The SLITRK4-CNPY3 axis promotes liver metastasis of gastric cancer by enhancing the endocytosis and recycling of TrkB in tumour cells

Gastric cancer (GC) is a malignant tumour with high mortality, and liver metastasis is one of the main causes of poor prognosis. SLIT- and NTRK-like family member 4 (SLITRK4) plays an important role in the nervous system, such as synapse formation. Our study aimed to explore the functional role of SLITRK4 in GC and liver metastasis. The mRNA level of SLITRK4 was evaluated using publicly available transcriptome GEO datasets and Renji cohort. The protein level of SLITRK4 in the tissue microarray of GC was observed using immunohistochemistry. Cell Counting Kit-8, colony formation, transwell migration assays in vitro and mouse model of liver metastasis in vivo was performed to investigate the functional roles of SLITRK4 in GC. Bioinformatics predictions and Co-IP experiments were applied to screen and identify SLITRK4-binding proteins. Western blot was performed to detect Tyrosine Kinase receptor B (TrkB)-related signaling molecules. By comparing primary and liver metastases from GC, SLITRK4 was found to be upregulated in tissues of GC with liver metastasis and to be closely related to poor clinical prognosis. SLITRK4 knockdown significantly abrogated the growth, invasion, and metastasis of GC in vitro and in vivo. Further study revealed that SLITRK4 could interact with Canopy FGF Signalling Regulator 3 (CNPY3), thus enhancing TrkB- related signaling by promoting the endocytosis and recycling of the TrkB receptor. In conclusion, the CNPY3-SLITRK4 axis contributes to liver metastasis of GC according to the TrkB-related signaling pathway. which may be a therapeutic target for the treatment of GC with liver metastasis.


Introduction
Gastric cancer (GC) causes a considerable health burden and is a devastating global disease with high morbidity and mortality. Most GC cases are reported in East Asia, Eastern Europe, and South America [1][2][3]. There is organotropism in metastasis, as the liver is a preferred organ to which various types of tumours metastasize. Liver metastasis is a major obstacle to the treatment of GC, and most patients are first diagnosed at an advanced stage when metastasis has already occurred [4,5]. Currently, surgical or endoscopic resection and chemoradiotherapy are the main therapies for GC patients [6][7][8]. However, gastrectomy does not improve the survival of GC patients with metastases. In this situation, it is necessary to clarify the mechanisms of GC metastasis to explore new therapeutic strategies.
By analyzing a database of 6 GC samples with liver metastases, we screened out SLIT-and NTRK-like family member 4 (SLITRK4). SLITRK encodes a family of neuronal transmembrane molecules, including SLITRK 1-6. SLITRK plays an important role in the nervous system, including in neuronal migration, axonal path-finding, dendritic branching, neuron survival, and synapse formation. Existing reports have indicated that SLITRK gene mutations are closely associated with neuropsychiatric disorders [9][10][11][12][13]. SLITRK proteins are always enriched at the postsynaptic membrane [14].Six members of the SLITRK family are homologous to Slit, a classic axon guidance molecule, and Trks, which are neurotrophin receptors. SLITRK4 is a single transmembrane protein with extracellular leucine-rich repeat (LRR) domains at the N-terminal region that controls synapse formation via interaction with leukocyte common antigen-related receptor protein tyrosine phosphatases (LAR-RPTPs) [14][15][16].
To date, most studies on SLITRK4 remain limited to the promotion or inhibition of synapse formation, and its functional role in tumours remains unclear and requires further clarification [10].
In this study, SLITRK4 was found to be upregulated in clinical samples of GC patients with liver metastasis compared with paired normal tissues. High SLITRK4 expression was associated with a poorer clinical prognosis. Furthermore, according to in vivo and in vitro experiments, SLITRK4 was found to participate in tumour progression, invasion and metastasis of GC. Therefore, SLITRK4 could potentially serve as a novel biomarker of GC liver metastasis and a promising treatment target.

