Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS), and phosphate buffered saline (PBS) (pH = 7.2) were purchased from Gibco (Shanghai, China); ethanol (70%), isopropyl alcohol, and Triton X-100 were bought from Sigma-Aldrich (Shanghai, China); Cell Counting Kit-8 (CCK-8) were bought from Dongren Chemical Technology (Shanghai, China); bicinchonininc acid (BCA) protein assay kit was purchased from Thermo Fisher Scientific (Shanghai, China); antibodies were purchased from Cell Signaling Technology (Shanghai, China) and Abcam (Shanghai, China); the primers [circ_0084043, PCDH9, RAC1, Cyclin D1, Pyk2, FAK, MMP2, MMP9, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH)] were designed and synthesized by Sangon Biotech (Shanghai, China); the small interfering ribonucleic acid (siRNA) for circ_0084043 (si-circ_0084043) was designed and synthesized by Ribo Biotech (Guangzhou, China); the miR-134-5p mimics, miR-134-5p inhibitor, and miRNA negative control (miR-NC) were also purchased from Ribo Biotech; GV358-PCDH9 lentivirus and GV358-siRNA lentivirus were designed by Genechem (Shanghai, China); SYBR^® Premix Ex Taq™ II and PrimeScript™ RT reagent Kit with genomic DNA (gDNA) Eraser were bought from Takara Bio Inc. (Beijing, China); ultrapure water was obtained from EPED-20TF (Nanjing, China).

Cell lines A375, A875, SK-MEL-28, G361 (melanoma cell lines), and PIG1 (normal melanocyte line) were bought from the American Type Culture Collection (MD, USA). They were grown in DMEM supplemented with 10% heat-inactivated FBS as well as penicillin (100 IU/ml) and streptomycin (100 μg/ml). Cells were maintained in a CO₂ incubator at 37°C under a humidified atmosphere (95% air, 5% CO₂).

Before the experimental process, cells were divided into three groups: blank group (no lentivirus), control group (lentivirus with empty plasmid), and PCDH9 group (lentivirus with overexpressed PCDH9 plasmid).

Different parameters of lentiviral infections were determined for optimizing the experimental conditions: (i) cell seeding density at 0.5 × 10⁶ cell/ml, (ii) infectious time course at 12 h post-infection (hpi), (iii) observed time at 72 hpi, (iv) lentivirus titer of a control group of 2 × 10⁸ TU/ml, (v) lentivirus titer of a transfected group of 1×10⁹ TU/ml, and (vi) MOI (multiplicity of infection) of 20. Briefly, after seeding cells at an appropriate density in six-well plates (1 × 10⁵ cells per well), they were incubated overnight for 30%–40% adherence. They were then centrifuged for 30 s at a slow speed. After including viruses (MOI = 20), the fresh medium with polybrene was added to maintain 1 ml each, whereas the blank groups contained 1 ml of culture medium with 10% FBS. After being incubated for 12 h, the medium with lentiviruses was changed with a culture medium with 10% FBS (2 ml). Cells were observed by an inverted fluorescent microscope at 72 hpi and were screened by puromycin for post-infection assay.

Before the experimental process, cells were divided into three groups: blank group (no lentivirus), control group (lentivirus with empty plasmid), and PCDH9-siRNA group (lentivirus with siRNA plasmid).

Different parameters of lentiviral infections were determined for optimizing the experimental conditions: (i) cell seeding density at 0.5 × 10⁶ cell/ml, (ii) infectious time course at 12 hpi, (iii) observed time at 72 hpi, (iv) both lentivirus titers of the control group and transfected group of 1 × 10⁹ TU/ml, and (v) MOI was of 20. After seeding cells at the appropriate density in six-well plates (1 × 10⁵ cells per well), they were incubated overnight for 30%–40% adherence. They were then centrifuged for 30 s under slow speed, and after including the viruses (MOI = 20), the fresh medium with polybrene was added to maintain 1 ml each, whereas the blank groups contained 1 ml of culture medium with 10% FBS. After being incubated for 12 h, the medium with lentiviruses was changed with a culture medium with 10% FBS (2 ml). Cells were observed by an inverted fluorescent microscope at 72 hpi and were screened by puromycin for post-infection assay.

