Transcriptional Regulation of the Protocadherin β Cluster during Her-2 Protein-induced Mammary Tumorigenesis Results from Altered N-Glycan Branching*

Background: Her-2-induced mammary tumor onset is significantly delayed in GnT-V knock-out mice. Results: The gene expression of the Pcdhβ cluster is up-regulated in her-2-induced tumors with GnT-V deletion. Conclusion: Up-regulation of the Pcdhβ cluster is one of the mechanisms for the reduced her-2-mediated tumorigenesis resulting from GnT-V deletion. Significance: Our findings shed new light on the molecular mechanisms of the effects of GnT-V on mammary tumorigenesis. Changes in the levels of N-acetylglucosaminyltransferase V (GnT-V) can alter the function of several types of cell surface receptors and adhesion molecules by causing altered N-linked glycan branching. Using a her-2 mammary tumor mouse model, her-2 receptor signaling was down-regulated by GnT-V knock-out, resulting in a significant delay in the onset of her-2-induced mammary tumors. To identify the genes that contributed to this GnT-V regulation of early events in tumorigenesis, microarray analysis was performed using her-2 induced mammary tumors from wild-type and GnT-V-null mice. We found that 142 genes were aberrantly expressed (>2.0-fold) with 64 genes up-regulated and 78 genes down-regulated after deletion of GnT-V. Among differentially expressed genes, the expression of a subgroup of the cadherin superfamily, the protocadherin β (Pcdhβ) cluster, was up-regulated in GnT-V-null tumors. Altered expression of the Pcdhβ cluster in GnT-V-null tumors was not due to changes in promoter methylation; instead, impaired her-2-mediated signaling pathways were implicated at least in part resulting from reduced microRNA-21 expression. Overexpression of Pcdhβ genes inhibited tumor cell growth, decreased the proportion of tumor-initiating cells, and decreased tumor formation in vivo, demonstrating that expression of the Pcdhβ gene cluster can serve as an inhibitor of the transformed phenotype. Our results suggest the up-regulation of the Pcdhβ gene cluster as a mechanism for reduced her-2-mediated tumorigenesis resulting from GnT-V deletion.

Her-2 (neu/ErbB2), a 185-kDa transmembrane glycoprotein and a member of the epidermal growth factor (EGF) receptor family, is overexpressed in ϳ15-30% of human breast cancer and has been correlated with poor prognosis of cancer patients and therapeutic resistance (15,16). The oncogenic potential of her-2 in mammary tumorigenesis has been confirmed in transgenic mouse models with overexpression or mutation of her-2 under the transcriptional control of the mouse mammary tumor virus promoter, and oncogenesis induced by her-2 in these transgenic mice has been shown to be quite similar to human breast cancer (17,18). We recently showed that her-2 receptor signaling is modulated by GnT-V expression levels. Her-2-induced mammary tumor onset is significantly delayed in GnT-V knock-out mice coincident with the reversion of her-2-induced deregulation of acinar morphogenesis and a significantly reduced population of tumor-initiating cells (cancer stem cells) in isolated tumor cells with GnT-V deletion, resulting in reduced ability to form secondary tumors in NOD/SCID mice (9). These results indicate that GnT-V promotes mammary tumor development by regulating some early events during tumorigenesis.
In addition to altering growth factor receptor signaling, the function of cadherins is modulated by GnT-V expression levels. For example, knockdown of GnT-V reduces the expression of * This work was supported, in whole or in part, by National Institutes of Health ␤(1,6) branching on the extracellular EC2-3 domains of N-cadherin, increasing N-cadherin-mediated cell-cell adhesion (11), whereas the deletion of GnT-V increases E-cadherin localization in cell adhesion junctions and cell-cell adhesion in both polyoma middle-T and her-2-induced mammary tumors (9,13). The cadherins constitute a superfamily of single pass transmembrane glycoproteins mediating calcium-dependent cellcell adhesion that plays an essential role in regulating major cellular behaviors, including cell growth, motility, and differentiation (19,20). Several subgroups of cadherins have been defined based on shared properties and sequence similarity, including classical cadherins and protocadherins (21). Protocadherins (Pcdhs) are divided into clustered and non-clustered groups based on their genomic structures. Three closely linked protocadherin gene clusters, Pcdh␣, Pcdh␤, and Pcdh␥, have been identified in both mouse and human (22,23). Unlike classical cadherins that have five extracellular domains, a transmembrane domain, and a conserved cytoplasmic domain, Pcdhs have six extracellular domains encoded by one large exon and distinct intracellular domains (24). Protocadherins are predominantly expressed in the nervous system and appear to play an important role in regulating neuron development (21). However, epigenetic aberrations of protocadherin gene clusters caused by hypermethylation have been identified recently in some human tumors, including breast cancer (25)(26)(27)(28), indicating that Pcdh genes may serve as tumor "suppressors" that can inhibit tumor development.
To identify the genes that could be contributive to early breast tumor progression regulated by GnT-V (9), microarray analysis was performed using mammary tumor tissues from GnT-V wild-type and knock-out her-2 mice, and qRT-PCR was used to confirm transcript differences. Here, we present evidence that members of the Pcdh␤ gene cluster are implicated, functioning to suppress the malignant phenotype, in her-2-mediated mouse mammary tumorigenicity. We found that deletion of GnT-V caused enhanced gene expression of the Pcdh␤ cluster in GnT-V knock-out tumors that contributes to the reduced her-2-induced tumorigenesis. Increased gene expression of the Pcdh␤ cluster in GnT-V knock-out tumors was mediated by attenuated her-2-mediated signaling pathways caused by deletion of GnT-V. One of the downstream regulators of her-2 signaling, microRNA-21 (miR-21), was identified and implicated at least in part in the increased gene expression of Pcdh␤ cluster.

