Steven M. Johnson ¹, Helge Grosshans ¹, Jaclyn Shingara ², Mike Byrom ², Rich Jarvis ², Angie Cheng ², Emmanuel Labourier ², Kristy L. Reinert ¹, David Brown ², and Frank J. Slack ^1, *

¹ Department of Molecular, Cellular and Developmental Biology, Yale University, P.O. Box 208103, New Haven, Connecticut 06520
² Ambion, Inc., 2130 Woodward, Austin, Texas 78744

* Phone:   203-432-3492;   Fax:   203-432-6161
E-mail:  frank.slack@yale.edu

MicroRNAs (miRNAs) are regulatory RNAs found in multicellular eukaryotes, including humans, where they are implicated in cancer. The let-7 miRNA times seam cell terminal differentiation in C. elegans. Here we show that the let-7 family negatively regulates let-60/RAS. Loss of let-60/RAS suppresses let-7, and the let-60/RAS 3'UTR contains multiple let-7 complementary sites (LCSs), restricting reporter gene expression in a let-7-dependent manner. mir-84, a let-7 family member, is largely absent in vulval precursor cell P6.p at the time that let-60/RAS specifies the 1° vulval fate in that cell, and mir-84 overexpression suppresses the multivulva phenotype of activating let-60/RAS mutations. The 3'UTRs of the human RAS genes contain multiple LCSs, allowing let-7 to regulate RAS expression. let-7 expression is lower in lung tumors than in normal lung tissue, while RAS protein is significantly higher in lung tumors, providing a possible mechanism for let-7 in cancer.

Hundreds of noncoding, regulatory RNAs known as miRNAs (Bartel, 2004), are encoded in animal and plant genomes (Lagos-Quintana et al., 2001, Lau et al., 2001 and Lee and Ambros, 2001). miRNAs have emerged as important regulators of development and control processes such as cell fate determination and cell death (Abrahante et al., 2003, Brennecke et al., 2003, Chang et al., 2004, Chen et al., 2004, Johnston and Hobert, 2003, Lee et al., 1993, Lin et al., 2003, Moss et al., 1997, Reinhart et al., 2000, Slack et al., 2000 and Wightman et al., 1993). Mounting evidence shows that miRNAs are mutated or poorly expressed in human cancer (Calin et al., 2002, Gauwerky et al., 1989, Lagos-Quintana et al., 2002, McManus, 2003, Michael et al., 2003, Takamizawa et al., 2004, Tam, 2001 and Tam et al., 2002), suggesting that miRNAs may act as tumor suppressors or oncogenes. In animals, miRNAs usually control gene expression through complementary elements in the 3' untranslated regions (UTRs) of their target messenger RNAs (mRNAs) (Lee et al., 1993, Moss et al., 1997, Reinhart et al., 2000, Slack et al., 2000, Vella et al., 2004 and Wightman et al., 1993). However, the targets of few mammalian miRNAs are known.

lethal-7 (let-7), a founding member of the miRNA family, is required for timing of cell fate determination in C. elegans (Pasquinelli et al., 2000 and Reinhart et al., 2000). In wild-type animals, temporal upregulation of let-7 miRNA in the seam cells (Johnson et al., 2003 and Reinhart et al., 2000) is required for their terminal differentiation at the adult stage (Reinhart et al., 2000) when these cells exit the cell cycle, fuse together, and excrete cuticular alae (Sulston and Horvitz, 1977). In let-7 mutants, seam cells fail to exit the cell cycle and terminally differentiate at the correct time and instead divide (Reinhart et al., 2000), a hallmark of cancer. let-7 is conserved in many phyla and, like in C. elegans, is temporally regulated in Drosophila and zebrafish (Pasquinelli et al., 2000). In humans, various let-7 genes have been reported to map to regions deleted in human cancers (Calin et al., 2004), and let-7 is poorly expressed in lung cancers (Takamizawa et al., 2004), suggesting that let-7 miRNAs may be tumor suppressors. In support of this, overexpression of let-7 inhibited cell growth of a lung cancer cell line in vitro (Takamizawa et al., 2004). However, the mechanism by which let-7 regulates cell cycle exit in C. elegans and human cells is unknown.

C. elegans let-7, mir-48, mir-84, and mir-241 encode four developmentally regulated miRNAs that comprise the let-7 family (Lau et al., 2001, Lim et al., 2003 and Reinhart et al., 2000). This family displays high sequence identity, with particular conservation at the 5' end of the mature miRNAs (see Figures S1A and S1B in the Supplemental Data available with this article online). The functions of these family members are unknown. In this paper, we show a role for one of the C. elegans let-7 family miRNAs, mir-84, in vulval development, a model for dissecting RAS/MAP kinase signaling (Wang and Sternberg, 2001). We also show that C. elegans let-60/RAS is regulated by members of the let-7 family. let-7 and mir-84 are complementary to multiple sites in the 3'UTR of let-60/RAS. let-7 and mir-84 are expressed in a reciprocal manner to let-60/RAS in the hypodermis and the vulva, respectively. let-7 and mir-84 genetically interact with let-60/RAS, consistent with negative regulation of RAS expression by let-7 and mir-84. Our results also demonstrate that miRNAs regulate human RAS, a critical oncogene. We find that all three human RAS genes have let-7 complementary sites in their 3'UTRs that subject the oncogenes to let-7 miRNA-mediated regulation in cell culture. Lung tumor tissues display significantly reduced levels of let-7 and significantly increased levels of RAS protein relative to normal lung tissue, suggesting let-7 regulation of RAS as a mechanism for let-7 in lung oncogenesis.

