Coordinate Suppression of ERBB2 and ERBB3 by Enforced Expression of Micro-RNA miR-125a or miR-125b*

Deregulation of micro-RNAs (miRNAs) is emerging as a major aspect of cancer etiology because their capacity to direct the translation and stability of targeted transcripts can dramatically influence cellular physiology. To explore the potential of exogenously applied miRNAs to suppress oncogenic proteins, the ERBB oncogene family was chosen with a bioinformatics search identifying targeting seed sequences for miR-125a and miR-125b within the 3′-untranslated regions of both ERBB2 and ERBB3. Using the human breast cancer cell line SKBR3 as a model for ERBB2 and ERBB3 dependence, infection of these cells with retroviral constructs expressing either miR-125a or miR-125b resulted in suppression of ERBB2 and ERBB3 at both the transcript and protein level. Luciferase constructs containing the 3′ 3′-untranslated regions of ERBB2 and ERBB3 demonstrated ∼35% less activity in miR-125a- and miR-125b-expressing cells relative to controls. Additionally, phosphorylation of ERK1/2 and AKT was suppressed in SKBR3 cells overexpressing either miR-125a or miR-125b. Consistent with suppression of both ERBB2 and ERBB3 signaling, miR-125a-or miR-125b-overexpressing SKBR3 cells were impaired in their anchorage-dependent growth and exhibited reduced migration and invasion capacities. Parallel studies performed on MCF10A cells demonstrated that miR-125a or miR-125b overexpression produced only marginal influences on the growth and migration of these non-transformed human mammary epithelial cells. These results illustrate the feasibility of using miRNAs as a therapeutic strategy to suppress oncogene expression and function.

Micro-RNAs (miRNAs) 2 comprise a large class of regulatory noncoding RNAs capable of exerting pronounced influences upon the translation and stability of mRNAs (1)(2)(3)(4). Efforts to understand their biogenesis and target specificity have become active areas of research as their role in such processes as development, differentiation, and apoptosis comes into focus (1)(2)(3)(4). Additional impetus to understanding their biology has emerged from growing evidence implicating miRNA deregulation in cancer etiology (5)(6)(7)(8).
More than 400 human miRNAs have been identified to date with up to ϳ1000 miRNAs postulated to exist on the basis of bioinformatics analyses, leading to the speculation that a majority of human mRNAs are subject to miRNA regulation (9 -10). miRNAs are initially excised from a primary RNA transcript in the nucleus by the nuclear RNase III endonucleases Drosha working in cooperation with DGCR8 (4,11). Export to the cytoplasm is facilitated by exportin-5, where the ϳ70-nt pre-miRNA is further processed by the RNase III endonuclease Dicer into a mature ϳ22-nt double-stranded miRNA (1)(2)(3)(4). Subsequent incorporation of one strand into RISC (RNA-induced silencing complex) establishes the miRNA as competent to target mRNAs for translational arrest and, in some instances, transcript decay (1)(2)(3). Although targeting of a mRNA by a miRNA remains a poorly defined process, it appears to be largely mediated by complementarity between nucleotides 2 to 5 (numbered from the miRNA 5Ј end), occasionally referred to as the "seed sequences," with a target element in the transcript 3Ј-UTR (1-3, 9, 12).
Involvement of miRNAs in the oncogenic process has been supported by the observation that the 13q14 deletion characterizing more than half of all chronic lymphocytic leukemias results in loss of miR-15a and miR-16-1 genes (5). Additional studies using microarray profiling have demonstrated aberrant expression of miRNAs in human cancers, with deregulation of specific miRNAs distinguishing different cancer types (7,8). One breast cancer study concluded that two up-regulated miR-NAs (miR-21, miR-155) might act as oncogenes, whereas three other down-regulated miRNAs (miR-10b, miR-125b, and miR-145) might act as breast cancer suppressor genes (8). Of interest, these same breast cancer suppressing miRNA candidates, miR-10b, miR-125b (along with its homolog, miR-125a), and miR-145, were also identified in a recent study as being significantly down-regulated in ERBB2-amplified and -overexpressing breast cancers clinically matched against ERBB2-negative human breast cancers (13).
