A relatively rare aldehyde dehydrogenase 1 (ALDH1)–positive “stem cell–like” subpopulation of tumor cells has the unique ability to initiate and perpetuate tumor growth; moreover, it is highly resistant to chemotherapy and significantly associated with poor clinical outcomes. The development of more effective therapies for cancer requires targeting of this cell population. Using cDNA microarray analysis, we identified that the expression of the Caenorhabditis elegans lin-28 homologue (LIN28) was positively correlated with the percentage of ALDH1+ tumor cells; this was further validated in an independent set of tissue arrays (n = 197). Both loss-of-function and gain-of-function studies showed that LIN28 plays a critical role in the maintenance of ALDH1+ tumor cells. In addition, we found that there is a double-negative feedback loop between LIN28 and let-7 in tumor cells, and that let-7 negatively regulates ALDH1+ tumor cells. Finally, we report that a LIN28/let-7 loop modulates self-renewal and differentiation of mammary gland epithelial progenitor cells. Our data provide evidence that cancer stem cells may arise through a “reprogramming-like” mechanism. A rebalancing of the LIN28/let-7 regulatory loop could be a novel therapeutic strategy to target ALDH1+ cancer stem cells. Cancer Res; 70(22); 9463–72. ©2010 AACR.
Aldehyde dehydrogenase 1 (ALDH1) catalyzes the irreversible oxidation of a range of aliphatic and aromatic aldehydes to their corresponding carboxylic acids. High endogenous ALDH1 activity has been detected in normal stem and progenitor cells of various lineages including hematopoietic, mesenchymal, neural, mammary, and prostate (1–3). Recently, ALDH1 has been successfully used to identify a unique “stem cell–like” subpopulation of cells from tumors, which shares novel characteristics with normal embryonic and somatic stem/progenitor cells: self-renewal and multipotent differentiation (1). Preclinical studies have further shown that this subpopulation of cells is highly tumorigenic and resistant to chemotherapy (4–7). Moreover, these cells promote and mediate tumor metastasis in vivo (8–11). Importantly, a high percentage of ALDH1+ cells in most types of epithelial tumors, such as breast (1, 10, 12, 13), lung (14), pancreatic (15), bladder (16), ovarian (7), and prostate (17), is associated with a poorer clinical outcome for these patients. This provides robust clinical evidence that the ALDH1+ tumor cells play a critical role in cancer initiation and progression. Therefore, more effective cancer therapies may be developed by targeting this cell population. However, our knowledge of the cellular and molecular regulation of ALDH1+ cancer stem cells of human tumors is limited. In this study, an important ALDH1+ tumor cell regulatory loop between reprogramming factor LIN28 and miRNA let-7 was identified by high-throughput profiling and was then further characterized in human breast and ovarian tumors.
Materials and Methods
Patients and specimens
The ovarian cancer specimens for microarray analysis (n = 26; Supplementary Table S1) were collected at the University of Turin (Turin, Italy). Detailed information is provided in Supplementary Materials and Methods.
Cell lines and cell culture
T47D and MCF7 cells were purchased from the American Type Culture Collection (ATCC), A2780 and 2008 cells were obtained from the Ovarian Cancer Tissue and Cell Bank, and HeLa tet-on cells were purchased from Clontech.
RNA isolation and cDNA microarray analysis
Total RNA was isolated with Trizol reagent (Invitrogen). cDNA microarrays were performed on the human U133+ 2.0 GeneChip (Affymetrix). Detailed information is provided in Supplementary Materials and Methods.
The tissue microarray was constructed at the University of Helsinki. Detailed information is provided in Supplementary Materials and Methods.
Immunohistochemistry and image analysis
Immunohistochemistry was performed using the Vectastain ABC kit as described by the manufacturer (Vector). The following primary antibodies were used in this study: mouse anti-human ALDH1 (clone: 44/ALDH, 1:250; BD Pharmingen) and rabbit anti-human LIN28 (1:4,000; Abcam). Antibodies were incubated overnight at 4°C, and the immunoreaction was visualized using 3,3′-diaminobenzidine. The image was analyzed using Image-Pro Plus 4.1 software (Media Cybernetics).