Clinical samples
Human primary GC tissues, primary para-cancerous tissues, matched liver metastatic tissues, and liver tissues adjacent to liver metastases were obtained with the informed consent of patients from Renji Hospital, School of Medicine, Shanghai Jiao Tong University. Transcriptomic data from the PRJNA555812 project stored in the Sequence Read Archive (SRA) were used to analyse transcriptional changes in liver metastasis of gastric cancer. The Renji Cohort used in our study comprised 69 GC liver metastatic tissues collected from 2005 to 2011; data collected also included RNA sequencing analyses, tissue microarrays, and corresponding retrospective analyses.
Immunohistochemical staining was performed on tumour microarray (TMA) and paraffin sections. Briefly, the tissues were fixed in formalin, embedded in paraffin, and then sectioned into 5-μm-thick pieces. The sections were deparaffinized in xylene and rehydrated in graded ethanol for histopathological evaluation. The sections were placed in a microwave oven at a high temperature for 5 min, which was repeated two to three times for antigen retrieval. After treatment with 0.3% hydrogen peroxide/ phosphate-buffered saline for 30 min, the sections were blocked with 10% BSA for 1 h at room temperature. Then, the sections were incubated at 4 °C overnight with primary antibodies against SLITRK4 (1:200, Novus, NBP1-76,878) and CNPY3 (1:200, Proteintech, 15,215-1-AP) diluted at the optimal concentration. Next, sections were incubated with HRP-conjugated mouse or rabbit secondary antibodies at room temperature for 1 h. The sections were then incubated with DAB substrate solution (No. S21024-2, Thermo Scientific, USA) and counterstained in haematoxylin. All sections were observed and photographed using a Carl Zeiss microscope.

Cell culture
The human GC cell lines BGC823 and MKN28 and other GC cell lines were purchased from American Type Culture 1 3 Collection (ATCC, Manassas, VA, USA). Cells were cultured in RPMI-1640 medium (Gibco, New York, USA) supplemented with 10% (v/v) foetal bovine serum (Gibco, New York, USA) and 1% (v/v) antibiotics (streptomycin and penicillin, Sigma-Aldrich, St. Louis, USA) at 37 °C in a humidified incubator in 5% CO2.

Quantitative Real-time PCR
Total RNA was extracted using RNAiso Plus (Takara, #9108Q) and reverse transcribed using the PrimeScript RT-PCR kit (Takara, #RR600A) according to the manufacturer's instructions. Quantitative real-time PCR assays were performed using SYBR Premix Ex Taq (Takara, Japan) on a 7500 Real-time qPCR system (Applied Biosystems) at the recommended thermal cycling settings: an initial cycle at 95 °C for 15 min followed by 40

Western blotting
The lysates of protein samples were collected in RIPA Lysis Buffer (YOCHE., YSD0100) on ice for at least 20 min and centrifuged at 12 000 × g for 20 min at 4 °C. The supernatant was diluted with 5 × loading buffer and boiled for 10 min. Proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. Subsequently, the membranes were washed with TBST and blocked for 1 h at room temperature using 5% non-fat milk solution diluted in TBST. The membranes were then incubated with primary antibodies (SLITRK4 (1:1

Cell Counting Kit-8 (CCK-8) viability assay
To measure cell proliferation, the indicated cells were seeded in triplicate in 96-well plates at a moderate density (10 4 cells/ml). Next, 100 μl serum-free RPMI-1640 medium containing 10% Cell Counting Kit-8 (CCK-8) reagent (Dojindo, Japan) was added to each well, and samples were incubated for 1 h at 37 °C for 4 days. Then, the absorbance was measured at a wavelength of 450 nm using multifunctional enzyme labels.

Colony formation assays
For the colony formation assays, cells were seeded in sixwell plates and cultured for 10 days. The colonies were washed twice with PBS, fixed in 4% paraformaldehyde (PFA) for 10 min, and stained with crystal violet (0.5% w/v) for 30 min at room temperature, after which the number of colonies was calculated visually using Image J. This experiment was repeated at least twice.

Transwell migration assays
To measure cell migration, the indicated cells (2.5×10 5 cells/ ml) were seeded into the upper chamber of an 8 μm-pore size Transwell chamber (BD, 353,097), where they were resuspended in serum-free RPMI-1640. Medium containing 10% FBS was added to the lower chamber. After incubation for 48 h, cells were fixed in 4% PFA and then stained with 0.5% (w/v) crystal violet to observe cells attached to the backside of the upper chamber. Non-invading cells were removed using a cotton swab, while invading cells on the underside of the filter were counted under a light microscope. The number of invading cells was measured by Image J. The experiments were repeated at least twice.

Mouse model of liver metastasis
In all, 10 6 mouse GC cells (sh-SLITRK4 and sh-NC BGC823) were suspended in 20 µl PBS and then inoculated into the spleen of each nude mouse. After 4 weeks, the mice were sacrificed, and their livers were dissected and fixed in 4% PFA.

