Metabolic Reprogramming Promotes Metastasis of Acidosis-Adapted Pancreatic Ductal Adenocarcinoma Cells Via AMPK/YAP/MMP1 Axis

We generated acidosis-adapted PDAC cells (PANC-1 and SW1990) by continuously culturing under pH 6.5 medium. Metabonomic assay (LC-MS and LC-MS) was used to detect the metabolites. Transcriptomic microarray was applied to reveal the gene expression prole. The in vivo metastatic potential was assessed by tail vein injection. Wound healing assay and transwell assay were applied to evaluate cell migration and invasion capacity. CCK8 and colony formation were performed to determine cell proliferation. Flow cytometry was used to test cell cycle and apoptosis.


Results
These acidosis-adapted PDAC cells had stronger invasion and migration ability, accompanied by increased EMT markers expression. In addition, the proliferation rate of acidosis-adapted PDAC cells increased while apoptosis decreased under acidic medium. Metabonomic analysis showed that the glycolysis and AMP was decreased, and ATP was increased in acidosis-adapted cells. The metabolic shift led to the inactivation of AMPK. Transcriptomic analysis revealed that the differentially expressed genes in acid-adapted cells were enriched in extracellular matrix modi cation, Hippo signaling and others. MMP1, a matrix metalloproteinase, was the most upregulated gene in acidosis-adapted cells, mediated by the YAP/TAZ pathway. Knockdown of MMP1 signi cantly reduced invasion and metastasis capacity of acidosis-adapted cells. The AMPK activator MK8722 could reduce both YAP and MMP1 level in acidosis-adapted PDAC cells.

Conclusion
Altogether, the present study showed that the metabolic reprogramming towards elevated ATP production promotes invasion and metastasis of acidosis-adapted PDAC cells via depressing AMPK/Hippo signaling, thus upregulating MMP1.

Background
To adapt to the adverse living environment caused by the unlimited proliferation, the metabolic mode of tumor cells presents a background dependent exibility 1 . Tumor cells tend to aerobic glycolysis to meet rapid proliferation, which generates a large amount of lactic acid. In addition, proton and carbon dioxide are respectively produced in the process of ATP consumption and oxidative phosphorylation. All these acidic materials are expelled into the extracellular microenvironment and retained in this space due to the hypoperfusion, leading to pH value of tumor microenvironment decreased to 6.4-7, namely acidosis 2 .
Acidosis is a selective pressure for tumor cells, and the one that survive have additional survival advantages, such as enhanced proliferation, metastasis and drug resistance 3 . Hepatocellular carcinoma cells proliferation and metastasis increased under acidosis condition explained well for this phenomenon 4 . Furthermore, Faes S et al. revealed that the acidic tumor microenvironment made mTORC1 inhibitors ineffective 5 . Moreover, lower PH value promotes malignant glioma 6 , breast cancer 7 , colon cancer 8 , lung cancer 9 progression. Our previous work found that the acid sensing ion channel ASIC2 promotes colorectal cancer cell invasion under acidosis 10 , and acid-adapted colon cancer cells have stronger metastasis capacity 11 . As the widely recognized pro-tumor effect, acidosis has been applied as a marker for prognosis prediction 12 and treatment target 13 . However, the underlying mechanism of acidosis should be further demonstrated, especially for chronic acidosis.
The acidic microenvironment can also in turn induce a metabolism shift. Minnan Zhao and colleagues found that acidosis-adapted colorectal cancer cells (HCT15, HCT116 and LoVo) maintain a higher level of reduced glutathione to reduce the acid-induced reactive oxygen species via the upregulation of CD44 and glutathione reductase (GSR) 14 . Additionally, the acidosis-adapted colorectal cancer cell (HCT116), cervical cancer cell (SiHa) and pharynx squamous cell carcinoma cell (FaDu) all exhibited a turnover from glycolysis to glutamine metabolism, which was mediated by HIF2α 15 . The same research group further revealed that acidic microenvironment profoundly reprograms the metabolism of these cancer cells toward fatty acid oxidation by inducing histone deacetylation and downregulating acetyl-CoA carboxylase ACC2 16,17 . As metabonomics technology developed, it can reveal signi cant metabolic changes under acute or chronic acidosis. Recently, it was reported that acute acidic treatment (pH 6.8 for 36 hours) increased the de novo purine nucleotide biosynthesis activity in glioma stem cells by upregulating glucose-6-phosphate dehydrogenase G6PD or H6PD expression 18 . AMP-activated protein kinase (AMPK) is an energy sensor that monitors the AMP:ADP:ATP ratio in eukaryotic cells. It has generally been considered a tumor suppressor as it is a putative substrate of tumor suppressor, LKB1 19 .
AMPK phosphorylates angiomotin like 1 (AMOTL1), an adaptor protein in the Hippo-Yap pathway, and thus blocks Yes1 associated transcriptional regulator (YAP) activity, dampening cancer cells proliferation and survival 20 . Further work should be done to reveal the metabolic reprogramming under chronic acidosis and its correlation with tumor progression, especially in the context of pancreatic ductal adenocarcinoma (PDAC), one of the most lethal malignancies 21 .
In this study, we generated the acidosis-adapted PDAC cells by continuously culturing under pH 6.5 medium treatment for 3 months. We found that cells proliferation and metastasis ability increased, and MMP1 expression was profoundly upregulated. Furthermore, glycolysis was signi cantly inhibited, and intracellular ATP level and other metabolites increased in acidosis-adapted PDAC cells. Inhibition of AMPK signaling by increased ATP level then increased the YAP and MMP1 expression, promoting metastasis. Our study demonstrated that metabolic reprogramming promotes metastasis of acidosisadapted PDAC cells via AMPK/YAP/MMP1 axis, enriching the mechanism by which acidic environment promotes tumor progression.

