The Role of PIWIL4, an Argonaute Family Protein, in Breast Cancer*

P-element-induced wimpy testis (PIWI) proteins bind to PIWI-interacting RNAs and play key roles in the biogenesis and functions of PIWI-interacting RNAs. It has been reported that PIWI proteins are essential for stem cell self-renewal and germline development in diverse organisms and that they are ectopically expressed in multiple forms of cancer. However, the role of PIWI in cancer remains elusive. Here we report that one of the four PIWI proteins in humans, PIWIL4, is highly expressed in both breast cancer tissues and the cytoplasm of MDA-MB-231 cells derived from breast cancer. Reducing PIWIL4 expression drastically impairs the migration ability of MDA-MB-231 cells, significantly increases their apoptosis, and mildly affects their proliferation. Our transcriptome and proteome analysis reveal that these functions are at least partially achieved via the PIWIL4 regulation of TGF-β and FGF signaling pathways and MHC class II proteins. These findings suggest that PIWIL4 may serve as a potential therapeutic target for breast cancer.

20 pairs of clinical samples were purchased from the tissue bank of the Institute of Health Sciences, Chinese Academy of Sciences. The local ethics committee approved the study, and the regulations of this committee were followed.
RNA Extraction and Quantitative Real-time PCR-Total RNA was isolated using TRIzol (Invitrogen) according to the protocol of the manufacturer. For reverse transcription, we used 1 g of RNA reverse transcriptase and the ABI high-capacity kit (Life Technologies, 4368814). Real-time PCR reactions were performed according to the protocol of the Bio-Rad real-time PCR system (iQ TM SYBR Green Supermix and CFX96 TM real-time system). Primers of GAPDH were designed as the real-time PCR control. Quantitative PCR primers are listed in supplemental Table S1 (33,34).
PIWIL4 cDNA Cloning-The PIWL4 cDNA primers were designed as follows: forward, 5Ј-CGCGGATCCATGAGTGG-AAGAGCCCG-3Ј; reverse, 5Ј-CGCGGATCCTCACAGGTA-GAAGAGATGG-3Ј. Total RNA was used for cDNA synthesis by SuperScript III reverse transcriptase (Invitrogen, 18080044) according to the protocol of the manufacturer. The cDNA was used as a template for amplification by Phusion high-fidelity DNA polymerase (New England Biolabs, M0530L) in PCR and cloned into the pMDTM19-T vector by a cloning kit (Takara, 6013).
PIWIL4 shRNA Knockdown Analysis-RNAi vector pSUPERpuro was purchased from Promega. Three short hairpin DNA sequences were designed as in supplemental Table S1. The DNA sense and antisense sequences were annealed in a pairwise fashion and cloned into the pSUPER-puro vector.
MDA-MB-231 cells were transfected with Lipofectamine 2000 (Life Technologies, 11668019) according to the instructions of the manufacturer. Transfected cells were selected using 0.8 g/ml puromycin. The silencing effect on the PIWIL4 gene was assessed by quantitative PCR and Western blotting analysis according to the aforementioned protocols.
Nuclear-Cytoplasmic Fractionation-Approximately 10 6 cells were washed with PBS twice, followed by centrifugation at 1000 ϫ g for 3 min. The resulting supernatant was discarded. The pellet was added to 150 -200 l of buffer B (10 mM Hepes, 10 mM KCl, 2 mM MgCl 2 , 0.1 mM EDTA, 0.2% Nonidet P-40, protease inhibitor, and 1 mM DTT), incubated on ice for 30 min, and centrifuged at 4°C at 13,200 rpm for 6 min. Cytoplasmic proteins were collected as the supernatant to new tubes and stored at Ϫ80°C until use. The nuclear pellet was washed with 500 l of wash buffer (10 mM Hepes, 20 mM KCl, 2 mM MgCl 2 , 0.1 mM EDTA, protease inhibitor, and 1 mM DTT) and centrifuged at 4°C at 3000 rpm for 5 min. The supernatant was discarded, and the nuclear pellet was dissolved in 50 -80 l of extraction buffer (20 mM Hepes, 0.64 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 2.5% glycerol, protease inhibitor, and 1 mM DTT), vortexed for 15 s, incubated on ice for 30 min, vortexed for 10 -15 s at 10-min intervals, and centrifuged at 4°C and 13,200 rpm for 20 min. The supernatants, which contained nucleus proteins, were collected in new tubes and stored at Ϫ80°C for later use. Separation of the cytoplasmic and nuclear fractions was verified by Western blotting for ␤-tubulin and TBP (Cell Signaling Technology, anti-␤-tubulin and anti-TBP antibody).