Immunofluorescence (IF)
For cell IF staining, we seeded 10 4 cells/well into 8-well plate firstly. Next, cells were fixed with 4% PFA and blocked with 5% BSA in PBS and incubated with primary antibody (SLITRK4 (1:200, Novus, NBP1-76,878) and CNPY3 (1:50, Santa cruze, sc-515151) and secondary antibodies. DAPI was used for Nuclei staining for 10 min at RT. Finally, using confocal microscopes to acquire images. For IF staining, paraffin sections were dewaxed with gradient ethanol and subjected to antigen retrieval in the citrate-based buffer. Then, the sections were blocked with 5% BSA for 1 h, stained with primary antibody (SLITRK4 (1:200, Novus, NBP1-76,878) and TrkB (1:200, ImmunoWay, YT4744)) at recommended dilutions at 4 °C overnight and secondary antibody at room temperature for 1 h. Finally, DAPI was used to counterstain nuclei for 5 min. Confocal microscopes were used to capture digital images.

Collection of conditional culture supernatant
First, we constructed the 293 T cells of Lenti-vector and lenti-CNPY3, CNPY3 ORF were cloned into the episomal expression vector with EX-Z1880-Lv242-B. Then, 293 T cells were screening with puromycin (2 μg/ml) for 7 days. Then, we seed cells into 10 cm dish, and the cells were treated with 6-8 ml of DMEM containing 2% FBS after the cells were attached. Finally, the culture supernatants were collected in 15 ml EP tubes, centrifuged at 2000 rpm, 4 °C for 10 min, and stored in -80 °C.

Co-Immunoprecipitation
For immunoprecipitation assays, cells were lysed in IP buffer, and cell lysates were immunoprecipitated with the indicated antibody or control IgG overnight and then incubated with protein A/G magnetic beads (Invitrogen, 10003D, USA) over night at 4 °C. The beads were washed three times in lysis buffer, after which the immunoprecipitated protein complexes were resuspended in 2 × loading buffer followed by western blot analysis.

Cell surface biotinylation assay to examine
TrkB internalization and recycling [18,19] Cells were starved overnight, washed in cold PBS, and then incubated with 0.2 mg/ml sulfo-NHS-SS-biotin (#21,331; Thermo Scientific, Waltham, MA, USA) in D-PBS on ice for 45 min. Then, the cells were washed three times in 50 mM Gly on ice for 5 min to quench the free biotin. Labelled cells were incubated at 37 °C for 30 min in culture medium containing 10% FBS to allow protein trafficking. Then, the medium was aspirated, and the dishes were placed on ice and washed with cold PBS. Biotin was removed from the cell surface by stripping with 50 mM L-GSH at 4 °C three times for 10 min each time. Cells not treated with L-GSH served as the control group to examine the total biotinylated TrkB. Endocytosis was detected according to the accumulation of the Fig. 1 SLITRK4 expression was upregulated in GC with liver metastasis and was associated with poor clinical prognosis. A. Significantly expressed genes (P < 0.01, log 2 FC > 3) were screened between the liver metastasis (LM) tissues and adjacent liver (AL) tissues and between LM and primary gastric cancer (GC) tissues from the Renji cohort. B. Heatmap of the cluster analysis applied to the targeted genes in six paired GC and LM samples. C. RNA-seq analysis showed the fold change of relative expression of the SLITRK gene family. D and E. Relative SLITRK4 mRNA expression levels were analysed in GC tissues and paired noncancerous tissues from the GSE65801 (D) and GSE30601 datasets (E). F. Levels of relative SLITRK4 mRNA expression were analysed in paracancerous tissues (PC), GC tissue, and LM tissue. G. Representative IHC staining for SLITRK4 in PC tissue, GC tissue and LM tissue. Scale bar, 50 μm H. Kaplan-Meier analyses of the overall survival of GC patients with liver metastases according to SLITRK4 expression by IHC using the log-rank test. (n = 69, p < 0.001). *p < 0.05. **p < 0.01. ***p < 0.001 ◂ sulfo-NHS-SS-biotin-labelled cargo proteins within the cells, which were protected from stripping by L-GSH. Next, the cells were lysed in lysis buffer and centrifuged, and the supernatant was collected. Except for an aliquot removed to measure the amounts of total cargo proteins, the remaining supernatant was added to prewashed streptavidin-agarose beads (#88,816; Thermo Scientific, Waltham, MA, USA) and rotated at 4 °C overnight. The beads were collected by centrifugation, and the supernatant was collected. Then, the beads were washed 3 times with RIPA buffer at 4 °C, and the sulfo-NHS-SS-biotin-labelled proteins were extracted by heating in loading buffer at 100 °C. Finally, the levels of total and internalized biotinylated TrkB were examined by immunoblotting. The internalization ratio was determined by calculating the ratio of internalized biotinylated TrkB to total biotinylated TrkB.
To measure the proportion of internalized TrkB that recycled back to the cell surface, the internalized fraction was returned to 37 °C for 15 min. Cells were then placed on ice, and biotin was removed from recycled proteins by a second reduction with L-GSH. Cells were lysed and processed as described above to measure the level of internalized receptor.