EXPERIMENTAL PROCEDURES
Cell Lines and Materials-Human breast carcinoma cell lines MDA-MB231 and SK-BR3 were from the American Type Culture Collection (Manassas, VA). Mouse her-2 tumor cells with different GnT-V backgrounds were isolated from her-2 tumor tissues as described in our previous report (9). For silencing experiments, miRCURY locked nucleic acid-modified anti-miR-21 or control miRCURY knockdown oligonucleotides were purchased from Exiqon. Silencer select neu/ErbB2 siRNA and scrambled control siRNA oligonucleotides were from Applied Biosystem (Ambion). pSuper vectors containing GnT-V siRNA and scrambled control siRNA were constructed and described in our previous report (14). 5-Aza-2Ј-deoxycyt-idine, PD98059, and wortmannin were purchased from Sigma; Lipofectamine TM 2000 reagent was from Invitrogen. Anti-Pcdh␤4 and -Pcdh␤7 were products of Abcam. Antibodies against ErbB2/neu, ERK, phospho-ERK, PKB, and phospho-PKB and HRP-labeled anti-rabbit IgG and anti-mouse IgG were from Santa Cruz Biotechnology. Anti-c-myc tag (clone 9E10) was from Millipore.
Mouse Breeding and Tumor Tissue Isolation-All procedures used for this study were approved by the Institutional Animal Care and Use Committee of the University. The her-2/GnT-V(ϩ/ϩ) and her-2/GnT-V(Ϫ/Ϫ) mice were produced by breeding her-2/neu transgenic mice with GnT-V knock-out mice and genotyped by PCR as our previously described (9). Mice with tumors were euthanized at 10 weeks after the first detection of a palpable tumor. Mammary tumor tissues were collected from three her-2 mice with wild-type GnT-V(ϩ/ϩ) (mouse identification numbers 2319, 2328, and 2323) and GnT-V-null (Ϫ/Ϫ) (mouse identification numbers 2318, 2205, and 2414) backgrounds, respectively; immediately frozen in liquid nitrogen; and stored at Ϫ80°C.
Microarray-Total RNA was isolated from tumor tissues using TRIzol reagent and cleaned using RNeasy columns (Qiagen). Using a random hexamer incorporating a T7 promoter, double-stranded cDNA was synthesized from total RNA. cRNA was generated from the double-stranded cDNA template through an in vitro transcription reaction and purified using the Ambion WT Expression kit and sample cleanup module. cDNA was then regenerated through a random primed reverse transcription using a dNTP mixture containing dUTP. Fragmented and biotinylated cDNA was used for hybridization with an Affymetrix Mouse GeneChip Gene 1.0 ST Array according to the manufacturer's protocol.
Expression Analysis-Gene expression alterations were determined using the PARTEK Genomics Suite. The CEL files were imported from the Affymetrix Expression Console and background-corrected and quantile-normalized, and probe summarization was performed using Robust Multichip Analysis (RMA). A gene summarization was performed on the data that estimates the intensity of individual genes by averaging the intensities of all the probe sets comprising the gene followed by an n-way analysis of variance using a mixed model and methods of moment to equate analysis of variance mean sum of squares to their expected values. The data were then analyzed using a two-sample t test for significance at p ϭ 0.05 and a -fold change cutoff of 2.0. To assess the possible functional connections between the differentially expressed genes, a pathway analysis, which assesses statistically overrepresented functional terms within a list, was conducted using Ingenuity Pathways Analysis (IPA) for all comparisons. The probability that a specific set of genes has a significant number of members in a canonical pathway is assigned a p value, which is calculated by Fisher's exact test (right tailed). The p value indicates the probability of observing the fraction of the focus genes in the canonical pathway compared with the fraction expected by chance in the reference set with the assumption that each gene is equally likely to be selected by chance.
qRT-PCR Analysis-TRIzol was used to isolate total RNA from tumor tissues and cell lines. Reverse transcription reac-tions were performed using cDNA synthesis kit (Bio-Rad). Primers used in the qRT-PCR analysis are listed in supplemental Table S1. Real time reactions were performed using the iQ TM SYBR Green Supermix (Bio-Rad) as reported previously (29). All PCRs were performed in triplicate samples and repeated at least two times.
Genomic DNA Purification, Bisulfite Modification, and Methylation-specific PCR-Purification and bisulfite treatment of genomic DNA samples were performed using the DNeasy Tissue kit and the EpiTect Bisulfite kit (Qiagen), respectively, according to the manufacturer's instructions. Methylation-specific PCR was carried out using the following cycling conditions: 95°C for 5 min; 40 cycles at 95°C for 30 s, 50°C for 30 s, and 72°C for 45 s; and a final cycle at 72°C for 5 min. The primer sequences used for methylation-specific PCR are listed in supplemental Table S2. The PCR products were isolated on a 1.5% agarose gel and visualized by ethidium bromide staining.
Real Time PCR Detection of miR-21-Mature miR-21 was detected using the miR-Q method as described previously (30,31). In brief, 500 ng of total RNA was used for reverse transcription of mature miR-21 using TaqMan reverse transcription reagents (Applied Biosystems) with a specific reverse primer (RT6-miR-21, tgtcaggcaaccgtattcaccgtgagtggttcaaca). The miR-cDNA generated was then quantitated using SYBR Green PCR Master Mix (Bio-Rad) with another specific primer (short-miR-21-rev, cgtcagatgtccgagtagagggggaacggcgtagcttatcagactga) and one pair of universal primers (MP-fw, tgtcaggcaaccgtattcacc and MP-rev, cgtcagatgtccgagtagag). The amplification was performed by a first step at 95°C for 10 min followed by 40 cycles of 15 s at 95°C, 10 s at 59°C, and 20 s at 72°C. Amplification of Gapdh was used as an internal control.