Apoptosis Assay-We stained cancer cells with the FITC Annexin V apoptosis detection kit I (BD Biosciences, 556547) according to the protocol of the manufacturer and analyzed early-and late-stage apoptosis by FACS (FACS Aria TM IIII, BD Biosciences).
Wound Healing and Transwell Migration Assays-5 ϫ 10 4 cells were seeded in a 96-well plate, and a wound was introduced with a wound maker kit (Essen Bioscience). Wounded monolayers were washed twice with PBS to remove non-adherent cells. Cells were cultured in 2% FBS for 48 h, and the wound healing tracks were recorded with an IncuCyte Zoom live-cell analysis system (Essen Bioscience). The transwell assay was done using Corning FluoroBlok TM cell culture inserts (Falcon, 351152) according to the protocol of the manufacturer.
For RNA sequencing, ϳ10 6 MDA-MB-231 cells with or without shPIWIL4 -3 treatment were collected. Total RNA was isolated using TRIzol (Invitrogen, 15596026) according to the protocol of the manufacturer. mRNAs and small RNAs were sequenced with an Illumina HiSeq3000 platform (Jing Neng Co., Shanghai, China). HTSeq software was used for the statistics of the original mRNA deep sequencing data of the control and shPIWIL4 -3-treated samples. DESeq software was used to screen differentially expressed genes compared with the control (p Յ 0.05 or -fold change Ն 2).
Gene Ontology and Pathway Analysis-A gene ontology and pathway analysis was conducted using an online tool. Gene ontology level 2 and 3 categories were selected for mRNA sequencing and mass spectrometry analysis, respectively.
Genome Mapping-The small RNA sequences were generated with Illumina HiSeq3000, and we trimmed off linkers with the FASTX toolkit. We selected sequences with sizes ranging from 12-43 nucleotides for analysis. The reference genome was human genome hg38. The gene annotation information was based on the Ensembl annotation. The miRNA annotation was based on version 21 of miRBase. Repeats and transposon annotation were based on RepeatMasker. Complete rRNA references were from the NCBI. One mismatch was allowed for mapping sequences against rRNA, tRNA, small nucleolar RNA, and small nuclear RNA.
Identification of Known piRNAs-Human known piRNAs in piRNABank and piRBase served as references to identify known piRNAs. Only the small RNAs that had an identical sequence to these reference piRNA sequences were defined as known piRNAs.
Effect of PIWIL4 Knockdown on Known piRNAs-The abundance of piRNAs was scaled according to the total mapped reads for comparison across different conditions. The shPIWIL4 treatedto-nontreated control ratio was calculated to measure the effect of PIWIL4 knockdown on piRNAs. In total, we detected ϳ200 small RNAs annotated as piRNAs in piRNABank and piRBase, of which ϳ20 were affected (fold change Ն 2 or Յ 1 ⁄ 2) when PIWIL4 was knocked down.

Results
The PIWIL4 Gene Is Highly Expressed in both Breast Cell Lines and Breast Cancer Samples-It has been reported that PIWI proteins have aberrant and ectopic expression in cancers, such as the expression of PIWIL2 in breast cancer (17)(18)(19)(20)(21)(22)35). Because all PIWI proteins are necessary for germ line development and stem cell self-renewal, we systematically examined the expression of three known active PIWI genes in humans, piwil1, piwil2, and piwil4, in six different types of human breast cancer cell lines and 20 pairs of normal and breast cancer samples from 20 patients by quantitative RT-PCR ( Fig. 1). piwil1 and piwil2 are expressed in four and two cell lines, respectively, at significantly higher levels than in a normal breast cell line (MCF-10A, a physiologically negative control) or a mouse breast cancer line (4T1, Fig. 1, A and B), consistent with a previous report of PIWIL2 expression in breast cancer cells (35). In addition, PIWIL1 is expressed at significantly higher levels compared with the normal tissue controls in two of 20 breast cancer samples (Fig. 1D). PIWIL2 is more highly expressed than PIWIL1 in one cancer cell line, MDA-MB-231 (Fig. 1B), and four of 20 breast cancer samples (Fig. 1E). Remarkably, PIWIL4 is expressed at very high levels in five of six cancer cell lines, much higher than PIWIL1 and PIWIL2 (Fig. 1C). Furthermore, PIWIL4 is significantly expressed in nine of 20 breast cancer samples, with five samples displaying a more than 50-fold upregulation (Fig. 1F). This indicates that, among the three PIWI genes known to be active, PIWIL4 has the best correlation with breast cancer.