Before cell transfection, A375 and A875 cells were plated into six-well plates and cultured until 60%–80% confluence, and then, the indicated 10 μM oligonucleotides (si-circ_0084043, si-NC, miR-134-5p mimics, miR-NC, miR-134-5p inhibitor, and miR-inhibitor-NC) were transfected into cells by using Lipofectamine RNAiMAX (Invitrogen) and the cDNA plasmids by using Lipofectamine 3000 (Invitrogen) according to the manufacturer’s protocol.

Cells were seeded into 96-well plates at a density of 2 × 10⁵ cells per well and treated by a non-transfected plasmid, transfected with the empty plasmid, and transfected with PCDH9 overexpressing plasmid as explained above. After incubation at 24, 48, 72, and 96 h, 10 μl of CCK-8 was added to each well, and cells were incubated for another 4 h at 37°C. The level of colored formazan derivative was analyzed on Thermo Scientific Multiskan FC (Vantaa, Finland) at a wavelength of 450 nm. The viable cells were directly proportional to the formazan production, and the percentage of viable ones was calculated.

Apoptosis was measured using the flow apoptosis assay. Cells were seeded in a six-well plate at a density of 1 × 10⁶ cells per well and treated as explained above. At the end of the treatment, cells were centrifuged (1,000 rpm for 10 min). After that, the cells were washed twice by cold PBS and resuspended in 1 ml of 1× Binding Buffer; to 100 μl of this solution, 5 μl of FluoresceinIsothiocyanate (FITC) Annexin V (BD, USA) and 5 μl of Propidium Iodide (PI) (BD, USA) were added. After being vortexed gently and incubated in the dark at room temperature for another 15 min, 400 μl of 1× Binding Buffer was added to the final solution. Finally, samples were analyzed using the BD FACSCanto II flow cytometer. On the basis of the FITC Annexin V staining procedure, cells can be divided into apoptosis and non-apoptosis groups.

The cell cycle was measured using the flow apoptosis assay. Cells were seeded in six-well plates at a density of 1 × 10⁶ cells per well and treated as explained above. At the end of the treatment, cells were centrifuged (1,000 rpm for 10 min) and then vortexed with 5 ml of cold 75% ethanol; cells were incubated at −20°C for 2 h, washed twice to remove ethanol (first washed in PBS and then washed in Stain Buffer), and centrifuged again (1,000 rpm for 10 min). The cells were resuspended in 0.5 ml of PI/RNase Staining Buffer (BD, USA), and after 15 min of incubation at room temperature, samples were analyzed using the BD FACSCanto II flow cytometer.

Cells were seeded into six-well plates at a density of 1.5 × 10⁵ cells per well and were incubated under cell culture conditions for 24 h until confluent condition. The scratched wounds were created by a sterile 10-μl pipette tip. The suspension cells were washed twice with PBS, and then, DMEM (2% FBS) was added for incubation. The wounds were viewed using an inverted microscope DMI3000B (Leica, Germany) at 0, 24, and 48 h. ImageJ (Maryland, USA) was applied to measure the wound space.

