1. Introduction
Lung cancer is the leading cause of cancer-related death worldwide; approximately 85% of all lung cancers are non-small cell lung cancers (NSCLC) [
1,
2,
3,
4]. Despite advances in early detection and standard treatment, NSCLC is often diagnosed at an advanced stage and with a poor prognosis; the overall cure and survival rate for NSCLC remains low at 19%, particularly in locally advanced stage IIIA cancer [
5]. Even after complete primary tumor resection, about 45% of the early-stage NSCLC patients develop local recurrences or distant metastases within 8 to 18 months [
6]. Therefore, the treatment and prevention of NSCLC can be improved with a better understanding of the biology and the mechanisms of metastasis.
Cancer metastasis is manifested by a highly complex cascade of processes, starting with the invasion of the tumor cells from a primary site into the surrounding tissues and continuing as intravasation into the circulatory system and extravasation to a distant organ, where the disseminated tumor cells that survive may initiate the progressive outgrowth of secondary tumors in a metastasis-receptive niche. However, metastasis is an inefficient process [
7]. Approximately millions of cells per gram are disseminated from the primary tumors per day, but only a few become capable of transmigrating and surviving in a distant organ. One key limitation to successful metastasis is the death of the cells that occurs as they become detached from the extracellular matrix (ECM), which is known as anoikis, and from the neighboring cells, and undergo cell rounding, which is known as amorphosis [
7,
8].
Cancer cells have evolved multifaceted mechanisms, including the epithelial-to-mesenchymal transition (EMT), to safeguard against anoikis and amorphosis. Moreover, the recent work by several groups highlights the ability of the detached cells to form clusters or aggregates is another critical factor that can enhance the metastatic capacity of cancer cells [
9,
10,
11]. Although metastasis has long been conceived of as a single-cell process, multicellular cell clusters, termed circulating tumor cell (CTC) clusters, of 2 to more than 10 cells tethered together have been directly observed in several steps of the metastasis cascade, including the systemic circulation of the tumor cells in the bloodstream. Aceto et al. showed that the CTC clusters appeared to be derived from the oligoclonal clumps of primary tumor cells rather than the coalescence of single CTCs in the circulation [
9].
CTC clusters are associated with poorer prognoses in many cancer types [
12]. Indeed, in different mouse models, multicellular aggregates give rise to between 50 and 97% of the metastases. The formation of clusters induces multiple molecular properties, including the increase in stem cell-like traits, evasion from targeting by natural killer cells, and resistance to metabolic stress, among others. However, the underlying spatiotemporal mechanism by which the detached cells tether together to form aggregates is poorly understood. According to several studies, canonical cell adhesion proteins, including cadherin, are involved in cancer cell cluster formation [
13,
14]. In addition, plakoglobin was shown to hold CTCs together [
9]. However, these studies were mostly performed under adhesive 2D and 3D conditions that could not replicate the in vivo tumor microenvironment that became stiffer during the progression toward advanced cancer.
Here, we examined spatiotemporal cell interaction in the in vivo cancer pathological context by employing nonadhesive 3D poly-2-hydroxyethyl methacrylate (poly-HEMA) culture. The 3D cultures of cancer cells in poly-HEMA hydrogel, which prevents cell spreading and cell attachment to the substratum due to its superhydrophilic nature, have been used for many years to mimic in vitro 3D cancer tissue architecture, as cell aggregates in poly-HEMA are pathologically similar to clusters isolated from a patient’s CTC, ascitic fluid and pleural effusion [
15,
16,
17]. We report that the clustering of free-floating NSCLC A549 cells in nonadhesive 3D poly-HEMA culture depends on actin-rich protrusions, which is intensely studied for cell migration under 2D culture conditions [
18]. Furthermore, our current study is the first to indicate that AQP3, a unique member of the water channel aquaporin (AQP) family [
19], is essential for forming protrusions by acting as a key regulator of actomyosin cytoskeleton remodeling through caspase 3 activation. We also discuss the implications of these findings in the context of multicellular metastasis in a hydrodynamic tumor microenvironment.