Construction of pcDNA3.1/myc-His/Pcdh␤ Expression Plasmids and Cell Transfection-Mouse Pcdh␤4 cDNA clone (clone identification number 6401988) and Pcdh␤19 cDNA clone (clone identification number 9055902) were purchased from Source BioScience (Cambridge, UK). These cDNA clones were used as PCR templates to amplify the ORFs of Pcdh␤4 and Pcdh␤19 with the forward primer containing a HindIII site and the reverse primer containing an XbaI site. The sequences of forward and reverse primers were 5Ј-cccaagcttacaatggagacagcgcta-3Ј and 5Ј-ctgtatctagaactattcaacatgt-3Ј, respectively, for Pcdh␤4 and 5Ј-tggaagcttactatggagaatcaagag-3Ј and 5Ј-acatctagaattacagtccctaaat-3Ј, respectively, for Pcdh␤19. The final PCR product was ligated into a pcDNA3.1/myc-His expression vector (Invitrogen) digested with HindIII and XbaI according to the manufacturer's instructions. The resulting vectors were confirmed by sequencing and digested with HindIII and XbaI to release the DNA fragment, which was inserted into the pcDNA3.1 vector upstream from the myc-His tags.
Cell transfections were performed in 6-well plates with Lipofectamine 2000 according to the manufacturer's instructions using 4 g of recombinant plasmids. 24 h after transfection, cells were selected for 3 weeks using G418 (800 g/ml), and nonclonal populations of transfected cells were used for all experiments.
Colony Formation and Anchorage-independent Growth Assay-For the colony formation assay, cells were transfected with 4 g of expression plasmids in 6-well plates and trans-ferred into 100-mm culture dishes the next day. Selection was performed 2 days after transfection with 800 g/ml G418 for 2 weeks. Colonies were stained with crystal violet and counted in 5-10 random fields under a phase-contrast microscope.
An assay of cell growth in soft agar was performed using 24-well culture plates (27). The wells were coated with two layers of agar in different concentrations. The lower layer was 0.7% agar in 0.9% sodium chloride, whereas the upper layer was 0.35% soft agar in complete culture medium. 3 ϫ 10 4 cells were added into the upper layers of the wells. Plates were incubated at 37°C in 5% CO 2 for 2-3 weeks. Then the numbers of the colonies that developed in soft agar were counted in 5-10 random fields under a microscope.
Implantation of Tumor Cells in NOD/SCID Mice-All procedures were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the University of Georgia. Subconfluent tumor cells were harvested and resuspended in serum-free Hanks' balanced salt solution in a 70-l volume containing 2 ϫ 10 6 cells. After NOD/SCID mice (The Jackson Laboratory) were anesthetized with isoflurane, a 70-l single cell suspension mixed with 30 l of Matrigel (BD Biosciences) was injected into flanks of mice of 6 -8 weeks age using a 27-gauge needle (9). Tumor formation was monitored by palpation, and tumor size was measured with calipers once a week.
Western Blotting-Subconfluent cells were harvested and lysed. Total cell lysates containing 30 g of protein were used for Western blotting as described in our earlier report (6).
ALDEFLUOR Assay-To detect tumor cells with high aldehyde dehydrogenase activity, the ALDEFLUOR assay was performed using an ALDEFLUOR kit from STEMCELL Technologies as described previously (9,32). In brief, dissociated single cells (1 ϫ 10 6 cells/ml) were incubated in ALDEFLUOR assay buffer containing aldehyde dehydrogenase substrate (1.5 M) at 37°C for 30 min. In each experiment, a fraction of cells was stained under identical conditions with a specific aldehyde dehydrogenase inhibitor, diethylaminobenzaldehyde (15 M), as a negative control. After staining with propidium iodide, aldehyde dehydrogenase-positive (tumor-initiating cells) and -negative cells (non-tumor-initiating cells) were analyzed using flow cytometry.
Immunochemical and Fluorescent Staining-Immunochemical staining were performed using a VECTASTAIN Elite ABC kit (Vector laboratories) following the manufacturer's instructions. For fluorescent staining, cells were cultured on chamber slides, fixed with 4% paraformaldehyde in PBS for 10 min, and permeabilized with 0.05% Triton X-100. After blocking with 10% goat serum, cells were stained with primary antibodies followed by incubation with secondary fluorescence-conjugated anti-mouse or rabbit IgG (1:250). After washing with PBS, the chamber slides were mounted, and the cells were subjected to fluorescence microscopy.