Cells were rinsed by cold PBS and lysed by trypsin, after which they were thoroughly vortexed and centrifuged for 5 min (1,000 rpm), and the supernatant was discarded and refreshed with cold PBS; after spinning for 5 min (1,000 rpm), TRIzol was respectively added, and after 5 min of incubation at room temperature, chloroform was added (0.2 ml of chloroform to 1 ml of TRIzol) according to the amount of TRIzol; after being vortexed vigorously (15 s) and after 3 min of incubation at room temperature, the sample was spun at 4°C for 15 min (12,000×g); aqueous phase was taken. RNA was precipitated by isopropyl alcohol (0.5 ml of isopropyl alcohol to 1 ml of TRIzol in homogenized step) after 10 min of incubation at room temperature. The supernatant was discarded, and then, the RNA, which was at the bottom of the tube, was washed with 75% ethanol (1 ml of ethanol to 1 ml of TRIzol in homogenized step). The sample was centrifuged at 4°C for 5 min (7,500×g) several times until all ethanol was removed. After being dried for 8 min, the RNA sample was dissolved in diethyl pyrocarbonate (DEPC)-treated water by pipette tip (1 μl of RNA to 39 μl of DEPC-treated water, 1:40 dilution). The A₂₆₀/A₂₈₀ of RNA extracted was above 1.8.

The reverse transcription of the prepared RNA sample was performed by MasterCycler Gradient PCR (Thermo Fisher Scientific, USA) using a PrimeScript RT reagent kit (TaKaRa, Japan) according to the SYBR Green qPCR assay introduction. Quantification of genes was performed with the 2^−ΔΔCT method, as described previously (42): The sample was cycled (95°C for 5 s and 60°C for 30 s) 40 times by ABI7500Fast Real-time PCR System Amplifier (Thermo Fisher Scientific, USA). The primers designed for genes (circ_0084043, PCDH9, GAPDH, Pyk2, Cyclin D1, MMP2, MMP9, RAC1, and FAK), and their amplicon sizes are shown in Supplementary Table S1 .

After treatments, cells were lysed in a buffer containing 20 mM Tris-HCl (pH = 7.5), 0.9% NaCl, 0.2% Triton X-100, and 1% protease inhibitor cocktail. The analyses were conducted on cells between the third and the sixth passages.

Total protein concentration was detected by the BCA protein assay kit. Equal amounts of protein (30 μg) of each sample were loaded and separated in a 10% acrylamide sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to immobilon poly(vinylidene fluoride) (PVDF) membranes. Membranes were washed three times with Tris-HCl–buffered saline containing 5% Tween 20 (TBST), blocked with a 5% non-fat dry blocker, and incubated with the primary antibodies at 4°C overnight. Next, membranes were washed three times with TBST and then incubated with the secondary antibodies. Protein bands were visualized using Immobilon Western Chemiluminescent Substrate, and the protein signals were detected by Azure c500 Infrared Western Blot Imaging System from Azure Biosystems (CA, USA). Quantification of protein expression was made using the software ImageJ (Maryland, USA) and Origin 2020 for visualization (Northampton, MA, USA).

Immunohistochemistry (IHC) was performed on collected tissues. Forty-five normal human skin tissues, 30 humans pigmented nevus tissues, and 30 human MM tissues were selected. These tissues were ethically acquired from the outpatient clinic of the Affiliated Hospital of Guangdong Medical University, and all of them are Chinese population (Han people) with personal identifiers redacted. These MM tissues were from patients diagnosed with MM by skin pathology from 2012 to 2019, and the pigmented nevus tissues came from patients who underwent surgical resection in dermatologic surgery at the same time. The age, sex, site of onset, Breslow thickness, ulcer condition, and clinical stage of patients with MM included in the sample are shown in Table 1 . All the tissues were prepared as paraffin specimens. Deparaffinization included two 100% xylene changes (xylene I for 10 min and xylene II for 10 min) followed by rehydration with a graded series of ethanol (anhydrous ethanol I for 5 min; anhydrous ethanol II for 5 min; 95%, 85%, and 75% ethanol all for 5 min) and then rinsed under distilled running water for 3–5 min. Antigen retrieval consisted of 2 min of incubation of slides in citric acid retrieval solution heated to 98°C with a commercial steamer following a cool down step to room temperature (cold water and ice pack were added); slides were transferred into a wet box and were then rinsed three times with PBS. After protein blocking, primary antibodies (1:200) were incubated at 4°C overnight. After being at room temperature for 30 min, the slides were washed three times for 3 min each with PBS. After removing PBS and protein blocking, secondary antibodies (1:1000) were added at room temperature for 1 h. The slides were then washed three times for 3 min each with PBS. After removing PBS, 1 drop of prepared diaminobenzidine (DAB) solution (1 ml of A:1 drop of B:1 drop of C) for DAB staining was added, and the slides were observed under a microscope. After being rinsed in running water for 10 min, hematoxylin was added for 1 min, and then, the slides were washed with water for 5 min. The slides were then dehydrated in a series of ethanol (75%, 85%, 95%, and 100%) and 100% xylene changes and mounted with a coverslip with dry neutral resin.

