LncRNA H19-Derived miR-675-5p Accelerates the Invasion of Extravillous Trophoblast Cells by Inhibiting GATA2 and Subsequently Activating Matrix Metalloproteinases

The invasion of extravillous trophoblast (EVT) cells into the maternal decidua, which plays a crucial role in the establishment of a successful pregnancy, is highly orchestrated by a complex array of regulatory mechanisms. Non-coding RNAs (ncRNAs) that fine-tune gene expression at epigenetic, transcriptional, and post-transcriptional levels are involved in the regulatory mechanisms of EVT cell invasion. However, little is known about the characteristic features of EVT-associated ncRNAs. To elucidate the gene expression profiles of both coding and non-coding transcripts (i.e., mRNAs, long non-coding RNAs (lncRNAs), and microRNAs (miRNAs)) expressed in EVT cells, we performed RNA sequencing analysis of EVT cells isolated from first-trimester placentae. RNA sequencing analysis demonstrated that the lncRNA H19 and its derived miRNA miR-675-5p were enriched in EVT cells. Although miR-675-5p acts as a placental/trophoblast growth suppressor, there is little information on the involvement of miR-675-5p in trophoblast cell invasion. Next, we evaluated a possible role of miR-675-5p in EVT cell invasion using the EVT cell lines HTR-8/SVneo and HChEpC1b; overexpression of miR-675-5p significantly promoted the invasion of both EVT cell lines. The transcription factor gene GATA2 was shown to be a target of miR-675-5p; moreover, small interfering RNA-mediated GATA2 knockdown significantly promoted cell invasion. Furthermore, we identified MMP13 and MMP14 as downstream effectors of miR-675-5p/GATA2-dependent EVT cell invasion. These findings suggest that miR-675-5p-mediated GATA2 inhibition accelerates EVT cell invasion by upregulating matrix metalloproteinases.

Non-coding RNAs (ncRNAs) that are not translated into protein have recently attracted much attention for their ability to fine-tune gene expression at epigenetic, tran-

Identification and Characteristics of Genes Expressed in EVT Cells from RNA Sequencing
From the cDNA libraries of EVT and chorionic villous trophoblast (CVT) cells, 17,194 protein-coding genes (mRNAs), 4370 lncRNAs, and 1560 mature miRNAs were defined as expressed. A statistical summary of clean reads from RNA sequencing of sample libraries is provided in Table S5A. First, to assess whether isolated cells from the first-trimester placenta have the characteristics of EVT cells, we paid attention to 12 mRNAs previously reported to be trophoblast cell markers (Table S5C). Cytotrophoblast cells express EGFR, TEAD4, and TP63; EVT cells have lost these markers and instead express HLA-G, MCAM, MYC, and CDH5 [20][21][22]. EVT cells are also characterized by the loss of genes encoding integrin subunits (ITGA6 and ITGB4), and the gain of ITGA1, ITGA5, and ITGB1 [23][24][25]. Six EVT cell marker genes, except for CDH5, were significantly upregulated in the isolated cells compared with CVT cells; five cytotrophoblast cell marker genes were significantly downregulated in the isolated cells (Table S5C). Our RNA-sequencing data of the expression of these trophoblast cell markers, except for CDH5, are consistent with previous reports on these expression changes at the mRNA level, suggesting that the isolated cells represent EVT cells.
Differentially expressed genes (DEGs) were identified based on a q < 0.05 and an absolute log2 fold change (log FC) ≥ 1. Of 23,124 genes, 5092 mRNAs, 422 lncRNAs, and 401 mature miRNAs were identified as differentially expressed. Details of the DEGs expressed in EVT cells can be found in Tables S5-S8. The chromosomal distribution of the DEGs is also shown in Figure 1B. The most highly expressed mRNAs, lncRNAs, and miRNAs identified in EVT and CVT cells are shown in Figure 1A and Tables S6-S8. The most remarkable point of this RNA-sequencing study is that all DEGs (i.e., ten mRNAs, seven lncRNAs, and 83 miRNAs) located in chromosome 14q32.2, which contain the eutherian-specific imprinted DLK1-DIO3 region [26], were significantly downregulated in EVT cells ( Figure 1B and Table S9). In terms of imprinted genes (http://www.geneimprint. com/site/genes-by-species), 68 genes were found to be DEGs; the percentages of the maternally and paternally imprinted genes that were significantly downregulated in EVT cells were 63% (19 out of 30 maternal genes) and 89% (23 out of 29 maternal genes), respectively (Table S10). cells were 63% (19 out of 30 maternal genes) and 89% (23 out of 29 maternal genes), respectively (Table S10).  Using Ingenuity Pathway Analysis (IPA), we performed Core Analysis to interpret our dataset of DEGs filtered by the IPA microRNA Target Filter tool (i.e., the dataset containing 351 miRNAs targeting 4783 mRNAs and lncRNAs). The top canonical pathways, as shown in Figure 1C were related to the reorganization of the actin cytoskeleton and modulation of cell adhesion/migration (e.g., signaling by Rho family GTPases and Tec kinase signaling). Most of the top enriched terms corresponding to Diseases/Functions based on significant activation z-score (absolute z-score > 2.0) were associated with invasion-or cancer-related functions (Table S11). The bioinformatics analysis revealed that DEG-containing molecular pathways/gene sets reflect the invasive differentiation process of trophoblast cells.
Another remarkable point of this RNA-sequencing study was that lncRNA H19 was encoded by a DEG (q = 1.29 × 10 −9 , log FC = 2.98) and the most highly expressed lncRNA (70.47%) in EVT cells ( Figure 1A and Table S7), whereas most of the differentially expressed lncRNAs (81%) were downregulated in EVT cells ( Figure 1B). The high expression level of lncRNA H19 in EVT cells revealed by RNA-sequencing was in good agreement with previous findings from single-cell RNA-sequencing and microarray experiments [27,28]. The lncRNA H19-derived miRNA miR-675-5p was relatively highly expressed in EVT cells, ranking 77th among the 401 differentially expressed miRNAs between EVT and CVT cells ( Figure 1D). There was no significant difference in the expression of the other lncRNA H19-derived mature miRNA miR-675-3p between EVT and CVT cells ( Figure 1D). We next focused on the lncRNA H19-derived miR-675-5p to investigate its role in EVT cell invasion.

