Down-regulated MicroRNAs in Gastric Carcinoma May Be Targets for Therapeutic Intervention and Replacement Therapy

Gastric cancer is one of the leading types of cancer with an annual death toll of 700,000 worldwide. Despite the fact that several agents are approved for its treatment, high percentage of recurrence and intractability of metastatic disease remain a major problem. The identification of new targets and modalities for treatment are therefore of high priority. We have searched the literature for microRNAs down-regulated in gastric cancer with efficacy in gastric cancer-related murine xenograft models after reconstitution therapy. Among the identified miRs were 25 miRs targeting transcription factors, seven of them regulating cell-cycle and apotosis-related targets, and five of them regulating GTPase-related targets such as GAPs and GEFs. According to criteria such as prognostic impact, functional data, and tractability, miR-133 b/a (MCL1) and miR-518 (MDM2) are suggested as potentially valuable targets for further evaluation and possible treatment of gastric cancer.

guanine nucleotide exchange factors (GEFs) and GTPases activating proteins (GAPs), in pathogenesis and metastasis of GC. We focus on miRs which are down-regulated in GC tissue in comparison to corresponding normal tissues and exhibit efficacy in preclinical in vivo models.

Roles of microRNAs in Oncology
microRNAs (miRs) are double-stranded RNAs comprising 22-25 nts and are generated from hairpin-containing precursor transcripts (12). They are transcribed in the nucleus by RNA polymerase II, processed, exported to the cytoplasm and released as 22-25 nts miR-duplexes (13,14). One of the strands of the RNA duplex is maintained (guide strand), the other strand (passenger strand) is degraded (13,14). Subsequently, the guide strand binds to the 3'-untranslated region (3'-UTR) of the mRNA of the corresponding targets and mediates their degradation and/or inhibition of their translation (13,14). In contrast to interfering RNA (RNAi) which targets a single type of mRNA, miRs can affect up to several hundreds of different mRNAs (15). Therefore, miRs can modulate several oncogenic and tumor-suppressive pathways and hence have the potential of rewiring tumor cells to a differentiated state (15). In humans, approximately 1,000 genes encoding miRs have been identified so far (16). They are positioned as separate or clustered genes, in introns or in coding regions (6). miRs can act as oncogenes as well as tumor suppressors (16). Their actual function can be context-dependent (16).
In a proof-of-concept experiment (POC) it was shown that expression of the miR-15/16 cluster in mice prevents chronic lymphocytic leukemia analogous to the human disease by targeting B cell lymphoma 2 (BCL2) (17). In another POC experiment an oncogenic role was identified for miR-221, which induced hepatocellular carcinoma in transgenic mice after liver-specific expression (18). miRs play a role in tumorigenesis, angiogenesis, tumor-stromal interactions, and metastasis (19). We have recently summarized their role in metastasis in breast, prostate, ovarian, lung, pancreatic and liver cancer (20)(21)(22)(23)(24)(25).
miRs -105, -138 target SOX9 and SOX4. miRs-105 and -138 ( Figure 1A) target sex determining region Y-related high mobility group 9 and 4 (SOX9 and SOX4) (36)(37)(38). Reduced levels of miR-105 and/or increased SOX9 expression are associated with poor survival in GC patients (36)(37)(38). miR-138 is reduced in GC patients with lymph node metastasis in comparison to patients without lymph node metastasis (38). miR-105 and -138 suppress proliferation, colony formation, migration, invasion and EMT of BGC-823 and SGC-7901 GC cells (36)(37)(38). Ectopically expressed miR-105 inhibits TG of SGC-7901 cells subcutaneously implanted into nude mice and their metastasis to the lungs (36,37). miR-138 transfected SGC-7901 cells show decreased TG in nude mice after subcutaneous implantation (38). SOX proteins are transcription factors of a gene family of 20 members in humans, which bind to the minor groove of DNA by their high motility group (HMG) domain. They have a regulatory function in development, cell-fate decision, sex determination and differentiation (39,40). Their deregulation leads to cancer initiation, progression and metastasis, angiogenesis and lymphatic angiogenesis and changes in the tumor microenvironment (39,40). SOX4 expression has been associated with an unfavorable prognosis in GC patients (41). Expression of SOX9 correlates with poor clinical outcome in the MSS/TP53+ subgroup of GC patients (42). miRs-136 and -489 target homeobox genes HOXC10 and PROX-1. miR-136 ( Figure 1A) is down-regulated in GC peritoneal metastasis tissue and is associated with worse prognosis (43). GC-9811P GC cells expressing exogenous miR-136 display unaltered proliferation or apoptotic rates but show reduced mobility and invasion. In vivo, a decrease in peritoneal metastasis formation was noted in a murine tumor model (43). Homebox protein HOXC10 was identified as a direct target of miR-136 (43). Knockdown of HOXC10 reduces migration and invasion of GC-9811P cells in vitro and abolishes metastasis in peritoneal colonization assays (43).