Methods
Cell lines and cell culture Human pancreatic cancer cell lines PANC-1 and SW1990 were obtained from the American Type Culture Collection (Manassas, VA, USA), all the cells were identi ed by STR and without mycoplasma contamination. Cells were cultured in Dulbecco's Modi ed Eagle Medium PH 7.4 or PH6.5 (HyClone, Logan, UT, USA). All culture media were supplemented with 10% FBS (Gibco, Rockville, MD, USA), 100 units/mL penicillin, and 100 μg/mL streptomycin (HyClone). All cells were grown in a 37 °C humidi ed atmosphere containing 5% CO 2 .

Metabonomic analysis
The metabonomic analysis was done by Oebiotech (Shanghai, China). Brie y, 5×10 7 cells were broken up using ultrasonic homogenizer in mixture of methanol and water (1/4, vol/vol). L-2-chlorophenylalanine (0.3 mg / mL) dissolved in methanol was used as internal standard. The dried supernatant was then dissolved in pyridine containing 15 mg/mL methoxyamine hydrochloride. The samples were then subjected to LC-MS and GC-MS analysis. The obtained GC/MS raw data in .D format were transferred to .abf format via software Analysis Base File Converter for quick data retrieval. Then, data were imported into software MS-DIAL, which performs peak detection, peak identi cation, MS2Dec deconvolution, characterization, peak alignment, wave ltering, and missing value interpolation. Metabolite characterization is based on LUG database. The original LC-MS data were processed by software Progenesis QI V2.3 (Nonlinear, Dynamics, Newcastle, UK) for baseline ltering, peak identi cation, integral, retention time correction, peak alignment, and normalization. Main parameters of 5 ppm precursor tolerance, 10 ppm product tolerance, and 5% product ion threshold were applied. Compound identi cation were based on precise mass-to-charge ratio (M/z), secondary fragments, and isotopic distribution using The Human Metabolome Database (HMDB), Lipidmaps (V2.3), Metlin, EMDB, PMDB, and self-built databases to do qualitative analysis. Variable Importance of Projection (VIP) values obtained from the OPLS-DA model were used to rank the overall contribution of each variable to group discrimination. A two-tailed Student's T-test was further used to verify whether the metabolites of difference between groups were signi cant. Differential metabolites were selected with VIP values greater than 1.0 and p-values less than 0.05.
Clinical PDAC samples PDAC specimens were collected from 61 patients with PDAC from 2011 to 2015 at the Second A liated Hospital of Chongqing Medical University after informed consent was obtained from all patients. The patients did not receive chemotherapy or radiotherapy before surgery. The diagnoses of PDAC were made independently by at least two histopathologists. This study was carried out according to the principles of the Helsinki Declaration and approved by the Ethical Committee of the Second A liated Hospital of Chongqing Medical University ((2019)133).