3. Materials and Methods
3.1. Cell Culture and Reagents
The human pulmonary adenocarcinoma A549 cells (The Korean Cell Line Bank, Seoul, Korea) of a human alveolar basal epithelial carcinoma cell line were maintained in Roswell Park Memorial Institute (RPMI-1640) medium supplemented with 10% fetal bovine serum (Gibco, Grand Island, NY, USA). The cells were cultured at 37 °C under a humidified atmosphere with 95% air and 5% CO2. Jasplakinolide and Y-27632 (Cayman Chemical, Ann Arbor, MI, USA) dissolved in dimethyl sulfoxide (Sigma-Aldrich, St. Louis, MO, USA) to reach a concentration of 1 mM. Rho activator II was obtained from Cytoskeleton Inc. (Denver, CO, USA). The cells were exposed to 30% deionized water and 5% sucrose to give an osmotic shock to induce hypotonic stress and hypertonic stress, respectively.
3.2. Imaging
Holotomographic images of the cells were taken on the 3D Cell-Explorer Fluo (Nanolive, Ecublens, Switzerland) using a low-power class I laser (0.2 mW/mm2, λ = 520 nm), a 60 × dry objective (NA = 0.8), and a USB 3.0 CMOS Sony IMX174 sensor with a typical quantum efficiency of 70% at 545 nm, dark noise (typical) of 6.6 e−, and a typical dynamic range of 73.7 dB. In the holotomographic image, the lateral (X and Y-axis) resolution was 200 nm, the Z-axis resolution was 400 nm, with a field of view of 90 × 90 × 30 µm, and the maximum temporal resolution was 0.5 fps 3D RI volume per second.
3.3. Time-Lapse Imaging
Live cell imaging was conducted in a Top-Stage Incubator system (Okolab, Pozzuoli, Italy) at 37 °C with 5% CO2 and humidifying conditions. The cells were cultured in FluoroDish cell culture dishes (World Precision Instruments Inc., Sarasota, FL, USA) for this experiment.
3.4. Image Analysis
Image rendering and export were performed with the STEVE v.1.7.3496 software (Nanolive). The backgrounds were subtracted during post-processing, and all the slices of the post-processed image were exported to RI volumes and transformed into the 3D tiff format. The RI volumes in the tiff format can be read by the software FIJI. Three-dimensional RI volumes of all the slices were transformed into 2D RI maps using maximum intensity projection and exported to a time-lapse video file.
3.5. Poly-HEMA Coating
First, 1.2 g of poly-HEMA (Sigma-Aldrich) was dissolved in 40 mL of 95% ethanol by mixing the solution overnight at 37 °C. Then, 50 μL or 3.2 mL of the poly-HEMA stock solution were added to 96-well plates and 10-cm dishes, respectively, under the tissue culture hood; the plates and dishes were swirled for 10 min using a plate rotator. The plates were left to dry overnight and then washed with phosphate-buffered saline (PBS) immediately before use.
