Exosomal ANGPTL1 Attenuates CRC Liver Metastasis by Regulating Kupffer Cell Secretion Pattern and Impeding MMP9 induced vascular leakiness.

Background: Angiopoietin-like protein 1 (ANGPTL1) has been proved to suppress tumor metastasis in several cancers. However, its extracellular effects on the pre-metastatic niches (PMNs) are still unclear. ANGPTL1 has been identied in exosomes, while its function remains unknown. This study was designed to explore the role of exosomal ANGPTL1 on liver metastasis in colorectal cancer (CRC). Methods: Exosomes were isolated by ultracentrifugation. The ANGPTL1 level was detected in exosomes derived from human CRC tissues. The effects of exosomal ANGPTL1 on CRC liver metastasis were explored by the intrasplenic injection mouse model. The liver PMN was examined by vascular permeability assays. Exosomal ANGPTL1 localization was validated by exosome labeling. The regulatory mechanisms of exosomal ANGPTL1 on Kupffer cells were determined by RNA sequencing. qRT-PCR, Western Blot, and ELISA analysis were conducted to examine gene expressions at mRNA and protein levels. Results: ANGPTL1 protein level was signicantly downregulated in the exosomes derived from CRC tumors compared with paired normal tissues. Besides, exosomal ANGPTL1 attenuated liver metastasis and impeded vascular leakiness in the liver PMN. Moreover, exosomal ANGPTL1 were mainly taken up by KCs and regulated the KCs secretion pattern, especially decreasing the MMP9 expression, which nally prevented the liver vascular leakiness. In mechanism, exosomal ANGPTL1 downregulated MMP9 level in KCs by inhibiting the JAK2-STAT3 signaling pathway. Conclusions: Taken together, exosomal ANGPTL1 attenuated CRC liver metastasis and impeded vascular leakiness in the liver PMN by reprogramming the Kupffer cell and decreasing the MMP9 expression. This study suggests a suppression role of exosomal ANGPTL1 on CRC liver metastasis and expands the approach of ANGPTL1 functioning. the protein level of ANGPTL1 in exosomes derived from tumor and normal tissues in CRC patients. Both In vivo and in vitro models were applied to characterize the effects of exosomal ANGPTL1 on CRC liver metastasis and PMNs formation. We demonstrated that exosomal ANGPTL1 attenuates CRC liver metastasis by regulating the kupffer cell secretion pattern and impeding vascular leakiness in the liver PMN.