Cycloheximide (CHX) chase experiments
Sh-NC and sh-SLITRK4 BGC823 cells were treated with 20 μg/ml of cycloheximide (CHX) to block new protein synthesis. The proteins were collected at 0, 1, 2, 4, 6 and 8 h after cycloheximide addition, and the TrkB protein level was detected by western blotting.

Statistical analysis
The software used for data processing and statistical analysis was GraphPad Prism 8.4.3 (GraphPad Software, San Diego, CA). Cumulative survival times were calculated using the Kaplan-Meier method and were analysed using the log-rank test. Data were shown as the mean ± SEM. Comparisons between groups were made using Student's t test or one-way ANOVA. P values less than 0.05 were considered statistically significant.

SLITRK4 expression was increased in human GC with liver metastasis and was associated with poor clinical prognosis
First, we collected surgically resected tumour tissues from 6 GC patients with liver metastases at Renji Hospital. RNA sequencing (RNA-seq) data, including data from primary gastric cancer (GC), para-cancerous (PC), liver metastasis (LM), and adjacent liver (AL) tissues, were analysed to determine differentially expressed genes. By comparing the differentially expressed genes according to mRNA expression profiles between LM and AL (n = 2141) and LM and GC (n = 348), we identified 19 differentially expressed genes with a log 2 FC > 3 and p < 0.01 (Fig. 1A). Using a heatmap cluster analysis of these selected genes in 6 paired GC and LM samples, we selected SLITRK4 for further study after excluding genes that was significantly differentially expressed as a result of no in situ expression (Fig. 1B). Moreover, of the SLITRK family, SLITRK4 had the most striking changes after the RNA-seq data were analysed (Fig. 1C). The GEO datasets (GSE65801 [20] and GSE30601 [21][22][23]) also supported the finding that SLITRK4 mRNA expression was significantly upregulated in GC compared with normal gastric mucosal tissues (Fig. 1 D&E).
We further confirmed the upregulation of SLITRK4 mRNA expression in GC and LM compared with matched PC tissues collected from the Renji cohort (n = 13; Fig. 1F). Consistently, representative immunohistochemical images of SLITRK4 in PC, GC and LM tissues demonstrated rich expression of SLITRK4 in GC and LM, especially in the membranes of GC cells (Fig. 1G). In addition, Kaplan-Meier analyses showed that patients with high SLITRK4 expression had significantly shorter overall survival (OS) times than those with low SLI-TRK4 expression (Fig. 1H) [24]. Taken together, these results revealed that the expression level of SLITRK4 was

SLITRK4 knockdown suppressed the proliferation, migration and invasiveness of GC cells in vitro
We first analysed the mRNA expression levels of SLITRK4 in a human normal gastric epithelial cell line (GES-1) and 7 GC cell lines (BGC823, MKN28, MKN45, MGC803, AGS, HGC27 and SGC7901) by quantitative real-time PCR (qRT-PCR). SLITRK4 was found high expressed in all 7 GC cell lines, and we selected BGC823 and MKN28 for the functional study ( Fig. 2A). By constructing sh-NC and sh-SLITRK4 lentiviruses and infecting BGC823 and MKN28 cells, stable cell lines were established. Western blotting (WB) analysis was performed to confirm the knockdown efficiency after puromycin screening (Fig. 2B). The results of the Cell Counting Kit-8 (CCK-8) assay revealed that viability and proliferation were significantly impaired in BGC823 and MKN28 cells transfected with sh-SLI-TRK4 (Fig. 2C). In addition, the colony formation assay revealed that silencing SLITRK4 drastically inhibited colony formation (Fig. 2D-E). Additionally, transwell assay was performed, and the results revealed that the migration and invasion abilities of the sh-SLITRK4 groups were suppressed compared with those of the control group (Fig. 2F-G). In conclusion, our data suggested that knockdown of SLITRK4 inhibited the proliferation, migration and invasiveness of GC cells in vitro.