miR-675-5p Accelerates EVT Invasion
To assess the possible contribution of miR-675-5p to EVT cell invasion, we performed a Matrigel-coated Transwell invasion assay using the EVT cell lines HTR-8/SVneo and HChEpC1b. Following the transfection of miR-675-5p (675-5p) mimic into the EVT cell lines, miR-675-5p significantly promoted the invasion of EVT cells compared with the control (Figure 2A). In addition, we assessed the effect of miR-675-5p transfection on cell growth. However, no significant difference in cell growth was identified between 675-5pand negative control (NC) mimic-transfected cells ( Figure S1A).

miR-675-5p-Mediated GATA2 Inhibition Accelerates EVT Cell Invasion by Upregulating MMP13 and MMP14
To determine whether miR-675-5p-mediated GATA2 inhibition accelerates EVT cell invasion, we analyzed invasion by cells in which GATA2 was downregulated via small interfering RNA (siRNA)-mediated knockdown (designated as siGATA2). First, we evaluated the efficiency of siGATA2 using qPCR and Western blotting. In EVT cell lines, siGATA2 inhibited the expression of GATA2 mRNA and protein ( Figure 2D). Next, we assayed the invasion of EVT cells treated with siGATA2. Following GATA2 knockdown, siGATA2 significantly promoted the invasion of these cells compared with the control ( Figure 2E), suggesting that GATA2 negatively regulates EVT cell invasion.
To evaluate whether MMP13 and MMP14 accelerate EVT cell invasion, we analyzed invasion by cells in which MMP14 and MMP13 were downregulated by siRNA mediated knockdown (designated siMMP13 and siMMP14, respectively). First, we assessed the efficiency of siMMP13 and siMMP14 using qPCR and Western blotting in EVT cells and found that siMMP13 and siMMP14 significantly inhibited both the mRNA and protein levels of MMP13 and MMP14, respectively ( Figure 4A). Next, we assayed the invasion of EVT cells treated with siMMP13 and siMMP14. Following MMP13 knockdown, siMMP13 significantly attenuated the invasive ability of HTR8/SVneo and HChEpC1b cells ( Figure 4B). Additionally, siMMP14 significantly inhibited the invasive ability of these EVT cells compared with the control values ( Figure 4B). Taken together, these findings suggest that both MMP13 and MMP14 promote EVT cell invasion as downstream effectors of the miR-675-5p/GATA2 axis. To evaluate whether MMP13 and MMP14 accelerate EVT cell invasion, we analyzed invasion by cells in which MMP14 and MMP13 were downregulated by siRNA mediated knockdown (designated siMMP13 and siMMP14, respectively). First, we assessed the efficiency of siMMP13 and siMMP14 using qPCR and Western blotting in EVT cells and EVT cells treated with siMMP13 and siMMP14. Following MMP13 knockdown, siMMP13 significantly attenuated the invasive ability of HTR8/SVneo and HChEpC1b cells ( Figure 4B). Additionally, siMMP14 significantly inhibited the invasive ability of these EVT cells compared with the control values ( Figure 4B). Taken together, these findings suggest that both MMP13 and MMP14 promote EVT cell invasion as downstream effectors of the miR-675-5p/GATA2 axis.