Given this involvement of miRNAs with cancer development, the manipulation of cellular miRNA levels has emerged as a potential strategy for therapeutic intervention (14 -16). Efforts to predictably alter intracellular transcript profiles by increasing specific miRNA levels either through transfection or viral delivery methods have demonstrated the potential of this strategy to modulate cellular physiology (12,14). Conversely, attempts to reduce miRNA levels using biologically stable antisense moieties such as 2ЈO-methyl oligonucleotides or "antagomirs" have also proven capable of altering intracellular transcript profiles (15,16).
A bioinformatics comparison of the 3Ј-UTRs of ERBB2 and ERBB3 revealed a consensus target element for both miR-125a and miR-125b; thus, the present study was undertaken to assess ERBB2 and ERBB3 suppression after retroviral delivery and overexpression of these miRNAs in an ERBB2-amplified and -overexpressing human breast cancer cell line, SKBR3. Overexpression of miR-125a or miR-125b reduced ERBB2 and ERBB3 at both the transcript and protein level in these cells, leading to reduced ERK1/2 and AKT signaling. Functionally, miR-125aor miR-125b-overexpressing SKBR3 cells displayed diminished plating and anchorage-dependent growth in addition to markedly reduced cell migration and invasion capacities. Parallel studies performed on MCF10A cells demonstrated that miR-125a or miR-125b overexpression produced only marginal influences on the anchorage-dependent growth and migration of these non-transformed and ERBB2-independent human mammary epithelial cells.
Micro-RNA Data Base and Retroviral Expression of miRNAs-Nucleotides 2-7 from the 5Ј end of miRNA of the 321 human miRNA listed in the Sanger miRBase 7.0 release were searched for complementarity to the 3Ј-UTRs of ERBB2 (Ensemble gene ENSG00000141736) and ERBB3 (Ensemble gene ENSG00000065361). Full-length human miRNAs miR-125a and miR-125b together with ϳ200 bp of flanking sequence were amplified from human genomic DNA and cloned into the pTOPO-D entry vector (Invitrogen). Cloned fragments were verified by sequencing, and recombinase was cloned into the pMSCVpuro expression vector (Clontech).
3Ј-UTR-luciferase Reporter Constructs and SKBR3 Transient Transfection Assays-EST clones containing the 3Ј-UTR sequences from ERBB2 and ERBB3 cDNA were commercially obtained (Invitrogen) and used as templates for PCR amplification of the full-length or deleted (nt 1-44 and 1-26 for deleted ERBB2 and ERBB3 UTR constructs, respectively) 3Ј-UTR products. PCR products were cloned into the XbaI site of the luciferase reporter pGL3 (Promega, Madison, WI) to yield pERBB2wtUTR, pERBB2⌬UTR, pERBB3wtUTR, and pERBB3⌬UTR containing the wild type and deleted 3Ј-UTRs from ERBB2 and ERBB3, respectively. The 3Ј-UTR inserts from all constructs were sequence-verified. Transient transfection assays were performed in 12-well plates seeded with ϳ1.5 ϫ 10 5 SKBR3-125a, SKBR3-125b, or SKBR3-puro cells per well using Lipofectamine 2000 (Invitrogen) according to the manufacturer's recommendations. For each transfection experiment, equal volumes of a given DNA-lipid complex formulation were distributed in duplicate to each of the SKBR3-125a, SKBR3-125b, and SKBR3-puro cultures with at least three separate transfection experiments performed for each 3Ј-UTR construct. The ␤-galactosidase expression vector pCH110 (Amersham Biosciences) was co-transfected with each experiment for normalization of transfection efficiency. Because both the pCH110 and pGL3 co-transfected constructs use the SV40 promoter, potential promoter differences resulting from miR-125a or miR-125b overexpression were presumed nullified.