Background
The prevalence and mortality of colorectal cancer (CRC) are really high worldwide [1]. Distant metastasis is the leading cause of cancer-associated death of CRC [2]. The 5-year survival rate is 92% for patients with local disease, while it sharply declines to 53% and 11% for patients with regional and distant metastasis [3]. The liver is the most common site of CRC metastasis. It is reported that ≤ 25% of CRC patients have synchronous colorectal liver metastases (CLM) upon diagnosis, and up to half of CRCs will eventually lead to liver metastasis [4]. Despite advances in surgical technique and targeted therapy, the prognosis for CLM patients is still poor. It is urgent to explore the mechanism of liver metastasis and seek a new strategy for CLM treatment.
In our previous study [5], we found some of the angiopoietin-like proteins (ANGPTLs) were downregulated in CRC tissue, among which ANGPTL1 was the most signi cant one. ANGPTLs are a family of proteins that are similar to angiopoietins in structure, including ANGPTL1 to ANGPTL8 [6]. Besides angiogenesis regulation, they were also reported to affect in ammation [7], metabolism disorders [8], hematopoiesis [9], and cancer development [10,11]. Early studies showed that Angiopoietin-like protein 1 (ANGPTL1) acts as an anti-angiogenic factor and a tumor suppressor [12,13]. ANGPTL1 is downregulated in various cancers [5,6], and several studies have proved its inhibitory role in tumor growth and metastasis [14][15][16][17].
Our previous study also demonstrated that ANGPTL1 overexpression inhibited the migration and invasion of CRC cells, leading liver metastasis suppression. Besides, low expression of ANGPTL1 was related to poorer prognosis in CRC patients [5].
Nevertheless, previous researches about ANGPTL1's function limited in primary tumors [5,15,17]. As a secretory protein [13], the biological effects of extracellular ANGPTL1 on the metastatic organs are still under investigation. Increasing evidence shows that the primary tumor-secreted factors and exosomes can enhance metastasis by promoting a supportive microenvironment in the metastatic organs, named the pre-metastatic niches (PMNs), such as vascular leakiness, in ammation, and immunosuppression [18]. Exosomes are small extracellular vesicles ranging from 50 to160 nm in size that carrying proteins, nucleic acids, and lipids [19]. Recently, tumor-derived exosomes have been reported to be involved in PMNs formation [20,21]. As the earliest event during PMNs evolution, vascular permeability is always regulated by tumor-derived exosomes [22]. For instance, CRC-derived exosomal miR-25-3p can promote PMNs formation by inducing vascular permeability [23]; breast cancer-derived exosomal miR-105 can destroy vascular endothelial barriers to promote metastasis [24]. Interestingly, ANGPTL1 has been identi ed in exosomes derived from saliva [25], urine [26], and ovarian cancer cells [27]. But the function of exosomal ANGPTL1 in CRC is still unknown.
In this study, we focused on the role of exosomal ANGPTL1 in CRC metastasis. We studied the protein level of ANGPTL1 in exosomes derived from tumor and normal tissues in CRC patients. Both In vivo and in vitro models were applied to characterize the effects of exosomal ANGPTL1 on CRC liver metastasis and PMNs formation. We demonstrated that exosomal ANGPTL1 attenuates CRC liver metastasis by regulating the kupffer cell secretion pattern and impeding vascular leakiness in the liver PMN.

Patients and specimens
The clinical CRC and paired normal tissues were obtained from CRC patients (n = 8) in the Second A liated Hospital of Zhejiang University School of Medicine. This project was approved by the ethical committee of the Second A liated Hospital of Zhejiang University School of Medicine. Informed consent was obtained from all patients.

Exosome collection and characterization
Exosomes were collected by sequential ultracentrifugation. CRC cells were cultured in the exosomedepleted (160,000 × g, 16 hours) complete medium for 72 hours. The supernatants were collected and centrifugate at 500 × g for 10 min to remove cell contamination, then at 3,000 × g for 20 min to remove apoptotic bodies and large cell debris, followed by centrifugation at 12,000 × g for 20 min to remove large microvesicles. Next, exosomes were collected by 100,000 × g centrifugation for 70 min (Beckman Ti70). The exosome pellet was resuspended in 20 mL of phosphate-buffered saline (PBS) and collected by 100,000 × g ultracentrifugation for 70 min (Beckman Ti70). Exosome preparation was veri ed by Transmission electron microscopy (TEM). Exosome size was measured by dynamic light scattering (DLS) analysis using Zetasizer Nano ZSE (Malvern Panalytical, Shanghai, China).
For tissue-derived exosome collection, the CRC tumors and paired normal tissues were cut into 1 mm × 1 mm pieces, and cultured in 15 mL of FBS-free RMPI-1640 medium for 24 hours. Then the supernatant was harvested for further isolation of exosomes.

Exosome treatment and labeling
Puri ed exosomes were injected into the mouse retro-orbital venous sinus in a total volume of 100 µL PBS. For in vivo education experiments, mice received 5 µg of exosomes every other day for 21 days. Retro-orbital injection of PBS was used in control groups. For in vitro education, exosomes (10 µg/mL) were added into the culture medium of ImKC for 3, 6, 12, 24 hours. For exosome-tracking experiments, exosomes were labeled using PKH67 membrane dye (Sigma, Shanghai, China), followed by 100,000 × g ultracentrifugation for 70 min and labeled exosomes were resuspended in PBS. In experiments involving the evaluation of exosome incorporation, labeled exosomes were injected retro-orbitally into the mice or added into the culture medium (CM) of ImKC 24 hours before immuno uorescence analysis for exosome cells.