Discussion
We performed RNA-sequencing analysis of EVT cells isolated from first-trimester human placentae to profile the gene expression signatures of mRNAs as well as ncRNAs (i.e., miRNAs and lncRNAs). We identified approximately 6000 DEGs between EVT and CVT cells. Bioinformatics analysis revealed that DEG-containing signaling pathways/gene sets are associated with the migratory/invasive phenotype of trophoblast cells. This EVT phenotype revealed by RNA-sequencing is consistent with previous findings from microarray experiments [27,43]. Single-cell RNA-sequencing technology has recently revealed unique gene signatures of EVT cells in healthy pregnancy and preeclampsia [28,44,45]. Although the regulatory mechanisms of ncRNAs in the EVT-associated lncRNA-miRNA-mRNA network will require further elucidation, the present study will be a valuable resource for future ncRNA research on EVT cells.
We originally hypothesized that miR-675-5p may play a role in EVT cell invasion during early pregnancy. In the present study, we showed that miR-675-5p, an EVT-associated miRNA, was involved in the regulation of EVT cell invasion. Our data further revealed

Discussion
We performed RNA-sequencing analysis of EVT cells isolated from first-trimester human placentae to profile the gene expression signatures of mRNAs as well as ncRNAs (i.e., miRNAs and lncRNAs). We identified approximately 6000 DEGs between EVT and CVT cells. Bioinformatics analysis revealed that DEG-containing signaling pathways/gene sets are associated with the migratory/invasive phenotype of trophoblast cells. This EVT phenotype revealed by RNA-sequencing is consistent with previous findings from microarray experiments [27,43]. Single-cell RNA-sequencing technology has recently revealed unique gene signatures of EVT cells in healthy pregnancy and preeclampsia [28,44,45]. Although the regulatory mechanisms of ncRNAs in the EVT-associated lncRNA-miRNA-mRNA network will require further elucidation, the present study will be a valuable resource for future ncRNA research on EVT cells.
We originally hypothesized that miR-675-5p may play a role in EVT cell invasion during early pregnancy. In the present study, we showed that miR-675-5p, an EVT-associated miRNA, was involved in the regulation of EVT cell invasion. Our data further revealed that miR-675-5p accelerated EVT cell invasion by direct suppressing GATA2 and subsequently activating MMP13 and MMP14.
Regarding the function of lncRNA H19-derived miR-675-5p in trophoblast cells, although some researchers have reported on its role in cell proliferation, we could not find a significant association between miR-675-5p and the proliferation of EVT cell lines in this study ( Figure S1A). Gao et al. showed that miR-675-5p inhibited trophoblast cell proliferation via direct inhibition of NOMO1 using JEG3 cells [16], whereas Keniry et al.
demonstrated that miR-675-5p attenuated trophoblast cell proliferation by suppressing Igfr1 using mouse trophoblast stem cell models [17]. By contrast, there are few reports on the role of miR-675-5p in trophoblast cell invasion. This work is the first to demonstrate that the enrichment of miR-675-5p in EVT cells promoted EVT cell invasion. The function of miR-675-3p, the other strand from MIR675, in trophoblast cells remains unknown.
Within the context of trophoblast cell invasion, many studies have reported that various miRNAs and their target genes are associated with trophoblast cell invasion [12,46,47]. In this study, miR-675-5p promoted trophoblast cell invasion by direct suppressing GATA2. The transcription factor GATA2 belongs to the mammalian GATA family of zinc-finger transcription factors, which consists of six proteins (GATA1-6) involved in a variety of physiological and pathological processes [48]. GATA2 and GATA3 are trophoblast intrinsic master regulators of gene expression in all trophoblast progenitors [49]. Both GATA2 and GATA3 are constitutively expressed in all trophoblast cell types in the developing human placenta; they coordinate gene regulatory networks to generate specialized trophoblast cell types during placentation. Furthermore, dual Gata2/Gata3 conditional gene deletion study revealed the importance of functional redundancy of Gata2 and Gata3 in mouse trophoblast development [50]. Since the present study focused on the regulatory role of miR-675-5p/GATA2 axis in EVT cell invasion, the possible involvement of miR-675-5p/GATA2 axis in the GATA2/GATA3-dependent transcriptional mechanisms in EVT cells requires further investigation. Although various studies have investigated the role of miR-675-5p in cancer cell invasion [18,[51][52][53], the promotion and inhibition of cancer cell invasion by miR-675-5p remain controversial. Zheng et al. reported that miR-675-5p promoted glioma cell invasion, migration, and proliferation via negative regulation of the tumor suppressor gene RB1 [51]. Zhou et al. showed that miR-675-5p accelerated cell invasion, migration, and proliferation of esophageal squamous cell carcinoma cells by inhibiting REPS2, which is involved in growth factor signaling [18]. In that sense, RB1 and REPS2 may be potential miR-675-5p target genes in EVT cell invasion. In contrast, He et al. reported that miR-675-5p suppressed cell invasion, migration, and proliferation of non-small cell lung cancer cells by targeting the proto-oncogene gene GPR55 [52]. Additionally, miR-675-5p has been shown to inhibit cell invasion, migration, and the proliferation of papillary thyroid cancer cells by suppressing MAPK1 [53]. It is likely that the dual roles of miR-675-5p in cancer cell invasion depend on the type of cancer, as target mRNA and protein levels are differentially modulated.
MMP14 is highly expressed in the human placenta, especially in EVT cells in the firsttrimester [74][75][76]. Wang et al. showed that MMP14 accelerated EVT cell invasion via Notch1 and PI3K/AKT signaling using HTR-8/SVneo cells [77]. Majali-Martinez et al. showed that MMP14 promoted the invasion of primary trophoblast cells [78]. Our findings of EVT cell invasion following MMP14 activation are consistent with these previous studies. Given that MMP14 activates pro-MMP13 on the cell surface in the presence of active MMP2 [62], EVT cell invasion may be synergistically enhanced by a combination of MMP13 and MMP14.