Northern and Western Blotting-Total RNA extracted using Trizol (Invitrogen) from retrovirally infected SKBR3 and MCF10A cells was analyzed by Northern blotting. Methods and probes described previously (18) were used to detect ERBB2 (4.6 kilobases) and ERBB3 (4.9 kilobases) transcripts relative to control glyceraldehyde-3-phosphate dehydrogenase transcripts. Levels of enforced and endogenous expression of miR-125a and miR-125b were detected using 15% Tris borate-EDTA urea polyacrylamide gels (Invitrogen), with RNA transferred onto Hybond Plus membranes (Amersham Biosciences) as previously described (18) and probed with antisense miR-125a or miR-125b DNA oligonucleotides that were 32 P end-labeled. Loading for small RNA species was controlled by probing with a 32 P end-labeled antisense snU6 DNA oligonucleotide (5Ј-ttgcgtgtcatccttgcgcagg-3Ј). Total protein extracted from retrovirally infected SKBR3 sublines was analyzed by Western blotting using methods and antibodies described previously (19) to detect endogenous ERBB2 and ERBB3 receptor levels relative to ␤-actin and phospho-Ser-ERK1/2 and phospho-Ser-AKT relative to total ERK1/2 (42/44 kDa) and AKT.
Cell Proliferation, Migration, and Invasion Assays-All cell proliferations assays were replicated in three independent experiments using the retrovirally infected cell pools (SKBR3-125a, SKBR3-125b, SKBR3-puro, MCF10A-125a, MCF10A-125b, and MCF10A-puro). In each experiment, cells were trypsinized, counted, and plated into duplicate wells of 6-well plates at 2 ϫ 10 5 cells per well. Total cell counts were obtained and averaged on days 2, 4, and 6 after plating. All cell migration and invasion assays were similarly replicated three times using independently infected cell pools. Cells were trypsinized, pelleted, and resuspended in media without serum or growth factors. After counting, 1 ϫ 10 4 SKBR3 cells in 0.5 ml of medium were added to either 8-m porous chamber inserts or 8-m porous BioCoat Matrigel chamber inserts (BD Biosciences) and placed in wells filled with 0.75 ml of medium supplemented with 10% fetal calf serum as chemoattractant for both migration and invasion assays. MCF10A cells are capable of migration but not Matrigel invasion; after counting, 3 ϫ 10 4 MCF10A cells in 0.5 ml of medium were added to 8-m porous chamber inserts placed in wells filled with 0.75 ml of medium supplemented with 5% fetal calf serum and growth factors (see above) as chemoattractant. After 20 h of incubation, cells remaining on the insert top layers were removed with cotton swab scrubbings as recommended by the manufacture. Cells on the membrane lower surfaces were fixed in 100% methanol for 2 min followed by staining (2 min) in a 1% toluidine blue solution prepared in phosphate-buffered saline. Inserts were washed several times in distilled water before air drying; membranes were photographed using an Olympus I ϫ 70 microscope at 70ϫ, and cell counts were determined after totaling five random fields (each field capturing ϳ6% of total membrane area) and expressed as a percentage of the control cell number.

Conserved miR-125 Targeting Element Identified within the 3Ј-UTRs of ERBB2 and ERBB3 and Retroviral Overexpression of miR-125 Species in Human Mammary
Epithelial Cells-To explore the strategy of externally delivered miRNAs to promote the down-regulation of proteins with known oncogenic properties, the ERBB family, consisting of ERBB1/EGFR (epidermal growth factor receptor), ERBB2, ERBB3, and ERBB4 family members, was chosen for miRNA targeting. A bioinformatics search of ERBB family 3Ј-UTRs for complementarity to miRNA seed sequences (nt 2-7 from the 5Ј end), as listed in the Sanger miRBase data base (Version 7.0 release), focused attention on miR-125a and miR-125b. The most proximal 3Ј-UTR regions of both ERBB2 and ERBB3 were found to exhibit perfect complementarity with the 5Ј nucleotides, nt 1-8, of miR-125a and miR-125b. Interest in the miR-125 family was also supported by its evolutionary link to lin-4, the first identified miRNA species functionally known to regulate Caenorhabditis elegans cellular proliferation (21,22) as well as recent identification of this miRNA family as candidate breast cancer suppressors (8). As shown in Fig. 1A, the maximum free energy predicted hybridization configurations between the putative 3Ј-UTR miR-125 target elements in ERBB2 and ERBB3 and the seed sequences in miR-125a or miR-125b all exhibit similar stabilities at ⌬G ϳ Ϫ14 kcal/mol, determined by mFold analysis and consistent with authentic miRNA targeting (20).