Animal model
To analyze the role of exosomal ANGPTL1 in CRC liver metastasis, 6-8-week-old scid-beige mice (SLAC Laboratory Animal Co. Ltd., Shanghai, China) were pre-educated with exosomes for 21 days (every other day). Then, 2 × 10 6 SW620 cells were injected into the mice spleen as described in the previous study [28,29]. A small animal IVIS Lumina Imaging System (Caliper Life Sciences, Hopkinton, MA) was used for liver lesion monitor. All mice were sacri ced at 1 month, and the livers were harvested for Hematoxylin and eosin (H&E) staining and analysis of metastases. For liver macrophages elimination, liposome clodronate was injected via tail vein in a dose of 0.2 mL/20-25 g as a tool to suppress macrophage function by inducing apoptosis [30]. Liposomes containing PBS were injected as a control. All animal experiments were approved by the Institutional Ethics Committee of the Second A liated Hospital Zhejiang University School of Medicine.

In vivo vascular permeability assay
After pre-education with exosomes for 21 days, mice were injected with the recombinant mouse MMP9 (rmMMP9; 50 ug/kg body weight; R&D) intravenously. The rmMMP9 was preactivated using 1 mM aminophenylmercuricacetate (AMPA, Sigma, Shanghai, China) for 2 hours at 37 °C. One hour after rmMMP9 injection, FITC-Dextran (~ 70KD; 100 mg/kg; Sigma, Shanghai, China) was injected through the tail vein. After one hour, mice were sacri ced and perfused with PBS to remove the excess dye. Liver tissues were embedded in Tissue-Tek O.C.T. Compound (Sakura; Torrance, CA, USA) to make frozen blocks for sectioning and immuno uorescent staining. Stained sections were viewed and photographed with a uorescence microscope. The intensity of uorescence was measured using ImageJ software (ImageJ software v1.8.0) Endothelial permeability HUVECs (2 × 10 4 ) were seeded on transwell lters (0.4 µm pore size; Corning, Shanghai, China). After reaching con uence, HUVECs were treated with CM from ImKC educated by PBS, Ctrl-Exo or ANGPTL1-Exo (with or without rmMMP9, 100 ng/mL) for 48 hours. Then, FITC-Dextran (1 mg/mL) was added to the top well. 40µL medium in the bottom well was taken for uorescence measurement every 30 minutes using a SpectraMax microplate reader (SpectraMax i3, Molecular Devices, USA) at 488 nm excitation and 520 nm emission. The uorescence intensity represents the passage of FITC-Dextran, which indicates the permeability of HUVECs layer.

qRT-PCR
Total RNAs were isolated from cells and mouse livers with TRIzol reagent (Invitrogen, USA) and evaluated using NanoDrop 2000 spectrophotometer (Thermo Scienti c, Pittsburgh, PA, USA). qRT-PCR was conducted using a standard SYBR-Green PCR kit protocol (YEASEN, Shanghai, China) with a 7500 Fast Real-Time PCR System (Life Technologies, Shanghai, China). The primers were synthesized by Tsingke biological technology (Hangzhou, China). The sequences of all primers are listed in Table 1.

RNA sequencing
Total RNA of ImKC cells educated by PBS or exosomes for 24 hours in vitro was extracted and subjected to BGI (Huada Genomics Institute Co. Ltd, Guangzhou, China) for RNA sequencing. The DEGseq R package was used to analyze differentially expressed genes based on the conditions of a fold change (FC) ≥ 1 and Q-values ≤ 0.001. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed to explore the signi cant pathways. All mRNA sequencing data were uploaded to NCBI (accession number PRJNA656088).