To further establish a functional correlation between the expression of PIWI proteins and breast cancer, we analyzed a cancer genomics data bank, cBioPortal (36,37). Our analysis showed that all three human PIWI genes have genetic alterations in breast cancer. We found that 2%, 7%, and 2% of breast patients cancer patients have amplification, deletion, missense mutation, and truncation of PIWIL1, PIWIL2, and PIWIL4, respectively (supplemental Fig. S1). Moreover, 2%, 2%, and 4% of PIWIL1, PIWL2, and PIWIL4 patients have mRNA up-regulation (supplemental Fig. S1). When only considering TNBC patients, the percentage is increased to 11% (n ϭ 82). The 4% of PIWIL4-positive patients is apparently lower than the 9% of 20 patients who we examined directly. This could be due to inherently lower sensitivity associated with surveying the expression of many genes in more than 5000 tumor samples from 20 can-cer studies. Alternatively, it could also reflect that 20 patients is a small sample size.
To investigate further correlation between PIWIL4 and breast cancer, we conducted a Kaplan-Meier survival analysis based on clinical data from a bioinformatics website, including 351-1616 patients for each analysis. Patients who have the upper tertile level of PIWIL4 expression have lower overall survival, distance metastasis-free survival, and post-progression survival than patients with the lower 67% of PIWIL4 expression (Fig. 2). This analysis further correlates PIWIL4 to breast cancer.  in the vector (OE) by centrifugation, followed by Western blotting, which revealed that PIWIL4 is present in the cytoplasm (Fig. 3A). Immunofluorescence microscopy results confirmed that PIWIL4 is present in the cytoplasm and further showed that it is not co-localized with the P body (Fig. 3B). Surprisingly, not all MDA-MB-231 cells showed a PIWIL4 signal when stained with anti-PIWIL4 antibody against 639 -839 amino acid residues (in exons 15-20) of PIWIL4 (Fig. 3, B and  S2). This indicates that these cells either do not express PIWIL4 or express aberrant PIWIL4 isoforms that lack 639 -839 amino acid residues. To discriminate between these two possibilities, we reverse-transcribed PIWIL4 mRNAs isolated from MDA-MB-231 cells and sequenced the resulting cDNAs. These analyses revealed that there are five variants of PIWIL4 transcripts in MDA-MB-231 cells (Fig. 3C). All of the variants have deletions or insertions that cause premature termination before the 639 -839 amino acid residues of PIWIL4 (Fig. 3D). The inserted sequences in PL4L9, PL4L13, and PL4L15 are all intron sequences of the PIWIL4 gene. Correspondingly, there are three aberrant PIWIL4 protein isoforms (Fig. 3D). These observations indicate the instability of the MDA-MB-231 genome, which is a typical situation in many cancer cell lines.
PIWIL4 Inhibits MDA-MB-231 Cell Apoptosis and Promotes Migration-To reveal the effect of ectopic PIWIL4 expression in cancer cells, we knocked down PIWIL4 expression using the pSuper-shRNA knockdown system with three different shRNA sequences against PIWIL4 mRNA and analyzed the effect of the knockdowns on the function of MDA-MB-231 cells. Fig. 4A indicates the three shRNA-corresponding regions in PIWIL4 mRNA (at the exon 2-exon 3 junction, exon 7, and exon 17, respectively). All of these three shRNAs effectively reduced both PIWIL4 mRNA and PIWIL4 protein levels (Fig. 4B). Reduction of PIWIL4 weakly inhibited cell proliferation (Fig.  4C) and colony formation ability (Fig. 4D) but caused signifi-cantly increased apoptosis (Fig. 4F). Correspondingly, we found that PIWIL4 knockdown increased the levels of phosphorylated CHK2 and phosphorylated CDC2 (two G 2 phase checkpoint markers) and p27 (a cell proliferation marker) and drastically increased the level of cleaved caspase 3, an indicator of the activation of the caspase 3 pathway (Fig. 4E). These results validated the negative impact of PIWL4 knockdown on cancer cell survival.