let-60/RAS as a Target of the let-7 miRNA in C. elegans

The let-7 miRNA is temporally expressed in C. elegans (Johnson et al., 2003, Pasquinelli et al., 2000 and Reinhart et al., 2000) where it downregulates at least two target genes, lin-41 (Slack et al., 2000) and hbl-1 (Abrahante et al., 2003 and Lin et al., 2003), mutations in which lead to precocious terminal differentiation of seam cells. To better understand the role of let-7 in C. elegans seam cell differentiation and its potential role in humans, we sought to identify additional targets of let-7. We performed a computational screen for C. elegans genes with let-7 family complementary sites (LCS) in their 3'UTR (Grosshans et al., 2005). One of the top-scoring genes was let-60, encoding the C. elegans ortholog of the human oncogene RAS. We identified eight LCSs in the 3'UTR of let-60 with features resembling validated LCSs (Lin et al., 2003; Reinhart et al., 2000; Slack et al., 2000; Vella et al., 2004; Figure 1A). Many of the identified sites were found in the 3'UTR of let-60 from the closely related nematode C. briggsae (Stein et al., 2003; Figures 1A, S2A, and S2B), suggesting that they are likely to be biologically significant. An additional three sites were found in the let-60/RAS coding sequence as well as 10 other nonconforming 3'UTR sites that may also bind to let-7 family miRNAs (Figure S2A).

Potential LCSs in C. elegans let-60/RAS and in Mammalian RAS Genes

Figure 1. Potential LCSs in C. elegans let-60/RAS and in Mammalian RAS Genes

(A) C. elegans let-60/RAS mRNA 3'UTR, black arrows indicate sites with similarity between C.e. and C.b. and white arrows indicate nonsimilar sites. Shown below are predicted duplexes formed by LCSs (top) and miR-84 (bottom). let-7 and miR-84 are so similar that most let-7 sites are also potential miR-84 sites.

(B–D), Human NRAS, KRAS, and HRAS mRNA 3'UTRs have nine, eight, and three potential LCSs, respectively. Black arrows indicate sites conserved among mammalian species (in most cases human, rat, mouse, hamster, and guinea pig). Shown below are hypothesized duplexes formed by LCSs (top) and let-7a miRNA (bottom).

let-7(n2853ts) loss-of-function (lf) mutants express reduced let-7 miRNA and die by bursting at the vulva at the nonpermissive temperature (Reinhart et al., 2000 and Slack et al., 2000). Lf mutations in two previously identified targets of let-7, lin-41, and hbl-1 have the property of partially suppressing the let-7 lethal phenotype (Abrahante et al., 2003, Lin et al., 2003, Reinhart et al., 2000 and Slack et al., 2000). We found that postembryonic reduction of function of let-60 by feeding RNA interference (RNAi) also partially suppressed let-7(n2853) in a reproducible manner. While 5% of let-7 mutants grown on control RNAi survived at the nonpermissive temperature of 25°C (n = 302), 27% of let-7(n2853); let-60(RNAi) animals survived (n = 345) (Figures 2A and 2B). Thus, similar to other known let-7 targets, let-60 lf partially suppresses the let-7(n2853) lethal phenotype, suggesting that let-7 lethality may at least partially be caused by overexpression of let-60. However let-60(RNAi) did not appear to suppress the let-7 seam cell terminal differentiation defect and did not cause precocious seam cell terminal differentiation (data not shown). In addition, wild-type animals subjected to let-60(RNAi) did not display typical lethal and vulvaless phenotypes associated with let-60 alleles (Beitel et al., 1990, Han et al., 1990 and Han and Sternberg, 1990). Under our conditions, let-60(RNAi) resulted in  ~80% knockdown of let-60 mRNA (Grosshans et al., 2005), suggesting that the remaining let-60 is still sufficient for seam cell differentiation and vulval development. To verify the specificity of the let-60(RNAi), we showed that while let-7(mn112) adults all die, let-60(n2021); let-7(mn112) adults can live (see Experimental Procedures). Interestingly, let-7(n2853); let-60(RNAi) animals delivered a brood and could lay some eggs (data not shown), suggesting that the vulval-bursting phenotype of let-7 was not suppressed merely because of the lack of a vulva.

let-60 Is Regulated by the let-7 miRNA through Its 3'UTR

Figure 2. let-60 Is Regulated by the let-7 miRNA through Its 3'UTR

(A) Many let-7(n2853); let-60(RNAi) mutant animals do not burst through the vulva at the adult stage at the restrictive temperature (25°C), while all let-7(n2853) animals do (B).

(C–E) Wild-type animals carrying a reporter gene schematically shown in (C) show robust hypodermal expression of b-galactosidase at the L1 stage ([D], magnification 400×) but not at the adult stage ([E], composite of multiple images shot at 400×). Asterisk in (E) indicates an embryo with b-galactosidase activity inside the adult animal, providing an internal control for staining.

(F) Quantitative analysis of the expression pattern from five independent wild-type transgenic lines grown at 20°C. We observed at least 25% repression in all lines. A nonregulated lin-41 3'UTR missing its LCSs (pFS1031) (Reinhart et al., 2000), tested in duplicate is shown as a control.

(G) Downregulation of the reporter gene expression is lost in let-7(n2853) mutant worms grown at the permissive temperature, 15°C. The parental (N2) line was tested in triplicate; four isogenic let-7(n2853) mutant lines were tested. Error bars represent standard deviations.

let-60 and let-7 are both expressed in hypodermal seam cells (Dent and Han, 1998 and Johnson et al., 2003). We fused the let-60 3'UTR behind the Escherichia coli lacZ gene driven by the hypodermally expressing col-10 promoter (Figure 2C). We found that reporter gene activity is downregulated around the L4 stage (Figures 2D–2F), around the same time that let-7 is expressed in the seam cells (Johnson et al., 2003). In contrast, the same reporter gene fused to an unregulated control 3'UTR was expressed at all stages (Figure 2F; Reinhart et al., 2000, Slack et al., 2000, Vella et al., 2004 and Wightman et al., 1993). We found that reporter downregulation directed by the let-60 3'UTR depended on a wild-type let-7 gene, since downregulation failed in let-7(n2853) mutants (Figure 2G). Thus, multiple lines of evidence strongly suggest that let-60 is negatively regulated by let-7. First, the let-60 3'UTR contains multiple elements complementary to let-7; second, the let-60 3'UTR directs downregulation of a reporter gene in a let-7 dependent manner; third, this downregulation is reciprocal to let-7 upregulation in the hypodermis; and finally, let-60 loss of function partially suppresses the let-7 lethal phenotype.