Western blot and ELISA
The tissue lysates were prepared using RIPA lysis buffer, while the cell and exosome lysates were prepared using 3 × Loading Buffer. All samples were performed as depicted previously [5]. Exosome proteins were detected with the antibodies as follows: Alix  (1:1000, Proteintech Group, Rosemont, USA). All samples were further incubated with the peroxidaseconjugated secondary antibody (1:5000, Huabio, Hangzhou, China). Bands were visualized using enhanced chemiluminescence reagents (YEASEN, Shanghai, China) and scanned via a Tanon 5200 Chemiluminescent Imaging System (Tanon, Shanghai, China). Quantitative analysis of WB was performed with ImageJ software (ImageJ v1.8.0).
In order to detect the protein level in the culture medium of ImKC through ELISA, cells were cultured in the presence of exosomes (10 µg/mL) for 24 hours before supernatants were collected and the MMP9 level was measured by mouse MMP9 ELISA kit (R&D Systems, Minneapolis, MN, USA).

Immuno uorescence
For histological analysis, tissues were dissected and xed in a mix of 2% Paraformaldehyde (PFA) and 20% sucrose solution overnight, followed by embedding and section. For ZO-1 detection in HUVECs, cell immuno uorescence was conducted as described previously [31].
Cells were incubated with primary antibody ZO-1 (1:400, Proteintech Group, Rosemont, USA) at 4 °C overnight. Nuclei were stained with DAPI for 5 min. Cells were viewed under the Zeiss LSM 710 laser confocal microscope (Carl Zeiss, Germany) and analyzed using Zen software (ZEN 3.0).

Statistics
All data are presented as the mean ± standard error of the mean (SEM) or mean ± standard deviation of the mean (SD). Statistical analyses were performed with Student's t-test for comparisons between two groups and with ANOVA for more than two groups using Prism 8.0. P < 0.05 was considered statistically signi cant.