The use of viral constructs offers the possibility of rapidly producing high intracellular levels of mature miRNA via the endogenous miRNA processing pathway. To this end, ϳ300-bp fragments containing genomic miR-125a and miR-125b sequences were obtained by PCR amplification from human DNA and cloned into pMSCV, a retrovirus producing high transcript levels driven by its long terminal repeat. A human breast cancer cell line, SKBR3, dependent upon endogenous ERBB2 amplification and receptor overexpression for its growth and survival, and the non-transformed and ERBB2-independent human mammary epithelial cell line, MCF10A, were retrovirally infected and selected for viral integration by puromycin selection. As shown in the Northern blots of Fig. 1B, enforced expression of mature miR-125a and miR-125b was achieved in both cell lines at ϳ7-10-fold increases above their endogenous miRNA levels.
SKBR3 Overexpression of miR-125a or miR-125b Suppresses ERBB2 and ERBB3 Transcript and Protein Expression and Inhibits Signaling by ERK1/2 and AKT Pathways-After retroviral infection and ϳ10 days of puromycin selection, total RNA and protein extracts were prepared from pooled SKBR3 cells showing either miR-125a or miR-125b overexpression and from control cells similarly infected but with the pMSCV empty vector and puromycin selected. As shown in the Northern blots of Fig. 2A, ERBB2 transcript levels in the SKBR3-125a and SKBR3-125b cells were 52 and 80%, respectively, of SKBR3puro control levels after loading normalization by glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA levels. Likewise, ERBB3 mRNA levels in SKBR3-125a and SKBR3-125b cells were 67 and 90%, respectively, that of SKBR3-puro control levels after normalization. As shown in the Western blots of Fig. 2B, ERBB2 protein levels were reduced to 40 and 65% that of control (SKBR3-puro) levels in SKBR3-125a and SKBR3-125b cells, respectively, whereas ERBB3 protein levels were reduced to 60 and 80% that of control levels, respectively, after loading normalization by ␤-actin levels.
The downstream signaling response of ERBB2 and ERBB3 receptor activation and heterodimerization includes serine phosphorylation of ERK1/2 and AKT pathway intermediates. Thus, the basal activation status of these two signal transduction pathways were assessed by Western blotting protein extracts isolated from SKBR3-125a and SKBR3-125b cells relative to those from control SKBR3-puro cells. As shown in the Western blots of Fig. 2B, phospho-ERK1/2 (pERK1/2) and phospho-AKT (pAKT) levels relative to total ERK1/2 and total AKT levels were significantly reduced in the miR-125a-and miR-125b-overexpressing SKBR3 cells, with the former showing greater suppression than the latter, consistent with the more pronounced ERBB2 and ERBB3 protein suppression observed in the SKBR3-125a cells compared with the SKBR3-125b cells. JANUARY 12, 2007 • VOLUME 282 • NUMBER 2

JOURNAL OF BIOLOGICAL CHEMISTRY 1481
Analysis of ERBB2 and ERBB3 3Ј-UTRs for miR-125 Responsiveness-To examine the 3Ј-UTRs from ERRB2 and ERRB3 for responsiveness to miR-125a or miR-125b overexpression, the full-length 3Ј-UTR sequences from ERBB2 and ERBB3 cDNAs were cloned into the luciferase expressing pGL3-promoter vector using a unique XbaI site just downstream of the luciferase stop codon to create pERBB2wtUTR and pERBB3wtUTR expression constructs, respectively, for transient transfection into SKBR3-125a, SKBR3-125b, and SKBR3-puro cells. As shown in Fig. 3A, luciferase activities expressed off the pERBB2wtUTR or pERBB3wtUTR