The let-7 Family Member mir-84 Is Dynamically Expressed in the Vulval Precursor Cells

let-60/RAS is best understood for its role in vulval development (Wang and Sternberg, 2001); however, let-7 has not been reported to be expressed in the vulva. In C. elegans, let-7, mir-48, mir-84, and mir-241 comprise the let-7 family (Lau et al., 2001, Lim et al., 2003 and Reinhart et al., 2000; Figures S1A and S1B). Our previous work demonstrated that a let-7::gfp fusion faithfully recapitulates the temporal expression of let-7 and is temporally expressed in seam cell tissues affected in the let-7 mutant (Johnson et al., 2003). We examined the expression pattern of mir-84, the closest let-7 relative, by fusing 2.2 kilobases (kb) of genomic sequence immediately upstream of the miR-84 encoding sequence to the green fluorescent protein (gfp) gene. mir-84::gfp was first observed in the somatic gonad in larval stage 1 (L1). In L3 animals, strong expression was observed in uterine cells including the anchor cell (AC), and weak dynamic expression was observed in the vulval precursor cells (VPCs) (Figure 3). VPCs are multipotent ventral hypodermal cells that generate the vulva during L3 and later stages (Sulston and Horvitz, 1977). VPCs adopt one of three fates depending on EGF signaling from the AC (Wang and Sternberg, 2001; Figure 4H). The cell closest to the AC, P6.p, receives the most LIN-3/EGF signal (Katz et al., 1995) and adopts the primary (1°) fate through activation of a RAS/MAPK signal transduction pathway (Beitel et al., 1990, Han et al., 1990 and Han and Sternberg, 1990); P5.p and P7.p receive less LIN-3 as well as receiving a secondary lateral signal (Sternberg, 1988) from P6.p and adopt the secondary (2°) fate; P3.p, P4.p, and P8.p adopt the uninduced tertiary (3°) fate. mir-84::gfp expression was observed during the early to mid L3 stage in all the VPCs except for P6.p, in which expression was rarely observed (Figures 3A, 3P, S1C, and S1D). Subsequent VPC expression in the mid to late L3 stage was restricted to the daughters (Pn.px) of P5.p and P7.p with weaker GFP first appearing in the P6.p daughters just before their division into P6.pxx. Thereafter, equivalent expression was observed in the granddaughters (Pn.pxx) of P5.p, P6.p, and P7.p (Figures 3E and 3F). We note that mir-84::gfp expression was observed in all the VPCs except for P6.p at the stage when their fate in vulval development is determined by signaling from the AC (Figure 3O; Ambros, 1999), suggesting that mir-84 could play a role in vulval cell fate determination. In the L4 stage, GFP expression was maintained in the AC and other uterine cells, appeared weakly in hypodermal seam cells (A. Kerscher et al., submitted), and was upregulated to higher levels in many P5.p–P7.p descendants (Figure 3G). Interestingly, a second let-7 family member, mir-48, was also expressed in non-P6.p VPCs (A. Kerscher et al., submitted), suggesting the potential for redundancy between mir-48 and mir-84 in the VPCs.

mir-84::gfp and gfp60 Are Expressed Reciprocally in the Pn.p Cells

Figure 3. mir-84::gfp and gfp60 Are Expressed Reciprocally in the Pn.p Cells

(A, C, E, and G) In L3, mir-84::gfp is expressed in all Pn.p cells except P6.p during the time of vulval fate determination ([O], bracket) and is upregulated in P6.pxx cells thereafter (mL3, lL3, and vlL3 are mid, late, and very late L3, respectively).

(I, K, and M) gfp60, a fusion of gfp to the let-60 3'UTR, driven by the lin-31 promoter, is expressed in all Pn.p cells at late L2 and early L3 (I and J), is then restricted to P6.p (K and L) during vulval fate determination, and thereafter is expressed strongly in P6.px cells and weakly in P5.px and P7.px cells (M and N). Images on right are Nomarski images of animals on left.

(O) gfp60 and mir-84::gfp expression schematic from L2 Pn.p to mid L3 Pn.px.

(P–R), Quantification of expression data shown in (A–N). gfp54 is a fusion of gfp to the unc-54 3'UTR, driven by the lin-31 promoter.

Error bars represent standard deviations. See Experimental Procedures for details.

Figure 4. Overexpression of mir-84

In L3, mir-84(+++) animals display precocious division of P6.p and precocious vulval invagination (A). In L4, mir-84(+++) results in an everted and protruding vulva (B) and precocious alae formation (C).

(D–F), Overexpressing mir-84 in let-60(n1046gf) animals partially suppresses the let-60(gf) Muv phenotype, whereas an empty vector control and Dmir-84(+++) do not.

(G), Northern blot showing that mir-84(+++) animals express more miR-84, compared to wild-type animals. 5.8S RNA is shown as a loading control.

(H), Model of miR-84 modulation of let-60/RAS during vulval morphogenesis. During vulval cell fate specification a LIN-3/EGF signal emanating from the anchor cell is received by the LET-23/EGFR on the P5.p, P6.p and P7.p cells. The P6.p receives the most LIN-3 and activates a RAS/MAPK signal transduction pathway to adopt the 1° fate. P5.p and P7.p receive less LIN-3 and also receive a second, lateral signal (involving LIN-12/Notch) from the 1° cell that induces them to the 2° fate. We propose that miR-84 is expressed in the non-1° lineage to reduce expression of LET-60/RAS in these cells.

