The authors have declared that no competing interests exist.
Conceived and designed the experiments: LAG XX LNL NTN LX XJ KLvG MCFC. Performed the experiments: LAG XX LNL NTN LX. Analyzed the data: LAG XX LNL NTN LX. Contributed reagents/materials/analysis tools: LAG XX LNL NTN LX KLvG XJ MCFC. Wrote the paper: LAG XX LNL NTN LX KLvG XJ MCFC.
To study the individual functions of hyaluronan interacting proteins in prostate cancer (PCa) motility through connective tissues, we developed a novel three-dimensional (3D) hyaluronic acid (HA) hydrogel assay that provides a flexible, quantifiable, and physiologically relevant alternative to current methods. Invasion in this system reflects the prevalence of HA in connective tissues and its role in the promotion of cancer cell motility and tissue invasion, making the system ideal to study invasion through bone marrow or other HA-rich connective tissues. The bio-compatible cross-linking process we used allows for direct encapsulation of cancer cells within the gel where they adopt a distinct, cluster-like morphology. Metastatic PCa cells in these hydrogels develop fingerlike structures, “invadopodia”, consistent with their invasive properties. The number of invadopodia, as well as cluster size, shape, and convergence, can provide a quantifiable measure of invasive potential. Among candidate hyaluronan interacting proteins that could be responsible for the behavior we observed, we found that culture in the HA hydrogel triggers invasive PCa cells to differentially express and localize receptor for hyaluronan mediated motility (RHAMM)/CD168 which, in the absence of CD44, appears to contribute to PCa motility and invasion by interacting with the HA hydrogel components. PCa cell invasion through the HA hydrogel also was found to depend on the activity of hyaluronidases. Studies shown here reveal that while hyaluronidase activity is necessary for invadopodia and inter-connecting cluster formation, activity alone is not sufficient for acquisition of invasiveness to occur. We therefore suggest that development of invasive behavior in 3D HA-based systems requires development of additional cellular features, such as activation of motility associated pathways that regulate formation of invadopodia. Thus, we report development of a 3D system amenable to dissection of biological processes associated with cancer cell motility through HA-rich connective tissues.
A majority of patients who die from solid tumors each year suffer from bone metastases
The study of pathways that control cancer cell invasion or migration, and the development of new drug to prevent these processes requires systems that can measure the invasive properties of these cells
The bone marrow matrix consists of soft, gel-like connective tissue rich in hyaluronic acid (HA)
Another way that cells interact with HA is by degrading it, by expression and secretion of hyaluronidases (HAases). Relevant to PCa invasion are the HAases, Hyal-1 and Hyal-2, whose expression levels have been implicated in PCa metastasis
Importantly, HAases are druggable targets, indicating their potential use as targets to slow invasion and possibly prevent metastasis
We previously described the development of a HA-based hydrogel for culture of poorly adherent bone metastatic PCa cells
To study the development of invasive properties of PCa cells in a native connective tissue matrix, we used the LNCaP series of human PCa cell sublines that were developed to mimic the process of PCa bone metastasis. These cells are regarded as one of the better PCa progression models available because of their ability to form clinically relevant osteoblastic lesions in mice
Electrophoresis, cell culture, and transfection reagents and supplies were purchased from Invitrogen (Carlsbad, CA). TCM™ defined serum replacement was purchased from MP Biomedicals (Solon, OH). Hyaluronic acid (sodium salt, 500 KDa and 1.3 MDa) was generously donated by Genzyme Corporation (Cambridge, MA). RHAMM (HMMR) and CD44 antibodies were purchased from Novus Biologicals (Littleton, CO). Hyal2 and β-actin antibody were purchased from Abcam (Cambridge, MA). Equine-α-mouse-HRP secondary antibody was purchased from Cell Signaling Technology (Danvers, MA). Hyal1 antibody, bovine testicular hyaluronidase (BTH), β-mercaptoethanol (βME), glycine, Tween®20, bovine serum albumin (BSA), sodium formate, Coomassie brilliant blue, disodium cromoglycate (DSC), and deoxycholic acid were purchased from Sigma Aldrich (St. Louis, MO). Goat-α-rabbit-HRP secondary antibody, protease inhibitors (PI), 5X sample buffer, sulfo-NHS-SS-biotin, and avidin agarose were purchased from Thermo Scientific (Rockford, IL). Laemmli sample buffer and alcian blue were purchased from Biorad (Richmond, CA). Tris buffer, glacial acetic acid, methanol, low melting point (LMP) agar, dimethyl sulfoxide (DMSO), sodium dodecyl sulfate (SDS), Triton X-100, and sodium chloride (NaCl) were purchased from Fisher Scientific (Fair Lawn, NJ). Nonidet P-40 was purchased from Roche (Indianapolis, IL) and ethanol was purchased from Decon Labs, Inc (King of Prussia, PA). All compounds used were reagent grade or better.
