. This two-way formation of CSC is not exclusive of each other and may occur simultaneously in the cancerous condition. b The outcome of the research on CSCs is largely dependent on the method used for their isolation. There have been quite a few methods developed for isolating them and all these methods are based on one of the characteristic properties of CSC. (1) CSCs have been found to express different cell surface markers and fluorescent antibodies against these markers can be used to isolate these cells using flow cytometer. Since cancer usually has more than one type of subpopulation and these subpopulations have different surface markers, this method is unable to isolate all CSCs of a tissue. (2) In non-adherent conditions, CSCs have the ability to grow and form floating aggregates called spheres whereas the non-stem cells undergo anoikis [91]. This property can be used to enrich cells with stem-like properties by growing them in serum-free media supplemented with growth factors like EGF and basic FGF on ultra-low attachment plates. Although this method isolates different CSC subpopulations yet it is a lengthy process (~7–14 days) and some of the non-stem cancer cells may also continue to grow. (3) CSCs are chemoresistant owing to its resistance to apoptosis and its ability to activate ABC transporters [4]. The cancer cells when grown in the presence of a drug for a significant duration of time eventually leads to the death of most of non-stem cells. This method also allows for isolation of different CSC subpopulations but some of the resistant non-stem cancer cells (because of possible mutations) may continue to grow. (4) The cell population with the ability to efflux lipophilic dye Hoechst33342 (due to activation of ABC transporter system) is termed as sidepopulation. This sidepopulation has been found to be enriched in stem cell population. Thus, fluorescently labeled Hoechst dye may be used to enrich for CSCs by flow cytometry. The toxicity of Hoechst33342, however,limits its use in functional studies. CSCs can also be isolated for their ability to express aldehyde dehydrogenase (ALDH). In this method, an inactive fluorescent aldefluor is added to the cell medium which gets activated when ALDH acts on it. This fluorescence can thus be used to separate out CSC population by flow cytometry. (CSC-1 and -2 refer to two different CSC subpopulations)

Full size image

There is an inherent property of CSCs which allows it to restrict its proportion with respect to non-stem cancer cells. Even if isolated CSCs are grown in culture, eventually they will differentiate into tumor cells and return to the same proportion which existed prior to their isolation [22]. Since, CSCs do not exist independently and there is no technique which allows them to be grown continuously and separately from the other non-stem cancer cells so the first step in any research on CSCs is to isolate these cells. The isolation methods based on CSC properties are highlighted in Fig. 1b.

Initially, it was believed that CSCs represent a small subpopulation in the cancer tissue but later it was shown that its proportion is highly variable even in the same cancerous condition. For instance, CSC subpopulation can vary by 500-fold in leukemia patients [23]. This may be attributed to the differential origin/stage of the cancer. The variability in the number of CSCs is a challenge when designing therapeutics as it may cause variable response to therapies in patients. Hence, animal models that take into account the differential CSC count must be used for designing therapeutics [24]. On the brighter side, it helps in the prognosis of disease as the research has shown that more is the CSC count, higher is the chance of that tumor to develop chemoresistance and metastasize. Hence, more radical therapy can be designed for patients with higher CSC count.

Another point of contention about CSCs is the interconvertibility of CSCs and non-stem cancer cells. Gupta et al. have validated that the subpopulations of breast cancer tissue are in equilibrium which is maintained by the interconversion from one state to another [22]. The interconversion phenomenon and equilibrium was also observed in breast cancer (MCF7) and colon cancer cells (SW260). Here, they further proved that EMT was responsible for this interconversion and TGFβ, a known EMT inducer, could accelerate the conversion to CSCs while inhibition of TGFβ had an opposite effect [25]. As many miRNAs like miR-200, -205, -34 [26–28] have been implicated in EMT regulation, the interconversions of CSCs and non-CSCs may also be affected by miRNAs. These studies have highlighted a novel concept of de novo generation of CSC from non-stem cancer cells. Therefore, for cancer therapies to be effective, agents that kill both CSCs and non-CSCs need to be used in tandem as agents that kill only one subpopulation are bound to be ineffective. This interconvertibility was thought to be a challenge to CSC concept as extremely fast and random interconversions would have meant that CSC state is transient [24] and thus, studying them or designing therapeutics against them would have been futile. But the studies have shown that rate of generation of CSCs from non-CSCs is very low and CSC is a phenotypically distinct state which remains constant under fixed microenvironments [22].

miRNAs and CSCs

The role of miRNAs in CSCs of various cancers is discussed in the following sections.