HOXC10 is a member of the human HOX gene family comprising 39 genes, which contain a homeobox gene sequence in the range of 180 bp with DNA binding capacity of the corresponding protein moiety (44). Homeobox genes and their proteins regulate patterns of anatomical development in animals and humans. HOXC10 promotes cell proliferation and migration of GC cells through nuclear factor kappaB (NFĸB), mitogen-activated protein kinase (MAPK) pathways and up-regulation of pro-inflammatory cytokines such as tumor necrosis factor α and β (TNFα,β), interleukin 6 (IL6) and epidermal growth factor (EGF) (45)(46)(47)(48). miR-489 ( Figure 1A) expression is reduced in GC tissues in comparison to corresponding normal tissues (49). miR-489 inhibits proliferation of AGS GC cells and tumorigenicity of AGS cells in nude mice after subcutaneous implantation (49). Prospero homeobox protein 1 (PROX-1) was identified as a direct target of miR-489 (49). Expression of PROX-1 is associated with progression and prognosis in GC (49). PROX-1 is a mediator of tumor cell proliferation, angiogenesis and lymphangiogenesis and has been identified as a marker for lymphocytic endothelium (50,51). miR-27b targets NR2F2. miR-27b ( Figure 1B) inhibits proliferation, migration, and invasion of MGC-803 GC cells (52). TG and metastasis to the liver of MGC-803 cells expressing exogenous miR-27b are inhibited after implantation into the stomach or the spleen of nude mice (52). Nuclear receptor subfamily 2, group F, member 2 (NR2F2), also referred to as chicken ovalbumin upstream promoter transcription factor-II (COUP-II) has been identified as a direct target of miR-27b (52). NR2F2 is an orphan nuclear receptor with a DNA and ligand binding domain and context-dependent onogenic and tumor suppressive functions (53,54). NR2F2 has crucial functions in cell differentiation and cell metabolism (53,54). In GC, NR2F2 has been found to activate embryonic pathways through up-regulation of expression of Cadherins 6 to 11 (55). However, in GC also a tumor-suppressive function has been reported for NR2F2 through inhibition of proliferation, invasion, and metastasis (56). miR-31 targets E2F2. miR-31 ( Figure 1B) is down-regulated in GC and GC-derived cell lines in comparison to adjacent normal tissues and GES-1 cells (57). miR-31 over-expression in SGC-7901 and MGC-803 cells leads to cell-cycle arrest in the G1 phase, increase in apoptotic index and inhibition of migration and invasion (57). miR-31-mimic transduced SGC-7901 cells exhibit decreased TG after subcutaneous implantation into nude mice and decreased liver metastasis after tail vein injection (57). Transcription factor E2F2 has been identified as a direct target of miR-31 (57). Knockdown of E2F2 reduces cell migration and invasion of SGC-7901 and MGC-803 cells (57). E2F2 binds to retinoblastoma protein (Rb) and after its dissociation it can act as a transcriptional activator or repressor (58). However, the data for the role of E2F2 in GC cancer are controversial since increased expression of E2F2 also was associated with a favorable prognosis (59).