RNA extraction and real-time quantitative PCR (qPCR)
Total RNA was extracted from each group with TRIzol Reagent (Takara, Dalian, China). The cDNA was obtained by using PrimeScript RT Master Mix (Takara, Japan). 50ng of each cDNA was ampli ed as a template, and qPCR was performed in a CFX96 real-time PCR system (Bio-Rad, Hercules, CA, USA) with SYBR ® Premix Ex TaqTM II kit (Takara, Japan). The PCR primers used are listed in supplementary table 1.
mRNA expression microarray RNA from acidosis-adapted PANC-1 cells and parental cells was extracted as mentioned above. Then the RNA was incubated with Agilent mRNA+LncRNA microarray (Capitalbiotech, Beijing, China). The differentially expressed genes were screened by the criterion: fold change larger than 2 and p value less than 0.05. Then these genes were applied to enrichment analysis by KEGG and GO database. were suspended in 100μl PBS and injected into nude mice from the tail vein. The mice were sacri ced 4 weeks after injection and the liver and lung were collected for pathological test. The lungs and livers were dissected into 5mm thick pieces and each piece was subjected for dehydration, embedding. Pathological section with 4μm thickness were made and stained with H&E. The metastatic foci were identi ed by a pathologist.
Wound healing assay After cell fusion rate in the 6-well plate reached more than 90%, then remove the cell medium. A 10μL pipette tip was used to draw a straight line in the cells along the sterilized ruler. Cells were washed with PBS for 3 times. Cell migration was observed and recorded at 0 h and 24 h.

Cell migration and invasion assays
Twenty-four-well transwell chambers were used for this assay (BD Falcon, Franklin Lakes, NJ). Brie y, 3×10 4 cells were plated into each upper chamber with 8-μm pores and cultured in 200μL serum-free DMEM. The lower chambers were lled with 500μL complete DMEM medium. After incubation for 24 hours at 37°C, cells that had migrated in the lower chambers were xed with 4% polyformaldehyde and stained with 0.1% crystal violet. The number of cells was counted in 5 distinct areas at ×100 magni cation. The results represent the average cell number in 3 wells per cell line. Then, the upper surface of the polycarbonate lter was coated with 10% Matrigel TM (BD Biosciences, Franklin Lakes, NJ), and 5×10 4 cells were added to detect cell invasion. The other conditions were the same as those in the migration assay.
Immuno uorescence assay Cells were xed with 4% polyformaldehyde for 20 min, and blocked with goat serum for 1 h at room temperature. Cells were then incubated with an anti-E-cadherin or β-actin antibody for 1 h at room temperature. After three washes in PBS, cells were incubated with a Cy3-conjugated secondary antibody (Beyotime, China) for 1 h. The nuclei were stained with DAPI for 5 min. Images were captured on a confocal microscope (Leica SP5, Germany).

Cell proliferation assay
Cells were plated in 96-well plates (2000 cells/well) in triplicate and measured using Cell Counting Kit-8 (CCK-8) (Dojindo Laboratories, Kumamoto, Japan). Cell proliferation was determined every 24 h for three days following the manufacturer's protocol. The absorbance at 450 nm was measured with a microplate reader (Bio-Rad, Hercules, CA, USA).

Colony formation assay
Long-term survival of cells was assessed by their ability to form colonies. 100 cells were seeded in a 96well plate. After 15 days, colonies were xed and stained with 0.1% crystal violate (Invitrogen, Carlsbad, CA, USA) before counting.