In terms of RNA sequencing, we found significant downregulation of DEGs in the chromosome region 14q32.2 (i.e., the DLK1-DIO3 region). The DLK1-DIO3 cluster contains three paternally expressed protein coding genes (DLK1, RTL1, and DIO3) and multiple maternally expressed ncRNAs (e.g., MEG3 and 54 clustered miRNA precursors (designated as C14MC miRNAs)) [26]. These imprinted genes are inversely regulated by differentially methylated regions [79][80][81]. Liu et al. reported that the degree of activation of the Dlk1-Dio3 region is correlated with pluripotency levels of mouse stem cells [82]. Several C14MC miRNAs may form a feedback loop by suppressing PRC2 formation for gene methylation of this region, resulting in the activation of all genes encoded by this region in pluripotent stem cells. In this study, the inactivation of the DLK1-DIO3 region in EVT cells suggests the loss of stemness of CVT cells and subsequent differentiation into EVT cells. Greife et al. reported epigenetic silencing across the DLK1-DIO3 imprinted gene cluster due to aberrant epigenetic regulation (i.e., unique changes in DNA methylation and repressive histone modifications), which resulted in the simultaneous downregulation of oppositely expressed and imprinted genes of the region in urothelial carcinoma [83]. The epigenetic regulatory mechanism and functional role(s) of the gene downregulation across the DLK1-DIO3 region in EVT cells remain to be elucidated. Although C14MC miRNAs are expressed predominantly in human placental trophoblast cells (i.e., placenta-associated miRNAs), there are two other placenta-associated miRNA clusters in chromosome 19: the chromosome-19 miRNA cluster (designated as C19MC; 46 miRNA precursors) and the miR-371-3 cluster (three miRNA precursors in the vicinity of C19MC) [84,85]. The C19MC miRNAs are paternally imprinting genes [86]. As described above, the expression levels of most C14MC miRNAs (83 out of 99 mature miRNAs) were significantly downregulated in EVT cells ( Figure 1B and Table S9), whereas the expression levels of most of the C19MC and miR-371-3 cluster miRNAs did not differ significantly between EVT and CVT cells, although a decreasing tendency was observed for EVT cells (data not shown).
This study has some limitations. We employed two EVT cell lines (HTR8/SVneo and HChEpC1b) for in vitro functional analysis since the numbers of obtainable first-trimester placentas after legal abortions were being limited in our institutes. Therefore, future work should perform using primary first-trimester EVT cells. In addition, the specific roles of lncRNA H19, except for its role as a primary miRNA of MIR675, in EVT cells are still unknown. LncRNA H19 is located in an imprinted region of human chromosome 11, along with the IGF2 gene [87]. H19 and IGF2 are reciprocally imprinted (maternally and paternally expressed, respectively) and regulated by the H19 differentially methylated region (H19-DMR). H19 and IGF2 are important for proper placental development function [87,88].
There are a few reports on lncRNA H19 promoting EVT cell invasion by acting as an endogenous miRNA sponge [89,90]. The dysregulation of lncRNA H19 and miR-675-5p in EVT cells during first-trimester pregnancy may cause inadequate EVT cell differentiation, invasion, and subsequent placenta dysfunction [90]. Epimutation at the H19-DMR in humans results in congenital imprinting disorders (i.e., Beckwith-Wiedemann and Silver-Russell syndromes) [91,92]. Placental epimutation at H19-DMR has also been reported among pregnancy complication cases [93]. Besides, during the isolation process of EVT cells for RNA-sequencing, each step may affect the expression signatures of EVT cells. Thus, it would be helpful to compare the results from RNA-sequencing with those from in vivo gene expression analysis (e.g., in situ hybridization) for the proper interpretation of the gene expression signatures.
In summary, we performed RNA-sequencing analysis of EVT cells isolated from the first-trimester placenta and found significant downregulation of DEGs in the chromosome region 14q32.2 (i.e., the DLK1-DIO3 region). RNA-sequencing analysis also revealed that the lncRNA H19 and its derived miRNA miR-675-5p were abundantly expressed in EVT cells. We evaluated the possible role of miR-675-5p in EVT cell invasion using EVT cell lines. In vitro analysis demonstrated that the lncRNA H19-derived miR-675-5p accelerated EVT cell invasion by direct suppressing GATA2. Moreover, MMP13 and MMP14 were shown to be downstream effectors of miR-675-5p/GATA2. Thus, our study suggests that the miR-675-5p-GATA2-MMP14/MMP13 axis may play an important role in EVT cell invasion ( Figure 5). tiation, invasion, and subsequent placenta dysfunction [90]. Epimutation at the H19-DMR in humans results in congenital imprinting disorders (i.e., Beckwith-Wiedemann and Silver-Russell syndromes) [91,92]. Placental epimutation at H19-DMR has also been reported among pregnancy complication cases [93]. Besides, during the isolation process of EVT cells for RNA-sequencing, each step may affect the expression signatures of EVT cells. Thus, it would be helpful to compare the results from RNA-sequencing with those from in vivo gene expression analysis (e.g., in situ hybridization) for the proper interpretation of the gene expression signatures.
In summary, we performed RNA-sequencing analysis of EVT cells isolated from the first-trimester placenta and found significant downregulation of DEGs in the chromosome region 14q32.2 (i.e., the DLK1-DIO3 region). RNA-sequencing analysis also revealed that the lncRNA H19 and its derived miRNA miR-675-5p were abundantly expressed in EVT cells. We evaluated the possible role of miR-675-5p in EVT cell invasion using EVT cell lines. In vitro analysis demonstrated that the lncRNA H19-derived miR-675-5p accelerated EVT cell invasion by direct suppressing GATA2. Moreover, MMP13 and MMP14 were shown to be downstream effectors of miR-675-5p/GATA2. Thus, our study suggests that the miR-675-5p-GATA2-MMP14/MMP13 axis may play an important role in EVT cell invasion ( Figure 5).