miRs-33b and -494 target c-MYC. miR-33b ( Figure 1B) is down-regulated due to DNA methylation in GC samples in comparison to adjacent normal tissues. Its decrease correlates with disease progression and GC patients with metastases have lower levels of miR-33b than those without metastases (60). Over-expression of miR-33b inhibits proliferation, migration, and invasion of the GC cell lines MGC-803 and HGC-27 (60). Tumorigenicity of subcutaneously implanted HGC-27 cells in nude mice is suppressed following their transduction with miR-33b (60). The transcription factor c-MYC was identified as a target of miR-33b (60). miR-494 ( Figure 1B) is down-regulated in GC samples and cell lines and is negatively correlated with survival (61). GC-derived AGS cells that are transfected with miR-494 exhibit delayed G1/S entry (61). c-MYC was identified as a direct target of miR-494 (61). Tumor burden of subcutaneously implanted AGS cells into nude mice was attenuated by intraperitoneal injection of a miR-494 mimic (61). c-MYC is a transcription factor with basic helix-loop-helix and leucine zipper structural motifs, which activates many proliferative genes (62). In addition, c-MYC has an impact on cell growth, ribosomal RNA processing and other crucial physiological processes (63)(64)(65)(66). c-MYC has been found to be deregulated in GC (67).
miR-34 targets YY1. Ying and Yang1 (YY1), a member of the GLI-Kruppel class of zinc finger transcription factors, was identifed as the target of miR-34 ( Figure 1B) family, which consists of miRs-34a, -b and -c (68). YY1 contributes to tumor sphere formation in GC-derived SC-M1 cells and up-regulates pluripotency genes cluster of differentiation 44 (CD44), octamer-binding transcription factor 4 (OCT4), SOX2 and homeobox protein NANOG. It is also involved in maintainance of the cancer stem cell (CSC) phenotype (68). Knock-down of YY1 in SC-M1 cells leads to expression of epithelial markers E-Cadherin and plakoglobin (a catenin protein family member homologous to β-catenin), and to decrease of mesenchymal markers N-Cadherin and vimentin (68). Transfection of SC-M1 cells with miR-34b and 34c inhibits tumor growth after subcutaneous implantation and lung metastases after tail vein injection in nude mice (68). miR-34a inhibitors expressed in SC-M1 cells lead to increased lung metastasis in the experimental metastasis assay system in nude mice (68). YY1 activates and represses many genes involved in cellular differentiation, DNA repair, cell division, survival, apoptosis, and autophagy with a context-dependent function as a tumor driver or suppressor (69,70). YY1 expression contributes to gastric carcinogenesis and its nuclear expression correlates with survival in patients with early-stage gastric adenocarcinoma (71). YY1 also is involved in the regulation of cancer immune cell resistance by up-regulating PD-L1 (72).
miR-145 targets ETS-1. Transcription factor E26 transformation specific-1 (ETS-1) has been identified as a target of miR-145 ( Figure 1B) (73). ETS-1 is the cellular homolog of the v-ETS oncogene, and is highly expressed in GC tissues and cell lines. Its expression inversely correlates with miR-145 levels (73). In SGC-7901 and MKN-45 GC cell lines, ectopic expression of miR-145 has no influence on proliferation, but inhibits migration, invasion, and tube formation (73). Transfection of ETS-1 restored the decrease in migration, invasion, and angiogenesis in SGC-7901 and MKN-45 cells mediated by miR-145 (73), whereas knockdown of miR-145 promoted migration, invasion, and angiogenesis (73). Knockdown of ETS-1 phenocopies the effects of miR-145 over-expression in GC cells in vitro (73). Transfection of SGC-7901 cells with miR-145 has no influence on TG in vivo, but decreases lung metastasis after tail vein injection (73). The ETS family of transcription factors is composed of 28 members in humans and regulates genes involved in extracellular matrix (ECM) remodelling, migration, and invasion (74). ETS-1 induces genes such as matrix metalloproteinases 1 and 9 (MMP-1, -9) and urokinase plasminogen activator (uPA) (75) and is involved in EMT and angiogenesis (76). GC patients with tumors that express ETS-1 have poorer prognosis than those without expression of ETS-1 (77).