Evaluation of various protein expressions in MM was evaluated by semi-quantitative analysis, according to the staining intensity and the percentage of positive cells. The score standards of staining intensity were as follows: no coloration, 0; low intensity (light yellow), 1; medium intensity (light brown), 2; and high intensity (dark brown), 3. Five fields of view were randomly selected under a microscope (400×), and 500 cells were counted as one unit; meanwhile, the percentage of positive cells was calculated. The percentage scores were as follows: <5%, 0; 6%~25%, 1; 26%~50%, 2; 51%~75%, 3; and >75%, 4. The score standards were the product of staining intensity and percentage of positive cells: 0, negative (−); 1 to 4, positive (+); 5 to 8, moderately positive (++); and 9 to 12, strongly positive (+++).

A375 cell was transfected with si-circ_0084043 and si-NC. After 48 h, the total RNA was extracted with TRIzol reagent. An amount of 3 µg of total RNA per sample was used as input material for the small RNA library. Following the manufacturer’s recommendations, sequencing libraries were generated using NEBNext^®Multiplex Small RNA Library Prep Set for Illumina^® (NEB, USA), and index codes were added to attribute sequences to each sample. Briefly, NEB 3′ SR Adaptor was directly and specifically ligated to the 3′ end of miRNA, siRNA, and piwi-interacting RNA (piRNA). After the 3′ ligation reaction, the SR RT Primer hybridized to the excess of 3′ SR Adaptor (that remained free after the 3’ ligation reaction) and transformed the single-stranded DNA adaptor into a double-stranded DNA molecule. This step is important to prevent adaptor-dimer formation. Moreover, double-stranded RNAs (dsRNAs) are not substrates for ligation-mediated by T4 RNA Ligase 1 and therefore do not ligate to the 5′ SR Adaptor in the subsequent ligation step. 5′-End adapter was ligated to 5′ ends of miRNAs, siRNA, and piRNA. Then, first-strand complementary DNA (cDNA) was synthesized using M-MuLV Reverse Transcriptase (RNase H^-). PCR amplification was performed using LongAmp Taq 2X Master Mix, SR Primer for Illumina, and index (X) primer. PCR products were purified on an 8% polyacrylamide gel (100 V, 80 min). DNA fragments corresponding to 140–160 bp (the length of small non-coding RNA plus the 3′ and 5′ adaptors) were recovered and dissolved in an 8-μl elution buffer. At last, library quality was assessed on the Agilent Bioanalyzer 2100 system using DNA High Sensitivity Chips.

According to the manufacturer’s instructions, the clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq SR Cluster Kit v3-cBot-HS (Illumia). After cluster generation, the library preparations were sequenced on an Illumina Hiseq 2,500/2,000 platform, and 50-bp single-end reads were generated.

The wild-type and mutant fragments of circ_0084043 were inserted into pSI-Check2 vectors. A375 and A875 cells were plated into 24-well plates and cultured at 37°C overnight. Then, the cells were cotransfected with the luciferase reporter vector and miR-134-5p mimic or miR-134-5p inhibitor using Lipofectamine 3000 (Invitrogen). After 48 h of transfection, luciferase activities were detected by the Dual-Luciferase Reporter Assay System (Promega, WI, USA).

