Human CLS1 cells were isolated from an 87-year-old male patient with lung adenosquamous carcinoma who provided written informed consent, according to the protocol NTUH-IRB201103028RC approved by National Taiwan University Hospital Research Ethics Committee (Taipei, Taiwan) (26). These cells were cultured in RPMI-1640 medium supplemented with 10% FBS and 1 mM sodium pyruvate (all from Thermo Fisher Scientific, Inc.) at 37°C with 5% CO₂ in a humidified incubator. For the establishment of pH 6.6 and pH 7.4 subclones by long-term acidosis treatment, cells were cultured in media adjusted to pH 6.6 as described previously (27) for 2 months, while the control pH 7.4 subclone cells were cultured in pH 7.4 media. Mother stocks of the acidotic and control cells were prepared upon reaching the 48th passage. All experiments were performed using cells at passage 50. In experiments assessing short-term acid treatment, the cells were cultured in the corresponding pH media for 24 h.

pH 7.4 or pH 6.6 cells were seeded in 6-well plates at a density of 2,000 cells/well, and live images were captured every 40 min over a period of 96 h using an inverted phase-contrast light microscopy equipped with the ASTEC-real time cell monitoring system (Astec Co., Ltd.). Cell doubling times were analyzed by visual inspection of the serial images and calculated as the time between two completed cytokinesis events.

pH 7.4 or pH 6.6 cells (20,000 cells/well) were seeded in 6-well plates as triplicates, and cultured for 0, 24, 48, 72 or 96 h at 37°C before harvesting. Cells were stained with trypan blue (Thermo Fisher Scientific, Inc.) for 30 sec at room temperature to exclude non-viable cells, and counted using a Bright-Line™ hemocytometer (Reichert, Inc.) under a DMi1 inverted light microscope (Leica Camera AG).

To analyze the cellular mitochondrial metabolic rate, pH 7.4 or pH 6.6 cells (1,000 cells/well) were seeded in 96-well plates and cultured for 24, 48, 72 or 96 h at 37°C, before incubation with 20 µl 5 mg/ml MTT (Merck KGaA) diluted in 180 µl complete medium (RPMI-1640 medium with 10% FBS and 1 mM sodium pyruvate) for 2 h. MTT was subsequently removed, and 100 µl/well DMSO (Bionovas Biotechnology Co., Ltd.) was added and cells were incubated for 15 min. Optical absorbance was measured at 570 nm using an Elx800 reader (BioTek Instruments, Inc.).

All animal protocols were reviewed and approved by the National Taiwan University College of Medicine Institutional Animal Care and Use Committee (approval no. 20170550; Taipei, Taiwan). A total of 60 female nude mice (BALB/cAnN.Cg-Foxn1^nu/CrlNarl mice; age, 4 weeks; mean weight, 14.71±1.21 g) were purchased from National Applied Research Laboratories, National Laboratory Animal Center (Taipei, Taiwan), and kept at 22±2°C and 55±10% humidity with a 12/12-h light/dark cycle, with free access to food and water. In experiment #1, 8 mice were implanted subcutaneously with 10-cell inoculates of pH 6.6 or pH 7.4 cells to simulate metastatic colonization at opposite flanks, while the same procedure was performed with 1-cell inoculates in another 8 mice. In experiment #2, 18 nude mice were implanted subcutaneously with 10 pH 6.6 or pH 7.4 cells at opposite lower flanks, while 1 pH 6.6 or pH 7.4 cell was implanted at opposite upper flanks of the same recipient mouse. In experiment #3, the same procedure as in experiment #2 was performed using 26 nude mice. Briefly, pH 6.6 or pH 7.4 cells at passage 50 were resuspended in complete culture medium, before accurately selecting either 10 or 1 cells under a Zeiss Axio Observer D1 inverted fluorescence microscope (Zeiss AG) and packing these together by natural gravity in 5 µl Growth Factor Reduced, Phenol Red-Free Matrigel™ (Corning, Inc.; final concentration, 50%) for subcutaneous injection into the 5-week-old nude mice using 4% isoflurane for induction and 2% isoflurane for maintenance of anesthesia. Body weight was recorded every other day starting from the time when a palpable tumor could be detected (18 to 20 days post-implantation; Fig. S1). Tumor volume was measured every other day using a caliper and calculated as 1/2 (length x width²), where length refers to the longest dimension and width refers to the smallest dimension perpendicular to the length. Intolerable distress caused to the animals, such as continued declining weight loss ≥20% for up to 2 days, or a xenograft tumor volume reaching 2,000 mm³, were set as the humane endpoints of the present study. Mice were euthanized at week 6 post-implantation in a CO₂ chamber for 5 min (CO₂ flow set to 30% chamber volume displacement per min). The mice were observed for an additional 5 min to determine death as judged by stopped breathing, heart arrest and loss of blink reflex. Dissection was performed to thoroughly inspect any subcutaneous tumors that may have developed. The tumors were extracted, and their volumes were recorded. No further metastasis was observed. The tumor burden observed was: maximum tumor diameter, 9.44 mm; maximum sum of diameter, 16.84 mm; and maximum tumor volume, 308.225 mm³. The samples were fixed in 4% paraformaldehyde at room temperature for 24 h, followed by histological analyses. Notably, mouse no. 18 of experiment 2 was euthanized at day 35 post-tumor inoculation after observation of clinical signs including dullness, lethargy and failure to respond to stimuli as suggested by the ethical guidelines of the National Taiwan University College of Medicine Institutional Animal Care and Use Committee (Taipei, Taiwan).