Breast cancer

The characteristic markers of breast CSCs include CD44+ CD24− [29], CD44+ CD24− ESA+ [29], and ALDH+ [30]. The first report on the role of miRNA in CSCs came out in 2007 when Yu et al. showed that let-7 regulates the stemness and differentiation of breast CSCs [31]. The authors observed that the exogenous expression of let-7 reduced mammosphere formation by inhibiting the expression of HRAS which is responsible for the self-renewal property of the cells and HMGA2 which affects the differentiation capability of the cells.

miR-93 has also been shown to inhibit the stem cell phenotype in basal breast cancer cell lines like HCC1954, SUM159 [32]. Surprisingly, this effect of miR-93 in luminal cell lines like MCF7 is just the opposite where it promotes the stem cell phenotype. This study highlights the complexity of regulation by miRNAs where they can have completely opposite functions in different cellular contexts. Several evidences in the literature reveal that miR-93 regulates the CSC population by simultaneously targeting a number of stem cell regulatory genes including JAK1, SOX4, STAT3, AKT, E2H1, and HMGA2. Furthermore, knockdown of STAT3/SOX4 decreases the proportion of breast CSC suggesting that these genes may mediate the function of miR-93 in the regulation of CSC self-renewal [32]. miR-93 did not cause apoptosis but instead induced differentiation of these CSCs thereby making them sensitive to cytotoxic drugs. The ectopic expression of miR-93 in these cells prevented the relapse of the tumor in NOD/SCID mice.

Interestingly, in a study by Ibarra et al. in 2007, both let-7 and miR-93 were found to be depleted in mouse progenitor cells, thereby pointing towards the evolutionary conserved functions of these miRNAs [33]. In fact, miR-93 is one of the few miRNAs reported to regulate CSCs in multiple types of cancer. However, the target of miR-93 responsible for the regulation of the CSC is not the same in these cancer types (Table 2). This may be because all the studies have focused on a specific target rather than taking into account that a single miRNA may target multiple mRNAs. It would nevertheless be interesting to explore the therapeutic potential of such miRNAs as they play a role in multiple cancers.

Table 2 miRNAs regulating cancer stem cells in multiple cancers
Full size table

CSCs are hypothesized to be the first cells that initiate metastasis. In breast cancer, miR-7 was shown to prevent brain metastasis by inhibiting the transcription factor, KLF4. Although, there was not much difference in the level of miR-7 in CSC and non CSC population of MDA-MB-231 but it was significantly reduced in CSC when compared with non-CSC subpopulation of metastatic variants of MDA-MB-231 [34]. This implies that the downregulation of miR-7 may be responsible for the metastatic property of CSCs.

Colorectal cancer

The markers used to isolate colon CSCs include CD133+ [35, 36], EpCAM+ [37], CD44+ [37], CD166+ [37], and ALDH+ [38].

In colon CSCs, miR-140 and -215 are reported to be significantly upregulated [39, 40]. While miR-140 caused chemoresistance to 5′fluoruracil (5-FU) and methotraxate (MTX) [39], miR-215 was responsible for the chemoresistance against MTX and Tomudex (TDX) [40]. The effect of miR-140 on 5-FU resistance was because of suppression of HDAC-4. For MTX resistance, the authors speculated that it could be because of another validated target of miR-140, CXCL12 [41]. miR-215-induced chemoresistance was affected through its target denticleless protein homolog (DTL). Interestingly, the function of both these miRNAs was dependent on p53. Another miRNA which is upregulated in colon CSCs is miR-21 [42]. Its expression is 3- to 7-fold upregulated in HCT-116 and HT-29 CSCs (enriched by drug resistance). This miRNA targets TGFΒR2 and activates Wnt pathway, an essential pathway for maintaining the stem-like properties of cells.