To explore whether PIWIL4 also has a role in the migration and metastasis of cancer cells, we first conducted the in vitro wound healing assay (i.e. the scratch assay) and the transwell migration assay on MDA-MB-231 cells. Knocking down PIWIL4 with each of the three anti-PIWIL4 shRNAs significantly inhibited the migration ability of these cells by either the scratch assay (Fig. 4G) or the transwell assay (Fig. 4H). These results indicate that PIWIL4 promotes the migratory ability of MDA-MB-231 cells.
To further investigate whether the PIWIL4-dependent migratory ability of MDA-MB-231 cells reflects their epithelialto-mesenchymal transition, we examined the expression of a key epithelial marker, E-cadherin, and a key mesenchymal marker, N-cadherin, in MDA-MB-231 cells by Western blotting analysis. These cells normally express only N-cadherin but not E-cadherin (Fig. 4E), indicating that these cancer cells have acquired the mesenchymal fate. Remarkably, knocking down PIWIL4 abolished N-cadherin expression but induced strong expression of E-cadherin (Fig. 4E). These data indicate that the PIWIL4-deficient MDA-MB-231 cells have abandoned the mesenchymal fate and have reacquired key features of the epithelial fate. Therefore, PIWIL4 is required for the epithelial-tomesenchymal transition and acquisition of the migratory ability of MDA-MB-231 cells.
PIWIL4 Activates TGF-␤ and FGF Signaling in MDA-MB-231 Cells-To investigate the molecular mechanism underlying PIWIL4 function in MDA-MB-231 cancer cells, we examined the transcriptome and proteome of MDA-MB-231 cells with and without shPIWIL4 -3 knockdown by deep sequencing of mRNAs and mass spectrometry of total cellular lysate, respectively. We found that, among 26,057 mRNAs that can be detected by at least one read in MDA-MB-231 cells via deep sequencing, 332 mRNAs were significantly down-regulated but 400 mRNAs were up-regulated when PIWIL4 was knocked down, as summarized in the heat map in Fig. 5A. The 60 most up-and down-regulated genes (38 up-regulated and 22 downregulated) are shown with names in Fig. 5B.
To detect the impact of PIWIL4 knockdown on the cellular protein profile, we conducted mass spectrometry analyses of MDA-MB-231 cells and MDA-MB-231 cells with PIWIL4 knocked down by the three different shRNAs. The three PIWIL4 knockdown samples show nicely overlapping protein profiles (49.74% overlap between shPIWIL4 -1 and shPIWIL4 -2, 50.41% overlap between shPIWIL4 -1 and shPIWIL4 -3, 50.98% overlap between shPIWIL4 -2 and shPIWIL4 -3, and 36.47% were found in all three samples), with 2571 proteins detected in all three PIWIL4 knockdown samples (Fig. 5C). Of the 2571 proteins, 2364 proteins were also present in MDA-MB-231 cells without PIWIL4 knockdown. Only 207 proteins were not detected in normal cancer cells. This suggests that reducing PIWIL4 expression does not significantly activate the expression of new proteins. However, 1288 proteins were not detectable in any of the three PIWIL4 knockdown samples. This indicates that PIWIL4 promotes the expression of many proteins in MDA-MB-231 cancer cells.
To investigate which mRNAs are down-regulated and which proteins become undetected under the PIWIL4 deficiency condition, we conducted a pathway enrichment analysis of the 332 most down-regulated mRNAs and the 1288 proteins that are only detected in MDA-MB-231 cancer cells without PIWIL4 knockdown. These analyses showed that MAPK-ERK, TGF-␤, and FGF were the most enriched pathways (Fig. 5E and supplemental Fig. S4 A). Full lists are shown in supplemental Tables S2  and S3 and include TGF␤R1, TGF␤R2, FGFR2, TGF␤1, and  TGF␤3, which are no longer detectable in the three PIWIL4 knockdown samples.
To validate the down-regulation of the above proteins, we measure the mRNA expression of TGF␤R1, TGF␤R2, FGFR2, TGF␤1, and TGF␤3 by quantitative RT-PCR. Indeed, all of the five genes were down-regulated in all three different PIWIL4 knockdown cell samples (Fig. 5F). The down-regulation of the TGF pathway components is further supported by gene ontology analysis of 1288 proteins specifically present in normal cancer cells and 207 proteins specifically present in PIWIL4 knockdown cells (totally 1495 proteins), which showed that the most enriched gene ontology term is transferase activity and transferring phosphorus-containing groups (Fig. 5G), which is frequently involved in protein activity, including the TGF-␤, FGF, and MAPK-ERK signaling pathways. Supplemental Fig. S3 shows the most significantly enriched biological processes, cellular components, and molecular functions as ranked by their p values, which reveals that cell division is also an enriched biological process (the full list is presented in supplemental Table  S4). Taken together, we conclude that PIWIL4 promotes MDA-MB-231 epithelial-to-mesenchymal transition, migratory ability, and proliferation and inhibited apoptosis partially by activating the TGF-␤ and FGF signaling pathways.