PCDH9-Flag was overexpressed in A375 and G361 cell lines by lentivirus infection and cultured by DMEM with 10% FBS in 10-cm petri dishes as mentioned above. After discarding the cell medium and washing once by iced PBS, the cells were lysed in a buffer containing TBS with 1% Triton X-100 and phenylmethanesulfonyl fluoride (PMSF), 1 ml in each dish. The solution was collected in 1.5-ml Eppendorf tubes, and after centrifugation (1,000 rpm for 15 min), the supernatant was collected. For the input sample preparation, 80 μl of protein solution with 20 μl of 5× loading buffer was boiled. The leftover solution was incubated with 2 μl of anti-Flag antibody at 4°C overnight. Protein G Agarose beads (40 μl each tube) were added, and the solution was shaken for 3 h at 4°C. Then, after being centrifuged (2,500 rpm for 5 min), the supernatant was removed. The sediments were washed five times by lysis buffer with PMSF (1 ml each time) and were boiled for 10 min at 100°C after adding loading buffer (40 μl). The control group was incubated with mouse immunoglobulin G (IgG).

All animal experiments were subject to approval by the Institutional Review Board of Guangdong Medical University (Zhanjiang, China). Six-week-old nude female mice were kept in the specific pathogen free (SPF) animal room. A375 and A375 cells (5 × 10⁶) transfected with PCDH9 overexpressing plasmid, PCDH9 KD plasmid, or empty plasmid in 150 μl of PBS were inoculated subcutaneously in the mice. After 35 days, the mice were sacrificed to remove the tumors, and the tumor size of each group was measured.

Data are presented as the mean ± SD from at least three independent experiments. Statistical comparisons between two groups were made using Student’s t-tests. Differences among three or more groups were compared by one-way ANOVA, followed by least significant difference post-hoc tests (Originlab 2020, Northampton, MA, USA); p < 0.05 was considered statistically significant, whereas p < 0.01 was considered highly statistically significant.

PCDH9 was overexpressed by lentivirus with PCDH9 plasmid ( Figures 1A, B ) and knocked down by lentivirus with siRNA ( Figures 2C, D ). The relative expression of selected genes (Pyk2, Cyclin D1, RAC1, MMP2, and MMP9 except for FAK) varied with PCDH9 expression, but their effectiveness differed. Pyk2, Cyclin D1, and PCDH9 exhibited a positive correlation, whereas RAC1, MMP2, MMP9, and PCDH9 negatively correlated in both A375 and G361 cells ( Figures 1C, D ). The results of quantitative real-time PCR (qRT-PCR) were confirmed by Western blotting analysis. Except for FAK, the protein expressions of Pyk2, Cyclin D1, MMP2, MMP9, and RAC1 had a significant correlation with PCDH9 in A375 and G361 cells ( Figure 2 ). PCDH9 and Pyk2/Cyclin D1 exhibited a positive correlation, whereas PCDH9 and RAC1/MMP2/MMP9 negatively correlated ( Figure 2 ). Therefore, the overexpression of PCDH9 upregulated the gene expressions of Pyk2 and Cyclin D1 and downregulated RAC1, MMP2, and MMP9, whereas KD of PCDH9 downregulated expression levels of Pyk2 and Cyclin D1 and upregulated RAC1, MMP2, and MMP9.

The overexpression of PCDH9 reduced the proliferation of melanoma cells. Overexpressed PCDH9 groups showed lower viability than blank/control groups ( Figures 3A, C ). As time passed, the viability of melanoma cells trended to stabilize, but the inhibition rate of cell proliferation in overexpressed PCDH9 groups was higher than that in the blank/control groups, and the differences were significant at 24, 48, 72, and 96 h ( Figures 3B, D ).