Trophoblast Cell Collection
Human placentae from pregnant women who provided informed consent were obtained using protocols approved by the Jichi Medical University Ethics Committee and the Nippon Medical School Ethics Committee. First-trimester placental tissues (seven weeks of gestation, n = 3) were obtained after legal abortions. The gestational age was determined by the last menstrual period and confirmed by ultrasound measurement in the first-trimester of pregnancy.
EVT cells growing from explanted human chorionic villi were isolated as described previously [8,97]. Briefly, placental tissues were washed with RPMI-1640 and dissected to remove decidual tissues and fetal membrane. Small fragments of chorionic villi were teased apart and soaked in culture medium (RPMI-1640 containing 10% FBS and 100 U/mL Penicillin-Streptomycin (Gibco, Carlsbad, CA, USA)). Villous fragments were then minced into fine pieces with surgical blades. The fine pieces of villi were placed in collagen type 1-coated dishes (cat. no. 4020-010; Iwaki, Tokyo, Japan) for 4 h. After 48 h, outgrowth cells from adherent villous tips were dispersed using TrypLE Express (cat. no. 12604-021; Gibco, Carlsbad, CA, USA), passed through a nylon strainer with pore diameter 40 µm (cat. no. 352340; Becton Dickson, Franklin Lakes, NJ, USA) to remove the chorionic villous parts, and placed again in collagen type 1-coated dishes. Following the removal of nonadherent cells and debris by washing with phosphate-buffered saline (PBS), residual cells were considered isolated EVT cells. The fine pieces of villous fragments were defined as first-trimester chorionic villous trophoblast (CVT) cells; they had cytotrophoblast cells, syncytiotrophoblast, and sparse stromal cells and fetal endothelial cells.