Microarray Analysis of Her-2-induced Mammary Tumors-
We recently demonstrated that deletion of GnT-V reduces the size of the compartment of tumor-initiating cells in a her-2 mouse model, consequently leading to an inhibition of her-2induced mammary tumor onset (9). To systematically study the mechanisms by which deletion of GnT-V inhibited her-2-induced tumor onset, microarray analyses were performed on the her-2-induced mammary tumors isolated from mice with wildtype (WT) and GnT-V-null (KO) genotypes. Three tumor tissues (T1-3) were collected, each from three her-2 mice with wild-type GnT-V(ϩ/ϩ) and GnT-V-null (Ϫ/Ϫ) backgrounds, respectively, and the GnT-V expression was confirmed by RT-PCR using total RNA (supplemental Fig. 1A). Using the normalized data from the Mouse GeneChip Gene 1.0 ST Array, gene lists were generated using the PARTEK Genomics Suite for differentially expressed genes in these two groups of tumors. There were 142 genes (0.4% of the total 35,556 genes) differentially expressed in GnT-V KO tumors compared with wild-type tumors using a 2-fold change as the ratio threshold. Among the 142 genes identified, 64 genes were up-regulated, and 78 genes were down-regulated (from 2.0-to 6.5-fold) (supplemental Table S3). Of the 142 differentially expressed genes, 57 genes (40.1%) were found to be significantly different (p Ͻ 0.05) ( Table 1). An agglomerative clustering diagram (heat map) was generated from the microarray data of her-2 tumors based on the differentially expressed genes (  Table S4 and were validated by qRT-PCR as shown in supplemental Fig. 2. Gene Expression of Pcdh␤ Cluster Is Up-regulated in GnT-V KO Tumors-The functions of many cell surface receptors have been shown to be altered by the GnT-V expression level, including integrins (4, 6), cadherins (11,12), and EGF family receptors (EGF receptor/ErbB1/her-1 and ErbB2/her-2) and TGF␤ receptors (9,13,14). As listed in supplemental Table S5, compared with wild-type tumors, her-2-induced mammary tumors showed little change in the transcript abundances of other glycosyltransferases and lectins except for Gcnt1, the enzyme that synthesizes core 2-O-glycans. Gcnt1 was significantly down-regulated (Ͼ3-fold) in GnT-V-null tumors based on the microarray data. However, GnT-V expression was reduced by only 2.1-fold in GnT-V KO tumors, although undetectable levels of both GnT-V mRNA and N-linked ␤(1,6) branching were observed by RT-PCR and leukocytic phytohemagglutinin staining, respectively (supplemental Fig. 1, A and D) (9), indicating either that some probes in the GnT-V probe set used in the microarray analysis target the areas of GnT-V mRNA that is transcribed from the non-deleted portions of the GnT-V gene (5) or that nonspecific hybridization generated by a few of the probes in the GnT-V probe set may have occurred. Among the 142 differentially expressed genes in GnT-V KO tumors, neither integrins nor growth factor receptors were observed, indicating that deletion of GnT-V had no detectable effects on the transcript levels of these receptors (supplemental Table S6). Many differentially expressed genes, however, were found in the cadherin family, including classic cadherin 19 (Cdh19) and the Pcdh␤ cluster (Pcdh␤2, Pcdh␤4, Pcdh␤7, Pcdh␤18, and Pcdh␤19) as shown in Fig. 1A (indicated by arrows) and Table 2. Interestingly, all gene members of the Pcdh␤ group were up-regulated to a different degree in GnT-V-null tumors (Table 2), a result validated by qRT-PCR analyses ( Fig. 1, B and C). These results suggested that the protocadherin ␤ gene cluster may function to impede tumor progression and could be involved to delay the onset of her-2-induced mammary tumors in the mice with no GnT-V expression (9).
To test whether the expression of Pcdh␤ was also increased at the protein level, tumor sections were used for immunochemical staining using antibodies against Pcdh␤4 and Pcdh␤7 (Fig. 1D). Consistent with increased gene expression, GnT-V KO tumors in which leukocytic phytohemagglutinin binding was totally suppressed due to deletion of GnT-V showed increased Pcdh␤4 and Pcdh␤7 staining compared with wildtype tumors, indicating increased protein expression of Pcdh␤4 and Pcdh␤7 in GnT-V KO tumors.
To test whether increased expression of Pcdh␤ genes in GnT-V KO tumors was indeed caused by GnT-V expression differences, her-2 tumor cell lines, including GnT-V WT cells, KO cells, and rescued KO cells (KO cells transfected with GnT-V), were established from tumor tissues (9), and transcript levels of Pcdh␤2, Pcdh␤4, Pcdh␤18, and Pcdh␤19 were measured by qRT-PCR because these four genes showed significant -fold changes (-fold change Ն2.0, p Ͻ 0.05) in GnT-V KO tumors based on the microarray data. As show in Fig. 2A, increased expression of the Pcdh␤ genes was observed in the cells derived from GnT-V-null tumors compared with WT tumor cells, consistent with the results obtained from microarray analyses of tumor tissues. After treatment with swainsonine, an inhibitor of N-linked ␤(1,6) branching, WT cells showed increased Pcdh␤ gene expression (Fig. 2B); reintroduction of GnT-V cDNA into KO tumor cells significantly attenuated the expression of Pcdh␤ genes in these cells (Fig. 2C), supporting the hypothesis that differential expression of Pcdh␤ genes in GnT-V KO tumors was due to altered expression levels of GnT-V. A recent study has shown that Pcdh gene clusters, including the Pcdh␤ family, are expressed at lower levels in human breast cancer tissues than in normal tissues, and these lower expression levels were linked to human breast tumorigenesis (26). To confirm and extend the observation concerning the differential expression of the Pcdh␤ gene cluster in mouse mammary tumors after deletion of GnT-V, Pcdh␤ gene expression was measured in the human breast cancer cell line MDA-MB231 after knockdown of GnT-V by siRNA. As shown in Fig. 2D, three members of the human PCDH␤ cluster (PCDH␤3, PCDH␤6, and PCDH␤13), which were shown to be significantly down-regulated in human breast cancer tissues (26), were dramatically enhanced in MDA-MB231 cells with GnT-V knockdown, further confirming the ability of GnT-V expression levels to regulate the expression of the Pcdh␤ cluster during breast tumorigenesis.