For long-term acidosis treatment, two independent transcriptome analysis experiments with two biological replicates each were performed for both pH 6.6 and pH 7.4 cells. CLS1 cells treated with either 0-h (short term, pH 7.4) or 24-h acidosis (short term, pH 6.6) served as short-term treatment references. Total RNA was extracted using TRIzol^® reagent (cat. no. 26073; Thermo Fisher Scientific, Inc.) and an RNeasy Mini Kit (cat. no. 74106; Qiagen, Inc.). Only samples with an integrity number >9.0 were used to construct the sequencing libraries using the TruSeq Stranded mRNA Library Prep Kit (cat. no. 20020595; Illumina, Inc.). Briefly, mRNA purification was performed using oligo(dT)-coupled magnetic beads. Double-stranded cDNAs were synthesized using the SuperScript™ II Reverse Transcriptase kit (cat. no. 18064-014; Thermo Fisher Scientific, Inc.) and random primers. The reverse transcription temperature protocol was 10 min at 25°C, 15 min at 42°C and 15 min at 70°C, then holding at 4°C. The quality of the libraries was assessed on the Agilent Bioanalyzer 2100 system and a Real-Time PCR system (Agilent Technologies, Inc.). The loading concentration of the final libraries was 290 pM as measured by the Qubit 2.0 Fluorometer Q32866 (Thermo Fisher Scientific, Inc.). Sequencing was performed using an Illumina™ NovaSeq 6000 System (Illumina, Inc.) with the NovaSeq 6000 S1 Reagent Kit (300 cycles; cat. no. 20012863; Illumina, Inc.) at a depth of ~40 million reads per sample of 150-bp paired-end reads. Bases with low quality and sequences from adapters in raw data were removed using the program FASTQ (version 0.20.0) (28). The filtered reads were mapped with Bowtie 2 (https://bowtie-bio.sourceforge.net/bowtie2/index.shtml) and aligned to the human reference genome hg19 as it provided the most comprehensive annotation available at the time of analysis. The software FeatureCounts v2.0.1 (29) in the Subread package v2.0.1 (https://subread.sourceforge.net/) was applied for quantification of the gene abundance. Differentially expressed genes were identified by DESeq2 version 1.28.0 (30) for the samples with biological replicates or by EdgeR version 3.36.0 (31) for the samples without biological replicates (0 and 24 h acidosis samples). Significant differential expression of genes was defined as a posterior probability of being equally expressed (PPEE) <0.05 and log₂ |posterior fold change|>1. Data were deposited in the National Center for Biotechnology Information Gene Expression Omnibus database (accession no. GSE200546; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE200546).