Low passage number PCa cells were maintained in Corning (Lowell, MA) tissue culture 75 mm flasks at 37°C in 5.0% (v/v) CO2 in T-medium supplemented with 5% (v/v) fetal bovine serum (FBS) and 100 U/ml penicillin G sodium and 100 µg/ml streptomycin sulfate in 0.085% (w/v) saline (PS). Medium was changed every 2–3 days. Cells were passaged at approximately 80% confluency, judged by eye, using 0.25% (w/v) trypsin with ethylenediaminetetraacetic acid (EDTA) 4•Na. LNCaP sublines
To synthesize the HA aldehyde (HAALD), oxidation reaction of HA (1.3 MDa) was performed in the presence of sodium periodate under aqueous conditions. Adipic dihydrazide (ADH)-modified HA (HAADH) was synthesized by carbodiimide-mediated coupling of ADH with the carboxyl groups of HA (500 KDa) in aqueous conditions. Detailed procedures for the synthesis and characterization of both of these HA derivatives was reported in our previous study
HAALD and HAADH were dissolved separately in Dulbecco’s phosphate buffered saline (DPBS) to a concentration of 10 mg/mL overnight at 37°C. 2% (w/v) LMP agar in DPBS was prepared and was melted in a 70°C heatblock (Labnet International, Woodbridge, NJ) at least 1 hour prior to use. Dissolved HA derivates were sterilized by UV irradiation at 254 nm for 15 minutes prior to use.
PCa cells were released from their flask using 0.25% (w/v) Trypsin EDTA 4NA and the total number of cells was counted using a hemocytometer (Hausser Scientific, Horsham, PA). 100,000 (for 24-well plate) or 600,000 (for 6-well plate) cells were pelleted for each culture. Cell culture inserts (Millipore, Billerica, MA, pore size: 0.4 mm, diameter 12 mm for 24-well plate, 30 mm for 6 well plate) were pre-wet in T-medium then placed in the wells of a 24-well or 6-well culture plate (Beckton–Dickenson Labware, Franklin Lakes, NJ). For 2D cultures, cell pellet was mixed with T-medium (1 mL for 24-well plate, 4 mL for 6-well plate) and plated.
For 3D HA hydrogel culture, cell pellet was mixed with 100 µl (for 24-well plate) or 300 µl (for 6-well plate) of HAALD. An equal volume of HAADH was added and the culture was mixed well to evenly disperse cells. The hydrogel solution then was pipetted into pre-wet cell culture inserts and allowed to solidify for approximately 10 minutes at 37°C. T-medium (1mL for 24-well plate, 4 mL for 6-well plate) supplemented with 1% (v/v) PS and either 5% (v/v) FBS or 2% (v/v) TCM™ was added around the insert and the culture was incubated at 37°C, 5% (v/v) CO2. This concentration of TCM™ is consistent with previous studies supplementing C4-2 culture
For 3D agar culture, cell pellet was mixed with 85 µl (for 24-well plate) or 510 µl (for 6-well plate) DPBS warmed to 37°C. Pre-melted 2% (w/v) LMP agar was added to DPBS/cell mixture to a final concentration of 0.3% (w/v) LMP agar. Agar culture was incubated at room temperature for approximately 5 minutes before pipetting into the cell culture insert to prevent agar gel from leaking through the membrane pores. Once plated in insert, the agar gel was allowed to solidify for approximately 10 minutes at room temperature. Solidification was performed at room temperature to expedite gelation of soft agar. T-medium (1 mL for 24-well plate, 4 mL for 6-well plate) was subsequently added around the insert and the culture was incubated at 37°C, 5% (v/v) CO2. For all cultures, half medium changes were performed every two days.