SLITRK4 knockdown suppressed liver metastasis and colonization of GC cells in vivo
Subsequently, we established a nude mouse model of GC with liver metastasis, in which GC cells were injected into the spleens of nude mice. We first confirmed that BGC823 cells had the strongest ability to form metastatic tumours in the liver compared with the other GC cell lines tested. Therefore, in the next study, we selected BGC823 cells to compare the effect of SLITRK4 on liver metastasis (Fig. 3A). Sh-SLITRK4 and sh-NC BGC823 cells were injected into the spleens of nude mice, and the volumes of liver metastases were measured after 25 days. Consistent with the in vitro results, liver metastases in the sh-SLITRK4 group were obviously attenuated compared with those in the sh-NC group, as reflected by the smaller and lower number of nodules on the liver surface ( Fig. 3B-C). To confirm SLITRK4 knockdown in liver metastases in vivo, RT-PCR of liver metastases from the sh-SLITRK4 and sh-NC groups was also performed (Fig. 3D). In addition, H&E staining and IHC for CK19 and PCNA were used to verify that the liver had been invaded by gastric tumour cells in mice with SLITRK4 knockdown compared with sh-NC mice. These images indicated that after the expression of SLITRK4 was knocked down, the size and number of liver metastases and expression CK19 and PCNA were significantly decreased ( Fig. 3E-G). In conclusion, our data suggested that SLITRK4 promoted tumorigenicity and liver metastasis of GC.

SLITRK4 directly interacted with CNPY3 in GC cells and tumour tissues
To further explore the mechanism of SLITRK4-mediated GC invasion and metastasis, we performed co-immunoprecipitation (Co-IP) experiments to screen and identify SLITRK4-binding proteins and combined the results of bioinformatics predictions (Fig. 4A and supplement Figure 1A). We then focused on Canopy FGF Signalling Regulator 3(CNPY3). CNPY3 is primarily located in the extracellular space and endoplasmic reticulum (ER), where it plays an important biological role in tumour metastasis. We performed immunofluorescence (IF) staining to reveal the co-localization of SLITRK4 and CNPY3 on the GC cells (Fig. 4B). Moreover, IHC for SLITRK4 and CNPY3 in serial sections of GC tissue demonstrated that CNPY3 and  (Fig. 4C). Based on the GC tissue microarray analysis, the correlation analysis of the H-score of SLITRK4 and CNPY3 (H-score = Σ pi (i + 1)) confirmed that SLITRK4 was closely correlated with CNPY3 (Fig. 4D). These data were consistent with the observation in the heatmap of the correlation analysis of the SLITRK4 and CNPY3 staining intensity (Fig. 4E). The GEO datasets (GSE65801 and GSE30601) also supported that CNPY3 mRNA expression was significantly upregulated in GC compared with normal gastric mucosal tissues (Fig. 4 F&G). We further used tissues collected from the Renji cohort and confirmed the overexpression of CNPY3 mRNA in GC and LM compared with matched PC tissues (Fig. 4H). In addition, the survival correlation was analysed using a Kaplan-Meier analysis, which revealed that patients with high CNPY3 expression had worse OS than those with low CNPY3 expression (Fig. 4I). Consistently, patients with high expression of both CNPY3 and SLITRK4 had the worst prognosis compared with the other groups (Fig. 4J). Thus, we speculated that the interaction between SLITRK4 and CNPY3 played an important role in the progression of GC with liver metastasis.