constructs were suppressed by 60 -72% in the miR-125a-and miR-125bexpressing SKBR3 cells relative to luciferase activity expressed in the SKBR3-puro control cells. To confirm the involvement of miR-125a and miR-125b in this reporter gene suppression, 3Ј-UTR deletion constructs (⌬UTR) were made that eliminated the conserved CUCAGGGA-targeting elements from the ERBB2 and ERBB3 UTR reporters; bases 1-44 were deleted to create pERBB2⌬UTR, whereas bases 1-26 were deleted to create pERBB3⌬UTR. In experiments where the full-length and deleted UTR constructs for ERBB2 and ERBB3 were each transiently transfected into SKBR3-125a and SKBR3-125b cells, luciferase activity from pERBB2⌬UTR was ϳ1.4fold greater than that from pERBB2wtUTR, whereas the luciferase activity from pERBB3⌬UTR was ϳ1.7-fold greater than activity from pERBB3wtUTR, as shown in Fig. 3B. Also seen in Fig. 3B is the much lower but still slightly increased (1.1-fold) luciferase activity resulting from the UTR deletion constructs (relative to their respective wild type UTR constructs) in the SKBR3-puro control cells, possibly reflecting the endogenous levels of miR-125a and miR-125b in these cells (as shown in Fig. 1B).
Overexpression of miR-125a or miR-125b Impairs Anchorage-dependent Growth of SKBR3-The anchorage-dependent growth consequences of enforced miR-125a or miR-125b overexpression were compared in the ERBB2-dependent SKBR3 and ERBB2-independent MCF10A cell lines. All experiments were performed on pooled puromycin-selected cells within 2-3 weeks after their retroviral infection and initial selection and while under puromycin selection pressure. As shown in Fig. 4A, within 2 days of plating an equal number of viable SKBR3-125a, SKBR3-125b, and SKBR3-puro cells, only 50 and 75% of the miR-125aand miR-125b-overexpressing cells, respectively, remained viably attached and growing relative to the SKBR3-puro control cells. After this initial 2-day interval, growth rates estimated by linear extrapolation out to day 6 indicated an apparent 60% slower growth rate by the SKBR3-125a and SKBR3-125b cells relative to the SKBR3-puro cells. In contrast to the impaired anchorage-dependent growth associated with miR-125a or miR-125b overexpression in the SKBR3 cells, Fig. 4B shows that the similarly produced and miRNA-overexpressing MCF10A-125a and MCF10A-125b cells exhibited little if any impaired plating and growth within the first 2 days relative to control FIGURE 1. Minimum free energy states of hybridization between miR-125a or miR-125b with conserved ERBB2 and ERBB3 3-UTR target elements and retroviral overexpression of miR-125a and miR-125b in SKBR3 and MCF10A breast epithelial cells. A, predicted hybridization of miR-125a and miR-125b with the conserved ERBB2 and ERBB3 target element CUCAGGGA (underlined), as determined by the mFold Server. Non-bonding bases protrude above and below the duplexed UTR and miRNA bases, respectively, with mapping of the target element into the ERBB2 and ERBB3 3Ј-UTRs as indicated. B, Northern analyses of total RNA prepared from SKBR3 and MCF10A cells after infection and puromycin selection using retroviral constructs expressing either miR-125a (SKBR3-125a, MCF10A-125a), miR-125b (SKBR3-125b, MCF10A-125b), or empty vector controls (SKBR3-puro, MCF10A-puro). Northern blots in the left two panels were probed for miR-125a, whereas Northern blots in the right two panels were probed for miR-125b. Mature and pre-miRNA species are noted, and equal RNA loading was confirmed by probing for the small RNA species, U6 RNA.