Lentiviral transduction and generation of stable cell lines
Two individual lentiviral short hairpin RNA (shRNA) clones targeting LIN28 were purchased from Open Biosystems. eGFP shRNA and nontarget shRNA were used as controls. The pSin-EF2-LIN28 lentiviral expression vector was purchased from Addgene. Lentiviral vector and packing vectors were transfected into the packaging cell line 293T (ATCC) using FuGENE 6 Transfection Reagent (Roche). The medium was changed 8 hours after transfection, and the lentivirus-containing medium was collected 48 hours later.
Protein isolation and Western blots
Cells were lysed in 200 μL mammalian protein extraction reagent (Pierce). Then, 15 μg of total protein were separated by 10% SDS-PAGE under denaturing conditions and transferred to a polyvinylidene difluoride membrane (Millipore). Membranes were blocked in 5% nonfat milk (Bio-Rad) and then incubated with an anti-LIN28 primary antibody (1:10,000; Abcam), followed by incubation in anti-rabbit secondary antibody conjugated with horseradish peroxidase (HRP; 1:10,000; Amersham Biosciences) together with an HRP-conjugated primary antibody to β-actin (1:10,000; Sigma). Immunoreactive proteins were visualized using the LumiGLO chemiluminescent substrate (Cell Signaling).
ALDEFLUOR assay and fluorescence-activated cell sorting analysis
ALDH1 activity was detected using the ALDEFLUOR assay kit (StemCell Technologies) as described by Ginestier and colleagues (1).
Mammosphere cultures were performed as described by Dontu and colleagues (18). Detailed information is provided in Supplementary Materials and Methods.
Detailed information is provided in Supplementary Materials and Methods.
Quantitative real-time reverse transcription-PCR and Taqman miRNA assay
Detailed information is provided in Supplementary Materials and Methods.
let-7–responsive sensor construction and transfection
The let-7 sensor was constructed by introducing two copies of let-7b perfect complement sequences into the 3′ untranslated region (UTR) of the Renilla luciferase gene in the psiCHECK-2 vector (Promega).
miRNA in situ hybridization and image analysis
In situ detection of miRNA expression was performed on tissue microarray sections by locked nucleic acid (LNA) probes (Exiqon). Detailed information is provided in Supplementary Materials and Methods.
Retroviral transduction and stable cell line generation
The retroviral human miRNA expression vector (Supplementary Fig. S1A) was purchased from GeneService. Detailed information is provided in Supplementary Materials and Methods.
Tet-on–inducible cell lines
Stable cell lines inducibly expressing let-7b were generated using the retrovirus-based RevTet-On system (Clontech). The human genomic sequence of let-7b and an upstream reporter gene (DsRed) were cloned into the pRevTRE response vector downstream of the tetracycline-responsive element (TRE; Supplementary Fig. S1B), and then pRevTet-On and pRevTRE-DsRed-let-7b were separately introduced into HeLa cells by retroviral gene transfer. The reporter gene expression was monitored by fluorescent microscopy (Supplementary Fig. S1C) and fluorescence-activated cell sorting (FACS) analysis (Supplementary Fig. S1D), and let-7b expression was measured using real-time reverse transcription-PCR (RT-PCR; Applied Biosystems).
Transfection of let-7 mimic and inhibitor oligonucleotides
Pre-miR miRNA precursor and control oligos were purchased from Ambion, and miRCURY LNA miRNA inhibitors and control oligos were purchased from Exiqon. Transfections were performed using the Lipofectamine RNAiMAX transfection reagent (Invitrogen), and then cells were incubated in the medium containing the transfection mixture for 72 hours.
3′UTR reporter construct and assay
The full-length sequence of human LIN28 3′UTR was cloned from human genomic DNA. The PCR products were ligated to the PCR2.1 TOPO cloning vector (Invitrogen) and subcloned into psiCHECK-2 reporter vector (Promega). Mutagenesis of miRNA binding sites on reporter vectors was performed by the approach of overlap extension by PCR. Detailed information is provided in Supplementary Materials and Methods.
Mammary gland epithelial cell isolation, infection, and colony-forming assay
Detailed information is provided in Supplementary Materials and Methods.