CNPY3/SLITRK4 axis promoted GC proliferation and migration
To explore the effect of CNPY3 on the progression of GC cells, we first knocked down CNPY3 in BGC823 and MKN28 cells and found that CNPY3 knockdown could inhibit the proliferation, migration and invasion of GC cells in vitro (Supplement Figure 2). Then, we began to study whether CNPY3 could regain the effect of SLITRK4 knockdown on the proliferation and migration of GC cells by the stimulation of conditional culture supernatants with overexpressing CNPY3 or CNPY3 recombinant protein (rCNPY3) (Fig. 5A and Supplement Figure 1D). The results of the colony formation and transwell assays indicated that CNPY3 conspicuously enhanced cell colony formation, invasiveness and migration of sh-NC BGC823 cells, whereas these effects were all weakened in sh-SLITRK4 BGC823 cells (Fig. 5B-E).
The consistent results were found under the stimulation of rCNPY3 (Fig. 5F-I). These results demonstrated that the stimulation of conditional culture supernatant of overexpress CNPY3 or rCNPY3 cannot rescue the effects of the knockdown of SLITRK4. Therefore, we speculated that CNPY3 might exert pro-cancer effects depending on SLITRK4.

SLITRK4-TrkB complex was found both in vitro and in GC liver metastases
The SLITRK family and the neurotrophic factor receptor family have similar structures and functions. Tyrosine Kinase receptor B(TrkB) is one of the neurotrophic factor receptors that affects downstream tyrosine kinase signalling pathways that function in biological processes. Some studies have shown that high TrkB expression is correlated with GC progression and metastasis [25][26][27]. Therefore, we performed Co-IP and found the interaction between SLITRK4 and TrkB (Fig. 6A). The GEO datasets (GSE65801 and GSE30601) also supported that TrkB mRNA expression was significantly upregulated in GC compared with normal gastric mucosal tissues (Fig. 6 B-C). IHC studies revealed that the TrkB expression in the GC and LM tissues was significantly higher than that on human PC tissues (Fig. 6D). We also performed IF studies to confirm the co-localization of SLITRK4 and TrkB on the human LM tissue of GC (Fig. 6E). Moreover, IHC for SLITRK4 and TrkB in serial sections of GC tissue microarray demonstrated that expression pattern of SLITRK4 was highly consistent with that of TrkB expression (Fig. 6F). In addition, the correlation analysis of H-score of SLITRK4 and TrkB (H-score = Σ pi (i + 1)) confirmed that expression of SLITRK4 was positively correlated with TrkB ( Fig. 6G-H). Therefore, SLITRK4

The SLITRK4-CNPY3 axis enhanced the TrkB pathway by promoting TrkB receptor internalization and recycling
We next aim to explore the role of TrkB in SLITRK4-CNPY3 axis in GC. Firstly, multiplex immunofluorescence analysis was performed on human LM of GC tissues to confirm the positional relationship of CNPY3/SLITRK4/ TrkB. It was found that they were highly overlapping in LM tissues (Fig. 7A). Subsequently, Co-IP experiments were performed to examine the endogenous CNPY3/SLI-TRK4/TrkB complex (Fig. 7B). Then, we knocked down SLITRK4 or CNPY3 in BGC823 and MKN28 cells and found that SLITRK4 or CNPY3 knockdown significantly attenuated the phosphorylation level of the TrkB protein and downstream signalling molecules such as PLCG 1 and PKCγ. Simultaneous knockdown of CNPY3 and SLITRK4 reduced the phosphorylation levels of TrkB, PLCG 1 and PKCγ to the greatest extent (Fig. 7C). Minseok Song et al. found that SLITRK5, as a coreceptor of TrkB, participates in endocytosis and TrkB recycling [28]. We next analysed whether the effect of SLITRK4 on TrkB is related to endocytosis and recycling. The rate of TrkB endocytosis in sh-NC and sh-SLI-TRK4 BGC823 cells was examined by precipitating biotinylated and internalized TrkB receptor from cell lysates with streptavidin beads and immunoblotting with an anti-TrkB antibody. The results showed that SLITRK4 knockdown led to significant restraint in the internalization of TrkB, while the level of recycling was also significantly diminished ( Fig. 7D-E). Then, we examined TrkB protein stability in sh-NC and sh-SLITRK4 BGC823 cells after the addition of cycloheximide (CHX) at different time points (Fig. 7F). Our results suggested that the degradation of TrkB was more quickly in sh-SLITRK4 BGC823 cells (Fig. 7G). These findings revealed that SLITRK4 played a unique role in the endocytosis and recirculation of TrkB and that the SLITRK4-CNPY3 axis enhanced the TrkB pathway by promoting TrkB receptor endocytic recycling. (Fig. 7H).