MCF10A-puro cells. Extrapolating their growth rates out to day 6 indicated an apparent 25% slower growth rate by the MCF10A-125a and MCF10A-125b cells relative to the control MCF10A-puro cells, suggesting a mild negative influence by these miRNAs on cellular proliferation mechanisms independent of ERBB2 status and possibly related to the reported tumor suppressing function of miR-125a and miR-125b (8).
Overexpression of miR-125a or miR-125b Impairs Migration and Invasiveness of SKBR3-Because previous studies have linked ERBB2 receptor activation and ERBB3 heterodimerization with cellular motility and invasiveness (23,24), a Boyden chamber assay format was used to measure the effects of mir-125a and mir-125b overexpression on cellular migration and invasion. As shown by the graph and images in Fig. 5A, SKBR3-125a and SKBR3-125b cells exhibited a greater than 50% reduction in their motility relative to control SKBR3-puro cells. In contrast and as shown in Fig. 5B, MCF10A-125a and MCF10A-125b cells exhibited less than a 10% reduction in their motility relative to control MCF10A-puro cells. Invasiveness, requiring both cell motility and proteolytic properties enabling advancement through a solid and proteinaceous extracellular matrix like Matrigel, is thought to more directly assess cell metastatic potential. Although immortalized and capable of motility comparable with SKBR3 cells, the nontransformed MCF10A cells are incapable of significant invasiveness through an extracellular matrix. Although the invasiveness of SKBR3 cells may be lower than that of other much more aggressive breast cancer cell lines (e.g. MDA-231), it is easily measurable as are the effects of mir-125a or mir-125b overexpression on SKBR3 invasiveness. As shown in Fig. 5C, SKBR3-125a and SKBR3-125b cells displayed pronounced impairments in their invasive potentials, with the former showing 75% reduction and the latter showing 67% reduction in their proportions of invading/migrating cells relative to SKBR3-puro controls.

DISCUSSION
As presented here, retroviral overexpression of miR-125a or miR-125b in the ERBB2-dependent human breast cancer cell line, SKBR3, produced suppression of its anchorage-dependent growth potential and even greater inhibition of its motility and invasive capabilities. Similar retroviral overexpression of these miRNAs in non-transformed and ERBB2-independent MCF10A cells produced marginal effects on its anchorage-de- -125a or miR-125b. A, Northern blots using total RNA isolated 2 weeks after retroviral infection and puromycin selection from control (SKBR3-puro) or miRNA-overexpressing cell pools (SKBR3-125a, SKBR3-125b) probed for ERBB2 or ERBB3 transcripts. ERBB2 and ERBB3 mRNA band intensities normalized by glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA are shown below the lanes as percent of the respective control value (100%). B, Western blots of total protein extracted from the same SKBR3 cell pools as above were probed for ERBB2 or ERBB3 receptor content relative to ␤-actin protein and phospho (p)-Ser-ERK1/2 or phospho-Ser-AKT relative to total ERK1/2 and AKT protein content. ERBB2 and ERBB3 band intensities normalized by ␤-actin are shown below the lanes as percent of the respective control value (100%).  Influence of miR-125a and miR-125b overexpression on ERBB2and ERBB3 3-UTR-regulated reporters in SKBR3 cells. A, SKBR3 cells overexpressing miR-125a (SKBR3-125a) or miR-125b (SKBR3-125b) and control SKBR3 cells (SKBR3-puro) were transiently transfected with luciferase reporters containing either the wild type ERBB2 3Ј-UTR (pERBB2wtUTR) or wild type ERBB3 3Ј-UTR (pERBB3wtUTR). Luciferase activities of pERBB2wtUTR and pERBB3wtUTR in SKBR3-125a and SKBR3-125b cells were normalized by their respective activities in control cells (SKBR3-puro) and plotted as percent activity relative to control cells. Open bars display ERBB2 3Ј-UTR activity, whereas solid bars display ERBB3 3Ј-UTR