miR-93, miR-328, and miR-451 are the downregulated miRNAs in CSC subpopulation of colon cancer. miR-93 inhibits the cell renewal and colony formation of SW116 CSC possibly by targeting HDAC8 and TLE4 [43]. miR-328 increases the sensitivity of the cells towards 5-FU and HCPT (hydroxycamptothecin) by targeting ABCG2 and MMP16 in CSC of colorectal cancer in SW116, SW480, and cancer tissues [44]. In colonospheres of colorectal cancer cell lines (DLD1, HT29, LS513, and SW620) and CSCs of colon cancer, the downregulation of miR-451 is required for their self-renewal property and chemoresistance towards irinotecan [45]. The self-renewal property was attributed to its indirect target COX2 and for the chemoresistance property, its direct target ABCB1 was found to be responsible.

Brain tumors

It is remarkable that in CSCs, majority of miRNAs have reduced expression and these downregulated miRNAs tend to either inhibit self-renewal or induce their differentiation. The brain CSCs are characterized by the expression of CD133 [46, 47] and CD15 [48] surface markers. In brain tumors, the expression of miR-451, -34, 199b-5p, and -124 was found to be reduced in CSCs.

One of the first reports of miRNA profiling in glioblastoma came from David Givol’s group in 2008. They found that miR-451 could inhibit neurosphere formation and reduce cell viability in glioblastoma cell line A172. Furthermore, miR-451 overexpression also enhanced the efficacy of imatinib mesylate, a drug used for inhibiting neurosphere formation. Although the target of miR-451 for this function is not identified, the authors validated two upstream regulators, SMAD3 and SMAD4, which enhance its expression [49].

The ectopic expression of miR-199b-5p resulted in not only reduced population of CD133+ cells but also reduced expression of several stem cell genes (NANOG, OCT4, KLF5) in medulloblastoma cells [50]. The literature reveals that miR-199b-5p targets HES-1 transcription factor which is a target of NOTCH2 and notch signaling is a regulator of stemness of the cells. It inhibited proliferation and induced differentiation in human medulloblastoma cells. In vivo also, miR-199b-5p resulted in reduced proliferation of the tumor. Furthermore, the high expression levels of miR-199-5p correlates with high patient survival [50]. The expression of miR-34 is also diminished in CSCs as it induces differentiation of glioma cells to astrocytes/oligodendrocytes, thus reducing their stemness [51].

miR-124 is another brain-enriched miRNA [52] which is downregulated in gliomas [53, 54] as well as medulloblastomas [55]. The overexpression of this miRNA and downregulation of its target SNAI2 in CD133+ cells inhibited the neurosphere formation and caused loss of stem cell genes (BMI1, NANOG, NESTIN) [56]. In a similar study, miR-128 has also been shown to inhibit the stem cell self-renewal factor BMI1 in gliomas [57]. Hence, these miRNAs may be used to suppress the “stem-like” characteristics of glioma cells.

Ovarian cancer

Ovarian cancer is the fifth most common cancer among women. The markers used to identify ovarian CSCs include CD133+ [58], CD44+ CD24− [59], and CD117+ [60]. So far, three miRNAs have been functionally validated for the regulation of ovarian CSCs; miR-200a,-199a, and -214. Earlier, miR-200a was established as the biomarker for ovarian cancer. In 2011, Wu et al. observed that miR-200a was downregulated in CSCs (CD133+ cells) of ovarian cancer and overexpression of miR-200a in these cells resulted in decreased migration and invasion [61]. They concluded that miR-200a targets SIP1 (ZEB2), a transcriptional repressor of E-cadherin and enhanced migration and invasion of cancer cells could be due to loss of miR-200a [62].