mir-84 Overexpression Causes Vulval and Seam Defects

We overexpressed miR-84 by generating transgenic animals harboring a multicopy array of a 3.0 kb genomic DNA fragment that spans from 2.2 kb upstream to 0.8 kb downstream of the miR-84 encoding sequence (called mir-84(+++)). These animals expressed elevated levels of miR-84 (Figure 4G) and displayed abnormal vulval development phenotypes, including protrusion and bursting of the vulva (40% of animals, n = 40; Figure 4B). They also displayed early division of P6.p and precocious vulval invagination in mid to late L3, a heterochronic phenotype we have not investigated further (Figure 4A). Consistent with mir-84::gfp expression in seam cells, we found that mir-84(+++) animals also exhibited precocious seam cell terminal differentiation and alae formation in the L4 stage (Figure 4C), a characteristic seen in precocious developmental timing mutants. In fact, let-7 overexpressing strains also exhibit precocious seam cell terminal differentiation in the L4 stage (Reinhart et al., 2000). In contrast, animals carrying an array containing a construct identical to mir-84(+++) except for a 75 nucleotide (nt) deletion of sequences encoding the predicted pre-mir-84 (Dmir-84(+++)) did not display any vulval or seam defects (data not shown), demonstrating that the phenotypes observed in mir-84(+++) are dependent on the miR-84 sequence.

We searched for LCSs in the 3'UTRs of all genes known to play a role in vulval development (Table S1). LCSs have the potential to bind all members of the let-7 family, including mir-84. Approximately 11 vulval genes contained at least one LCS (Table S1), raising the possibility that the let-7 family may regulate multiple genes in the vulva. In this analysis though, let-60/RAS stood out due to the high number of LCS sites.

mir-84 Overexpression Partially Suppresses let-60/RAS Gain of Function Phenotypes

let-60/RAS is active in P6.p following a lin-3 EGF signal from the anchor cell that activates a MAPK signal transduction cascade transforming P6.p to the 1° vulval fate (Han and Sternberg, 1990). Since mir-84 is expressed in all VPCs except P6.p, we examined the possibility that mir-84 negatively regulates expression of let-60/RAS in cells not destined to adopt the 1° fate. Activating mutations in let-60/RAS cause multiple VPCs (including the non-P6.p VPCs) to adopt 1° or 2° fates leading to a multivulva (Muv) phenotype (Han et al., 1990). We found that overexpression of mir-84 partially suppressed the Muv phenotype of let-60(gf) mutations. In our study 41% (n = 51) of let-60(ga89) (Eisenmann and Kim, 1997) animals displayed a Muv phenotype, while only 13% (n = 168) did so when also overexpressing mir-84 from a multicopy array (p << 0.0001, chi-square test). The same suppression was observed with a second let-60(gf) allele, let-60(n1046) (Han et al., 1990): 77% (n = 39) of let-60(n1046) animals displayed a Muv phenotype (Figure 4E), while only 50% (n = 113) did so when also overexpressing mir-84 (Figure 4D; p << 0.0001, chi-square test). let-60(n1046) animals displayed an average of 1.54 pseudovulvae per animal compared to an average of 0.66 pseudovulvae per let-60(n1046) animal overexpressing mir-84. For both let-60(gf) alleles, animals exhibiting low mosaicism for the myo-3::gfp coinjection marker were completely suppressed (data not shown), suggesting that the partial suppression was likely due to mosaicism of the transgeneic array. Neither an empty vector control (TOPO) (n = 111; p = 0.1435, chi-square test) nor the Dmir-84(+++) array (n = 129; Figure 4F), suppressed the Muv phenotype of let-60(n1046). For all let-60(gf) experiments, three independent lines behaved similarly (Figure S3C).

The let-60/RAS 3'UTR Confines Expression to P6.p

The promoter of let-60/RAS drives reporter expression in all VPCs (Dent and Han, 1998). However, the transgenic reporters used in this earlier work did not include the let-60 3'UTR. We fused GFP to the let-60 3'UTR and drove GFP expression in all the VPCs using the VPC-specific lin-31 (Tan et al., 1998) promoter (gfp60). In the late L2 and early L3 stages, GFP was expressed in all the Pn.p cells (Figure 3I), but by mid to late L3 stages, GFP was largely restricted to the P6.p cell (Figures 3K and 3Q), with some expression in the P5.p and P7.p cell descendants (Figure 3M). A similar fusion construct in which the let-60 3'UTR was replaced by the unregulated unc-54 3'UTR showed GFP expression in all Pn.p cells (Figure 3R). Since the lin-31 promoter is active in all Pn.p cells (Tan et al., 1998), this result demonstrates that the let-60/RAS 3'UTR is sufficient to downregulate a reporter gene in the non-P6.p cells.

We replaced the let-60 3'UTR with the unregulated unc-54 3'UTR in a let-60 genomic DNA fragment. While we could generate viable lines using a let-60::let-60(+)::let-60 3'UTR construct at 10 ng/ml (data not shown), we were unable to generate viable transformants using this let-60::let-60(+)::unc-54 3'UTR construct, even at 0.1 ng/ml. We did not try lower concentrations of DNA, but our result suggests that the removal of the let-60 3'UTR may severely overexpress let-60 and cause lethality.

let-60/RAS Is a Likely Target of mir-84 in the Vulva

Previous work has demonstrated that VPCs are sensitive to the levels of let-60/RAS (Beitel et al., 1990 and Han and Sternberg, 1990). Animals carrying extra copies of the wild-type let-60/RAS gene display a Muv phenotype, where non-P6.p VPCs can adopt the 1° fate. Our data strongly suggest that mir-84 negatively regulates let-60 in non-P6.p VPCs. First, mir-84 is complementary to multiple sites in the let-60 3'UTR. Second, mir-84 is expressed in a reciprocal manner to let-60 in the VPCs. miR-84 is largely absent from P6.p, at the same time as the let-60 3'UTR confines GFP expression mainly to the P6.p cell lineage. Finally, mir-84 overexpression partially suppresses the effects of activating mutations in the let-60 gene. We propose that mir-84 modulates the expression of let-60/RAS in non-P6.p VPCs to reduce flux through the RAS/MAPK signaling pathway and hence decrease the likelihood that these cells will also adopt the 1° fate (Figure 4H). However, mir-84 is clearly not the only regulator of let-60/RAS in non-P6.p cells: daf-12(rh61) mutants do not express mir-84 in any VPC (n = 60 animals), and yet daf-12(rh61) animals do not display a Muv phenotype (data not shown). Other known factors, e.g., synmuv genes (Berset et al., 2001, Ceol and Horvitz, 2004, Hopper et al., 2000, Lee et al., 1994, Wang and Sternberg, 2001, Yoo et al., 2004 and Yoon et al., 1995) or unknown factors may also regulate let-60/RAS signaling in these cells.