miR-320 targets FOXM1. Human GC exhibits low expression of miR-320 ( Figure 1B) (83). Forkhead box protein M1 (FOXM1) was identified as a target of miR-320 (83). Over-expression of miR-320 in the GC cell lines AGS, BGC-823 and HGC-27 leads to inhibition of FOXM1 expression and up-regulation of cyclin-dependent kinase inhibitor 1B (p27KIP1) (83). Conversely, suppression of miR-320 in BGC-823 GC cells leads to increased TG in nude mice through altered FOXM1-p27KIP1 signaling (83). FOXM1 is frequently over-expressed in various cancers, functions as a proto-oncogene and plays a role in cell-cycle progression (84,85). In GC, FOXM1 facilitates cell migration by inducing cathepsin (86) and cooperates with urokinase plasminogen activator (uPA) to promote disease progression (87). Expression of FOXM1 is associated with poor survival in patients with GC (88).
miR-520b, -1297 target CREB1. miR-520b ( Figure 1B) inhibits migration and invasion of GC cells BGC-823 and MKN45 (92). Expression of miR-520b is dependent on zinc transcription factor GATA-binding factor 6 (GATA6) (93), which in turn is down-regulated in metastatic GC specimens and GC-derived cell lines (92). Down-regulation of GATA6 correlates with poor prognosis in GC (92). cAMP response element binding protein 1 (CREB1) has been identified as a direct target of miR-520b. Ectopic expression of miR-520b inhibits migration and invasion of GC cell lines BGC-823 and MKN45 in vitro and in vivo in nude mice (93). miR-1297 ( Figure 1B) also targets CREB1 and shows significantly lower expression in GC tissue samples compared to adjacent normal tissue (94). Lower expression of miR-1297 correlates with larger tumor size, lymph node metastases, advanced tumor/lymph node/metastasis (TNM) type and poor survival of GC patients (94). Conversely, upregulation of miR-1297 leads to inhibition of proliferation and colony formation in vitro in GC cell lines and to suppression of TG in nude mice after subcutaneous implantation (94). Increased CREB1 expression rescued the in vitro effects of miR-1297 in GC cell lines. Independently, it has been shown that CREB1 promotes GC cell proliferation, colony formation, invasion and migration, and TG in vivo (95). CREB1 knockdown in GC cells leads to inhibition of cyclin D1 (CCND1), BCL2 and MMP9 (95). High expression of CREB1 is associated with increased tumor stage, metastasis, and poor outcome in GC patients (96).
miR-19b targets BCL3. miR-19b ( Figure 1C) is downregulated in GC tissues and cell lines and is associated with clinico-pathological factors and poor prognosis in patients (106). miR-19b inhibits cell proliferation and cell-cycle progression of GC cell lines MGC-803 and SGC-7901 (106). B-cell lymphoma 3 (BCL3) was identified as the target of miR-19b (106). Over-expression of miR-19b inhibits TG of MGC-803 GC cells subcutaneously implanted in nude mice (106). Also, silencing of BCL3 abolishes the effects of miR-19b inhibition on cell proliferation and cell-cycle progression (106). BCL3 acts as an oncogene in haematologic malignancies such as B-cell chronic lymphocytic leukemias. But it is also deregulated in solid tumors, modulating proliferation and cell death (107). BCL3 regulates transcription by forming complexes with p50 or p52 homodimers of the NFĸB signaling pathway, and also by association with other transcription factors (108). However, the role of BCL3 in GC remains to be explored in more detail.