Cell cycle analysis
For cell cycle detection, 1.0×10 6 cells were harvested and xed with 75% ethanol at 4°C overnight after being washed with PBS. Then, cells were washed with PBS and stained with propidium iodide. FACSCalibur instrument (BD Biosciences, Franklin Lakes, NJ, USA) and CellQuest software were used to analyze cell cycle.

Cell apoptosis assay
Cells were obtained at a density of 1.0×10 6 cells/mL and washed by PBS, then incubated with reagents from the Annexin-V-FITC Apoptosis Detection Kit (Neobioscience, Shenzhen, China) according to the manufacturer's protocol. Then the cells were analyzed by FACS Vantage SE ow cytometer (BD Biosciences, San Jose, CA, USA).

Immunohistochemistry (IHC)
Following depara nization and rehydration, tissue samples (5mm slices) were incubated in 0.3% H 2 O 2 in methanol for 30 min at 37 °C to block endogenous peroxidase. The sections were then boiled in 10 mmol/L citrate buffer (pH 6.0) for 2 min in an autoclave. tissue samples were incubated with anti-MMP1 (Abcam, UK) overnight at 4 °C. MMP1 was detected using HRP-conjugated anti-rabbit secondary antibody (ZSGB-BIO, China) and visualized with DAB. The negative control only incubated with secondary antibody. The intensity of staining (brown color) was semi-quantitatively scored as follows: 1, weak; 2, medium; 3, strong; and 4, very strong. The percentage of maximally stained tumor cells in each section was recorded (0, <5%; 1, 5-30%; 2, 30-50%; 3,>50%). High expression of MMP1 was de ned as a combined score for the intensity and area of staining that was larger than 6. The results were veri ed by two pathologists independently.

Statistical analysis
All values were presented as means ± standard deviation. All data were obtained from at least three repetitions of each experiment. Prism 8.0 (GraphPad, USA) software was used to analyze the data. All data are shown as mean ± SD. Student's t-test was used to analyze differences between two groups. Oneway ANOVA was used to compare three or more groups. Probability values less than 0.05 was considered statistically signi cant.

Results
Reduction in AMP level activates AMPK signaling in acidosis-adapted PDAC cells Acidosis-adapted PDAC cells (PANC-1-AA and SW1990-AA) were obtained by culturing in DMEM acid medium of pH6.5 for three months, and the control cells were cultured in DMEM at pH7.4 (PANC-1-NA and SW1990-NA). Non-targeting metabonomic analysis, including LC-MS and GC-MS, was applied to reveal the metabolic changes in acidosis-adapted PDAC cells (SW1990 and PANC1). The GC-MS revealed that 99 and 69 metabolites were respectively changed in SW1990 (Fig1. a) and PANC1 cells ( Supplementary   Fig1. a). Accordingly, LC-MS revealed that 105 and 99 metabolites were respectively changed in SW1990 (Fig1. b) and PANC1 cells (Supplementary Fig1. b). The common decreased metabolites include adenosine monophosphate (AMP), fructose-6-phosphate (F6P), fructose-1-phosphate (F1P), fructose-1,6bisphosphate (FBP), glucose-6-phosphate (G6P) (Fig1. c). And the few common increased metabolites include glucose, metanephrine and myo-inositol (Fig1. d). The metabolites only increased in SW1990 cells include arginine, serine, glutamine, isoleucine, guanosine, ornithine, citrulline, glutathione, and valine (Fig1. d). The metabolites that were only increased in PANC1 cells includes phosphatidylcholine (PC), phosphosphingolipids (PS), glutaric acid, fumaric acid, tromethamine, testosterone, and methylamine (Fig1. d). The increased metabolites in PANC1 cells were enriched in choline metabolism in cancer, glycerophospholipid metabolism, glycine, serine and threonine metabolism and arginine/proline metabolism (Fig1. e). In contrast, the metabolites that were increased in SW1990 cells were mainly enriched in amino acid metabolism pathways such as arginine biosynthesis, glutathione metabolism, cysteine and methionine metabolism, valine/leucine/isoleucine biosynthesis (Fig1. f). As AMP was decreased in acidosis-adapted PDAC cells and the altered metabolites were related to energy production, we detect the intracellular ATP level and found that the ATP level was higher in acidosis-adapted cells (Fig1. g&h). These results revealed that the PDAC cells have different compensatory metabolic ways to increase intracellular ATP level after inhibiting glycolysis. As the ATP/AMP ratio tightly regulates AMPK signaling, we detected the phosphorylation level of AMPK and found that the phosphorylation level of AMPK was lower in acidosis-adapted cells than parental cells (Fig1. i). Altogether, we found that the metabolic reprogramming in acidosis-adapted PDAC cells inhibited AMPK signaling.