ANGPTL1 level was downregulated in CRC derived exosomes.
To study the ANGPTL1 expression in CRC-derived exosomes, we rst detected the ANGPTL1 protein level in exosomes secreted by CRC tumors (TDEs) and paired normal tissues (NDEs). We collected 8 pairs of samples from our oncology center and isolated the tissue-derived exosomes by ultracentrifugation. The characteristic of TDEs and NDEs were accessed by TEM, DLS, and Western Blot (WB) (Fig. 1a-1c). We observed a cup-shaped structure in both TDEs and NDEs, with the diameter of most particles around 150 nm. WB analysis of the exosomal proteins showed that ANGPTL1 level was signi cantly decreased in TDEs than in paired NDEs (P = 0.03, Fig. 1d and 1e, see the detailed ANGPTL1 index of each sample in additional le 4 Table S1). What's more, the ANGPTL1 level in TDEs varied among CRC patients, indicating the patient heterogeneity of exosomal ANGPTL1 levels. The above results suggested that ANGPTL1 was downregulated in CRC derived exosomes, and exosomal ANGPTL1 may be involved in CRC progression.
In our previous study [5], ANGPTL1 overexpression inhibited the migration and invasion of CRC cell and hindered liver metastasis. To further investigate the function of exosomal ANGPTL1 in CRC, SW620-ANGPTL1 cells were used to enrich the exosomal ANGPTL1. Exosomes were collected from the CM of SW620 (named Ctrl-Exo) and SW620-ANGPTL1 (named ANGPTL1-Exo) through ultracentrifugation (Additional le 1 Fig. S1a and S1b). WB analysis con rmed that ANGPTL1 was abundant in ANGPTL1-Exo (Additional le 1 Fig. S1c).
We then employed an intrasplenic injection mouse model to investigate the role of exosomal ANGPTL1 in liver metastasis. Firstly, SCID-Beige mice were injected retro-orbitally every other day for 21 days with PBS, Ctrl-Exo, or ANGPTL1-Exo in a process that de ned as "education" (Fig. 2a). After that, 2 ⋅ 10 6 SW620 cells were injected into the mouse spleens. The IVIS Lumina imaging system was used to monitor the liver metastasis weekly. At the fourth week, we found the uorescence intensity in mouse livers was lower in the ANGPTL1-Exo education group, compared with the Ctrl-Exo education group (P = 0.02, Fig. 2b). Then mice were sacri ced and their livers were harvested for H&E examination (Fig. 2c). The liver macrometastatic burden was measured by lesion areas. We found a heavier metastatic burden in mouse livers educated by Ctrl-Exo than by PBS or ANGPTL1-Exo (PBS vs Ctrl-Exo, P < 0.01; Ctrl-Exo vs ANGPTL1-Exo, P < 0.01; Fig. 2c). The results suggested that exosomal ANGPTL1 attenuated liver metastasis induced by tumor-derived exosomes in CRC.
It is known that cancer cells spread through the body in multiple steps [32]. Our intrasplenic mouse model allowed tumor cells to disseminate through the portal circulation immediately after injection [29], making tumor extravasation the rst step during liver metastasis. Vascular permeability is one of the most critical factors in this process, as well as a typical characteristic of PMNs [33]. Furthermore, ANGPTL1 was proved to regulate angiogenesis [6]. Thus, we focused on exosomal ANGPTL1's function on liver vascular permeability. The effect of Ctrl-Exo and ANGPTL1-Exo on endothelial barriers was then detected in vivo. Mice educated by PBS or exosomes were injected with FITC-Dextran through tail veins. More uorescence was detected in the liver educated by Ctrl-Exo than by PBS, while less uorescence was detected in the liver educated by ANGPTL1-Exo (P < 0.001, Fig. 2d), which indicated that exosomal ANGPTL1 could impede the vascular leakiness in liver PMNs induced by tumor-derived exosomes.
3. Exosomal ANGPTL1 was mainly taken up by KCs and regulated KCs secretion pattern.
To explore the mechanism of how exosomal ANGPTL1 inhibits CRC liver metastasis, we rstly try to determine the cells that take up CRC derived exosomes in liver. The PKH67 labeled Ctrl-Exo and ANGPTL1-Exo were injected intravenously. Twenty-four hours post injection, mouse livers were harvested and liver frozen sections were analyzed by immuno uorescence. We found the cells which took up exosomes were mainly F4/80 positive, a surface marker of KCs (Fig. 3a). Besides, the labeled exosomes failed to fuse with other cells in the liver microenvironment, such as αSMA + hepatic stellate cells or CD31 + endothelial cells (Additional le 2 Fig. S2a and S2b). To validate the KC-speci c localization of exosomes, we treated the mice with liposome clodronate, known to deplete macrophages, by via tail vein injection.
We found that 48 hours treatment was enough to ablated most of the F4/80 + cells in the mouse liver (Additional le 2 Fig. S2c). Then, the PKH67 labeled exosomes were injected into the mice. We found there was no exosome remained in livers after macrophage ablation (Fig. 3b), indicating that KC is the predominant cell taking up CRC derived exosomes. In addition, we also examined the exosome uptake of KCs in vitro. PKH67 labeled Ctrl-Exo and ANGPTL1-Exo were co-cultured with ImKC and were both taken by ImKC in 2 hours (Additional le 2 Fig. S2d) and stably fused into ImKC in 4 hours (Fig. 3c).
The above results implied that exosomal ANGPTL1 may inhibit CRC liver metastasis through regulating KCs. To further investigate the effect of exosomal ANGPTL1 on KCs, we educated ImKC with PBS, Ctrl-Exo, and ANGPTL1-Exo in vitro for 24 hours and analyzed gene expression by mRNA sequencing. GO enrichment analysis showed that the changing genes between ImKC educated by Ctrl-Exo and ANGPTL1-Exo mainly belonged to the extracellular region (Fig. 3d), which suggested that the secretion pattern of KCs was regulated. Among the top20 genes (Fig. 3e), Mmp9, Lif, Cxcl2, Csf3, Il1a, and Ccl5 were veri ed by qRT-PCR (Fig. 3f, P < 0.05). The six genes were signi cantly downregulated in ImKC educated by ANGPTL1-Exo, and all of them were reported to be involved in PMNs formation. Taken together, exosomal ANGPTL1 was mainly taken up by KCs and regulated KCs secretion pattern, which may result in liver PMNs remodeling.