Our combined results provide strong evidence that let-7 and mir-84 regulate let-60/RAS expression through its 3'UTR in seam and vulval cells, cells in which they are all naturally expressed. Given that the 3'UTR of let-60/RAS contains multiple let-7/mir-84 complementary sites, we propose that this regulation is direct.

Numerous miRNAs are altered in human cancers (Calin et al., 2002, Calin et al., 2004, Michael et al., 2003 and Tam et al., 2002) and three of the best understood miRNAs, lin-4 (Lee et al., 1993), let-7 (Reinhart et al., 2000), and bantam (Brennecke et al., 2003), all regulate cell proliferation and differentiation. The closest human homologs of let-7 and mir-84 are the H.s. let-7 family miRNAs (Lagos-Quintana et al., 2002 and Pasquinelli et al., 2000). let-60/RAS is the C. elegans ortholog of human HRAS, KRAS, and NRAS (Figures S3A and S3B), which are commonly mutated in human cancer (Malumbres and Barbacid, 2003), including lung cancer. We found that all three human RAS 3'UTRs contain multiple putative let-7 complementary sites with features of validated C. elegans LCSs (Figures 1B, 1C, and 1D). Many of these are conserved in rodents, amphibians, and fish (Figure S4), suggesting functional relevance. The presence of putative LCSs in human RAS 3'UTRs suggests that mammalian let-7 family members may regulate human RAS in a manner similar to the way let-7 and mir-84 regulate let-60/RAS in C. elegans.

Human RAS Expression Is regulated by let-7 in Cell Culture

Microarray analysis on six different cell lines revealed that HepG2 cells express let-7 at levels too low to detect by microarray analysis (data not shown). We transfected HepG2 cells with a double-stranded (ds) RNA that mimics the let-7a precursor. A similar approach has been used to study other miRNAs in mammalian cells (Chen et al., 2004 and Lewis et al., 2003). Consistent with the prediction that RAS expression is negatively regulated by let-7, immunofluorescence with a RAS-specific antibody revealed that the protein is reduced by approximately 70% in HepG2 cells transfected with exogenous let-7a miRNA relative to the same cells transfected with a negative control miRNA (Figures 5A and 5B). The protein expression levels of GAPDH and p21CIP1 were largely unaffected by the transfected let-7a and negative control pre-miRNAs (Figures S5A and S5B), indicating that let-7a regulation is specific to RAS. To confirm that the RAS antibody is specific to RAS protein in the transfected cells, HepG2 cells were also independently transfected with two exactly complementary siRNAs targeting independent regions of NRAS. Both siRNAs reduced cell fluorescence by more than 60% as compared to negative control siRNA-transfected cells (Figures S5C and S5D).

The Presence of let-7 Influences the Expression of RAS in Human Cells

Figure 5. The Presence of let-7 Influences the Expression of RAS in Human Cells

(A) HepG2 cells were transfected with 10 and 30 nM of a let-7 or negative control precursor miRNA. Immunofluorescence using an antibody specific to NRAS, VRAS, and KRAS revealed that the let-7 transfected cells have much lower levels of the RAS proteins.

(B) Quantification of the RAS antibody fluorescence from replicates of the transfections shown in (A).

(C) HeLa cells were transfected with 100 nM let-7 inhibitor or negative control inhibitor. RAS immunofluorescence revealed that cells transfected with the let-7 inhibitor have increased levels of the RAS proteins relative to the negative control transfected cells.

(D) Quantification of the RAS antibody fluorescence from replicates of the transfections shown in (C).

We predicted that cells expressing native let-7 may express less RAS protein and that inhibition of let-7 may lead to derepression of RAS expression. To test this, we transfected HeLa cells, which express endogenous let-7 (Lagos-Quintana et al., 2001 and Lim et al., 2003), with antisense molecules designed to inhibit the activity of let-7 (Hutvágner et al., 2004 and Meister et al., 2004). Reducing the activity of let-7 in HeLa cells resulted in an ~70% increase in RAS protein levels (Figures 5C and 5D). These results, combined with the reciprocal experiment using pre-let-7 miRNAs discussed above, strongly suggest that let-7 negatively regulates the expression of RAS in human cells.

We fused the 3'UTR of human NRAS and KRAS to a luciferase reporter gene and transfected these constructs along with transfection controls into HeLa cells. NRAS contains two naturally occurring 3'UTRs that utilize alternative polyadenylylation and cleavage sites, such that one of the 3'UTRs is 2.5 kb longer than the other. We found that while the long NRAS 3'UTR strongly repressed reporter expression compared to an unregulated control 3'UTR (Figures 6A and 6B), the short NRAS 3'UTR led to only slight, but reproducible, repression of the reporter. The short 3'UTR contains four LCSs, while the long form contains nine LCSs. The KRAS 3'UTR also repressed the luciferase reporter (Figures 6A and 6B), while HRAS was not tested. Our results demonstrate that the 3'UTRs of NRAS and KRAS contain regulatory information, sufficient to downregulate the reporter. Little is known about the exact characteristics that convey functionality to a miRNA complementary site, and future work may reveal the differences between the NRAS and KRAS 3'UTRs.

The 3'UTRs of NRAS and KRAS Enable let-7 Regulation

Figure 6. The 3'UTRs of NRAS and KRAS Enable let-7 Regulation

(A) Cartoon showing the NRAS short (NRAS s) NRAS long (NRAS l) and KRAS 3'UTRs. Arrows indicate LCSs as described in Figure 1. The blackened areas indicate the sequence cloned behind the reporter.