3.6. RNA Sequencing
Total RNAs were isolated from different cell lines using Trizol (Invitrogen, Carlsbad, CA, USA). Total RNA quantity and quality were verified spectrophotometrically (Nano-Drop 2000 spectrometer; Thermo Scientific, Wilmington, DC, USA) and electrophoretically (Bioanalyzer 2100; Agilent Technologies, Palo Alto, CA, USA). To prepare Illumina-compatible libraries, a TruSeq RNA library preparation kit (Illumina, San Diego, CA, USA) was used according to the manufacturer’s instructions. In brief, mRNA purified from total RNA using polyA selection was chemically fragmented (50-bp fragment libraries) and converted into single-stranded cDNA using random hexamer priming. After this, the second strand was generated to create double-stranded cDNA that was ready for TruSeq library construction. Short double-stranded cDNA fragments were then connected with sequencing adapters, and suitable fragments were separated by agarose gel electrophoresis. Truseq RNA libraries were built by PCR amplification and quantified using quantitative PCR (qPCR) according to the qPCR Quantification Protocol Guide. qPCR data were qualified using the Agilent Technologies 2100 Bioanalyzer (Agilent technologies). Libraries were sequenced (101-nt paired-end sequencing) using a HiSeq™ 2000 platform (Illumina). To estimate expression levels, the RNA-Seq reads were mapped to the human genome using TopHat (version 1.3.3) [
36]. The reference genome sequence (hg19, Genome Reference Consortium GRCh37) and annotation data were downloaded from the UCSC website (
http://genome.uscs.edu (accessed on 15 April 2021)). The transcript counts at the gene level were calculated, and the relative transcript abundances were measured in fragments per kilobase of transcript per million mapped reads (FPKM) using Cufflinks software (version 1.2.1; Seattle, WA, USA) [
37]. FPKM is computed similarly to RPKM, except it accounts for the scenario in which only one end of a pair-end read is mapped [
38]. Using this approach, the expression levels were measured for 37,396 Ref-Seq genes uniquely aligned based on RNA sequencing reads. Raw data were extracted as FPKM values across all samples, and samples with zero values across more than 50% of uniquely aligned genes were excluded.
3.7. siRNA-Mediated Knockdown of AQP3
The transient knockdown of
AQP3 was performed using Lipofectamine
TM RNAiMAX (ThermoFisher, Rockford, IL, USA). The cells were plated in a 6-well plate at a density of 3 × 10
5 cells per well and cultured overnight at 37 °C. The following day, the cells were transfected with
AQP3 siRNA (sequence available in
Supplementary Table S2) or non-targeting control siRNA (OriGene, Rockville, MD, USA) using 7.5 μL of Lipofectamine
TM RNAiMAX according to the manufacturer’s instructions. The final concentration of the siRNA used per well was 25 pmol. After incubating for 24 h, the cells were divided into conventional 2D and poly-HEMA 3D cultures and incubated further for 24 h for the following experiments.
3.8. qRT-PCR
Total RNA was extracted from the cultured cells using the PureLink
TM RNA Mini Kit (Invitrogen). The first-strand cDNA was synthesized using oligo-dT primers and M-MLV reverse transcriptase (Invitrogen). qRT-PCR reactions were performed in triplicates at a final volume of 20 μL containing TB Green Premix Ex Taq II (Takara, Shiga, Japan), 10 ng of cDNA, and 20 pmol of each primer. qRT-PCR was performed using a 7500 real-time PCR system (Applied Biosystems, Foster City, CA, USA) at 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s and 60 °C for 34 s. The glyceraldehyde 3-phosphate dehydrogenase gene (
GAPDH) was used as an internal control in each reaction. Specific amplification was verified by performing a melting curve analysis (55–95 °C, 0.5 °C/s). The quantification of relative gene expressions was performed using the ΔΔCT method. The expression level of each gene was normalized to that of
GAPDH in the same sample. Genes and their primers are listed in
Supplementary Table S3.
3.9. Western Blot Analysis
Cells were lysed with RIPA buffer (Santa Cruz Biotechnology, Santa Cruz, CA, USA) on ice for 30 min, and the lysates were centrifuged at 13,000 g at 4 °C for 15 min. The supernatants were incubated with 4 × Laemmli sample buffer (Bio-Rad, Hercules, CA, USA) at 95 °C for 5 min. The samples were then separated with SDS-PAGE gel and immunoblotted with the antibody against AQP3 (Alomone Labs Ltd., Jerusalem, Israel, 1/200), GAPDH (BioLegend, San Diego, CA, USA), or β-actin (Santa Cruz Biotechnology) or α-tubulin (Santa Cruz Biotechnology). β-actin, GAPDH, and α-tubulin were used as loading controls.