CD44, a marker used to identify CSC subpopulation in ovarian cancer has been validated as target for miR-199a [63]. Overexpression of miR-199a led to a decrease in the expression of stemness genes of CSCs (OCT4, NANOG). Also, the multidrug resistance gene ABCG1 was reduced in miR-199a transfected cells, thus making the transfected cells more susceptible to cisplatin, adriamycin, and paclitaxel. It also considerably suppressed the proliferation, invasion, and migration of CSCs in vitro. In xenograft models too, it had similar effects on the inhibition of tumor growth. Interestingly, CD44 is marker for CSCs of not only ovarian cancer but also for CSCs of breast, pancreatic, and prostate cancer. It would be interesting to see if miR-199a can target CD44 in different tissues. If that happens, miR-199a could be a universal miRNA regulating stemness of multiple cancers. In hepatic cancer, miR-199a-3p has been shown to target CD44 to inhibit the proliferation of CSC cells [64].

miR-214 had been implicated in enhanced invasion [65] and increased resistance of ovarian tumors [66, 67]. Since both chemoresistance and metastasis are properties of CSC, Xu et al. studied its role in ovarian CSCs where they found that miR-214 inhibits p53, thereby rescuing the p53-mediated suppression of NANOG (stemness marker). miR-214 expression increased the chemoresistance of the cells to cisplatin and doxorubicin [68].

Liver cancer

Hepatocellular carcinoma (HCC) is the most common liver cancer. The cell surface markers which are expressed by CSCs of liver include CD133 [69, 70], EpCAM [71], CD90 [72], and OV6 [73]. Ji et al. discovered that in HCC patients, miR-181a-1, -181a-2, -181b-1, -181b-2, -181c, and -181d were significantly upregulated in stem cell fraction (EpCAM+ and AFP+) of these tumors when compared with non-stem cell population [74]. Functionally, miR-181 led to increase in the stem cell population in Huh1 and Huh7 cells. The authors validated that in order to maintain stem cell phenotype, miR-181 targets CDX2 and GATA6 to suppress cell differentiation and targets NLK to activate Wnt signaling pathway. Meng et al. also found increased miR-181 expression in CSCs of HCC [75]. The authors validated RASSF1A, TIMP3, and NLK as the targets of miR-181. In addition to miR-181, Meng et al. also found upregulation of let-7 in CSCs which inhibited SOCS1 and CASP3 thereby making CSCs chemoresistant to sorafenib and doxorubicin [75].

miR-150 was another miRNA downregulated in CSCs [76]. The ectopic expression of this miRNA reduced CD133+ cell population and inhibited tumorsphere formation. The authors proved that its effects on self-renewal of CSCs are partially mediated by MYB.

Prostate cancer

The prostate CSCs have been isolated by selecting for cells which are CD44+ [77], CD133+ [78], and β-integrin+ [79]. In prostate cancer samples and xenograft models, miR-34a is significantly reduced in the CSCs (CD44+ cells). The enforced expression of miR-34a inhibits the colonogenic ability, tumor regeneration, and metastasis of the prostate cancer cells by targeting CD44 [80]. Interestingly, CD44 has also been found to be targeted by miR-199a in ovarian cancer stem cells [63] and by miR-199a-3p in HCC [64]. The delivery of miR-34 also reduced the prostate tumor of NOD/SCID mice by 50 %.

In an independent miRNA profiling study, mir-143 was found to be downregulated by 8-fold in CSCs of PC3 cell line [81]. Its expression level increased gradually with differentiation of these cells, thereby pointing towards a potential role in differentiation. The downregulation of miR-143 resulted in decreased migration and invasion in vitro and repressed metastasis in vivo. FNDC3B was validated as the target of miR-143 which is responsible for its role in differentiation. FNDC3B was previously shown to be downregulated in highly metastatic cancer. Previously, Zhang et al. had reported that overexpression of miR-143 resulted in repression of FNDC3B and thus enhanced the metastasis of HCC [82]. It is remarkable that although the CSCs are believed to aid in metastasis yet surprisingly, PC3 CSCs resulted in decreased migration and invasion ability of the cells. The authors speculated that maybe the differentiation of these cells resulted in a more aggressive metastatic phenotype.