RNA Sequencing and Data Analysis
Total RNA was extracted from each sample using RNAiso Plus (cat. no. 9109; Takara Bio, Shiga, Japan). The 26 bp single-end reads for miRNA sequencing and the 151 bp paired-end reads for mRNA/lncRNA sequencing were obtained as fastq files and imported into CLC Genomics Workbench (version 8.0.1; Qiagen, Venlo, The Netherlands) for the analysis of raw data, including trimming, filtering, and the cleaning up of contaminated reads based on default parameters (a minimum quality score of 0.05 (equivalent to Phred quality score of 30)). Clean 26 bp and 151 bp reads were aligned to the human genome sequence retrieved from the Ensembl database (Assembly GRCh38.p13, database version 82.38 and 101.38) and the microRNA database miRbase (release 21), respectively, using the CLC Genomics Workbench "Map Reads to Reference" tool (length fraction = 0.95 and similarity fraction = 0.9).

Quantitative Polymerase Chain Reaction (qPCR)
Total RNA was isolated from the samples using RNAiso Plus. Then, qPCR was performed with the Applied Biosystems 7300 Real-Time PCR System (Waltham, MA, USA) according to the manufacturer's instructions. TB Green Premix Ex Taq (cat. no. RR420; Takara Bio, Shiga, Japan) was employed for the quantitative analysis of mRNA level. To normalize the expression levels of mRNAs, 18S ribosomal RNA (designated as 18S) was used. Three independent analyses of the expression levels of each mRNA and miRNA were performed. Primers for the lncRNA H19 and mRNAs are shown in Table S1.

Western Blotting
Cell lysates were obtained using mammalian protein extraction reagent (cat. no.

Cell Invasion Assay
The cell invasion assay was performed as described previously [8]. Briefly, growth factor-reduced BD Matrigel (cat. no. 354230; Becton Dickinson, Franklin Lakes, NJ, USA) was coated onto cell culture inserts (Falcon Transparent polyethylene terephthalate (PET) Membrane/24 well, 8.0 µm pore size; cat. no. 353097; Corning, NY, USA). The Matrigel concentration was 200 µg/mL (2%). These upper chambers were set onto the lower chambers (Falcon 24-well tissue-culture (TC)-treated Cell Polystyrene Permeable Support Companion Plate; cat. no. 353504; Corning, NY, USA). The cells were transfected with miRNA mimics or the siRNAs described above. After transfection for 24 h, cells (3.5 × 10 4 HTR-8/SVneo or 7.0 × 10 4 HChEpC1b) in 250 µL of each culture medium without FBS were placed in the upper chamber. Then, 800 µl of each culture medium with FBS (5% for HTR-8/SVneo; 10% for HChEpC1b) was added to the lower chamber. After incubation for 24 h (HTR-8/SVneo) or 48 h (HChEpC1b), the non-invading cells at the top of each Transwell chamber were scraped off using cotton swabs. The invading cells at the bottom of the membrane were fixed with 2.5% glutaraldehyde (Wako, Osaka, Japan)-PBS, stained with 0.5% crystal violet (Wako, Osaka, Japan), and counted manually using a light microscope (KX4, Olympus, Tokyo, Japan) under 200× (HTR-8/SVneo) or 100× (HChEpC1b) magnification. The average numbers of invading cells in five areas were counted to reflect invasive ability. Three invasion assays were conducted under each experimental condition.