Exosomal ANGPTL1 dependent MMP9 decrease in KCs normalized vascular leakiness induced by CRC derived exosomes.
Matrix Metallopeptidase 9 (MMP9) is intimately involved in regulating vascular integrity in PMNs [22]. To explore the effect of exosomal ANGPTL1 dependent MMP9 decrease on vascular leakiness, we rst veri ed the MMP9 downregulation in KCs. The ELISA analysis showed that MMP9 was signi cantly decreased in the CM of ImKC educated by ANGPTL1-Exo than by Ctrl-Exo (Fig. 4a, P < 0.05). We also con rmed MMP9 decrease in ANGPTL1-Exo educated mouse livers (Additional le 3 Fig. S3). Next, we accessed the permeability of endothelial monolayer after treatment with the CM of ImKC by measuring the passage of FITC-Dextran (70 KD). We found that the CM of ANGPTL1-Exo educated ImKC signi cantly decreased uorescent probe leakiness as compared to the CM of Ctrl-Exo educated ImKC (Fig. 4b). Active rmMMP9 was added into the CM of ImKC educated by ANGPTL1-Exo and resulted in more uorescent probe passing through the endothelia cell (EC) layer (Fig. 4c). The same tendency was observed in the in vivo vascular permeability assay and rmMMP9 treatment deteriorated liver vascular leakiness in the mice educated by ANGPTL1-Exo (P < 0.001, Fig. 4d).
Since the tight junction proteins (TJs) were proved to regulate the EC layer permeability [34], we examined the TJs levels in ECs. The results showed the ZO-1 and Claudin-5 was upregulated in HUVECs layers under the treatment with CM of ImKC educated by ANGPTL1-Exo than by Ctrl-Exo, while it abolished by rmMMP9 ( Fig. 4e and 4f). Immuno uorescence analysis also revealed that rmMMP9 dampened the ZO-1 upregulation in HUVEC monolayers after treatment with CM of ANGPTL1-Exo educated ImKC (Fig. 4g). All the above data above indicated that MMP9 was involved in exosomal ANGPTL1 dependent vascular leakiness prevention.