Cell-cycle and Apoptosis
miR-29c targets CCND2. miR-29 ( Figure 2) is a family consisting of three members (29a-, b-and c) with miR-29c being most down-regulated in GC (112). Low expression levels of miR-29c are associated with aggressive and progressive phenotypes of GC. In HGC-27 and MGC-803 GC cells, over-expression of miR-29 a, -b and -c suppresses cell growth, induces arrest of G1/S transition, and promotes apoptosis in vitro (112). Cyclin D2 (CCND2) and matrix metalloproteinase-2 (MMP2) were identified as direct targets of miR-29c (112). Intratumoral injection of synthetic miRs mimicking miRs-29a, -b andc inhibit TG of subcutaneously implanted HGC-27 cell xenografts in mice (112). CCND2 forms complexes with cyclin-dependent kinases 4 and 6 (CDK4,6) and is required for G1/S transition (113). MMP2 is up-regulated in different types of tumors including GC (114). Expression of CCND2 is associated with shorter survival in patients with GC (113). Based on TCGA-derived data, miR-29c is down-regulated in GC specimens in comparison to corresponding normal tissues, in contrast to miRs-29a and -b (Figure 3).
Mir-125b, -133b/a target MCL1 and BCL-xL. miR-125b ( Figure 2) is down-regulated in GC and decreased levels of miR-125b correlate with advanced clinical stage, lymph node metastases, and poor clinical outcome in GC (115). Induced myeloid leukemia cell differentiation protein (MCL-1) was identified as a target of miR-125b (115). miR-125b inhibits proliferation of MGC-803 cells and TG in nude mice after subcutaneous implantation (115). miR-133b/a ( Figure 2) is down-regulated by histone modification of its promoter in GC cells (116). miR-133b/a suppresses GC cell proliferation and promotes apoptosis, delays tumor formation, and reduces tumor size of GC xenografts in nude mice (116). MCL1 and BCL-xL (B-cell lymphoma-extra large) were identified as targets of miR-133b/a (116). MCL1 is a member of the antiapoptotic BCL2 family, regulates cell growth, promotes EMT of GC cells, and its expression predicts poor outome in GC patients (117,118). BCL-xL is a multifunctional antiapoptotic protein (119). BCL2 and BCL-xL mediate resistance to receptor-kinase targeted therapy in lung and gastric cancer (120). Based on TCGA data, miRs -125b, -133a and miR-133b are down-regulated in GC tissues in comparison to corresponding normal tissues (Figure 3). miR-137 targets CDK6. miR-137 ( Figure 2) expression correlates with better prognosis in GC patients (121). Ectopic expression of miR-137 in GC cells inhibits proliferation in vitro and in vivo by targeting cyclin-dependent kinase 6 (CDK6) (121). CDK6 inhibitor PD0332991 mediates an antiproliferative effect in GC cell lines, arrests them in the G1 phase and affects numerous oncogenic pathways such as p53, phosphinosite-3 kinase (PI3K)/AKT, RAS/ERK, c-jun-N-terminal kinase (JNK)/MAPK, WNT/β catenin, and SMAD signaling (122). CDK4,6 inhibitor palbociclib together with aromatase inhibitor letrozole has been approved for treatment of estrogen receptor positive breast cancer (123,124). According to data derived from TCGA, miR-137 is down-regulated in GC tissues in comparison to corresponding normal tisses ( Figure 3). miR-181d targets CYLD. miR-181d ( Figure 2) is downregulated in GC and is a predictor for bad prognosis (125). miR-181d targets the cylindromatosus gene (CYLD) and its expression inhibits proliferation and invasion of HGC-27 GC cells (125). In addition, it inhibits EMT and the PI3K/AKT pathway (125). Over-expression of miR-181 inhibits xenograft growth of HGC-27 cells after subcutaneous implantation into nude mice (125). CYLD is a lysine 63 deubiquitinylase, which modulates multiple signaling pathways such as NFĸB and MAPK by its catalytic activity on key intermediates of these pathways (126). In GC, down-regulation of CYLD by methylation of its promoter has been noticed (127).  tissues (128). Over-expression of miR-376-3p in SGC-7901 cells enhances apoptosis and inhibits TG of xenografts in nude mice (128). Over-expression of miR-376-3p decreases the levels of BCL2 (B-cell lymphoma 2) and increases the levels of pro-apoptotic protein BCL2 antagonist of cell death (BAD). However, it was not resolved whether BCL2 is a direct target of miR-376-3p (128). BCL2 has been shown to promote cell proliferation of GC cells (129). High Ki67/BCL2 index can predict disease free and overall survival in intestinal type GC (130).