Statistical analysis was performed using the SPSS statistics software package. All results were expressed as mean ± SD, with significance at P < 0.05.
LIN28 expression is positively correlated with a higher percentage of ALDH1+ tumor cells
To explore the molecular mechanisms regulating ALDH1+ tumor cells, we chose 26 human ovarian tumor specimens (Supplementary Table S1), in which the ALDH1 expression was characterized by immunohistochemistry, and the percentage of ALDH1+ tumor cells was scored by two independent investigators. Using the median value of ALDH1+ cells observed in these tumors (7.5%) as cutoff point, the tumors were divided into two groups: ALDH1 low (n = 13, ALDH1+ = 2.6 ± 2.3%) and ALDH1 high (n = 13, ALDH1+ = 48.8 ± 37.5%). An Affymetrix microarray was then used to characterize the transcriptional signature of these tumors. In summary, we found that the expression of 59 genes was significantly different between the ALDH1-high and ALDH1-low groups (P < 0.002; Supplementary Table S2; Fig. 1A). Interestingly, we found that expression of the homologue of Caenorhabditis elegans lin-28 (LIN28) was remarkably higher in the ALDH1-high group (P = 0.00007). LIN28 is an evolutionarily conserved RNA binding protein, which is highly expressed in embryonic stem (ES) cells, progenitor cells, and developing tissues but not in most adult organs (19–24). Together with OCT4, SOX2, and NANOG, LIN28 can act as a reprogramming factor, reprogramming somatic cells to induced pluripotent stem cells (iPS cells; ref. 25). LIN28/LIN28B has also been identified as an oncogene that is upregulated/reactivated in tumors, promoting transformation (26–30). These results suggest that LIN28 may be one of the genes involved in the regulation of ALDH1+ tumor cells. To validate this result, we analyzed LIN28 expression and the percentage of ALDH1+ tumor cells in an independent ovarian cancer tissue array set (n = 197) using immunohistochemistry (Fig. 1B and C). Consistent with the cDNA microarray result, the percentage of ALDH1+ tumor cells was significantly higher in LIN28+ tumors (n = 19) compared with LIN28− tumors (n = 178; P = 0.007; Fig. 1D). Similar correlation was also observed in a breast cancer tissue array set (n = 69; P = 0.045). This positive correlation of LIN28 expression with ALDH1+ tumor cells indicates that LIN28 may play a functional role in maintenance of ALDH1+ cells in human tumors.
LIN28 plays a functional role in the maintenance of ALDH1+ tumor cells
To further examine whether LIN28 plays a role in the regulation of the ALDH1+ cell population, we chose two cancer cell lines (A2780 and T47D) that highly express LIN28. Using lentiviral shRNA vectors, LIN28 was specifically knocked down in these cell lines (Fig. 2A). The percentage of ALDH1+ cells in these two cell lines was measured by the ALDEFLUOR assay, which was originally developed to detect ALDH1 activity in hematopoietic tissues, and that has since been successfully applied to detect progenitor and cancer stem cells in nonhematopoietic tissues such as the mammary gland and breast cancer (1). We found that the knockdown of LIN28 expression in these cells significantly decreased the brightly fluorescent ALDH1 (ALDH1br) cell population (Fig. 2B), and that this occurred in a dose-dependent manner; the two independent shRNA clones knocked down LIN28 expression to different levels, leading to different levels of ALDH1+ cells in each population. Similar observation was also found in vivo (Supplementary Fig. S3). To confirm this observation, we also ectopically expressed LIN28 in a LIN28− cell line (MCF7; Fig. 2C). Consistent with the shRNA study, enforced expression of LIN28 remarkably increased the ALDH1br population (Fig. 2D). Then, we examined the self-renewal capability of MCF7 cells using a mammosphere assay. Ectopically expressing LIN28 produced a significant increase in the number of spheres, especially in the third generation (Fig. 2E and F). Finally, other cancer stem cell markers such as CD133 and CD24/CD44 were also examined (Supplementary Fig. S4). Taken together, these results show that LIN28 plays a functional role in the maintenance of the ALDH1+ cancer stem cell population in tumors.