Cellular component (CC), molecular function (MF) and biological process (BP) functions were analyzed using the GO database (http://geneontology.org/) using Protein Analysis Through Evolutionary Relationships 15.0 (https://pantherdb.org/) with a significance threshold of P<0.05 and false discovery rate <0.05. KEGG signaling pathway analyses were subsequently performed using KEGG Orthology-Based Annotation System 3.0 (32) for the Homo sapiens reference list, with a significance threshold of P<0.05 and corrected P<0.05. PPI analyses were performed using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database version 10.5 (https://string-db.org; >0.5 confidence score; species was limited to Homo sapiens).

For H&E staining, xenograft samples were fixed in 4% paraformaldehyde at room temperature for 24 h and then embedded in paraffin. Sections (4-µm thick) were de-paraffinized with xylene (Merck KGaA) for 5 min at room temperature and rehydrated with a gradient of 95, 85, 70 and 50% ethanol for 5 min each at room temperature prior to a 2-min incubation at room temperature with H&E Y solution (ScyTek Laboratories, Inc.). Images of all sections were captured using a Zeiss Axio Imager Z2 upright fluorescence microscope (Zeiss AG).

For immunohistochemistry staining, human lung cancer tissue microarray slides (LC10014a, LC10013c and LC814a; 10% formalin-fixed for 24 h at room temperature, paraffin-embedded; 5-µm thick) were purchased from US Biomax, Inc.; TissueArray.Com LLC as approved by National Taiwan University Hospital Research Ethics Committee (approval no. 202201055RIND; Taipei, Taiwan). Xenograft samples including both paired and unpaired tumors were fixed in 4% paraformaldehyde for 24 h at room temperature, paraffin-embedded and cut as sections of 4-µm thick. All slides were de-paraffinized with xylene (Merck KGaA) for 5 min at room temperature and rehydrated with a gradient of 95, 85 and 70% ethanol for 5 min each at room temperature. Antigen retrieval was performed in 10 mM citric acid buffer at 100°C for 20 min followed by 3% H₂O₂ treatment for 20 min at room temperature. Blocking using Protein block (normal goat serum reagent as supplied by the manufacturer; cat. no. HK112-9KE; BioGenex) was performed for 30 min at room temperature. The slides were incubated with anti-human ITGA4 antibodies (cat no. 350695; dilution, 1:500; United States Biological) for 2 h at room temperature and then washed with PBS buffer three times. The slides were incubated with HRP rabbit/mouse antibodies (undiluted; cat. no. K5007: Dako; Agilent Technologies, Inc.) for 1 h at room temperature. 3,3-diaminobezidine chromogenic staining was used for detection prior to 5 sec hematoxylin counterstaining at room temperature. LC814a included 6 cases of small cell carcinoma and 17 cases each of squamous cell carcinoma and adenocarcinoma, as well as their matched lymph node metastases. LC10014a included 50 cases of adenocarcinoma, and LC10013c included 48 cases of adenocarcinoma with the matched adjacent normal lung tissues (NATs) and 4 normal lung tissues. There were 13 overlapping cases among these three lung tissue arrays, which were annotated and analyzed only once to avoid repetition. Images of all sections were captured using a Zeiss Axio Imager Z2 upright fluorescence microscope (Zeiss AG) and analyzed using StrataQuest 6.0.1.216 software (TissueGnostics GmbH) for at least three randomly selected fields.

In vitro vasculogenic mimicry assays were performed on 96-well plates coated with 50 µl Growth Factor Reduced, Phenol Red-Free Matrigel™ per well. pH 6.6 and pH 7.4 cells were seeded at a density of 4×10⁴ cells/well and cultured for 10 h. Images of the wells were captured using a Zeiss Axio Observer D1 inverted fluorescence microscope (Zeiss AG). Three independent experiments with three repeats each were performed. Three randomly selected images per sample were analyzed using Image J version 1.53k (National Institutes of Health) for vasculogenic mimicry structures. Five parameters were analyzed: Total tube length, number of branching points, number of loops, percentage of total loop area per well and number of tubes.