Acidosis-adapted PDAC cells exhibited enhanced migration and invasion ability.
We then test the cellular behavior of acidosis-adapted cells. Wound healing test was used to detect the difference of migration ability between normal cultured PDAC cells and acidosis-adapted PDAC cells. We found that acid-adapted PDAC cells migrated faster than that of normal cells. The healing rate of PANC-1-AA was about three times that of PANC-1-NA, while SW1990-AA was almost two times that of SW1990-NA (Fig2. a&b). Transwell assays showed that, compared with PANC-1-NA cells, the number of invasive cells of PANC-1-AA was obviously increased by about 50% (Fig2. c). Similar to PANC-1 cells, the invasive cells of SW1990-AA increased 30% to SW1990-NA (Fig2. d). As is well known, epithelial-mesenchymal transition (EMT) is an essential factor in promoting invasion and metastasis, so we detected the expression of EMT markers. As exhibited in Fig2. e, mRNA level of N-cadherin, Vimentin, α-SMA, Snail, Zeb1 and Zeb2 were signi cantly increased in PANC-1-AA and SW1990-AA cells. Besides, Snail, Vimentin and β-catenin protein level increased as well (Fig2. f&g), but E-cadherin protein level decreased in acidadapted PDAC cells (Fig2. g). Therefore, acidosis-adapted PDAC cells exhibited enhanced migration and invasion ability.

Acidosis-adapted PDAC cells exhibited enhanced cell proliferation
Previous studies revealed that acid-adapted colon cancer and cervical cancer cells could grow at a similar or higher rate in comparison to their parental cells 22 . Still, it is not clear whether chronic acidic microenvironment can affect the proliferation of PDAC cells. The results of CCK8 test showed that, compared with the NA cells in normal medium, the proliferation of PANC-1-AA and SW1990-AA cells were signi cantly improved in acidic medium (Fig3. a&b). The colony formation assay also re ected similar results (Fig3. c&d). Moreover, we compared the cell cycle of normal cells and acidosis-adapted cells. The number of PANC-1-AA cells in S phase was signi cantly more than that in PANC-1-NA cells, but the number of G1 phase cells was markedly reduced. There was no signi cant difference in the number of G2 phase cells between NA and AA cells. In SW1990 cells, the number of acidosis-adapted cells in S phase was more than that in normal culture, but there was no signi cant difference in other phases (Fig3. e&f). All cell cycle pictures of above cells were in Supplementary Fig2. On the contrary, the apoptosis of PANC-1-AA and SW1990-AA cells was less than that of PANC-1-NA and SW1990-NA cells (Fig3. g&h). Therefore, acidosis-adapted cells have enhanced proliferation and less apoptosis than parental cells.