Pancreatic cancer

Pancreatic cancer has one of the worst prognosis of all the cancers and is also considered to be therapy resistant [83]. CSC discovery in pancreatic cancer has provided alternative approach to study this cancer. The CSCs of pancreas are characterized by their expression of CD44 [84], CD24 [84], and ESA [84]. One of the major miRNAs deregulated in CSCs of pancreatic cancer is miR-34. miR-34 is a p53 regulated miRNA and its expression is usually abrogated in the cells which lack p53 [85–87]. In cancerous condition where p53 is inactivated, the levels of miR-34 are significantly reduced [88]. Moreover, about 15 pancreatic cancer cell lines have at least a 2-fold reduction in miR-34a expression in comparison to its expression in normal pancreatic cell lines [87]. In 2009, Ji et al. found that overexpression of miR-34 in MIAPaCa-2 cell line led to decreased colonogenic growth, decreased invasion, and also sensitized these cells to chemotherapy and radiotherapy. This made the authors speculate the role of miR-34 in CSCs of pancreatic cancer. The low levels of miR-34 in CSCs of MIAPaCa-2 cells added weight to their speculation. They further went on to show that overexpression of miR-34 caused about 87 % reduction in the tumorsphere formation by CSCs of MIAPaCa-2 and BxPC3 cells. This effect was because of the downregulation of NOTCH1 and BCL2 which significantly affected the self-renewal potential of these cells. In vivo too, miR-34 could inhibit tumor formation in nude mice and reduce the number of tumor initiating cells [89].

In another study, miRNA profiling was carried out in CSCs of two pancreatic cancer cell lines Capan-1, HPAC, and one normal cell line hYGIC6 [90]. Fourteen miRNAs showed differential expression in CSCs of all three cell lines when compared with their respective non-CSCs, and 22 miRNAs showed differential expressions only in the CSCs of Capan-1 and HPAC cell lines. These 22 miRNAs may determine whether the cell will become cancerous but further experimentation is needed to validate this speculation.

miRNA therapeutics against CSC

miRNAs have been demonstrated to target all the properties of CSCs be it self-renewal, differentiation, or chemoresistance. Thus, there are many ways in which miRNAs may inhibit the function of CSCs (Fig. 2). Some of these are discussed below:

Fig. 2
figure 2

miRNA therapeutics against CSCs, The differential expression of miRNAs in CSCs and their ability to regulate the different properties of CSC make them potential therapeutic candidates. The miRNAs are particularly interesting because of the following functions in CSC: (1) Prevents chemoresistance: miRNAs like 140 [39], miR-215 [40], and miR-451 [45] have been shown to make colon CSC cells amenable to chemotherapy. (2) Induces apoptosis: Certain miRNAs like miR-150 can specifically target surface markers of CSCs and bring about their apoptosis [76]. (3) Induces differentiation: The overexpression of miRNAs like let-7 [31] and miR-93 [32] can induce differentiation of breast CSCs, thus they lose their potential to self-renew. Once differentiated, they are no longer resistant to therapy and are also rendered ineffective to cause any relapse. miR-34 plays a similar role in glioblastoma [51]. (4) Inhibits self-renewal: Reintroduction of miR-34 has been shown to prevent self-renewal and sphere formation of CSCs in prostate [98] and pancreatic cancer [89]. The prevention of self-renewal limits the growth potential of the cancer and makes the drug treatment more effective. (check mark, induces; cross mark, inhibits)

Full size image

Proliferation

miRNAs which target the self-renewal potential of CSC would restrict the proliferation of cancer cells. miR-128 is significantly downregulated in glioblastoma patients. miR-128 targets BMI1 which is implicated in self-renewal and maintenance of stem cells [57], thus ectopic expression of miR-128 may exert therapeutic effect by inhibiting cell proliferation in glioblastomas.