(B) Relative repression of firefly luciferase expression standardized to a transfection control, renilla luciferase. pGL3-Cont is the empty vector.

(C) Induction of firefly luciferase expression when reporter plasmids with 3'UTR domains corresponding to KRAS and NRAS are cotransfected with an inhibitor of let-7, relative to a control inhibitor.

As with the endogenous RAS experiments described above, we performed the reciprocal experiment wherein we cotranfected HeLa cells with the RAS 3'UTR reporter constructs and the let-7a antisense inhibitor molecule (or a control scrambled molecule). Cells transfected with the let-7a inhibitor relieved repression exerted on the reporter relative to the control transfections (Figure 6C). Since we observe a loss in the extent of downregulation when let-7 is inhibited, these results strongly suggest that let-7 regulates NRAS and KRAS in human cells through their 3'UTRs.

let-7, RAS, and Lung Cancer

Like let-60/ras, human RAS is dose sensitive, since overexpression of RAS results in oncogenic transformation of human cells (McKay et al., 1986 and Pulciani et al., 1985). It is plausible that loss of miRNA control of RAS could also lead to overexpression of RAS and contribute to human cancer. Indeed, recent work has mapped let-7 family members to human chromosomal sites implicated in a variety of cancers (Calin et al., 2004). In particular let-7a-2, let-7c and let-7g have been linked to small chromosomal intervals that are deleted in lung cancers (Calin et al., 2004), a cancer type in which RAS misregulation is known to be a key oncogenic event (Ahrendt et al., 2001 and Johnson et al., 2001).

We utilized a miRNA microarray to examine expression levels of members of the let-7 gene family in tissue from 21 different cancer patients, including 12 lung cancer patients with squamous cell carcinomas (stage IB or IIA). We found that the let-7 miRNAs were reduced in expression in a number of the tumors relative to the normal adjacent tissue (NAT) samples from the same patients (Figure 7A). Interestingly, we found that let-7 was expressed at lower levels in all of the lung tumor tissues (Figure 7) but only sporadically in other tumor types. A similar finding was independently discovered (Takamizawa et al., 2004). On average, let-7 was expressed in lung tumors at less than 50% of what it was expressed in the associated normal lung samples (Figure 7A). We used Northern analysis to measure let-7c in the tumor and NAT samples for the two patients from which we purified enough RNA (samples represented by the first and fifth lung cancer bars in Figure 7A). Consistent with the microarray results, Northern analysis verified that the expression of let-7c was 65% lower in the tumor of patient 1 and 25% lower in the tumor of patient 5 (Figure 7B). Seven of eight examined samples also had on average 30% less let-7g expression in the tumor tissue (Figure S6). The miRNA arrays used to compare the lung tumors and NAT included probes for 167 total miRNAs; the expression of the vast majority of these were unchanged in the lung tumors (J.S. and D.B., unpublished data) indicating that let-7 might be important in lung cancer. In theory, downregulation of let-7 could result in upregulation of RAS and thus induce or accentuate oncogenesis.

let-7 Is Poorly Expressed in Lung Tumors

Figure 7. let-7 Is Poorly Expressed in Lung Tumors

(A) Expression of let-7 in 21 breast, colon, and lung tumors relative to associated NAT. Fluorescently labeled miRNA was hybridized to microarrays that included probes specific to let-7a and let-7c. Fluorescence intensities for the tumor and NAT were normalized by total fluorescence signal for all elements, and the relative average signal from the let-7 probes in the tumor and normal adjacent samples are expressed as log ratios. Note, let-7a and let-7c had similar profiles, suggesting crosshybridization between the two closely related miRNAs.

(B) Confirmation of differential let-7c expression in tumor versus normal adjacent samples. The Northern blot was assayed sequentially with radio-labeled probes specific to let-7c and 5S rRNA. The relative expression of let-7c in NAT versus tumor tissue, normalized to the 5S signal, is shown in patients corresponding to lung sample 1 and 5 from (A).

(C) Correlation between RAS protein and let-7c expression in tumor and normal adjacent tissue samples from three lung squamous cell carcinomas. GAPDH and RAS proteins were measured from crude extracts of tumor and normal adjacent tissues using Western analysis. The two proteins were assessed simultaneously by mixing the antibodies used for detection. The small RNA Northern blot was assayed sequentially with radio-labeled probes specific to let-7c and U6 snRNA. NRAS mRNA in the tumor and normal adjacent tissues samples was measured by real-time PCR. The real-time data were normalized based on the real-time PCR detection of 18S rRNA in the various samples. The relative expression of NRAS in the normal adjacent tissues was taken to be 100%, and the Ct value of NRAS in the tumor samples was used to assign the relative expression of NRAS in the tumor samples.

To test this hypothesis, we isolated total RNA and total protein from the tumor and normal adjacent tissues of three new lung cancer patients with squamous cell carcinoma. The RNA samples were split and half was used for Northern analysis to measure let-7c and U6 snRNA. The other halves of the RNA samples were used for real-time PCR to measure the NRAS mRNA, 18S rRNA, and B-actin mRNA. The protein samples were used for Western analysis to assess RAS and GAPDH protein levels. As seen in Figure 7C, RAS protein was present in the tumors at levels at least 10-fold higher than in the normal adjacent samples from the same patients. Consistent with the miRNA array results for other lung cancer samples, all three lung tumor samples had 4- to 8-fold lower levels of let-7 than did the corresponding NAT samples. Interestingly, the first and third lung cancer samples had similar levels of NRAS mRNA in both the tumor and NAT while the second sample pair had significantly higher levels of NRAS mRNA in the tumor sample. In our limited analysis, RAS protein levels correlate poorly with NRAS mRNA levels but very well with let-7 levels, suggesting that the expression of the oncogene is significantly influenced at the level of translation, consistent with the known mechanism of let-7 in invertebrates.

The reciprocal expression pattern between let-7 and RAS in cancer cells closely resembles what we saw with let-7 and RAS in C. elegans and in our human tissue culture experiments. The correlation between reduced let-7 expression and increased RAS protein expression in the lung tumor samples suggests that one or more members of the let-7 gene family regulates RAS expression in vivo and that the level of expression of the miRNA might be an important factor in limiting or contributing to oncogenesis.