miR-518 targets MDM2. miR-518 ( Figure 2) is downregulated in GC tissues and cell lines (131). In GC cell lines MKN45 and HGC-27 miR-mimics trigger apoptosis in vitro and in vivo (131). Mouse double minute 2 homolog (MDM2) has been identified as a target of miR-518 (131). MDM2 is an E3 ubiquitin-protein ligase which ubiquitinylates p53, triggering its degradation (131). miR-518 up-regulates p53, pro-apoptotic BCL2-associated protein X (BAX), activates caspase 3 and down-regulates BCL2 in GC cells (131). MDM2 up-regulation has been frequently observed in intestinal GC and correlates with nodal and distal metastases (132). An association between MDM2 amplification and tumorigenesis has been noted (133). Targeting p53/MDM2 interaction is an important effort for cancer drug discovery approaches (134). miR-1254 targets SMURF1. miR-1254 ( Figure 2) is downregulated in human GC tissues and in GC-derived cell lines (135). Expression of miR-1254 inhibits proliferation, migration, and invasion of MGC-803 and SGC-7901 GC cell lines (135). SMAD ubiquitintion regulatory factor 1 (SMURF1) was identified as a direct target of miR-1254 (135). SMURF1 is up-regulated in human GC tissues and cell lines (135). SMURF1 expression could partially reverse the effects of miR-1254 on GC cells (135). miR-1254 inhibits EMT and negatively regulates PI3K/AKT signaling by down-regulatiing SMURF1 (135). In vivo, miR-1254 mimics decrease TG of MGC-803 and SGC-7901 xenografts in nude mice (135). SMURF1 is an E3 ubiquitin ligase which regulates cell-cycle, apoptosis, metastasis, senescence, and genomic stability. It can act, in a context-dependent manner, as a tumor promoter or suppressor (136). According to TCGA-derived expression data, miR-1254 levels are very low in GC and also in corresponding normal tisses. Therefore, the degree of deregulation in GC cannot be determined based on these data (Figure 3).

Small GTPases and Proteins Modulating Their Activity
Let-7a targets RAB40C. Expression of the microRNA (lethal7-precursor) Let-7a (Figure 4)  and cell lines (137). Ectopic expression of let-7a mediates inhibition of proliferation, anchorage-dependent cell growth, and G1 arrest (137). BGC-823 cells transfected with let-7a mimics exhibit reduced TG after subcutaneous implantation into nude mice (137). RAB40C was identified as a direct target of miR Let-7a (137). RAB40C is a member of the RAS family of GTPases and is expressed in brain neurons and glia, is involved in vesicle transport in oligodendrocytes and plays an important role in tumorigenesis (138)(139)(140). Small GTPases, when activated transmit signaling to a variety of effector proteins after binding of GTP and hydrolyzing it to GDP (140). They are regulated by GAPs, GEFs (GTPase-activating proteins and Guanine nucleotide exchange factors) and RAS homology member family member (RHO) guanine dissocation inhibitors (GDIs). GAPs bind to activated G proteins and stimulate their GTPase activity, GEFs promote the exchange of GDP by GTP and GDIs sequester RHO-GTPases in the inactive state by inhibition of dissociation of inactive guanine nucleotides (140). The role of RAB-40C in GC remains to be investigated in further detail. Figure 4) is negatively correlated with GC metastasis (141). miR-31 mimics inhibit migration of BGC-823 cells and inhibition of miR-31 promotes migration of SGC-7901 cells (141). Reduced lung metastasis after tail vein injection in nude mice was found when SGC-7901 GC cells were transfected with miR-31 mimics (141). RAS homology family member A (RHOA) was identified as a direct target of miR-31 (141). RHOA is a small GTPase which can activate RHOassociated, coiled-coil containing protein kinase which phosphorylates myosin light chain (MLC) and thus exacerbates the migration and invasion capability of GC cells (142). RHOA is also associated with cytoskeleton regulation such as actin stress fiber formation and actomyosin contractibility, cell migration, and metastasis in GC (143). RHOA has potential clinical significance as a prognostic marker in GC (144). Together with cell division control protein 42 homolog (CDC42) and RAS-related botulinum toxin substrate 1 (RAC1), RHOA is one of the best studied GTPases. Tractability of RHOA as a therapeutic target is a challenging issue, as holds true for all GTPases.