MMP1 was signi cantly upregulated in acidosis-adapted PDAC cells
We analyzed the differentially expressed genes of PANC-1-NA and PANC-1-AA cells by microarray technology. We found that MMP1, ASIC2 and NANOG expression were signi cantly upregulated in PANC-1-AA cells, while the expression of LONRF2, BMP4 and COL5A1 were markedly decreased. Among them, the most obviously changed gene was MMP1 (Fig4. a). Interestingly, the glucose transporter SLC2A12 was also signi cantly upregulated in PANC-1-AA cells (Supplementary table 2), implying the redirection of increased glucose from glycolysis to pentose phosphate pathway, which provides the substrates for nucleic acid, fatty acids, sterols, nucleotides and non-essential amino acids synthesis 23 . GO analysis showed that the main difference between PANC-1-NA and PANC-1-AA cells were in cell adhesion and extracellular matrix organization (Fig4. b). qPCR and WB assays showed that both mRNA and protein level of MMP1 were increased either in PANC-1-AA cells or in SW1990-AA cells (Fig4. c). 24 . MMP1 expression level is increased in various malignant tumors and is negatively correlated with prognosis 25 . TCGA database showed that the expression of MMP1 in most cancer tissues was signi cantly higher than that in adjacent tissues (Fig4. d). However, the relationship between MMP1 and pancreatic cancer is rarely studied, especially in the acidic microenvironment of pancreatic cancer. We detected the expression of MMP1 in tissues of 61 patients with PDAC by immunohistochemistry (Fig4. e), and found the higher the expression of MMP1, the worse the prognosis (Fig4. f). The tumor size was larger in the MMP1 high group than MMP1 low (Fig4. g). Consistently, bioinformatic analysis revealed that the expression level of MMP1 was correlated with advanced tumor stage (Fig4. h) and poor prognosis (Fig4. i). Pathological characteristics of clinical samples revealed that high expression of MMP1 is associated with metastasis and acidic marker CA9 and acid sensing ion channels ASIC1 and ASIC2 (Table 1). These results showed that elevated MMP1 expression might at least partially mediate the enhanced invasion capacity of acidosis-adapted PDAC cells.

Activation of YAP/TAZ signaling upregulated MMP1 in acidosis-adapted PDAC cells
In order to explore the regulatory mechanism of the acidic microenvironment of MMP1 in PDAC cells, we reviewed the microarray results and found that Hippo signaling pathway was signi cantly changed (Fig6. a). We detected the expression and activation of YAP and TAZ, the key molecules of Hippo pathway. The results showed that expression levels of YAP and TAZ in PANC-1-AA and SW1990-AA cells were signi cantly higher than those in PANC-1-NA and SW1990-NA cells (Fig6. b). When YAP/TAZ is phosphorylated, their activities are inhibited. Non-phosphorylated YAP/TAZ will enter the nucleus as cotranscription factors and combine with TEADs to promote oncogene transcription 26 . Thus, we chose YAP/TAZ inhibitor ML-7 to block their activities. After treatment with ML-7, the activation level of YAP and TAZ were signi cantly reduced, while the expression of MMP1 and Snail in PANC-1-AA and SW1990-AA cells were signi cantly decreased (Fig6. c). Transwell assays showed that the invasion capacity of PANC-1-AA and SW1990-AA cells was reduced by about 50% after being treated with ML-7 (Fig6. d). The results of the scratch test were consistent with those of transwell (Fig6. e). Transwell and wound healing images were showed in Supplementary Fig. 4. In addition, YAP/TAZ inhibitor also signi cantly inhibited the proliferation of acid-adapted cells, compared with NA cells (Fig6. f). Moreover, we found a positive correlation between the expression of YAP and MMP1 in PDAC tissues (Supplementary Fig5). There is a predicted binding site of TEAD4 in the promoter region of MMP1, which is a recognized YAP/TAZ effector molecule 27