Our work reveals that the let-7 miRNA family negatively regulates RAS in two different C. elegans tissues and in two different human cell lines. Strikingly, let-7 is expressed in normal adult lung tissue (this work; Pasquinelli et al., 2000) but is poorly expressed in lung cancer cell lines and lung cancer tissue (this work; Takamizawa et al., 2004). The expression of let-7 inversely correlates with expression of RAS protein in lung cancer tissues, suggesting a possible causal relationship. In addition, overexpression of let-7 inhibited growth of a lung cancer cell line in vitro (Takamizawa et al., 2004), suggesting a causal relationship between let-7 and cell growth in these cells. The combined observations that let-7 expression is reduced in lung tumors, that several let-7 genes map to genomic regions that are often deleted in lung cancer patients, that overexpression of let-7 can inhibit lung tumor cell line growth, that the expression of the RAS oncogene is regulated by let-7, and that RAS is significantly overexpressed in lung tumor samples strongly implicate let-7 as a tumor suppressor in lung tissue and suggests a possible mechanism.

Plasmid Constructs

See Supplemental Experimental Procedures for details of plasmid construction,

http://www.cell.com/cgi/content/full/120/5/635/DC1/

C. elegans and Transgenic Reporter Analysis

All animal experiments were performed at room temperature or 20°C unless stated otherwise. All experimental plasmids were injected in animals at 50–100 ng/ml. Two different markers, rol-6 (100 ng/ml) and myo-3::gfp (50 ng/ml), were separately coinjected with PSJo84. myo-3::gfp (50 ng/ml) was coinjected with o84D84, and myo-2::gfp (5 ng/ml) was coinjected with GFP60 and GFP54. These animals are mosaic for the transgenes. To compare expression between individual lines, the percent expression of GFP in each of the Pn.p cells was normalized relative to the expression of the highest expressing Pn.p cell and represented as a fraction of the highest expresser for each individual line of animals. For each construct, the average of the lines was calculated along with the standard deviation for each construct represented as error bars (mir-84::gfp n = 239, gfp60 n = 42 and gfp54 n = 40). For the mir-84(+++) analysis, animals were examined using DIC optics to score seam cell and vulval anatomy. LacZ reporter analysis was as described (Vella et al., 2004). The lin-41 3'UTR missing its LCSs (pFS1031) was used as a control (Reinhart et al., 2000). RNAi methods were standard feeding procedures using synchronized L1s (Timmons et al., 2001). All RNAi experiments were done in parallel to an empty vector (L4440) feeding control. See Supplemental Experimental Procedures for details on the let-60; let-7 double mutant cross.

let-7/RAS Association in Mammalian Cells

HeLa S3 cells grown in D-MEM (GIBCO) supplemented with 10% fetal bovine serum (GIBCO) were cotransfected in 12-well plates using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol using 1.0 mg/well of Pp-luc-expressing plasmid (pGL3-Control from Promega, pGL3-NRAS S pGL3-NRAS L and pGL3-KRAS) and 0.1 mg/well of Rr-luc-expressing plasmid (pRL-TK from Promega). 24 hr posttransfection, the cells were harvested and assayed using the Dual Luciferase assay as described by the manufacturer (Promega). HeLa cells grown as above were transfected in 24-well plates with 30 pmol of Anti-miR let-7 or negative control 1 inhibitors (Ambion) using Lipofectamine 2000. Three days posttransfection, RAS expression was monitored by immunofluorescence using an FITC conjugated primary antibody against RAS protein (US Biological). The resulting fluorescent signal was analyzed using the appropriate filter set and was quantified using MetaMorph software. We typically measured the fluorescence intensity of 150–300 cells in one or a few viewing areas. The experiments with both the precursors and the inhibitors were performed three times. The photos represent single viewing fields from one of the experiments and are representative of the triplicate experiment. Identically grown HeLa cells were cotransfected in 24-well plates using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol using 200 ng /well of Pp-luc-expressing plasmid (pGL3-Control from Promega, pGL3-NRAS S pGL3-NRAS L and pGL3-KRAS). 48 hr posttransfection, the cells were harvested and assayed using the Luciferase assay as described by the manufacturer (Promega).

HepG2 cells grown in D-MEM (GIBCO) supplemented with 10% fetal bovine serum (GIBCO) were transfected with 15 or 5 pmol of Pre-miR Let-7c or negative control 1 Precursor miRNAs (Ambion) in 24-well plates using siPort Neo-FX (Ambion) according to the manufacturer’s protocol. Three days posttransfection, RAS expression was monitored by immunofluorescence as described above.

MiRNA Microarray Analysis

Total RNA from tumor and NAT samples from three breast cancer, six colon cancer, and twelve lung cancer patients was isolated using the mirVana RNA Isolation Kit (Ambion). Twenty micrograms of each total RNA sample was fractionated by polyacrylamide gel electrophoresis (PAGE) using a 15% denaturing polyacrylamide gel, and the miRNA fractions for each sample were recovered. The miRNAs from all of the samples were subjected to a poly(A) polymerase reaction wherein amine modified uridines were incorporated as part of ~40 nt long tails (Ambion). The tailed tumor samples were fluorescently labeled using an amine-reactive Cy3 (Amersham), and the normal adjacent tissue samples were labeled with Cy5 (Amersham). The fluorescently labeled miRNAs were purified by glass-fiber filter binding and elution (Ambion), and the tumor and normal adjacent tissue samples from the same patient were mixed. Each sample mixture was hybridized for 14 hr with slides upon which 167 miRNA probes were arrayed. The microarrays were washed 3 × 2 min (min) in 2× SSC and scanned using a GenePix 4000B (Axon). Fluorescence intensities for the Cy3- and Cy5-labeled samples for each element were normalized by total Cy3 and Cy5 signal on the arrays. The normalized signal intensity for each element was compared between the tumor and NAT samples from each pair of patient samples and expressed as a log ratio of the tumor to normal adjacent sample.