Discussion
Paget's 'seed and soil' hypothesis suggests that the development of metastases is largely dependent on the properties of seeds (cancer cells) and soil (microenvironment of target organs) [29]. Another pivotal discovery, the formation of premetastatic niches (PMNs), also provides an explanation for organotropic metastasis. The primary tumour can induce a supportive and receptive tissue microenvironment in distant organs before the colonization of tumour cells at those sites [30][31][32]. David Lyden and colleagues have demonstrated that exosomes from cancer cells can preferentially fuse with specific resident cells of target organs-lung, liver and brain-to establish the PMNs [33]. The liver is the primary metastatic site in gastric cancer; however, the mechanism of crosstalk between tumour cells and the microenvironment remains unknown. The coevolution of tumour and stromal cells in the tumour microenvironment (TME) has been a critical theory according to previous studies. The liver microenvironment provides PMNs for cancer cells to seed and survive after cross-communication between tumour cells and other relevant cells, such as cancer-associated fibroblasts and immune cells [34,35].
By exploring the biological function of SLITRK4 in GC with liver metastasis, we found an interacting protein, CNPY3. CNPY3 is a secreted protein that is mainly expressed in the ER and is also known as PRAT4A (protein associated with TLR4A) [36,37]. Faraz M et al. found that CNPY3 interacts with proteins of the LRR superfamily [38]. Then, we performed bioinformatics Fig. 5 The SLITRK4-CNPY3 axis promoted GC proliferation and migration. A. CNPY3 rescue experiment on sh-NC/sh-SLITRK4 BGC823 cells. B and C. Representative images of colony formation of sh-NC/sh-SLITRK4 BGC823 cells under the culture with supernatants of lenti-vector or lenti-CNPY3 (B). 500 BGC823 cells were seeded each well in six-well plates and cultured for 20 days. The number of colonies was counted and is shown as statistical data (C). Scale bar, 1 cm. D and E. Representative images of the invading sh-NC/sh-SLITRK4 BGC823 cells under the culture with supernatants of lenti-vector or lenti-CNPY3 (D). Scale bar, 50 μm. The number of invading cells was counted and is shown as statistical data(E). F and G. Representative images of colony formation of sh-NC/sh-SLITRK4 BGC823 cells under the stimulation of vehicle or rCNPY3 (F). 3,000 BGC823 cells were seeded each well in six-well plates and cultured for 7 days. The number of colonies was counted and is shown as statistical data (G). Scale bar, 1 cm. H and I. Representative images of the invading sh-NC/sh-SLITRK4 BGC823 cells under the stimulation of vehicle or rCNPY3(H). The number of invading cells was counted and is shown as statistical data(I). Scale bar, 50 μm. *p < 0.05. **p < 0.01. Data are presented as the mean ± S.D ◂ predictions and Co-IP to confirm the interaction between CNPY3 and SLITRK4. In addition, we found that CNPY3 was concentrated both in the tumour and in the TME, which probably helped establish premetastatic niches for the proliferation and metastasis of cancer cells. (Supplement Figure 1B-C). Our study showed that the proliferation and invasion abilities of GC cells were higher after they were cultured with the supernatant from CNPY3-overexpressing cells or adding with rCNPY3. However, this phenomenon was not observed in sh-SLITRK4 cells. We supposed that CNPY3, as a secreted protein, acts as a signalling molecule and combines with SLITRK4, which causes changes in the SLITRK4 downstream signalling pathway.
The intracellular region of SLITRK4 is the regulatory region for neurotrophic factors at the intracellular C-terminal domain, which has significant homology to tropomyosin-related kinases (Trks) [9]. TrkB signalling pathway functions in the proliferation, invasiveness and metastasis of various cancers [39][40][41][42][43][44]. Oestradiol can upregulate BDNF and then activate TrkB, which is expressed by cancer cells, promoting breast cancer brain metastases [45]. Current evidence suggests that TRK inhibitors are reliably effective in patients with positive neurotrophic tyrosine receptor kinase (NTRK) gene fusions [46][47][48][49][50]. In gastric cancer, some study suggested that cancer cells at the invasive front of primary tumours acquired proliferation ability and anoikis resistance via the BDNF/TrkB pathway [25,27]. Bongkun Choi found significant correlations of elevated BDNF/TrkB expression with bone metastatic potential in advanced-stage gastric cancers [26]. TrkB is also a well-studied target in the treatment of other solid tumour, blocking BDNF and TrkB can prevent CAF-driven acquired resistance to anlotinib in gastric cancer [25]. In our studies, we found that CNPY3/SLITRK4/TrkB complex may exist on LM of GC, and SLITRK4/CNPY3 knockdown influenced molecules downstream of TrkB, and thus, we focused on the crosstalk between SLITRK4 and TrkB.
Previous studies have suggested that SLITRK4 is closely related to the biological behaviour of endocytosis. Minseok Song reported that SLITRK5 can influence the signalling pathway downstream of the neurotrophin TrkB receptor and that SLITRK5 can act as a coreceptor for TrkB and is involved in the endocytosis and recycling of TrkB [28]. TrkB and SLITRK5 form a physical complex through their LRR domains, while SLITRK4 also contains LRR domains at the N-terminus that are similar to those of SLITRK5. Endocytosis plays key roles in a variety of processes, and endocytic trafficking is altered in cancer cells, which enhances tumour progression and metastasis. In colorectal cancer, the loss of Rab11 endosomes can accelerate tumorigenic signalling, causing tumour growth [51]. The endocytosis of TrkB is critical in the process of signal transduction, and internalized TrkB receptors can cause a series of cascade reactions, including tyrosine phosphorylation activation [52][53][54][55]. Once internalized, the activated receptor can be degraded via the lysosomal pathway or can be recycled. Our study demonstrated that interference with SLITRK4 expression can inhibit the internalization and recovery of TrkB in GC.