PIWIL4 Represses the Expression of MHC II Genes in MDA-MB-231-Tumorigenesis and development is a complex process involving not only cell proliferation but also tumor immune escape pathways. Combined gene ontology analysis of the 400 up-regulated mRNAs and 332 down-regulated mRNAs (total of 732 genes) revealed that genes related to tumor immune escape were enriched among the 62 gene ontology terms (supplemental Table S5). Especially the top 10 gene ontology terms are more related to cell adhesion, cell periphery, extracellular matrix structural constituents, single-multicellular organism cellular processes, and anatomical structure development (supplemental Fig. S4C). Supplemental Fig. S4, D-F, shows the top 10 most significant biological processes, cellular components, and molecular functions, respectively. Immunoresponse and cell adhesion are enriched among biological process, and the enriched molecules are mostly localized to the cell periphery, extracellular space, and cell surface, performing their roles of extracellular matrix structural constituent, antigen binding, and receptor binding. Moreover, cell proliferation and death are also in the enriched terms. Pathway enrichment analysis of the 400 up-regulated genes reveals that PIWIL4 represses the expression of MHC class II mRNAs, including HLA-DR␣, HLA-DP␣1, HLA-DO␣, HLA-DP␤1, cathepsin S, cathepsin E, dynamin 1, and CD74. This might lead to the upregulation of some immune responses involved in the neuronal cell adhesion molecule, the signal regulatory protein family, and the cGMP pathway, as evident in supplemental Fig. S4B (a full list is presented in supplemental Table S6). These observations are consistent with the enrichment of translation-related mechanisms among the 207 most up-regulated proteins under the PIWIL4 deficiency condition (Fig. 5D, a full list is presented in supplemental Table S7). These results imply that PIWIL4  MAY 13, 2016 • VOLUME 291 • NUMBER 20 might repress MHC class II, which might help cancer cells avoid immune recognition (38,39).

PIWI Proteins and Breast Cancer
MDA-MB-231 Cells Express a Small Number of piRNAs-As the first step to investigate whether PIWIL4 function in MDA-MB-231 cells is related to piRNA, we examined whether these cells express piRNAs under normal PIWIL4 expression and PIWIL4 knockdown conditions. We isolated small RNA (12-42 nucleotides in length) from these cells with or without shPIWIL4 knockdown, followed by RT-PCR and deep sequencing. We observed the presence of many small RNAs in MDA- MB-231 cells under both conditions. These include miRNAs, other small RNAs, and fragments of rRNAs and tRNAs peaked at 19-nucleotide length (Figs. 6, A and B). Interestingly, under the knockdown condition, the miRNA population become significantly enriched in abundance (from 38% to 60% of the total small RNA reads, reflecting a 58% increase) at the expense of rRNA and tRNA fragments. Although this change could be due to less degradation of rRNA and tRNA in the RNA preparation from the PIWIL4 knockdown cells, this possibility is unlikely because, if so, the "other small RNA" fraction should correspondingly show a 58% increase in its abundance. However, the other small RNA fraction is decreased from 18% to 14%, reflecting a 29% decrease. Therefore, the increase in miRNA abundance mostly, if not exclusively, reflects a role of PIWIL4 in repressing miRNA expression in MDA-MB-231 cells.
To search for piRNAs, we selected small RNAs in the piRNA size range (24 -32 nucleotides) for further analysis. Totally, 98,139 species of small RNAs are in this fraction from MDA-MB-231 cells without PIWIL4 knockdown and 55,624 species in this fraction from these cells with PIWIL4 knockdown. Among them, 61 species have previously been reported as human piRNAs (piRNABank and piRBase), with 50 and 53 species present in the normal and PIWIL4 knockdown MDA-MB-231 cells, respectively, representing 1% and 2% of the total small RNAs from the normal and knockdown cells (Fig. 6C).