Northern Analysis

mir-84 Northerns were performed as described (Johnson et al., 2003). For human tissues, 1 mg of total RNA from the tumor and normal adjacent tissues of patients 1 and 5 (Figure 7A) were fractionated by PAGE using a 15% denaturing polyacrylamide gel. The RNA was transferred to a positively charged nylon membrane by electroblotting at 200 mA in 0.5× TBE for 2 hr. The Northern blot was dried and then incubated overnight in 10 ml of ULTRAhyb-Oligo (Ambion) with 10⁷ cpm of a radio-labeled transcript complementary to let-7c. The blot was washed 3 × 10 min at room temperature in 2× SSC, 0.5% SDS and then 1 × 15 min at 42°C in 2× SSC, 0.5% SDS. Overnight phosphorimaging using the Storm system (Amersham) revealed let-7c. The process was repeated using a radio-labeled probe for 5S rRNA.

Lung Tumor Protein/Northern/mRNA Analysis

Total RNA and protein were isolated from tumor and normal adjacent tissue samples from three lung cancer patients using the mirVana PARIS Kit (Ambion). let-7 miRNA and U6 snRNA were measured using the Northern procedure described above. NRAS and B-actin mRNA as well as 18S rRNA were quantified by real-time RT-PCR using primers specific to each of the target RNAs. RAS and GAPDH protein were measured by Western analysis using the RAS antibody described above and an antibody for GAPDH (Ambion).

We thank S. Kim for the lin-31 promoter (PB255), T. Johnson and M. Gerstein for help with the computational analysis, A. Fire for pPD95.70, the Caenorhabditis Genetics Stock Center for strains, and K. Jeffers and J. Shelton for technical and intellectual contributions. We thank R. Goetsch, S. Hartman, and M. Snyder for HeLa S3 cells and D. Banerjee, K. Carter, and M. Stern for critical reading of this manuscript. H.G. was supported by a long-term fellowship from the Human Frontiers Science Program. F.J.S. was supported by NIH grant (GM62594) and NSF grant (IBN-03444429).

Document S1. Six Figures, One Table, and Supplemental Experimental Procedures.  Document_S1..pdf (932 K)
Document_S1..pdf    Help with PDF files (Opens New Window)
Acrobat PDF file 1.
http://www.cell.com/cgi/content/full/120/5/635/DC1/

In this new and detailed study by Steven Johnson, Helge Grosshans, Jaclyn Shingara, Mike Byrom, Rich Jarvis, Angie Cheng, Emmanuel Labourier, Kristy Reinert, David Brown, and Frank Slack, it is confirmed that let-7 microRNAs play a decisive role in C. Elegans maturation from the larval to the adult stage. For the first time, this effect is shown to occur with an inhibition of RAS gene expression. In humans, let-7 microRNA is shown to inhibit the oncogenic activity of RSA in human lung cells, and it is confirmed that human lung cancer cells alone are strangely deficient in let-7 gene expression, possibly allowing an increased oncogenic activity by unopposed RAS expression. The earlier work of Takamizawa J,  et al, 2004 indicated that the addition of let-7 microRNA to cultured human lung cancer cells resulted in a decrease in their growth activity in vitro. It is possible that human lung cancer is a let-7 RNA deficiency disease that could respond to let-7 RNA replacement reprogramming therapy.

Additional References:

1. Bracht J, Hunter S, Eachus R, Weeks P,  and Pasquinelli AE, "Trans-splicing and polyadenylation of let-7 microRNA primary transcripts", RNA, vol. 10: no.10, pp. 1586-1594 (October, 2004).

2. Mansfield JH, Harfe BD, Nissen R, Obenauer J, Srineel J, Chaudhuri A, Farzan-Kashani R, Zuker M, Pasquinelli AE, Ruvkun G, Sharp PA, Tabin CJ, and McManus MT, "MicroRNA-responsive 'sensor' transgenes uncover Hox-like and other developmentally regulated patterns of vertebrate microRNA expression", Nature Genetics  36, no. 10, pp. 1079- 1083 (October, 2004).

3. Persengiev SP, Zhu X, and Green MR, "Nonspecific, concentration-dependent stimulation and repression of mammalian gene expression by small interfering RNAs (siRNAs)", RNA, vol. 10, no. 1, pp. 12-18 (January, 2004).

4. Geiss G, Jin G, Guo J, Bumgarner R, Katze MG, and Sen GC, "A Comprehensive View of Regulation of Gene Expression by Double-Stranded RNA-Mediated Cell Signaling", J. Biol. Chem. vol. 276, pp. 30178-30182 (2001).

5. Hovsepian JA, and Frenster JH, "RNA-Induced Melting of DNA during Selective Gene Transcription".

6. Herstein PR, and Frenster JH, "Mated Models of Gene Regulation in Eukaryotes".

7. Takamizawa J, Konishi H, Yanagisawa K, Tomida S, Osada H, Endoh H, Harano T, Yatabe Y, Nagino M, Nimura Y, Mitsudomi T, and Takahashi T, "Reduced Expression of the let-7 MicroRNAs in Human Lung Cancers in Association with Shortened Postoperative Survival".

Links to Reprogramming and Neoplasia:

Links to RNA and Biological Causality:

Links to Euchromatin Activator RNA Reviews:
Links to Euchromatin Activator RNA Research:
Links to Ultrastructural Probes of DNase I-Sensitive Sites:
Links to RNA as a Therapeutic Agent:
Links to Hodgkin Lymphoma Immuno-Pathology:
Links to Activated T-Lymphocyte Immunotherapy:
Links to Medical Systems Biology:
Links to Selective Gene Transcription:
Links to RNA-Induced Epigenetics:
Links to RNA-Induced Embryogenesis:
Links to RNA and Biological Causality:
Links to Reprogramming and Neoplasia:

A Brief History of Activator RNA:

"Ultrastructural Probes of Active DNA Sites, and the RNA Activators of DNA". (PowerPoint Presentation).