Differentially Expressed Pcdh␤ Cluster in GnT-V KO Tumors Is Not Primarily Mediated by Altered DNA Methylation-Epigenetic aberrations are frequent events in human breast cancer development (26,33), and promoter methylation has been implicated in the regulation of Pcdh clusters, including the Pcdh␤ family (26). Also, an association between frequent CpG island methylation and her-2 expression status was observed in human breast cancer (33,34). To determine whether altered expression of the Pcdh␤ cluster in GnT-V KO tumors was caused by promoter methylation differences, the promoter methylation status of Pcdh␤ genes was analyzed. We first searched the mouse Pcdh␤ cluster for CpG islands around the transcription start site (TSS) (Ϫ1 to ϩ1 kb) using CpGPlot. As shown in supplemental Fig. 3, few CpG islands around the TSS area were found in most of the Pcdh␤ genes (Pcdh␤2, Pcdh␤4, Pcdh␤7, Pcdh␤18, and Pcdh␤19) that showed alterations in  JULY 20, 2012 • VOLUME 287 • NUMBER 30

N-Glycan Branching Regulates Pcdh␤ Gene Cluster Expression
GnT-V-null tumors, consistent with reports that the mouse genome contains lower amounts of CpG islands than the human genome (22,23). Interestingly, treatment of her-2 tumor cells with the demethylation drug 5-aza-2Ј-deoxycytidine increased expression of some Pcdh␤ genes, such as Pcdh␤4 and Pcdh␤18 (Fig. 3A), indicating that these genes can still be regulated by methylation status regardless of the lower frequency of CpG islands. Because no CpG islands around the TSS area of Pcdh␤4 and Pcdh␤18 were found, it is likely that demethylation activated certain transcription activators of these genes, indirectly leading to increased gene expression. Methylation-specific PCR targeting the area surrounding the TSS was performed for the Pcdh␤ genes that have predicted CpG island(s) around their TSS (Fig. 3B). Interestingly, Cdh1, whose gene expression is silenced due to promoter hypermethylation in many human tumors, including breast tumors (35), showed very low promoter methylation in the her-2-induced mouse mammary tumors. The same patterns of promoter methylation of Pcdh␤ genes were observed in both wildtype and GnT-V KO cells, however, consistent with the unchanged expression of methylation-related genes in KO tumors observed in microarray analysis data (data not shown). These results indicated that knock-out of GnT-V had little effect on methylation levels of Pcdh␤ genes. The altered gene expression of the Pcdh␤ cluster following deletion of GnT-V probably does not primarily result from changes in methylation status.

Impaired Her-2-mediated Signaling Pathways Are Involved in Differentially Expressed Pcdh␤ Cluster Transcripts in GnT-V KO Tumors-Deletion of
GnT-V had no effect on her-2 expression levels but did cause inhibition of her-2-induced oncogenic signaling pathways (supplemental Fig. 4) (9). Therefore, the differentially expressed Pcdh␤ cluster in GnT-V KO tumors may possibly result from attenuation of her-2-induced activation of either the PKB or ERK signaling pathways. To this end, inhibitors and activators of these two signaling pathways were used to study the effect of her-2 signaling on expression of the Pcdh␤ cluster. As shown in Fig. 4A, treatment of WT tumor cells with PD98059 or wortmannin, inhibitors of MAPK and PI3K/PKB signaling pathways, respectively, led to increased gene expression of the Pcdh␤ cluster in these cells. Stimulation of KO tumor cells with EGF and neuregulin, the ligand of her-1 (EGF FIGURE 1. Microarray analysis of gene expression in her-2-induced tumors. A, total RNA was isolated from three wild-type and GnT-V knock-out tumor tissues and used for microarray analysis. A heat map of 57 transcripts differentially expressed (p Ͻ 0.05) between wild-type and GnT-V knock-out tumors was generated from the microarray data. Red indicates a high expression level, whereas green indicates a low expression level relative to wild-type. T1-3 represents three tumor tissues collected from three mice with GnT-V wild-type and knock-out backgrounds, respectively. B, total RNA was isolated from two wild-type and two GnT-V knock-out tumors and used for validation of transcript levels of the Pcdh␤ cluster by qRT-PCR. For each transcript, the values are normalized to control (Gapdh or Rpl4) and expressed as means with error bars indicating Ϯ1 S.D. C, comparison of -fold change (KO/WT) of transcripts from microarray and qRT-PCR analyses. D, mammary tumor sections from 10-week-old mice were immunostained with leukocytic phytohemagglutinin (L-PHA), anti-Pcdh␤4, and anti-Pcdh␤7, respectively. receptor) and her-3, respectively, activated her-2 signaling pathways as evident by enhanced expression levels of phospho-ERK and phospho-PKB and attenuated gene expression of Pcdh␤ (Fig. 4B). These results indicated that her-2-induced signaling pathways were implicated in GnT-V-mediated gene regulation of the Pcdh␤ cluster. To further confirm that the MAPK and PI3K/PKB pathways involved in the regulation of gene expression of Pcdh␤ were her-2 oncoprotein-related, knockdown of her-2 was performed using her-2/neu siRNA oligonucleotides. As shown in Fig. 4, C-E, a 60% reduction of her-2 in WT tumor cells treated with her-2 siRNA was accompanied by attenuated her-2-induced downstream signaling (phospho-ERK and phospho-PKB), significantly enhancing Pcdh␤ gene expression. This result strongly supports the conclusion that gene expression of the Pcdh␤ cluster is regulated by her-2-induced signaling pathways. Therefore, the deletion of GnT-V observed in the GnT-V-null tumors attenuated the her-2-mediated signaling pathways, resulting in increased gene expression of the Pcdh␤ cluster.