Exosomal ANGPTL1 downregulated MMP9 in KCs by inhibiting the JAK2-STAT3 signaling pathway.
To further investigate how exosomal ANGPTL1 downregulated MMP9 expression in KCs, we conducted KEGG enrichment to analyze the RNA sequencing data. The results showed that the changing genes between ImKC educated by Ctrl-Exo and ANGPTL1-Exo were mainly enriched in IL-17, TNF, Toll-like receptor, and JAK-STAT signaling pathway (Fig. 5a). Since ANGPTL1 was reported to inhibit the JAK2-STAT3 pathway [17], we focused on the effect of exosomal ANGPTL1 on the JAK2-STAT3 pathway. WB analysis showed that ANGPTL1-Exo obviously inhibited the phosphorylation of STAT3 (Y705) and JAK2 (Y1008) in KCs at 6, 12, 24 hours ( Fig. 5b and 5c). MMP9 is known as one of the target genes of STAT3 [35]. To con rm if the JAK2-STAT3 signaling pathway involved in exosomal ANGPTL1 induced MMP9 downregulation in KCs, the recombinant mouse IL-6 was used to active ImKC, which is a speci c activator of STAT3. The WB results showed that IL-6 activated the phosphorylation of STAT3 (Y705) and increased the MMP9 mRNA expression in ImKC educated by ANGTPL1-Exo in 24 hours (Fig. 5d). The above data demonstrated that exosomal ANGPTL1 downregulated MMP9 in KCs by inhibiting the JAK2-STAT3 pathway, which may be the mechanism of exosomal ANGPTL1 dependent vascular leakiness prevention and liver metastasis attenuation. Discussion ANGPTL1 has been reported to suppress tumor metastasis in several cancers [6], while its extracellular effects on the PMNs are still unclear. Our present study was intended to determine the function of exosomal ANGPTL1 in CRC liver metastasis.
In this study, we found that ANGPTL1 expression was downregulated in the exosomes derived from CRC tumor tissues than paired normal colorectal tissues. The exosomes containing more ANGPTL1 proteins attenuated liver metastasis and impeded vascular leakiness. Further exploration showed exosomal ANGPTL1 regulated the KCs secretion pattern, especially decreased the MMP9 expression by inhibiting the JAK2-STAT3 pathway, which in turn normalized vascular leakiness in live PMN (Fig. 5e).
An increasing number of studies have found that ANGPTL1 could be secreted in exosomes. Sinha A et al. identi ed ANGPTL1 in human ovarian cancer cell-derived exosomes by mass spectrometry [27]. ANGPTL1 was also detected in exosomes derived from human saliva [25] and urine [26]. However, the role of exosomal ANGPTL1 is unknown. Our study demonstrated for the rst time that ANGPTL1 expression was downregulated in exosomes derived from CRC tumors than paired normal tissues. It was similar to its differential expression in CRC tumors and normal tissues [5,36], suggesting a possible suppression role of exosomal ANGPTL1 on CRC progression.
Our functional experiments in mouse models of CRC proved that the exosomal ANGPTL1 upregulation inhibited liver metastasis. Consistently, early studies have proved that ANGPTL1 hindered tumor metastasis in lung cancer [15], hepatocellular carcinoma (HCC) [17], and CRC [5]. But the mechanism varies. Actually, cancer metastasis is a multi-step process [32]. So far researches mostly focus on the impact of ANGPTL1 on the primary sites, such as tumor invasiveness and mobility inhibition, which prevent tumor cells from invading nearby tissues and moving through the vascular walls [14,15,17]. However, we paid our attention to the metastatic organs and found exosomal ANGPTL1 impeded liver vascular leakiness induced by CRC derived exosomes. Growing evidence indicates that exosomes serve as mediators for long-distance cell-to-cell communication and play a pivotal part in PMNs formation [19].
Several studies have showed increased vascular permeability at PMNs, including liver PMN, which is associated with an enhanced metastatic burden [33]. For example, the melanoma-derived exosomes increased the metastatic behavior of primary tumors by inducing vascular leakiness at pre-metastatic sites [37]; breast cancer-derived exosomes destroyed vascular endothelial barriers to promote metastasis [38]. Thus, it may be reasonable to imply that exosomal ANGPTL1 attenuated CRC liver metastasis through preventing liver vascular leakiness. But whether exosomal ANGPTL1 affected vascular permeability through a direct effect on the endothelial cells or other indirect way is still unknown.
In our study, we found that KC was the predominant cell that took up CRC-derived exosomes in the liver, which was also approved by Shao Yingkuan et al. [39]. It indicated that exosomal ANGPTL1 possibly remodeled the liver PMNs through KCs. Several studies have correlated KCs with PMNs formation and liver metastasis [40,41]. Once reprogrammed by tumor-derived exosomes, KCs exert their regurgitationfeeding activity on liver microenvironments via the secretion of cytokines and chemokines [39,42]. We detected an evident variation of the secretion pro le of factors from KCs upon exposure to exosomal ANGPTL1, which in uenced the liver PMN formation. Among these changing factors, we attributed MMP9 to exosomal ANGPTL1 dependent vascular leakiness prevention. ANGPTLs were reported to regulate the expression of MMP9 in osteosarcoma [43] and HCC [44]. And MMP9 is known to be involved in regulating vascular integrity in PMNs [22]. High levels of MMP9 in pre-metastatic lung promoted vascular remodeling, while genetic ablation of MMP9 normalized the aberrant vasculature in the lung PMN, impeding cancer metastasis [45]. The above evidence suggested that exosomal ANGPTL1 reprogranmmed KC and downregulated its MMP9 expression, thus preventing liver vascular leakiness and hindering CRC liver metastasis.
Exosomal cargos, like proteins or microRNAs, can regulate the recipient cells' physiological activities [46]. We demonstrated that exosomal ANGPTL1 inhibited the JAK2-STAT3 signal pathway, especially hindered STAT3 activation. The exosomal ANGPTL1 dependent MMP9 downregulation was reversed by IL-6 induced STAT3 activation. It indicated that exosomal ANGPTL1 downregulated the MMP9 expression by inhibiting the JAK2-STAT3 signaling pathway. Consistently, Qian Yan et al. also found ANGPTL1 repressed the JAK-STAT3 signaling in HCC [17]. There is substantial evidence con rming the JAK-STAT3 signaling pathway is involved in CRC development [47]. STAT3 (Signal transducer and activator of transcription 3) is a critical transcriptional factor that has been identi ed as a central regulator of tumor metastasis. STAT3 could be activated by IL-6 and promote gene transcription including MMP9 [35].
Therefore, we implied that exosomal ANGPTL1 may in uence liver vascular permeability through the JAK2-STAT3-MMP9 axis.
However, how exosomal ANGPTL1 inhibited the JAK2-STAT3 pathway still needs further investigation. The way that exosomes transfer the content or induce signals may involve ligand-receptor interaction or cytomembrane fusion [19]. Several studies have shown that ANGPTL proteins deliver their signals via integrin receptor-related pathways [15,17]. The previous study proved ANGPTL1 interacted with the integrin α1β1 receptor to suppress the downstream FAK/Src-JAK-STAT3 signaling pathway [17]. Besides, ANGPTL1 worked intracellular to represses the Src-JAK-STAT3 signaling [17]. Therefore, we implied that the possible mechanism of exosomal ANGPTL1 inhibiting JAK2-STAT3 pathway may be: (1) ANGPTL1 exists on the exosome membrane and binds the integrin receptors on the KCs membrane, suppressing the downstream JAK2-STAT3 signaling pathway; (2) exosomes fuse with the KCs membrane and ANGPTL1 is released into the cytoplasm, inhibiting the JAK2-STAT3 pathway through interaction with molecules involved in the pathway. These questions will be explored in our future research.