3.10. Immunocytochemistry
A549 cells were seeded on sterile glass coverslips, and immunocytochemical staining was performed. In short, the cells on coverslips were fixed with 4% paraformaldehyde for 10 min and permeabilized with 0.15% Triton-X 100 for 5 min. Then, the cells were blocked for 1 h with the blocking solution of 3% bovine serum albumin in PBS and incubated with the primary antibody against AQP3 for 2 h at room temperature. Subsequently, the cells were incubated with Fluorescein-conjugated anti-rabbit IgG (Sigma-Aldrich) for 60 min at room temperature. The subcellular organization of the actin microfilaments was assessed by incubating the cells with rhodamine-conjugated phalloidin (Molecular Probes, Eugene, OR, USA) at a dilution of 1:200 to reach the final concentration of 1.5 units/mL. Next, the cells were washed with PBS, and the coverslips were mounted on a glass slide in 10% Mowiol 4–88, 1 μg/mL 4′,6-diamidine-2-phenylindole dihydrochloride, and 25% glycerol in PBS with nuclei counterstained blue with 4′,6-diamidine-2-phenylindole dihydrochloride (DAPI). Then, the cells were observed under a confocal laser scanning microscope LSM800 (Zeiss, Oberkochen, Germany).
3.11. Scanning Electron Microscopy of Spheroids
The cell spheroids were collected using wide pipette tips and pooled into an Eppendorf tube. Following a PBS wash, the spheroids were incubated overnight in 2.5% glutaraldehyde (EMS, Hatfield, PA, USA), 1.25% paraformaldehyde (EMS), and 0.03% picric acid in 0.1 M sodium cacodylate buffer (pH 7.4) at 4 °C. The spheroids were then washed in 0.1 M cacodylate and post-fixed with 1% osmium tetroxide (OsO4)/1.5% potassium ferrocyanide (KFeCN6) for 1 h. The samples were then washed 2 times in PBS, dehydrated with ethanol, exposed to critical-point drying, placed on glass coverslips, and subjected to platinum sputtering before imaging. Images were acquired at 20 kV at 1000–1500× magnification using scanning electron microscopy (JSM 630/OA, JEOL Ltd., Tokyo, Japan).
3.12. Transmission Electron Microscopy of Spheroids
The fixed spheroids that were serially dehydrated with ethanol were subsequently infiltrated by a mixture of ethanol and propylene oxide at the ratio of 2:1, 1:1, 1:2, or 0:1 for 1 h, and then by a mixture of propylene oxide and epoxy resin (Structure Probe, Inc., West Chester, PA, USA) at the ratio of 2:1, 1:1, or 1:2 for 1 h. Then, the spheroids were embedded in epoxy resin and loaded into capsules to be polymerized at 60 °C for 72 h. Following the staining of the semi-fine thin 1-µm sections with toluidine and sodium tetraborate, thin-sectioning at 80 nm was performed using a Leica EM UC7 ultramicrotome (Leica Microsystems, Wetzlar, Germany). The resulting sections were collected on copper grids and contrasted in 1% uranyl acetate solution in distilled water for 1 h at room temperature in the dark and lead citrate. The images were acquired using a JEM-1400 Flash TEM (JEOL Ltd.) at 120 kV.
3.13. Boyden Chamber Assay
The migration of A549 cells was examined using a 6.5 mm Transwell (Corning, Glendale, AZ, USA). The cells were plated on the inserts and cultured at 37 °C in the upper chambers. After 20 h, the migrated cells that had crossed the inserts were fixed with 4% paraformaldehyde for 15 min and stained with 0.1% crystal violet (Sigma-Aldrich) for 10 min. The inserts were washed at least three times in PBS and the interior of the inserts was gently swabbed with a cotton swab to remove the non migrated cells. Then, the migrated cells counted as cells per field of view under phase-contrast microscopy.