To analyze in vivo vasculogenic mimicry, paraffin-embedded slides of xenograft samples (4% paraformaldehyde-fixed for 24 h at room temperature; 4-µm thick) were de-paraffinized with xylene (Merck KGaA) for 5 min at room temperature and rehydrated with a gradient of 95, 85 and 70% ethanol for 5 min each at room temperature. The slides were treated with a periodic acid-Schiff PAS-1 staining kit (ScyTek Laboratories, Inc.) according to the manufacturer's instructions. Slides were then processed as for immunohistochemical staining, including antigen retrieval, quenching, blocking, secondary antibody incubation, washing, chromogen detection, and counterstaining as described for immunohistochemical staining. Slides were incubated with polyclonal rabbit anti-human CD31 antibody (cat. no. ab28364; dilution, 1:500; Abcam) for 2 h at room temperature. Images of all sections were captured using a Zeiss Axio Imager Z2 upright fluorescence microscope (Zeiss AG). The angiogenic structures (CD31⁺/PAS⁺) were counted in four biologically paired whole tumors. Vasculogenic mimicry structures were defined by CD31⁻/PAS⁺ staining accompanied by the presence of red blood cells in the lumen, and counted from at least five whole tumors of either pH 6.6 and pH 7.4.

pH 7.4 or pH 6.6 cells (1×10⁶ cells/ml) were seeded 14 h before the experiment, then incubated in RPMI-1640 medium containing 2% FBS (both from Thermo Fisher Scientific, Inc.) and 7.5 µg/ml Hoechst 33342, with either DMSO (Mock group) or the ATP-binding cassette (ABC) transporter inhibitor reserpine (20 µM; Reserpine group) (all from Merck KGaA) for 2 h at 37°C in the dark, then treated with 0.05% Trypsin-EDTA for resuspension in sorting buffer containing 50 µg/ml propidium iodide (both from Thermo Fisher Scientific, Inc.) to exclude non-viable cells. Cells were subsequently sorted by dual wavelength analysis (blue, 424-444 nm; red, 675 nm) with an excitation wavelength of 350 nm using a BD FACS Aria II flow cytometer (Becton, Dickinson and Company). Data were analyzed using FCS Express 4 Plus (De Novo Software), and the fold-change of ABC transporter-dependent side population cells was calculated as [pH 6.6 (side population %^Mock-side population %^Reserpine)]/[pH 7.4 (side population %^Mock-side population %^Reserpine)].

Cell lysates were collected using a buffer containing 50 mM Tris/HCl (pH 8.0), 1% SDS, 1% deoxycholate and 10 mM EDTA, supplemented with cOmplete™ Protease Inhibitor Cocktail (cat. no. 11836145001; Roche Diagnostics) and phosphatase inhibitor cocktail 3 (cat. no. P0044; Merck KGaA), and the lysates were centrifuged at 4°C at 12,000 × g for 5 min. The protein concentration was determined using a BCA protein assay kit (Thermo Fisher Scientific, Inc.), and aliquots containing 20 µg total protein per lane were analyzed by 10% SDS-PAGE and blotted onto a PVDF membrane (Pall Corporation). The membranes were blocked with TBS buffer containing 0.05% Tween-20 and 5% skimmed milk for 30 min at room temperature. After a 2-h incubation at room temperature with the primary antibodies, including rabbit anti α-tubulin antibody (cat. no. GTX112141; dilution, 1:1,000; GeneTex, Inc.), rabbit anti-Oct4 antibody (cat. no. SC9081; dilution, 1:500; Santa Cruz Biotechnology, Inc.) and rabbit anti-mouse Nanog antibody (cat. no. RCAB002P; dilution, 1:500; ReproCELL, Inc.). The membranes were washed with TBS buffer containing 0.1% Tween-20 three times at room temperature before incubation with secondary antibodies (goat anti-rabbit HRP antibody; cat. no. GTX213110; dilution, 1:5,000; GeneTex, Inc.), for 1 h at room temperature. Bands were visualized using the ECL chemiluminescence method (WesternBright^® ECL HRP substrate; Advansta, Inc.) with an LAS-4000 luminescent image analyzer (FUJIFILM Wako Pure Chemical Corporation).