Conclusion
This study aimed to explore the function of exosomal ANGPTL1 in CRC liver metastasis. Our experimental results demonstrated that ANGPTL1 was downregulated in CRC derived exosomes. More importantly, exosomal ANGPTL1 attenuated CRC liver metastasis and impeded vascular leakiness. In mechanism, we found exosomal reprogrammed the Kupffer cell and decreased MMP9 expression through inhibiting the JAK2-STAT3 signaling pathway. These ndings suggest a suppression role of exosomal ANGPTL1 on CRC progression and expand the approach of ANGPTL1 functioning, enriching the mechanisms of CRC liver metastasis.   were analyzed using Mann-Whitney test. *P < 0.05, **P < 0.01, ****P < 0.0001. Scale bar represents 20um. Data are presented as the mean ± SEM of three independent experiments, and analyzed using t-test. *P < 0.05, **P < 0.01, ****P < 0.0001.

Figure 4
Exosomal ANGPTL1 dependent MMP9 decrease normalized vascular leakiness induced by CRC derived exosomes. a ELISA analysis of MMP9 level in the culture medium from IMKC educated by PBS, Ctrl-Exo or ANGPTL1-Exo for 24 hours. b, c Permeability of the HUVEC monolayers to FITC-Dextran (70 kDa) after exposure to CM from IMKC educated by PBS, Ctrl-Exo, ANGPTL1-Exo (b), or exposure to CM from IMKC