miR-148-3p targets DOCK6. miR-148-3p ( Figure 4) targets and reduces the levels of dedicator of cytokines 6, a guanine nucleotide exchange factor (DOCK6) (152). DOCK6 expression is increased in GC tissues and indicates poor prognosis (152). DOCK6 increases proliferation, migration, and invasion of SGC-7901 SM GC cells by activation of RAC1 and CDC42 (152). In nude mice, over-expression of miR-148-3p in SGC-7901 M cells decreased lung and liver metastasis after tail vein injection, while DOCK6 overexpression reversed inhibition of metastasis (152). DOCK6 expression is negatively correlated with with miR-148b expression in GC tissues (152). In mammals, 11 DOCK proteins have been identified as regulators of the actin cytoskeleton, cell adhesion and migration (153). The role of DOCK6 in GC needs further investigation.
miR-329 targets TIAM1. miR-329 ( Figure 4) is downregulated in GC and inhibits migration and invasion of GC cells MGC-803 and HGC-27 (154). In vivo, injection of HGC-27 cells together with miR-329 mimics inhibits TG in nude mice (154). T-cell lymphoma invasion and metastasis inducing protein 1 (TIAM1) was identified as a direct target of miR-329 (154). TIAM1 acts as a GEF for activation of RAC1, but also for CDC42, and to a lesser extent for RHOA (154). Protumoral activities of TIAM1 involve modulation of the cytoskeleton, cell polarity, membrane trafficking, cell migration, adhesion, cell growth, survival, carcinogenesis, and metastasis (155). But anti-cancer activity of TIAM1 has also been reported (155). Up-regulation of TIAM1 has been reported in GC and its involvement in cell migration and invasion (156,157). The propensity of GEFs and GAPs as targets for anti-cancer therapy has been discussed (158).

Therapeutic Aspects
This review focusses on miRs that are down-regulated in GC and affect the expression of transcription factors, cell-cycle and apoptosis-related proteins, and of small GTPases and their regulatory proteins with demonstrated efficacy in preclinical in vivo models. Down-regulation of miRs can be due to aberrant miR processing, methylation of CpG islands in the promoter region and adjacent regions, and other epigenetic mechanisms (164)(165)(166)(167)(168). Replacement therapy can be achieved with miR-mimics, synthetic double-stranded RNAs that mimic the endogenous RNAs. Chemical modifications such as introduction of 2'-O-methyl, phosphorothioate and locked nucleic acids (LNA) can improve stability and targeting to the disease site (169)(170)(171)(172). miR mimics or miRs can be injected intratumorally or systemically in designed formulations or as plasmid-or viralderived vectors (169)(170)(171)(172). Adeno-associated vectors can transduce target cells efficiently, do not integrate, are eliminated efficiently with minimal toxicity, and have emerged as vehicles for miR-replacement therapies (173,174). A POC experiment revealed that restoration of miR let-7 was able to inhibit lung cancer in an orthotopic mouse model expressing KRAS G12D, an activating mutation of KRAS (175). Further important POC experiments reconstituting expression of miR-29b and 708-5p have been described in mouse lung cancer models (176,177).
This review compiles published POC experiments for 37 selected in vivo miR replacement studies in GC-related in vivo models. The design of optimized delivery systems is of paramount importance in this context (169). Delivery systems based on modification of miRs with polyethylene glycol (PEG), neutral lipid emulsions, conjugation with N-acetyl-Dgalactosamine, chitosan-nucleic acid conjugates, glucosebased-polymers such as cyclodextrin, dendrimers, synthetic polyethylenimine based vehicles, poly (lactide or glycolide) particles or TargomiRs, bacterium-based vesicles with surface-based antibodies have been developed (169,178).