LIN28 modulates the biogenesis of miRNA let-7 in tumor cells
Next, we examined the mechanism by which LIN28 regulates the ALDH1+ cell population in tumors. Results from previous studies suggest that LIN28 maintains embryonic and somatic stem/progenitor cell pluripotency by blocking miRNA let-7 maturation (22, 31–36). However, it is still unknown whether upregulated/reactivated LIN28 also regulates let-7 maturation in cancer, because tumor cells may be absent in the molecular context for LIN28/let-7 regulation. To address this question, we used RNA immunoprecipitation and real-time RT-PCR to show that LIN28 was able to specifically bind to the let-7 precursor in A2780 cells (Fig. 3A and B). In addition, we found that knocking down LIN28 expression by shRNAs significantly increased expression of mature let-7 in a dose-dependent manner in A2780 cells (Fig. 3C and D). Finally, using a let-7 sensor assay, which contained a constitutively expressed reporter bearing sequences complementary to let-7 in the downstream 3′UTR, we showed that blocking LIN28 in A2780 cells remarkably increased let-7 activity (Fig. 3E and F). Similar observations were also found in T47D cells (Supplementary Fig. S5). Taken together, we conclude that as in embryonic and somatic stem/progenitor cells (22, 31–36), upregulated/reactivated LIN28 modulates miRNA let-7 maturation and activity in tumor cells.
let-7 modulates the ALDH1+ tumor cell population in cancer
Given that let-7 has been shown to regulate the differentiation of somatic stem/progenitor cells, as well as cancer stem cells, into differentiated cells (37), we hypothesized that LIN28 may maintain ALDH1+ cell populations by modulating miRNA let-7 maturation. To test this hypothesis, we first asked whether there was a negative correlation between let-7 expression and ALDH1+ cell numbers in tumors. Using the same tissue array set as above (Fig. 1B and C), let-7 expression was detected by in situ hybridization (Fig. 4A and B), and the hybridization signals were classified into two groups: no staining/weak cytoplasmic signals (let-7 low) or moderate/strong cytoplasmic signals (let-7 high). We found that, in contrast to LIN28, the percentage of ALDH1+ tumor cells in the let-7 low group (n = 141) was significantly higher than the let-7 high group (n = 56; P = 0.018; Fig. 4C). Similar correlation was also observed in a breast cancer tissue array set (n = 69; P = 0.016). These observations indicate that let-7 may be involved in the differentiation of ALDH1+ tumor cells in cancer. To further functionally test our hypothesis, we stably overexpressed let-7 in tumor cell lines using retroviral vectors (Supplementary Fig. S3A). As expected, increased let-7 expression dramatically reduced the number of ALDH1br cells (Fig. 5A). To confirm this observation, we generated an inducible cell line in which let-7 expression was controlled by the administration of doxycycline (Supplementary Fig. S1B–E). We found that the ALDH1br cell population was negatively associated with increased let-7 expression in a dose-dependent manner (Fig. 5B). Importantly, in our polyclonal inducible cells, only ∼30% of the cells responded to doxycycline treatment, as detected by the reporter gene DsRed (Supplementary Fig. S1D; Fig. 4C). This allowed us to separate the doxycycline-treated cells into two subpopulations: DsRed−/let-7low and DsRed+/let-7high (Fig. 5C). We found that the percentage of ALDH1br cells was >6-fold lower in the let-7–induced cells (DsRed+/let-7high) compared with the noninduced let-7 cells (DsRed−/let-7low; Fig. 5C). Additionally, using a let-7 inhibitor, we blocked endogenous let-7 expression in the above cell lines and found that downregulation of let-7 significantly increased the number of ALDH1br cells (Fig. 5D). Finally, we examined the effect of let-7 on the self-renewal ability of MCF7 cells using the mammosphere assay. In contrast to LIN28, overexpression of let-7 led to a decrease in the number of mammospheres, whereas a let-7 inhibitor led to an increase in the number of mammospheres in MCF7 cells (all P < 0.05; Fig. 5D–F). Taken together, these results show that the expression of mature let-7 modulates the ALDH1+ tumor cell population in cancer.
let-7 targets LIN28 expression in tumor cells
Interestingly, when we treated LIN28+ tumor cells (A2780 and T47D) with a let-7 mimic, we found that the expression of both LIN28 protein and mRNA was significantly decreased (Fig. 6B and C), suggesting that let-7 was able to regulate LIN28 via a feedback loop in tumor cells. We predicted that there was a conserved let-7 binding site in the LIN28 3′UTR by TargetScan (Fig. 6A); this was then confirmed experimentally. Using a LIN28 3′UTR reporter assay, we showed that transfection of a let-7 mimic was able to significantly reduce luciferase activity in the wild-type but not the let-7 binding site mutant LIN28 3′UTR reporters (Fig. 6E). The above results show the existence of a feedback regulatory loop between let-7 and LIN28 in tumor cells.