Invasion

The invasive property of CSC is vital for their metastatic activity and miRNAs which counter act this property offer a therapeutic advantage. For instance, the forced expression of miR-34 in prostate CSCs inhibits its sphere forming ability and invasion in vitro and prevents the proliferation and metastasis of prostate cancer in vivo [80]. In pancreatic cancer too, expression of miR-34 elicit similar response as it inhibits the colonogenic potential of the cells in vitro and reduces tumor initiating cells in vivo [89].

Apoptosis

miRNAs capable of inducing apoptosis in CSC subpopulation would be an excellent therapeutic prospect. miR-150 has been shown to cause apoptosis in CD133+ fraction of HCCs [76].

Terminal differentiation

Those miRNAs which induce terminal differentiation of CSCs cause loss of stem cell phenotype and thus make them amenable to chemotherapy and radiotherapy. let-7 [31] and miR-93 [32] induce differentiation of breast CSCs and miR-34 does so in glioblastomas [51].

Chemoresistance

The cancer therapies are rendered ineffective primarily for the chemoresistant properties of CSCs. Hence, the miRNAs which make the CSCs chemosensitive would make for an excellent therapy. Many miRNAs have been able to overcome this phenotype of CSCs. For instance miR-140 [39], miR-215 [40], and miR-451 [45] make the colon sphere cells susceptible to 5-FU, MTX, and irinotecan treatment, respectively.

The ideal therapy for cancer would be the one which targets both CSC and non-stem cancer populations, simultaneously. Since miRNAs can regulate both these populations, a combination of miRNAs may be a very potent therapy for cancer. Currently, a commercial venture called miRNA Therapeutics is working on eight miRNAs (five of which are undisclosed) to develop miRNA replacement therapies for treatment of cancer. Already, miR-34 (as MRX34) is in preclinical trials Phase1 and let-7 (as miR-Rxlet7) is in preclinical development for treatment of cancer. The regulation of CSCs by these miRNAs was not known at the time these studies were initiated but later it was revealed that both miR-34 [51, 80, 89] and let-7 [31] were significantly downregulated in CSCs. Thus it would not be wrong to hypothesize that along with reducing the bulk of the tumor, they will also kill a substantial percentage of CSCs.

Future directions

Cancer stem cells have a remarkable ability to maintain its proportion with respect to non-stem cells in the normal culture conditions [22]. The pure population of CSCs, when cultured in normal conditions, would eventually differentiate to non-stem cells and return to the same proportion as was in the original sample from which they were isolated. This causes an experimental constraint when culturing CSCs. They may however be cultured in serum-free media enriched with growth factors in non-adherent conditions [91] but in this process the cells do not retain a separate identity rather they grow as spheroids. Moreover, this method is expensive, takes about 7–14 days and yields fewer cells. Since, miRNAs have been shown to regulate differentiation of stem cells [92, 93], it would be a good strategy to explore miRNAs which prevent the differentiation of CSCs but maintain its properties of self renewal and chemoresistance. Such miRNAs would make it easier to culture CSCs and make them more amenable to further experimentation. Also, many tumors have been reported to have more than one type of CSCs [94]. Even the methods used to isolate the CSC subpopulation tend to pick only a subset of CSCs. This observation complicates the therapeutic research targeting CSCs since eradication of one subpopulation may not be sufficient as other CSC subpopulation may continue to cause tumor formation. Hence, there is a need to develop more robust method for identification of CSCs. Since miRNAs are differentially expressed in CSC subpopulation, they may be explored for their role in detection of CSC subpopulations in cancer patients.

All the miRNA studies in CSCs have not only used different cancer cell lines or patient samples at different stages of cancer but have also utilized different methods of isolating CSC population. Considering the variable functions that miRNAs accomplish in different cellular contexts, it might be difficult to extrapolate the findings of one study to another. Nevertheless, these studies have proved the potential of miRNAs in regulating CSCs and a more concerted approach in selection of cell lines, patient samples, and isolation techniques of CSCs is warranted in future studies.