Discussion
Studies have shown that tumor cells prefer forming acidic microenvironment because of high metabolic activity and insu cient perfusion. Tumor cells adapted to acidic microenvironment exhibited aggressive phenotype, which is achieved by acid-induced cell motility, extracellular matrix degradation, decreased immune responses and modi ed signaling pathways 28 . However, its speci c regulatory mechanism has not been fully clari ed. Our experiments elucidated that glycolysis was signi cantly inhibited, while intracellular ATP, glucose and other metabolites were increased, in acidosis-adapted PDAC cells. This metabolic reprogramming leads to AMPK inactivation. It should be noted that the glucose transporter SLC2A12 (GLUT12) rather than SLC2A1 (GLUT1) was also upregulated in acidosis-adapted PDAC cells.
GLUT12 was required for maximal androgen-mediated glucose uptake and cell growth in prostate cancer cells (LNCaP and VCaP) 29 . In addition, GLUT12 is highly expressed in triple-negative breast cancer (TNBC) samples and promotes TNBC tumor growth and metastasis in vitro and in vivo through regulating aerobic glycolysis 30 . This evidence implies that acidosis-adapted uptake more glucose to fuel their aggressive activities.
Although the intracellular glucose was elevated in acidosis-adapted PDAC cells, it was not used for glycolysis due to the many intermediate products of glycolysis were reduced. Pentose phosphate pathway (PPP) is the other way for glucose catabolism. The PPP branches after the rst step of glycolysis and consumes the intermediate glucose 6-phosphate (G6P) to generate fructose 6-phosphate (F6P) and glyceraldehyde 3-phosphate (G3P). The PPP supplies NADPH and ribose5-phosphate (R5P) instead of ATP 23 . These two metabolites are vital for cell survival and proliferation. R5P is a building block for nucleic acid synthesis. NADPH is the reducing power required to synthesize fatty acids, sterols, nucleotides and non-essential amino acids 31 . Moreover, NADPH mediates oxidized glutathione (GSSG) conversion to reduced glutathione (GSH) via glutathione reductase to reduce the reactive oxygen species.
Recently, the PPP has been reported to enhance the chemoresistance of breast cancer cells 32 . We have reason to propose that the glucose was re-directed to PPP as the increase in intracellular glucose, amino acids, phosphatidylcholine, phosphosphingolipids, and the reduction in glycolysis intermediates.
However, the mRNA level of the PPP enzymes was not altered in the acidosis-adapted PDAC cells (data not shown). This result indicates that there might be post-transcriptional regulation of this pathway.
The increased ATP thus inactivated the AMPK pathway, which is generally considered as a metabolic sensor to suppress cancer. AMPK could initiate a series of downstream pathways, including Hippo. We found that the Hippo pathway effector molecules YAP and TAZ were signi cantly activated in acidosisadapted PDAC cells. Then, YAP/TAZ bound to transcription factor TEAD4 in nucleus, and facilitated MMP1 expression, resulting in PDAC progression (Fig. 7). Previous studies demonstrated that other MMPs are regulated by acidic condition in different diseases. The expression levels of MMP3 and MMP13 were elevated under the degenerated intervertebral disc-like acidic condition 33 . MMP12 is also increased by extracellular acidosis in systemic sclerosis 34 . Acidic microenvironment induces MMP2, MMP3, MMP9 and MMP13 expression to promote breast or lung cancer progression 35 . MMP1 belongs to the zinc-dependent endopeptidase family. It has been proved to be closely relevant to proliferation, migration and invasion in various cancers 36 . MMP1 can also promote the development of pancreatic cancer 37, even become a potential diagnostic marker for pancreatic cancer 38 . However, there lacks of study on the regulation of MMP1 in the cancer acidic microenvironment. Our study revealed that the expression of MMP1 was increased by acidosis, enriching the research on the regulation of acidic microenvironment on MMP family.

Conclusion
In summary, the present study showed that metabolic reprogramming promotes metastasis of acidosisadapted PDAC cells via AMPK/YAP/MMP1 axis, enriching the mechanism by which acidic environment promotes tumor progression.        Proposed mechanisms of acidosis promoted PDAC progression. In acidosis-adapted PDAC cells, glucose enters the cells through GLUT12 transporter, which increases glucose-6-phosphate (G6P) and pentose phosphate pathway (PPP) metabolism, thus affecting amino acids or phosphatidylcholine (PC), phosphosphingolipids (PS) and others metabolism, increased ATP content and inhibited AMPK, so as to activate YAP/ TAZ through AMOTL1. YAP/TAZ translocated into nucleus and bound with TEAD4 to promote MMP1 transcription and increase expression level of MMP1, leading to PDAC cells metastasis and proliferation.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.