Cell Proliferation Assay
Evaluation of cell proliferation ability was performed using a CellTiter-Glo Luminescent Cell Viability Assay kit (cat. no. G7571; Promega, Madison, WI, USA). Briefly, 7500 HTR-8/SVneo or 15,000 HChEpC1b cells were placed in 96-well plates for 24 h and transfected with 675-5p or NC mimic (50 nM). After transfection for 24, 48, and 72 h, the cells were lysed using a reagent from the CellTiter-Glo Luminescent Cell Viability Assay kit, and the ATP activities of the cell lysates were measured using the Glomax-Multi Detection System (Promega, Madison, WI, US). ATP activity was measured five times for each experimental condition.

Luciferase Reporter Assay
The luciferase reporter assay for GATA2 was carried out as follows. To construct a reporter plasmid, we first cloned the 3 -UTR of human GATA2 into the pMIR-REPORT vector (Applied Biosystems, Waltham, MA, USA). Total RNA isolated from Jurkat cells was reverse-transcribed to cDNA using PrimeScript reverse transcriptase (cat. no. 2680A; Takara Bio, Shiga, Japan). The portion of the GATA2 3 -UTR (327 bp), including the target candidate sequence of miR-675-5p was then amplified from the cDNA with the following primers: GAGGGAGCTCACCCTTAGCAGCCCAGCAT (SacI site underlined) and ACGCACGCGTCACCAAGTCTCCAAGTCCTTGTT (MluI site underlined). After sequence verification, the GATA2 3 -UTR (327 bp) was cloned into pMIR-REPORT via the SacI and MluI restriction sites. This final construct was designated as pMIR-GATA2/wild. To construct a reporter plasmid with a mutated miR-675-5p recognition site for the GATA2 3 -UTR ( Figure 2C), an inverse PCR method was used [101]. The primers used for the inverse PCR were as follows: GATA2 3 -UTR-mutation, forward: tgagaggctgcctccacgcctgaccgctgc-CCAGGT, and reverse: gcagcggtcaggcgtggaggcagcctctcaGCGGTG (the complementary sequence is shown in lowercase). PCR amplification was carried out using the previously cloned vector ZERO Blunt TOPO containing the GATA2 3 -UTR. Plasmid DNA was digested by DpnI. Amplified DNA was transformed into Escherichia coli (One Shot TOP10: cat. no. C404003; Thermo Fisher Scientific, Waltham, MA, USA). After sequence verification, the mutated 3 -UTR sequence was cloned into pMIR-REPORT via the SacI and MluI restriction sites. The final construct was designated as pMIR-GATA2/mut.
For the reporter assay, HTR-8/SVneo cells were transfected with pMIR-GATA2/wild or pMIR-GATA2/mut and the control vector pRL-TK (Renilla luciferase expression plasmid), together with 20 nM 675-5p or NC mimic using Lipofectamine 2000 in 24-well plates. At 24 h after transfection, luciferase assays were performed using the Dual Luciferase Reporter Assay System (Promega, Madison, WI, USA). Firefly luciferase (from the pMIR-REPORT vector) and Renilla luciferase (from the phRL-TK vector) activities in the cell lysates were measured using the Glomax-Multi Detection System (Promega, Madison, WI, USA). Firefly luciferase activity was normalized to Renilla luciferase activity. Luciferase activity was measured three times for every experimental condition.
Luciferase reporter assays for BHLHE41, CHMP1A, and NOG were carried out in a similar manner. The primers for 3 -UTR amplification and inverse PCR are shown in Tables S2 and S3, respectively.

Statistics
All statistical analyses were performed using IBM SPSS Statistics (version 25; IBM, Armonk, NY, USA). The significance of between-group differences was assessed using a Student's t-test or analysis of variance followed by Tukey's test. A p-value < 0.05 was considered to indicate significance.

Conflicts of Interest:
The authors have declared that no conflict of interest exists.