Approaches towards applications of miR replacement therapy in cancer patients have focused on miR-34 (179) or miR-15/16 (180). miR-34 has been validated in numerous preclinical in vitro and in vivo studies as a candidate for replacement therapy (181). This holds true for colorectal, prostate, breast, liver, and lung cancer as well as osteosarcoma and haematological neoplasms (181). miR-34 inhibits crucial oncogenic and metastatic pathways such as EMT mediated by transcription factors SNAIL, SLUG and ZEB1,2, NOTCH-, WNT-, IL6R-and TGFβ/SMAD signaling (181). MRX 34 (Mirna Therapeutics), an amphoteric, lipid nanoparticle filled with miR-34 mimics, has been evaluated in a Phase I clinical study (NCT01829971) in 155 participants in several solid tumors and haematological malignancies. That study, however, had to be terminated due to immunerelated side effects in five patients (181,182). As outlined, in GC, miR-34 targets transcription factor YY1 and may be a possible agent for replacement therapy.
The other ongoing clinical study for miR-replacement therapy is based on epidermal growth factor receptor (EGFR)coated bacterial minicells (EnGene IC) expressing mimics corresponding to the miR-15/107 consensus sequence (miR-16 mimics). The delivery vehicels are nanocells also referred to as TargomiRs. The Phase I study in patients with metastatic pleural mesothelioma and advanced non-small cell lung carcinoma (NSCLC) is referred to as the mesomiR-1 study (180,183). In preclincal experiments, expression of miR-16 in mesothelioma cell lines had a tumor-suppressive effect through induction of apoptosis by down-regulation of BCL2 and CCND1 (184). In a phase I study, 27 patients were enrolled with an objective response and stabilization of disease in 15 patients (183).

Conclusion
Our comprehensive review of GC-associated miRs revealed 25 down-regulated miRs that target transcription factors, 7 targeting cell-cycle/apoptosis-related genes, and 5 directed against small GTPase-related and -associated molecules such as GEFs and GAPs. These might present as candidates for potential therapeutic intervention in GC patients, in the form of replacement therapy or inhibition of the corresponding upregulated targets. The applicability of such replacement therapies is still limited by technical hurdles as described above. For target prioritization issues such as prevalence, prognostic relevance, druggability and consistent or contextdependent functional properties are important parameters for ranking. In case of transcription factors, tractability is critical because mode of intervention is difficult because it would involve inhibition of DNA binding or of co-factor proteinprotein interactions (193). Also, modulation of the levels of transcription factors by interfering with ubiquitinylation and subsequent proteasome degradation might be considered. Proteolysis targeting chimeras (PROTACS) directed against intrinsically disordered regions of transcription factors with Cys-reactive inhibitors and another moiety engaging an E3 ubiquitin ligase are emerging agents for interference (194). The druggability of small GTPase related molecules, GEFs and GAPs has to be categorized as critical, despite of encouraging recent developments (195,196). The functions of TIAM1 (miR-329), ARHGAP10 (miR-337) and SMURF1 (miR-1254) are context-dependent as outlined. For NR2F2 (miR-27b), E2F2, RHOA (miR-31) and CYLD (miR-181d) contradictory functions have been published as described in the previous text. Prognostic data are available for miR-29 (CCND2), miR-105, -138c (SOX9, SOX4), miR-124, -130 (SRGAP1), miR-133 b/a miR-136 (HOXC10), miR-141, -200c (ZEB1, ZEB2), miR-145 (ETS1), miR-186 (HIF-1α), miR-320 (FOXM1), miR-489 (PROX1), miR-518 (MDM2) and miR-1297 (CREB1). Taken together miR-133b/a (MCL1) and miR-518 (MDM2) might present as promising targets. However, more GC specific target validation experiments are needed to derive firm conclusions. Broad drug discovery programmes for MCL1 and p53/MDM2 and clinical studies on other indications are ongoing (197)(198)(199).

Conflicts of Interest
FB, SA, and UB are and UHW was an employee of Roche.

Authors' Contributions
FB, SA, UB and UHW jointly designed and prepared the manuscript.