MiR-21 Is One of the Downstream Regulators of Her-2 Signaling Implicated in the Altered Gene Expression of Pcdh␤ Cluster-
Activation of her-2 signaling increases the expression of miR-21 (36,37), which contributes to the increased invasion and metastatic potential of breast cancer cells (36). To further study whether miR-21 was involved in her-2-regulated gene expression of the Pcdh␤ cluster, the expression of mature miR-21 and its effects on Pcdh␤ gene expression were investigated. Consistent with inhibited her-2 signaling, approximately 75% down-regulation of miR-21 levels as detected by qRT-PCR was observed in GnT-V-null tumors compared with wild-type tumors (Fig. 5A). To verify that reduced miR-21 expression was GnT-V-related, we silenced GnT-V by siRNA expression in two human breast cancer lines, MDA-MB231 and SK-BR3, and measured miR-21 levels. We found that miR-21 expression was dramatically inhibited in both cell types after GnT-V knockdown as shown in Fig. 5B. These results demonstrated that miR-21 was regulated by expression levels of GnT-V in the breast cancer cells. To investigate the role of miR-21 in the  altered gene expression of the Pcdh␤ cluster, we transfected her-2 tumor cells (wild-type GnT-V) with a locked nucleic acidmodified anti-miR-21 siRNA oligonucleotide to inhibit miR-21 expression (Fig. 5C). Interestingly, these treatments caused to varying degrees increases in Pcdh␤ cluster gene expression (Fig.  5D), indicating that the Pcdh␤ cluster is the likely target of miR-21 because the expression of these genes was altered when miR-21 levels were reduced. Therefore, increased gene expression of the Pcdh␤ cluster observed in GnT-V knock-out tumors most likely resulted at least to some extent from reduced expression of miR-21 caused by attenuated her-2 signaling. Based on computational target prediction by Targetscan, we investigated some of the predicted targets of miR-21 that either have been validated by experiments or showed significant changes by microarray analysis in our study (Table 3). From our microarray data, we identified several genes whose expression was altered, including Rasgrp1, Npas3, Reck, and Epha4, and changes in these genes were validated by qRT-PCR (data not shown). However, none of the Pcdh␤ family members was among them, suggesting that the Pcdh␤ gene cluster does not contain miR-21 binding sequences in their 3Ј-UTRs and that the altered gene expression of the Pcdh␤ cluster in the GnT-Vnull tumors is regulated indirectly by miR-21 expression. To further substantiate that altered expression of Rasgrp1, Npas3, Reck, and Epha4 resulted from reduced miR-21 expression, the expression of these predicted miR-21 targets were determined in tumor cells whose miR-21 expression was knocked down. As shown in Fig. 5E, the gene expression of Rasgrp1, Npas3, Reck, and Epha4 was all up-regulated after knockdown of miR-21, indicating that all genes, except for Rasgrp1, were indeed targets of miR-21 in the her-2 mammary tumors, which is likely responsible for the altered gene expression of the Pcdh␤ cluster in the GnT-V-null tumors.
Tumor-suppressive Functions of Pcdh␤ Cluster-Pcdh gene clusters have characteristics of suppressors of the transformed phenotype, and their epigenetic silencing has been implicated in some human cancer development, including breast cancer (26,27). To determine whether the differential expression of the Pcdh␤ cluster following deletion of GnT-V contributed to the delayed onset of her-2-mediated tumorigenesis, cell culture-based assays of colony formation and anchorage-independent cell growth commonly utilized for assessing tumor suppressor function were used to evaluate the ability of the Pcdh␤ cluster to inhibit tumor progression (27). Based on the microarray data and commercial availability of clones (Source Bio-Science), Pcdh␤4 and Pcdh␤19 were chosen for these experiments. Overexpression of either Pcdh␤4 or Pcdh␤19 in both mouse her-2 tumor cells (wild-type) and human MDA-MB231 cells, which was confirmed by qRT-PCR and immunofluorescent staining (supplemental Fig. 5), significantly inhibited colony formation after 2 weeks compared with control cells (Fig.  6A). Similarly, anchorage-independent cell growth in soft agar was markedly reduced in both cell lines after transfection of Pcdh␤4 and Pcdh␤19 (Fig. 6B), indicating inhibition of cell proliferation in these cells due to the exogenous expression of Pcdh␤ genes. To further confirm the ability of Pcdh␤ genes to inhibit tumor progression in vivo, we next tested the ability of tumor cells overexpressing Pcdh␤ to form tumors in NOD/ SCID mice. Slower tumor growth was observed in xenografts resulting from injection of either her-2 tumor cells (Fig. 6, C and  D) or MDA-MB231 cells (supplemental Fig. 6A), each of which was overexpressing either Pcdh␤4 or Pcdh␤19, compared with mock-transfected cells. These results indicated reduced tumorigenesis due to Pcdh␤ expression. Hematoxylin/eosin staining of tumor sections showed a reduced number of mitotic cells and less tumor necrosis in tumors resulting from injection of Pcdh␤-expressing cells (supplemental Fig. 6, B and C). Consistent with inhibited tumorigenesis, reduced populations of tumor-initiating cells as detected by the ALDEFLUOR assay (9) were also observed in both Pcdh␤-expressing her-2 tumor cells (1.47 versus 4.04% in control) and their xenograft tumors (2.04 versus 11.6% in control) in NOD/SCID mice (Fig. 6E). Our results are supportive of reports that Pcdh clusters function to suppress tumor progression and suggest that the delayed her-2-induced mammary tumor onset observed after deletion of GnT-V (9) might be attributed at least in part to enhanced expression of Pcdh␤ cluster transcripts.