The expression of PCDH9 significantly affected the apoptosis of melanoma cells as well. Overexpression increased apoptosis (p < 0.05), whereas KD decreased apoptosis (p < 0.05) in G361 cells ( Figure 3F ), and a similar phenomenon happened in A375 cells with more modest effects ( Figure 3E ).

Recent studies have found low PCDH9 expression in various cancer types (10, 43). The above phenomenon agreed with our investigation where alterations of PCDH9 could influence the viability of melanoma cells. Meanwhile, the percentage of apoptosis cells was also affected by PCDH9.

Regarding the cell cycle arrest, although we found differences between control and KD PCDH9 groups in G361 cells, there was no discrepancy between overexpressed/KD PCDH9 groups and other groups (blank and control groups) in both cell lines (A375 and G361) ( Figures 4A–D ). Therefore, the alteration of PCDH9 did not affect the melanoma cell cycle, and the cell cycle regulator Cyclin D1 modulation revealed no significant effect on the cell cycle. Previous investigations of hepatocellular carcinoma (HCC) found that PCDH9 suppresses HCC cells by inducing cell cycle arrest at G0/G1 phase (8). This result suggests that PCDH9 and Cyclin D1 may affect melanoma cells through other mechanisms.

Concerning cell migration, after quantifying the scratched boundary by ImageJ ( Figures 4E, F ), the results revealed that the scratch area decreased with the duration of cell culture in blank and control groups (p < 0.05), whereas the relative wound width did not change markedly in PCDH9 overexpressed groups (p > 0.05) ( Figures 4G, H ). The relative wound width of the scratched boundary was significantly different in PCDH9 overexpressed groups compared with blank and control groups after 24 and 48 h (p < 0.05) ( Figures 4I, J ). This phenomenon is similar to the correlation between PCDH9 and other tumor cells (HCC) (11, 43). To conclude, overexpressing PCDH9 could significantly inhibit melanoma cell migration by decreasing cell viability and increasing apoptosis.

The interaction between Pyk2 and other proteins could affect the function of Pyk2 as previously demonstrated (2, 40). In the current study, the expressions of Pyk2 and RAC1 in melanoma cells were significantly affected by the alteration of PCDH9 ( Figures 1B, C , 2B ). Therefore, we explored the interaction between PCDH9 and Pyk2/RAC1 by complex immunoprecipitation. As a result, PCDH9 and Pyk2/RAC1 were co-immunoprecipitated with PCDH9-FLAG in both cell lines (A375-PCDH9-Flag and G361-PCDH9-Flag), whereas there was no protein in the negative control ( Supplementary Figure S1 ). It was demonstrated that PCDH9 interacted with the protein Pyk2 and RAC1. On the basis of previous investigations, we hypothesized that PCDH9 may exert the function of Hace1 through a similar mechanism; however, Hace1 targets complex-bound RAC1 directly; whereas PCDH9 needs the help of Pyk2 (44). The results cited from the preprint https://doi.org/10.21203/rs.3.rs-95026/v1.

We verified that circ_0084043 was significantly upregulated in melanoma cells. The expression of circ_0084043 was tested in three melanoma cell lines (A375, A875, and SK-MEL-28) through qRT-PCR. The result showed that circ_0084043 was increased in A375, A875, and SK-MEL-28 cells compared to normal human epidermal melanocytes (PIG1) ( Figure 5A ). In addition, si-circ_0084043 could knock down circ_0084043 in melanoma cells (A375 and A875) ( Figure 5B ). Then, IHC results showed that the positive percentage of PCDH9 expression was lower in human melanoma tissue than in normal skin or/and pigmented nevus tissue, and the positive percentage of RAC1 was just the opposite; additionally, PCDH9 was mainly expressed in the cytoplasm, whereas a small amount was expressed in the nuclei ( Figure 5C ). PCDH9 was 100% expressed in normal skin (45/45) or/and pigmented nevus tissue (30/30), whereas only 23.3% (7/30) in melanoma tissue, which was lower than non-tumor tissue ( Table 2 ).