4. Conclusions
Our study demonstrates that cell detachment-induced AQP3 upregulation contributes to the extrusion of the cell surface to form protrusions through caspase 3, leading to the differential aggregation of substratum-detached cells important for multicellular metastasis in a manner dependent on the properties of the substratum.
The significance of our study is two-fold. First, there is increasing evidence showing that multicellular tumor cell aggregates are critical for cell survival following the loss of ECM attachment and dissemination through the circulatory system. The current study demonstrates that AQP3 contributes to tumor cell clustering through cell surface membrane protrusion. The AQP family comprises 13 mammalian members. While they primarily facilitate the passive transport of water across membranes, they also play a crucial role in tissue migration during embryonic development and wound healing. Furthermore, several studies have reported that this unexpected role for AQPs in cell migration is also implicated in tumor cell migration [
30,
39,
40,
41,
42]. Chae and colleagues reported that AQP5 promoted tumor invasion in NSCLC. However, the mechanism underlying the AQP5-mediated invasion has not been delineated. Indeed, our study is the first to elucidate the mechanism of AQP3 in influencing multicellular aggregation through protrusion-promoted coalescence under suspended cell growth conditions.
Second, protrusions have been extensively studied in tissue regeneration, cancer invasion and metastasis, and the environmental exploration of leukocytes [
43,
44,
45]. However, many in vitro studies are performed with cells in adhesive flat 2D culture, under which integrin-mediated adhesion to the ECM is preserved. However, cancer invasion and metastasis occur independently of cell adhesion to ECM, as evidenced by pathological clusters isolated from patients’ CTCs, ascitic fluid, and pleural effusion [
15]. A study using intravital imaging reported that CTCs with active transforming growth factor-β (TGF-β) signaling migrate as solitary cells, whereas the cells lacking TGF-β signaling invade lymphatics collectively, suggesting that TGF-β signaling regulates the mode of cancer cell motility [
46]. However, the mechanisms underlying tumor aggregate formation under cancer pathological conditions remain poorly studied. Indeed, our results demonstrate for the first time that protrusions are important in 2D cellular movement and also play a critical part in the 3D aggregates of cancer cells detached from the substratum via the downstream apoptosis executor caspase 3 and migration.
Finally, it will be interesting to elucidate the mechanism through which protrusions contribute to cell-cell cohesion in cancer clusters following substratum detachment. Considering the studies demonstrating that hydrostatic or osmotic pressure controls cell rounding, we hypothesize that protrusion-mediated cell aggregation under suspension conditions proceeds in two steps, that protrusions should first render floating cells migratory and then adhesive. We propose that hydrostatic pressure built up locally through AQP3 channels that extrude through cell surface protrusions, acting as pedals in a fluid environment and increasing intercellular interactions to overcome Brownian dispersion. Consistent with this proposal, Saadoum et al. reported that as the underlying mechanism by which non-endothelial cells overexpressing
AQP1 or
AQP4 showed accelerated cell migration, the AQPs at the protrusions at the leading edge led to rapid water fluxes, providing the space for actomyosin assembly and flow [
42]. However, it is still unclear how protrusions are adhesive and contribute to inducing the self-assembly of the floating cells. Cell clustering in directed multicellular migration can be subdivided into cohort aggregates, in which the cells are in tight contact with each other, or streaming aggregates, in which the coordinated aggregation is not always in direct physical contact [
34]. Although our study did not definitely confirm the type of aggregates in the A549 cell spheroids, it would be interesting to investigate the molecular components and viscoelastic properties of the protrusions. Addressing this idea in detail is beyond the scope of the current study. Furthermore, our findings strongly support the idea that protrusions are a useful target in anticancer drug development, particularly targeting advanced lung cancer characterized by highly motile EMT. However, the details of the spatiotemporal architecture of the protrusions in the A549 cells and the localization of AQP3 in protrusions remain to be examined.