Unless otherwise specified, all experiments were repeated independently at least three times. The mean values of the groups were compared, and the results are presented as the mean ± SD. GraphPad Prism 7 software (Dotmatics) and OriginPro 2023b software (OriginLab) were used for statistical analysis. Comparisons between different groups were analyzed by Student's unpaired t-test. In paired tumors, the germination times, volume, density, the percentage of ITGA4⁺ cells and ITGA4⁺ stained areas, as well as the analysis of endothelial angiogenic and vasculogenic mimicry structures, were analyzed by Student's paired t-test. In the lung tissue microarray, the mean intensity of expression per ITGA4⁺ cells was analyzed for tumor different nodal statuses and stages using one-way ANOVA with Tukey's multiple comparisons test. Two-way mixed ANOVA with the Bonferroni multiple comparisons test was used to analyze the percentage of ITGA4⁺ cells in the normal group vs. (NAT group vs. lung cancer within the same patient group). Two-way mixed ANOVA with the Bonferroni multiple comparisons test was used to analyze the mean intensity of expression per ITGA4⁺ cells in the no-metastasis group vs. (metastatic primary tumors group vs. lymph node metastases within the same patients group). P<0.05 was considered to indicate a statistically significant difference.

Tumor acidosis develops over long periods of time, involving serial regulatory processes in the cancer cell that may lead to the acquisition of a stable phenotype (15). To accurately reflect this scenario, a long-term acidosis cell model was established in the present study (Fig. 1A) using the CLS1 patient-derived primary cells of lung adenosquamous carcinoma reported in our previous study (26). These cells were acclimated for 2 months in culture media at pH 6.6 to emulate the cellular responses to acidosis relative to normal physiological culture conditions (pH 7.4) (33,34). Using time-lapse microscopy, both subclones acclimated at pH 6.6 and pH 7.4 were found to replicate synchronously without detectable abnormalities or differences in morphology (Fig. 1B), nor any significant differences in cell doubling time, population survival or metabolic rates (Fig. 1C-E). Cell lines from other cancer types, such as SAS (Tohoku University, Sendai, Japan) and OECM1 (National Defense Medical Center, Taipei, Taiwan) of oral squamous cell carcinoma origin, also exhibited good adaptation to the of acidotic conditions, displaying discernible changes in population survival rates or growth dynamics (data not shown). These finding suggested that the lung cancer cells became well adapted to the long-term acidosis regime, and subsequent analyses could be conducted with only minimal concern regarding discrepancies of growth dynamics.

Numerous studies have reported that an acidic microenvironment can trigger stemness-like properties in tumor cells (35,36). Although western blotting indicated no significant change in the protein expression levels of either of the stemness-associated transcription factors Oct4 and Nanog (Fig. 1F), the pH 6.6 subclones displayed a 1.85-fold increase in the proportion of reserpine-sensitive side population cells relative to the pH 7.4 subclones (Fig. 1G), suggesting a survival advantage of the pH 6.6 subclone associated with ABC transporter-dependent clearance of noxious agents while growing in adverse environments. Taken together, the present results indicated that lung cancer cells can adapt to long-term acidosis, preserving regular cellular morphology and growth dynamics. Acidosis per se may not be sufficient to promote the stemness properties of lung cancer cells; however, with an increased proportion of side population cells, acidotic cancer cells may display a survival advantage in adverse environments. The two CLS1 subclones obtained were referred to as pH 6.6 cells and pH 7.4 cells, and they were subsequently used as the model to study the lung cancer cell response following long-term acidosis.