Conclusion
In our study, we proposed a CNPY3-SLITRK4/TrkB positive feedback axis that participates in GC metastasis. Our study may contribute to an increased understanding of GC progression, and targeting this pathway may be a promising therapeutic strategy to suppress liver metastasis from GC. Data availability The Gene Expression Omnibus (GEO) datasets involved in the study include GSE65801 and GSE30601. GSE65801 contains gene expression microarray analyses of 32 GC tissues and 32 paired noncancerous tissues, while GSE30601 contains genome-wide DNA methylation profiles of 203 GC tissues and 94 paired noncancerous tissues. Transcriptomic data from the PRJNA555812 project stored in the Sequence Read Archive (SRA) were used to analyse transcriptional changes in liver metastasis of gastric cancer. The Renji Cohort used in our study comprised 69 GC liver metastatic tissues collected from 2005 to 2011; data collected also included RNA sequencing analyses, tissue microarrays, and corresponding retrospective analyses.

Declarations
Ethics approval and consent to participate This study involved human participants and was approved by the Research Ethics Committee of Renji Hospital, School of Medicine, Shanghai Jiao Tong University (No. (2017)114.). This study involved animal subjects and was approved by the Research Ethics Committee of East China Normal University (ID: 2012-1204).

Competing interests
The authors declare no competing interests. Fig. 7 The SLITRK4-CNPY3 axis enhanced the TrkB pathway by promoting TrkB receptor endocytic recycling. A. Multiplex immunofluorescence analysis of co-localization of SLITRK4/CNPY3/TrkB on human LM tissue of GC. (SLITRK4, green; TrkB, red; CNPY3, violet; DAPI, blue) Scale bar, 100 μm(middle) or 200 μm(upper and lower). B. Co-IP of SLITRK4/CNPY3/TrkB. Lysates of cells were subjected to immunoprecipitation with anti-SLITRK4, anti-CNPY3 or anti-TrkB antibody. C. Western blotting analysis of the effect of SLITRK4/CNPY3 knockdown on TrkB pathway activity in BGC823 and MKN28 cells. D. TrkB internalization and recycling levels were measured by cleavable surface biotinylation in sh-NC or sh-SLITRK4 BGC823 cells. The amount of surface biotinylated TrkB that was internalized or recycled was measured by western blot for TrkB after streptavidin pulldown. E. Quantification of internalization and recycling of TrkB through densitometric analysis of bands from western blots using Image J software. F. Sh-NC and sh-SLITRK4 BGC823 cells were harvested at different times after Cycloheximide (CHX) addition, and the TrkB protein level was detected by western blotting. G. The TrkB protein level in the cells collected at 0, 1, 2, 4, 6 and 8 h after CHX addition. Linear regression analyses revealed the degradation of TrkB in CHX-treated cells. H. SLITRK4 acts as a coreceptor of TrkB, and the interaction between SLITRK4 and CNPY3 can regulate the endocytic recycling of TrkB in GC ◂