activity. B, SKBR3-125a, SKBR3-125b, and SKBR3-puro cells were transiently transfected with luciferase reporters containing either wild type ERBB2 or ERBB3 3Ј-UTR sequences as in panel A or UTR reporters deleted of their respective 3Ј-UTR miR-125 target elements (pERBB2⌬UTR, pERBB3⌬UTR), as described under "Experimental Procedures." Open bars display relative ERBB2 3Ј-UTR luciferase activity (pERBB2⌬UTR/ pERBB2wtUTR), whereas solid bars display relative ERBB3 3Ј-UTR luciferase activity (pERBB3⌬UTR/pERBB3wtUTR). JANUARY 12, 2007 • VOLUME 282 • NUMBER 2 pendent growth and no significant impact on the motility of these non-invasive human breast epithelial cells. Such phenotypic alterations seen in the aneuploid and ERBB2-amplified SKBR3 breast cancer model and not observed in the diploid MCF10A breast epithelial model are consistent with inhibition of the overexpressed ERBB2 and ERBB3 transcript and protein levels in SKBR3, caused by enforced expression of either miR-125a or miR-125b, two closely related miRNAs that appear capable of targeting the cognate miR-125 target element, CUCAGGGA, found conserved within the proximal 3Ј-UTR sequences of human ERBB2 and ERBB3. The observed reduction in ERBB2 and ERBB3 protein levels may be reasonably attributed to miRNA effects on ERBB2 and ERBB3 transcript stability and translation, as a previous report indicates that miR-125a and mir-125b can bifunctionally mediate both transcript decay and translational inhibition (22). Although the molecular mechanisms explaining MCF10A cell immortalization remain unclear, these motile and proliferating cells express little or no ERBB2 or ERBB3 and do not respond to any ERBBtargeted therapeutics. The possibility that miR-125a and miR-125b target other growth regulating mechanisms independent of ERBB2 and ERBB3 may be surmised by the slight growth inhibitory response (25% reduction) observed in MCF10A-125a and MCF10A-125b cells relative to control MCF10Apuro cells and also the purported tumor suppressing function of miR-125a and miR-125b in human breast cancers independent of ERBB2 status (8).

ERBB2 and ERBB3 Suppression by miR-125a/b Micro-RNA
Although only a single ERBB homologue exists in invertebrates such a C. elegans and Drosophila, there are four ERBB paralogs expressed in higher organisms encoding distinct transmembrane receptors capable of homo-or heterodimerization, activated by numerous epithelial cell growth factor ligands and coordinately transducing a complex network of intracellular signaling pathways used for epithelial cell development, proliferation, and survival (25). Dysregulation of three of these four receptor-encoding paralogs (ERBB1/EGFR, ERBB2, ERBB3) singly or in combination can induce malignant transformation. About 25% of all human breast cancers are associated with amplification and overexpression of ERBB2, and the aggressive tumorigenic properties of such ERBB2-positive

ERBB2 and ERBB3 Suppression by miR-125a/b Micro-RNA
breast cancers are dependent on both homodimerization of ERBB2 receptors as well as their heterodimerization with either ERBB1/EGFR or ERBB3 receptors. In particular, overexpressed ERBB2-ERBB3 heterodimers are thought to be most critical for the aggressive growth and metastatic potential of ERBB2-positive human breast cancers, since it is the transphosphorylated ERBB3 partner of ERBB2 that can recruit up to six individual p85 subunits and constitutively activate intracellular phosphatidylinositol 3-kinase, induce serine phosphorylation of AKT, and thereby drive cell survival, proliferation, motility, and invasion mechanisms characteristic of this aggressive form of human breast cancer (23,24,26). Consequently, as seen here, the coordinated suppression of both ERBB2 and ERBB3 expression in an ERBB2-positive breast cancer cell line would be expected to most significantly inhibit phospho-AKT levels and impair not only its proliferative potential but also its motility and invasive capabilities.