Overcoming the adverse environment for growth is a primordial task for CTCs to successfully establish as a new metastatic colony (22). To test whether the survival advantage of acidotic cells may lead to increased metastatic colony incidence, a novel experimental model that tightly controlled the exact number and physical conditions of the inoculated cells was designed, to enable a rigorous comparison of the colonizing ability of the pH 7.4 and pH 6.6 cells. Since CTCs found in the pulmonary veins of patients with non-small cell lung cancer (NSCLC) are mostly clusters of ~10 cells (37), exactly 10 cells or one single cell from the pH 6.6 and pH 7.4 cells were selected under an inverted microscope, then packed together by natural gravity in 5 µl growth factor-reduced Matrigel (Fig. 2A). These cells were implanted subcutaneously in opposite flanks of nude mice, which were sacrificed to inspect the tumor growth either on day 42 post-inoculation or earlier if humane endpoint criteria had been reached. After performing three independent experiments, grown tumor masses were retrieved from 27 mice (Fig. S2), while 33 mice did not develop tumors (Fig. S3). As shown in Fig. 2B, the cumulative tumor incidence of the 10-cell and 1-cell inoculations of the pH 6.6 cells surpassed that of the pH 7.4 cells by 2.22- and 1.25-fold, respectively. When comparing paired tumors grown on the same mouse, the germination time of the pH 6.6 tumors was faster (20±1.5 days vs. 25±9.4 days post-inoculation for the pH 7.4 tumors), and their standard deviation was smaller, indicating a more coordinated pace compared with their pH 7.4 counterparts (Fig. 2C); they also exhibited a larger volume with a slightly higher density (Fig. 2D). Albeit without reaching statistical significance, the tumor incidence rate, germination time, volume and density all displayed consistent differences when comparing pH 6.6 tumors with their pH 7.4 counterparts, collectively supporting a more robust establishment of the pH 6.6 cells as developing colonies. This also indicates that, despite the low number of cells inoculated, the metastatic colonization model was able to provide reliable information on the role of long-term acidosis in promoting tumor germination and growth.

The molecular events underlying the metastatic advantage of acidotic cells were subsequently investigated. RNA sequencing analysis was employed to profile the transcriptomes of both the pH 6.6 and pH 7.4 cells (Gene Expression Omnibus series accession no. GSE200546; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE200546) (38). A total of 1,735 differentially expressed genes (>2-fold change; PPEE <0.05) were identified in two independent RNA sequencing experiments with two biological replicates each (Fig. S4). Of these genes, 142 showed consistent changes in all replicates (Fig. 3A), of which 44 genes were found to be upregulated and 98 genes were downregulated in pH 6.6 cells. Heatmap analysis of these 142 acidosis-responsive genes highlighted the differential expression of numerous ECM-related genes, including ITGA4, FBN1, FBN2, COL6A1, ITGB4 and VTN (Fig. 3B). In Table I, the 20 most upregulated and downregulated genes are listed. Furthermore, functional annotation of these 142 acidosis-responsive genes using KEGG database analysis identified 'ECM-receptor interaction' and 'cell adhesion molecules' as the most significantly overrepresented categories. Correspondingly, in the GO enrichment analysis of these acidosis-responsive genes, the terms 'cell adhesion' and 'regulation of cell communication' in the BP category, 'plasma membrane' and 'extracellular region' in the CC category, and 'extracellular matrix structural constituent' in the MF category were the main terms identified (Fig. 3C). Finally, STRING analysis of the protein-protein networks (>0.5 confidence score) revealed that these 142 acidosis-responsive genes gravitated around processes involving 'ECM organization', 'regulation of localization' and 'cellular response to chemical stimulus' (Fig. 3D). Taken together, these results suggested that a coordinated and largescale reorganization of the ECM occurs upon long-term acidosis of lung cancer cells.

To identify the direct participants of acidosis-induced ECM organization that may offer targetability in terms of future therapeutic interventions, the acidosis-responsive genes annotated to the two GO categories 'extracellular region' and 'plasma membrane', which were also included in the 'ECM organization' category in PPI analysis, were selected. Of the six targets identified, ITGA4 was found to be the most upregulated gene as determined by the fold change of the read counts (Fig. 4A). ITGA4 transcripts were present at low levels in pH 7.4 cells. They encode for the α4 subunit of the integrin protein family, which heterodimerizes with either the β1 or the β7 subunit to form a functional integrin to regulate cell motility and adhesion (39). High and steady levels of ITGB1 transcripts were found in both the pH 6.6 and 7.4 cell groups, while ITGB7 was expressed only at negligible levels (Fig. 4B). This suggests that the higher expression of ITGA4 may result in an increase of functional α4β1 integrin in the pH 6.6 cells. Immunohistochemical examination confirmed that the percentage of ITGA4-expressing cells was significantly higher in the pH 6.6 metastatic tumors, although the increase in the positively stained area was not statistically significant (Fig. 4C). Furthermore, ITGA4 proteins were demonstrated to be predominantly localized to the cell membrane of the pH 6.6 tumors (Fig. 4C; lower right panel). Taken together, these experiments suggested that acidosis induced the upregulation of ITGA4 mRNA, which is effectively translated into protein, and this higher level of ITGA4 protein expression was sustained in acidotic metastatic colonies.