DISCUSSION
Aberrant N-glycosylation has been documented during tumorigenesis and tumor progression (38). We recently reported that a specific posttranslational modification by deletion of GnT-V disrupts mammary acinar formation and delays her-2-induced mouse mammary tumor onset by down-regulating the relative size of the compartment of tumor-initiating cells (9). In the present study, we show by microarray analysis that 142 differentially expressed genes were observed in GnT-V knock-out her-2-induced tumors compared with wild-type GnT-V her-2-induced tumors, including one group in the cadherin gene cluster, termed Pcdh␤, which was significantly upregulated. Two members of this cluster were chosen and shown to suppress several characteristics of the transformed phenotype that could contribute at least in part to the reduced her-2mediated mammary tumorigenesis observed in GnT-V-null mice.
Pcdh gene clusters consist of three closely linked family members designated Pcdh␣, Pcdh␤, and Pcdh␥ (22). The mouse Pcdh␤ gene family contains 22 members (Pcdh␤1-22) that are organized in a tandem array on chromosome 18, and each member consists of a single exon that encodes the extracellular, transmembrane, and short cytoplasmic protein domains (22).
Although Pcdh clusters play potential roles in neuron development (21), little is known about the cellular functions of these proteins. Unlike classic cadherins, Pcdhs appear not to function via the strong interaction of their extracellular domains as do the classic cadherins because their interactions appear to be much weaker (39,40). Like the classic cadherins, Pcdhs are glycoproteins with N-linked glycan sequons located in their extracellular domains. Studies have shown that Pcdhs are proteolytically cleaved by ␥-secretase complex, producing soluble intracellular fragments that may enter the nucleus and affect gene expression (41,42). Of note, several studies have shown that the Pcdh superfamily may be implicated in tumor development and function as tumor suppressors. Dallosso et al. (27) reported that PCDH gene clusters, including PCDH␣, PCDH␤, and PCDH␥ , were epigenetically silenced in Wilms tumor, and ectopic expression of PCDH␥ showed growth inhibition of tumor cells in vitro. Another study found that the expression of PCDH cluster transcripts was reduced due to promoter hypermethylation in human breast tumors compared with normal breast tissues and concluded that aberrant DNA hypermethylation of multiple PCDH CpG islands is a common event in human breast cancer (26). In our study, up-regulation of gene expression of the Pcdh␤ cluster as detected by microarray anal- To further understand the role of altered gene expression of Pcdh␤ cluster in the inhibition of her-2-induced tumor development that we observed previously, we subcloned two family members of the Pcdh␤ cluster, Pcdh␤4 and Pcdh␤19, into an expression vector and stably transfected mouse and human breast cancer cells. Results obtained from these experiments strongly suggest that Pcdh␤ genes display some features of  genes that suppress or inhibit tumor progression. First, anchorage-independent cell growth in soft agar and colony formation ability after transfection, two commonly used methods to evaluate tumor suppressor function, were remarkably inhibited when Pcdh␤ was expressed in both mouse and human breast cancer lines. Second, in vivo tumor growth in NOD/SCID mice that resulted from injection of tumor cells expressing Pcdh␤ was significantly suppressed compared with that observed from injection of control mock-transfected cells. Third, a reduced proportion of tumor-initiating cells was observed in tumor cells with Pcdh␤ expression, consistent with the inhibited tumor growth of these cells injected in NOD/SCID mice. Based on these observations, it is reasonable to conclude that Pcdh␤ expression may function to inhibit some aspects of the transformed phenotype. The decreased tumorigenesis and the observed inhibition of her-2-induced breast tumor onset associated with GnT-V deletion (9), therefore, can be attributed at least in part to the increased gene expression of the Pcdh␤ cluster. Aberrant levels and patterns of DNA methylation have been found as ubiquitous events in many human cancers (43). Hypermethylation of promoter CpG islands, frequently observed in human breast cancer, is related to transcriptional silencing of some genes, including tumor suppressor genes (44). Although studies have shown that gene expression of PCDH clusters was epigenetically regulated through changes in promoter methylation patterns in some human cancers (26,27,33), the altered expression of the Pcdh␤ cluster caused by GnT-V deletion in our study appeared not to result from aberrant DNA methylation of Pcdh␤ genes. Despite the fact that fewer CpG islands are predicted in promoter areas of mouse Pcdh gene clusters compared with human PCDH clusters (22,23), higher DNA methylation levels as detected by methylationspecific PCR were still observed in her-2 tumors or cells derived  JULY 20, 2012 • VOLUME 287 • NUMBER 30 from these tumors after treatment with a methylation inhibitor, consistent with the previous observation that DNA methylation levels are increased in her-2-positive primary breast cancers (33). However, deletion of GnT-V had little effect on the DNA methylation status of Pcdh␤ genes because the same patterns of methylation in GnT-V knock-out tumors were observed as those in wild-type tumors. These results indicated that DNA methylation of the Pcdh␤ cluster was not regulated by GnT-V expression levels, and altered gene expression of the Pcdh␤ cluster caused by null GnT-V was not likely due to the changes in DNA methylation. Supporting this conclusion, no significant changes in expression of DNA methylation-related enzymes, such as DNMT1, DNMT3A, and DNMT3B, were observed in the comparison of GnT-V knock-out and wild-type tumors by microarray analyses.