The identification of these known piRNAs allowed us to conclude with reasonable confidence that at least some of the remaining 24-to 32-nucleotide small RNAs are piNRAs. However, these small RNA do not show the enrichment at either the 5Ј first position for U that is a signature of primary piRNA or at the 5Ј 10th position for A that is a signature of secondary piRNA (Fig. 6D). This could reflect that some of the small RNAs are not piRNAs. Alternatively, the signature may not be obvious because of the small number of species. To further search for sequence features of these small RNAs that may help us distinguish between the two possibilities, we mapped these small RNAs to the genome and found that they correspond to DNA repeats, transposons, intergenic sequences, and genes at proportions one would expect from piRNAs (Fig. 6E). These analyses support our conclusion that at least some of the small RNAs are likely piRNAs. Without experimental validation of their true identification, we can only call these small RNAs putative piRNAs. Combined data on the known piRNAs and putative piRNAs indicate that the MDA-MB-231 cells express a small number of piRNAs.
To assess how much of the expression of the putative piRNAs is affected by PIWIL4 knockdown, we first examined 24 known piRNAs (of the 61 known ones) that have at least 10 mappable reads in either normal or knockdown samples. Only three have a 2-fold or more decrease in PIWIL4 knockdown cells, and only one has a more than 2-fold increase, colored in red and blue, respectively, in supplemental Table S8. We then examined putative piRNAs with at least 10 mappable reads in both types of the cells. Of 1192 such piRNA candidates, only 333 show 2-fold or greater changes in abundance (Fig. 6F). These combined data on the known piRNAs and putative piRNAs indicate that the piRNA expression is not much affected by reducing the PIWIL4 level.

Discussion
PIWI proteins have been reported to be ectopically expressed in diverse types of cancer (17)(18)(19)(20)(21). However, most of these studies are at a correlative level. The role of PIWI expression in cancer remains unclear. Here we reported that PIWIL4 is widely expressed in breast cancer samples from different patients and in multiple breast cancer cell lines. Moreover, using a TNBC line (MDA-MB-231) as a model, we demonstrated that PIWIL4 expression promotes cancer cell survival, division, and, more significantly, migration. Our study provides a definitive demonstration of the function of PIWI proteins in cancer cells.
Furthermore, our transcriptome analysis revealed that PIWIL4 achieves its function in the breast cancer cells partially by activating TGF-␤, MAPK/ERK, and FGF signaling and repressing MHC class II expression (Fig. 7). The TGF-␤, FGF, and MAPK-ERK pathways are well known to play key roles in cancer. Specifically, TGF␤R1 and TGF␤R2 are known as receptors of TGF and other signaling molecules. They then activate downstream signal molecules during epithelial-to-mesenchymal transition of tumor cells (40). Some researchers suggested that the expression of FGF4 and FGFR2 in ovarian cancer stemlike cells/cancer-initiating cells promoted their tumor initiation capacity (41). Our mass spectrometry results show that PIWIL4 promotes the TGF-␤ and FGF signaling pathways in the breast cancer cell line MDA-MB-231, which links PIWIL4 function to these important signaling pathways. In addition, our transcriptome analysis demonstrates that PIWIL4 also upregulates the MAPK-ERK signaling pathway (supplemental Fig.  S4A). These findings reveal a new dimension of regulation of TGF␤ and FGF signaling in cancer formation. The PIWIL4 regulation of these classic cancer signaling pathways provides an intellectual framework for further investigation of how PIWIL4 promotes breast cancer cell migration, survival, and proliferation through these pathways.
It is intriguing to see that PIWIL4 also suppress the expression of MHC class II genes. Tumorigenesis is a complex process involving not only signaling but also tumor immune escape pathways. MHC class II molecules are constitutively expressed in professional antigen-presenting cells, which is an essential part of cell-mediated immunity, but may also be induced on other cells by interferon ␥ (42). The suppression of MHC class II genes might help cancer cells avoid immune recognition and reaction by the T cell pathway.
PIWI proteins have multiple roles in the piRNA pathway. They participate in piRNA biogenesis in germ cells and represses transposon activity by affecting local epigenetic states and transcription (4). Correspondingly, piRNAs have important roles in mRNA regulation in the mouse by rendering targeting specificity (6,43). Furthermore, some previous researchers suggested that piRNAs have independent functions, such as piR-823 having a role in breast cancer (5). Our analysis indicates that a small number of piRNAs are expressed in MDA-MB-231 cells. This raised the possibility that PIWIL4 works with piRNA to achieve its function. Further isolation of the PIWIL4-piRNA complex, identification of their regulatory targets, and analysis of their regulatory effect on the targets should shed light on how PIWI-mediated mechanisms function in cancer development.