The LIN28/let-7 loop controls mammary gland progenitor cell differentiation
Our studies show that the LIN28/let-7 loop regulates the ALDH1+ cancer stem cell population in tumors, suggesting that this regulatory loop may also play an important role in maintenance of stem/progenitor cells under normal physiologic conditions. To test this hypothesis, we used a mouse mammary gland epithelial cell model. A standard protocol was used to isolate the heterogeneous mouse mammary gland epithelial cells containing epithelial progenitor cells. Then, LIN28 was introduced into the mammary gland epithelial cells by lentiviral infection (Fig. 7A). Protein expression was confirmed by Western blot analysis (data not shown). After short-term (48–72 hours) cultures, the percentage of the ALDHbr population was analyzed by FACS. Enforced LIN28 expression significantly increased the percentage of ALDHbr cells in the population (Fig. 7B). Another mammary gland progenitor cell marker, CD24/CD49f, was also examined (Supplementary Fig. S6). To further confirm that LIN28 indeed led to an increase in the number of mammary gland progenitor cells, we quantified the number of progenitors (colony-forming cells) using a mammary colony-forming assay in which the number of clones reflects the self-renewal capacity, whereas the type of the clones indicates the bipotent differentiation ability (Fig. 7C). In summary, LIN28 significantly increased the numbers of all three types of colonies (Fig. 7D), suggesting that LIN28 was able to increase the self-renewal capacity of the mammary gland epithelial progenitor cells. In addition, we found that the enforced expression of let-7 led to the differentiation of progenitor cells and significantly decreased total colony numbers (Fig. 7E). Interestingly, let-7 selectively decreased progenitor cells, which had the ability to form the luminal or mixture type of colonies. Taken together, these results show that a LIN28/let-7 regulatory loop plays a functional role in the maintenance of mammary gland epithelial progenitor cells.
Recent research on iPS cells supports the hypothesis that cancer stem cells may arise through a “reprogramming-like” mechanism (38). By enforcing the expression of a set of genes, the so-called reprogramming factors, differentiated somatic cells can be converted to iPS cells, which have the same capabilities as ES cells to give rise to all tissue types of the body. Interestingly, most of these reprogramming factors are overexpressed or upregulated in certain types of human tumors, and at least some of them (e.g., c-MYC, KLF4, SOX2, and LIN28) are established or putative oncogenes (38). Moreover, five independent studies have shown that disabling p53, an essential tumor-suppressor gene, remarkably improves the efficiency of iPS cell production (38). Therefore, there may be overlapping mechanisms that control the functions and maintenance of iPS cells and cancer stem cells (38). LIN28, one of reprogramming factors discussed above, is very restricted in its expression; it is found only in ES cells, developing tissues, and tumors (19–24, 26–30). In human tumors, LIN28/LIN28B is upregulated and functions as an oncogene promoting malignant transformation and tumor progression (26–30). We have shown that LIN28 is positively correlated with the percentage of ALDH1+ tumor cells in cancer, suggesting that LIN28 may play a role in regulation of ALDH1+ tumor cells. In further functional studies, we have shown that LIN28 also functionally maintains this cell population. Our data provided evidence that cancer stem cells may arise through a reprogramming-like mechanism; reactivated reprogramming factors such as LIN28 may promote the conversion of epithelial cells to a more undifferentiated stage and furthermore maintain a small subpopulation of tumor cells in this stem-like stage. In agreement with our findings, Peng and colleagues (30) recently reported that LIN28 together with OCT4 identified a subpopulation of stem cell–like cells in ovarian carcinoma. However, several critical questions need to be addressed further, including the mechanisms by which these reprogramming factors are reactivated in tumors (because most of them are completely silenced after development), the interactions between these reprogramming factors during the “reprogramming” process, and cellular safeguard mechanisms such as p53 that respond to these reprogramming events.