Recognizing the mechanistic importance of ERBB2 heterodimer formation in human breast tumorigenesis, pharmaceutical companies are now developing second-generation kinase inhibitors that simultaneously target multiple members of the ERBB receptor family (Ref. 27; e.g. lapatinib/Tykerb, an inhibitor of ERBB2 and ERBB1/EGFR kinases (GlaxoSmithKline) and canertinib/CI-1033, an inhibitor of ERBB2, ERBB1/ EGFR, and ERBB4 kinases, Pfizer). Although there now exist promising clinical candidates capable of targeting the kinase functions of both ERBB1/EGFR and ERBB2, there are as yet no specific inhibitors of the docking or signal-initiating functions of ERBB3. It was, therefore, of considerable interest when a computational search among ERBB family members identified a conserved 8-nt miR-125 binding element, CUCAGGGA, encoded within the proximal 3Ј-UTRs of both ERBB2 and ERBB3. Free energy determinations for the predicted hybridization configurations between the putative 3Ј-UTR miR-125 target elements in ERBB2 and ERBB3 and the seed sequences in miR-125a or miR-125b are consistent with authentic miRNA targeting (20). Aside from identifying a conserved miR-125 targeting element in both ERBB2 and ERBB3, BLAST analysis also identified homologous U-rich regions in the 3Ј-UTRs of ERBB2 and ERBB3. Such U-rich 3Ј-UTR regions have been reported in other genes to be essential for the mediation of miRNA-induced transcript destabilization (28). Luciferase constructs containing the 3Ј-UTRs of ERBB2 and ERBB3 demonstrated significantly less activity in miR-125a-and miR-125b-overexpressing SKBR3 cells relative to controls. To confirm that the 3Ј-UTR elements containing the candidate miR-125a and miR-125b targeting elements actually mediated this reporter gene repression, deletion constructs missing these miR-125 targeting elements were generated and similarly transfected into SKBR3-125a, SKBR3-125b, and control SKBR3-puro cells.
These deletion experiments demonstrated reversal of the reporter gene repression caused by the intact (wild type) ERBB2 and ERBB3 UTRs but only in the context of miR-125a or miR-125b overexpression. Thus, whereas not explicitly identifying the 8-nt miR-125 binding element as responsible for this repressive influence, these deletion constructs localized miR-125 responsiveness to very small regions within the proximal 3Ј-UTRs of ERBB2 and ERBB3 containing the conserved CUCAGGGA targeting elements.
Targeting oncogenic ERBB2 and ERBB3 overexpression in a model breast cancer cell line like SKBR3 by enforced expression of miR-125a and miR-125b demonstrates the potential utility of rapid and efficient miRNA gene delivery by retroviruses to interrogate cancer cell phenotypes. However, as a therapeutic tool, delivery of miRNAs by viral or other means must be balanced against their potential to impair normal cellular functions by competing for easily saturated miRNA processing steps. Two such limiting steps appear to involve nuclear export by exportin-5 and the cytoplasmic Dicer partner and RISC component, TRBP. Sequestration of TRBP by transfected RNA has been shown to attenuate overall Dicer-RISC activity (29). Likewise, in vivo delivery and sustained intrahepatic viral expression of short hairpin RNAs has produced dose-dependent liver injury in mice, associated with down-regulation of liver-derived miRNAs that also compete for exportin-5 (30).
Commensurate with the successful down-regulation of ERBB2 and ERBB3 proteins levels mediated by overexpression of miR-125a and miR-125b, influences on such critical features of the malignant cell phenotype as proliferative growth, motility, and invasiveness were also observed. Investigational agents have now been identified that alter critical cancer gene programs by rapidly modulating intracellular miRNA profiles (18). Although delivery of a single miRNA can clearly produce a phenotypic disturbance, it has been suggested that miRNA regulation of a given transcript is most efficient when multiple miR-NAs work in concert (1)(2)(3). With such recent technological advancements as the delivery of multiple miRNAs from a single polycistronic construct (31), future efforts directed toward delivery of multiple and coordinately targeted miRNAs are anticipated to not only enhance the utility of miRNAs as research tools but also provide further rationale for the design of new therapies that modulate miRNA expression.