N-Glycan Branching Regulates Pcdh␤ Gene Cluster Expression
Although results from microarray analyses showed that deletion of GnT-V had little effect on expression of her-2 and other oncoproteins from the EGF family, such as her-3 and EGF receptor (her-1), consistent with our previous results (9), we found that two major downstream signaling pathways mediated by her-2 activation, the ERK and PI3K/PKB pathways, are significantly inhibited in GnT-V knock-out tumors (9). This result prompted us to investigate further whether aberrant her-2-mediated signaling pathways were implicated in altered gene expression of the Pcdh␤ cluster observed in GnT-V knock-out tumors. We observed that stimulation of GnT-V knock-out cells with growth factors (EGF and neuregulin) significantly inhibited gene expression of the Pcdh␤ cluster, whereas treatment of GnT-V wild-type tumor cells with PD98059 and wortmannin, inhibitors of ERK and PKB, respectively, remarkably enhanced gene expression of the Pcdh␤ family, indicating the direct involvement of these two pathways in regulating Pcdh␤ gene expression. The regulation of her-2-mediated signaling pathways on the gene expression of the Pcdh␤ cluster was further confirmed by silencing her-2 oncoprotein by her-2 siRNA expression, demonstrating that altered gene expression of the Pcdh␤ family closely followed changes in her-2 expression levels. These results strongly argue that her-2-mediated downstream signaling pathways negatively regulated the expression of the Pcdh␤ cluster. Deletion of GnT-V impaired her-2 signaling pathways, therefore attenuating the inhibition of her-2 signaling on the gene expression of the Pcdh␤ family. The involvement of other signaling pathways in regulating gene expression of the Pcdh␤ family in GnT-V knock-out tumors cannot be ruled out, however, because of the ability of GnT-V glycan products to modify the function of other glycoprotein receptors.
To further investigate the mechanisms governing how impaired her-2-mediated downstream signaling positively regulates gene expression of the Pcdh␤ cluster in GnT-V-null tumors, we explored the possible downstream effectors that are regulated by her-2-mediated signaling pathways that could be involved in the regulation of Pcdh␤ cluster expression in GnT-V-null tumors. MicroRNAs constitute an abundant class of non-coding RNAs of about 21-23 nucleotides that negatively regulate protein expression by targeting mRNA transcripts and mediate either translational repression or degradation of targeted mRNA (45,46). There is increasing evidence that microRNAs are key molecules involved in cancer initiation and progression (47,48). Functioning as an oncogene (49), miR-21 is one of the most studied microRNAs associated with cancer and is highly up-regulated in breast cancer (50,51). Studies have shown that her-2 signaling activates several transcription factors, such as Ets-1 and Ap-1, that stimulate the expression of miR-21 (36,37), and increased metastatic potential of her-2expressing breast cancer cells is mediated by up-regulation of miR-21 (36). Altered expression of miR-21 may, therefore, be one of the downstream regulators of her-2 signaling implicated in her-2-mediated up-regulation of the Pcdh␤ cluster. Consistent with impaired her-2 signaling pathways, the expression of miR-21 was significantly reduced in GnT-V knock-out tumors. Mostly importantly, silencing of miR-21 using a locked nucleic acid-modified anti-miR-21 resulted in the enhancement to different degrees of the expression of the Pcdh␤ gene cluster, indicating the involvement of miR-21 in the altered gene expression of Pcdh␤ caused by inhibited her-2 signaling at least in part. Interestingly, the effect of regulation of miR-21 on Pcdh␤ expression appeared to be indirect because Pcdh␤ genes are not among the predicted potential targets of miR-21 due to the lack of binding sequences for miR-21 in their 3Ј-UTRs. Although some of the miR-21 target genes that have been experimentally confirmed were not affected by reduced miR-21 expression in our study, including Pdcd4, Pten, Tpm1, and Spry2 (49), effects on the expression of other target genes, for example Npas3, a newly confirmed tumor suppressor gene (52); Reck, a membrane-associated inhibitor of metalloproteinases and recently confirmed miR-21 target gene (53); and Epha4, a member of the Eph receptor tyrosine kinases (54), were observed. These targets were further validated by miR-21 silencing experiments (Fig. 5E) and likely mediate miR-21 regulation of the expression of the Pcdh␤ gene cluster. Indirect regulation of genes by miR-21 has been documented; for example, an array expression analysis of MCF-7 cells depleted of miR-21 by siRNA identified miR-21 target genes that were subsequently experimentally validated, including targets that did not contain miR-21 binding sequences in their 3Ј-UTRs (55). Also, miR-21 was found to indirectly inhibit the expression of DNMT1 by targeting Ras-grp1, a critical upstream regulator of the Ras-MAPK signaling cascade, which modulates DNMT1 levels in lupus CD4 ϩ T cells (56). Another indirect regulation of Bcl-2 by miR-21 has also been reported in breast cancer (57).
The attenuation of her-2-mediated signaling pathways observed in GnT-V KO tumor cells is likely to be the result of aberrant N-glycosylation of her-2 and/or the ErbB family of receptors that can lead to altered ligand (EGF and neuregulin) binding, regulation of the endocytosis of signaling complexes, and/or inhibition of dimer and multimer formation among members of this family (13,58,59). Experiments are in progress to determine the mechanisms by which ␤(1,6) N-linked glycosylation regulates her-2 function.
In conclusion, we provide evidence that a gene cluster in the cadherin superfamily, Pcdh␤, functions to inhibit her-2-mediated mouse mammary tumorigenicity. Our results show that GnT-V expression levels regulate the expression of members of the Pcdh␤ cluster by affecting her-2-mediated downstream signaling pathways. This increased Pcdh␤ expression is mediated at least in part by miR-21, one of the downstream effectors regulated by her-2 signaling (Fig. 7). Because up to 30% of human breast tumors are her-2-positive (15,16), our findings shed new light on the molecular mechanisms of the suppressive effects of GnT-V deletion on mammary tumorigenesis and progression and illuminate GnT-V as a potential target for an inhibitor with therapeutic utility for her-2-positive human breast cancer.