In solid tumors, ITGA4 expression has been implicated in tumor interaction with vascular endothelial cells, which is crucial for proper metastatic colonization and angiogenesis (40,41). To examine the development of vascular structures, all paired tumors that were grown in the same recipient mouse were compared, in order to minimize inter-individual differences in the ability of vascularization. A similar extent of macroscopic neoangiogenic blood vessel formation was observed for pH 6.6 and pH 7.4 tumors (Fig. 5A; left panels). Both tumors were found to be irrigated internally with CD31-positive angiogenic microvessels. By contrast, only the pH 6.6 tumors displayed a large number of CD31⁻/PAS⁺ trabecular structures, indicative of the presence of vasculogenic mimicry in the acidotic tumors, while no such structures were observed in the paired pH 7.4 tumors (Fig. 5A). This observation was sustained even when comparing all grown tumors obtained from the experiment (Fig. S5A), and the accentuated expression of ITGA4 surrounding these structures (Fig. 5B) suggested a possible involvement of ITGA4 in the formation of vasculogenic mimicry. To verify whether long-term acidosis could promote vasculogenic mimicry in vitro, live microscopy was employed to compare the potential of vasculogenic mimicry of pH 6.6 cells vs. pH 7.4 cells. The numbers of vasculogenic mimicry loops and branch points formed by pH 6.6 cells were significantly higher, and their total lengths were longer (Fig. 5C). Parameters of potential trabecular space for nutrient infusion, as evaluated by the total loop area per well (Fig. S5B), as well as the quantity of tubes formed (Fig. S5C), were also found to be higher in the acidotic cells. Taken together, these results suggested that long-term acidosis could promote the formation of vasculogenic mimicry networks, thereby enhancing the ability of tumors to establish nutrient and metabolite exchanges in the new colony.

Since acidosis may affect the microenvironment beyond the intratumoral milieu, it was subsequently examined whether upregulation of ITGA4 also occurred in the adjacent tissues. Immunohistochemical staining indicated that ITGA4 protein expression in normal lung tissue was low, whereas in normal lung tissues adjacent to the tumors, the presence of ITGA4 was found to be significantly higher, accompanying an even higher expression in the tumor cells themselves (Fig. 6). A 2-fold upregulation of ITGA4 expression upon challenging the cells with a pH 6.6 environment for 24 h was also observed, suggesting that this increase in the mRNA level was a rapid response to environmental acidosis, which may be further maintained and amplified as acidosis is sustained (Fig. S6). These results indicated that the upregulation of ITGA4 is a phenomenon common to tissues in the tumor microenvironment, which is further amplified in the tumor cells as acidosis is prolonged.

Upregulation of ITGA4 expression has been associated with metastasis in several different tumor types (42-44). To determine the participation of ITGA4 in metastatic colonization, clinical lung cancer specimens were examined, and ITGA4 expression was compared across TNM descriptors. As shown in Fig. 7A, ITGA4 was widely expressed in all adenocarcinoma specimens tested. Notably, compared with the non-metastatic primary tumors, ITGA4 expression as evaluated by mean intensity per positively stained cells was significantly higher in the primary tumors that had already disseminated (N1 and N2+N3; Fig. 7B), whereas the level of ITGA4 expression was not significantly different between the N1 or N2 primary tumors and their corresponding lymph node metastases (Fig. 7C). Similarly, statistical analysis across different stages of lung cancer (Fig. 7D) also indicated that upregulation of ITGA4 expression occurred upon the tumor's entry into stage II (Fig. 7E), whereas neither more advanced stages, nor their corresponding metastases, showed any further significant increase in the ITGA4 level (Fig. 7F). Taken together, these results indicated that the upregulation of ITGA4 is a common feature of primary lung tumors that have developed metastatic potential, and acquisition of a higher level of ITGA4 may be a prerequisite for the successful completion of metastasis to form secondary tumors. Since the upregulation of ITGA4 persists within the lymph node metastases where acidosis may no longer be present, these data also suggested that acidosis may possess the potential to induce long-term regulatory changes that support the continued upregulation of ITGA4 even after extratumoral acidosis pressure is alleviated.