The present study also suggests that the maintenance of the ALDH1+ tumor cell population by LIN28 is mediated by the regulation of let-7, and that a LIN28/let-7 regulatory loop controls ALDH1+ cancer stem cells. The tumor suppressor role of let-7 in cancer was first shown by the Slack laboratory (39). They found that the let-7 family negatively regulates let-60/RAS in C. elegans by binding to multiple let-7 complementary sites in its 3′UTR (39). Moreover, having found that let-7 expression is lower in lung tumors than in normal lung tissue, whereas RAS protein is significantly higher in lung tumors, let-7 was proposed as a tumor suppressor gene (39–41). Increasing evidence indicates that let-7 plays a functional role in normal and cancer stem cell differentiation. First, in C. elegans, let-7 times seam cell, the stem cells that divide asymmetrically during each larval stage, terminal differentiation, possibly by acting as a regulator of multiple genes required for cell cycle and proliferation (42–45). Second, in ES cells, mature let-7 is poorly expressed, although its precursor transcripts are readily detected (46). This may oppose the actions of a family of cell cycle–regulating miRNAs that maintain self-renewal in ES cells (47). Third, in mammalian embryonic and somatic stem cells, let-7 interacts with two iPS genes, MYC and LIN28, and these autoregulatory loops [MYC/let-7 (27) and LIN28/let-7 (22, 31–36)] may control stem cell self-renewal and differentiation (37). Forth, Nishino and colleagues have shown that during aging, elevated levels of let-7b block HMGA2 and contribute to declining neural stem cell function; in contrast, HMGA2 maintains neural stem cell function in young mice through repression of the Ink4a/Arf locus (48). Fifth, Ibarra and colleagues have found that let-7 is depleted in a population of self-renewing mammary epithelial progenitor cells that can reconstitute the mammary gland. Enforced expression of let-7 leads to a loss of these self-renewing cells from mixed cultures, suggesting that let-7 plays a role in the regulation of progenitor cell maintenance (3). Finally, by comparing miRNA expression in self-renewing and differentiated breast tumor cells, Yu and colleagues (49) found that let-7 was markedly reduced in breast cancer stem cells but increased on differentiation. These findings show that let-7 plays a functional role in normal and cancer stem cells. In support of these notions, we have shown that let-7 plays an opposing function to LIN28 in regulating the ALDH1+ tumor cell population. Importantly, we reported a double-negative feedback regulating loop of LIN28 and let-7 in tumor cells. It has been widely reported that the let-7 family of genes is globally downregulated in cancer (37, 39, 46), which may be one of mechanisms of LIN28 reactivation in tumors. Finally, LIN28 may regulate the ALDH1+ cell population through a let-7–independent pathway (20). For example, LIN28 may directly regulate other reprogramming factors in cancer stem cells, such as OCT4 (30, 50).
In summary, we have identified a novel mechanism regulating ALDH1+ cancer stem cells, which could lead to new therapeutic strategies for targeting the ALDH1+ tumor cell population. For example, recent studies indicate that LIN28 may cooperate with a terminal uridylyltransferase (TUTase) to regulate let-7 maturation (23, 36). Given that polymerases are facile targets for pharmacologic inhibition by small chemical compounds, TUT4 may prove to be a novel target for manipulating the LIN28/let-7 regulation loop (23). Additionally, nanoparticle-delivered LIN28 siRNA or let-7 mimics may also be an attractive therapeutic strategy to target ALDH1+ cancer stem cell population in tumors.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant Support: Breast Cancer Alliance, Ovarian Cancer Research Found, Mary Kay Ash Charitable Foundation, National Cancer Institute grant R01CA142776, Ovarian Cancer SPORE grant P50CA83638-7951, and U.S. Department of Defense. X. Lin was supported by the China Scholarship Council.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received June 30, 2010.
- Revision received September 8, 2010.
- Accepted September 13, 2010.
- ©2010 American Association for Cancer Research.