As known, circRNAs work as a sponge of miRNA in tumor cells. To find more targets of circ_0084043, we performed a small RNA sequencing on A375 cells after circ_0084043 was knocked down. Compared with si-NC group, eight miRNAs differentially expressed in si-circ_0084043 group (P < 0.05), in which miR-134-5p, miR-4659b-5p, and miR-760 were upregulated and miR-202-5p, miR-33a-3p, miR-181b-3p, miR-217-5p, and miR-3143 were downregulated ( Figures 5D, E ). Previous studies reported that miR-134-5p was significantly downregulated in patients with melanoma compared with healthy control (45). In addition, we verified the upregulation of miR-134-5p in circ_0084043 KD cells by qRT-PCR. Therefore, we speculated that circ_0084043 could sponge miR-134-5p. Luciferase reporter assays were performed to test the hypothesis. The results indicated that miR-134-5p mimics could reduce the luciferase activity of cells transfected with wild-type sequence of circ_0084043 instead of mutant type sequence ( Figure 5F ). The binding site between miR-134-5p and circ_0084043 was predicted using RNAhybrid 2.2 ( Figure 5G ).

It is reported that circ_0084043 promoted the growth and metastasis of melanoma cells (41), and KD of circ_0084043 suppressed melanoma progression (46). We further explored the biological role of circ_0084043 and miR-134-5p in the progression of MM. Scratch wound assays showed that the migration abilities of melanoma cells decreased in both the circ_0084043 KD group and miR-134-5p mimic group ( Figures 6A–C ). The flow apoptosis assays demonstrated that the apoptosis of melanoma cells was induced in the circ_0084043 KD group and miR-134-5p mimic group ( Figures 6D, E ). Moreover, the miR-134-5p inhibitor could reverse the effects of inhibiting migration and promoting apoptosis after knocking down circ_0084043 ( Figure 6 ). These results indicated that circ_0084043 might promote melanoma progression in vitro by sponging miR-134-5p.

To further identify the anticarcinogenic of PCDH9 in melanoma, we established the subcutaneous xenograft tumor model using nude mice. We found that tumors in the PCDH9 KD group were significantly heavier than those in the control and blank group ( Figure 7 ). The nude mice with overexpression of PCDH9 did not grow tumors of significant size ( Figure 7 ). Therefore, animal experiments presented that PCDH9 could work as a tumor suppressor gene to inhibit the growth of MM.

The findings indicated that PCDH9 functioned as a tumor suppressor gene in melanoma development, and circ_0084043 is proved to be an oncogene. What is the mechanism of the circ_0084043-miR-134-5p axis promoting cancer in melanoma? We obtained the prediction results of the miR-134-5p target on the TargetScan and Starbase 3.0 database. Hence, we selected the following proteins for research based on the two databases: Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) analysis ( Supplementary Figures S2A, B ). Then, Western blot was carried out to search for the downstream mRNA targeted by the circ_0084043-miR-134-5p axis. As shown in Figure 8 , the levels of PCDH9, Pyk2, and c-Jun were increased after knocking down circ_0084043 in A375 and A875 cells, whereas RAC1, Cyclin D1, MMP2, and PD-L1 were downregulated. Overexpression of miR-134-5p had the same effect as KD of circ_0084043 except for MMP2. Cotransfection of si-circ_0084043 and miR-134-5p inhibitor could reverse the transfection of circ_0084043 alone except for MMP2. X-linked inhibitor of apoptosis protein (XIAP) is an inhibitor of apoptosis protein highly expressed in melanoma, and we found that its expression level was not affected by circ_0084043 and miR-134-5p. To sum up, the circ_0084043-miR-134-5p axis could regulate PCDH9 and the above tumor-associated proteins in melanoma cells.