Metabolic activity and cell counts were performed for the cell types and culture conditions described. To measure metabolic activity of cells, water soluble tetrazolium salt-1 (WST-1, Roche) reagent was utilized as described
For all invasion assays, phase contrast images were taken of each cell culture using a Nikon Eclipse TE2000-U (Tokyo, Japan) microscope and a 10X objective. The average number of invadopodia was determined by counting the total number of invadopodia per photographed field for three biological replicates. An “invadopodium” was defined as a thin cellular process extending outward from a cell cluster, and clear enough to be easily identified by eye. The percentage of merging clusters was calculated by counting both the number of clusters in physical contact with another cluster and the number of free clusters for three biological replicates. From these values, the percentage of merging clusters was calculated and considered an index of migration in 3D hydrogels. In samples where networks of cell clusters were observed, a single cell cluster was defined as a distinct, rounded area of the cellular network sometimes appearing to contain a higher cell density. For both types of quantification, care was taken to ensure consistency when counts were being performed.
Rheological characterization of hydrogel samples were performed on a stress-controlled rheometer (AR-G2, TA Instruments, New Castle, DE) with a 20 mm diameter standard steel parallel-plate geometry and at a 100 µm gap. Dynamic oscillatory time sweeps were performed at 25°C or 37°C for agar and HA gels respectively, and the storage (G‘) and loss (G“) moduli were recorded. 30 µl agar solution (0.3% w/v) or HAADH/HAALD mixture (1% w/v) was loaded into the geometry that was subsequently covered with mineral oil at the edge to prevent water evaporation. These mixtures were allowed to solidify
The cells combined from two wells of a 6-well plate culture were used for protein extraction experiments. BTH (180 µl at 10 mg/mL) was applied and cultures were incubated at 37°C for 30 minutes. The cultures were transferred to centrifuge tubes and cells were pelleted by centrifugation for 1 minute at 3000 RPM. Cells were washed once with DPBS. Radioimmunoprecipitation (RIPA) buffer (50 mM Tris buffer pH 8.0, 150 mM NaCl, 0.5% (w/v) deoxycholic acid, 1% (v/v) Nonidet P-40, 0.1% (w/v) SDS, ddH2O, 200 µl) containing PI was applied to the cell pellets. Lysates were incubated on ice for 45 minutes with occasional vortexing. Lysates were cleared by centrifugation at 13,000 rotations per minute (RPM) for 10 minutes. The total protein concentration from the lysates was determined using a bicinchoninic acid (BCA) protein assay (Thermo Scientific) following a standard protocol. Lysates were stored at −20°C.
Protein (50 µg) was mixed with RIPA buffer and 5X Lane Marker Sample Buffer and βME (3% (v/v) final) to a total volume of 25 µl. Samples were boiled for 5 minutes, then cooled and quickly spun. Samples and a protein ladder were electrophoresed on a NUPAGE 4–12% Bis-Tris gel using an X-Cell Surelock™ electrophoresis cell and MOPS running buffer. Samples were transferred to a nitrocellulose membrane using an X-Cell™ II transfer apparatus and a transfer buffer consisting of 1.5125 mg/mL Tris, 7.2 mg/mL glycine, and 0.01% (w/v) SDS at a setting of 21V for 1.5 hours.
The membrane was blocked in 3% (w/v) BSA in DPBS containing 0.1% Tween® 20 (PBST) for 1 hour at room temperature with gentle shaking. Primary antibodies (1∶10,000 for RHAMM, 1∶5000 for CD44 and β-actin, 1∶500 for Hyal1 and Hyal2) in 3% (w/v) BSA in PBST were incubated with the membrane for 2.5 hours at room temperature with shaking. The membrane was washed 3 times, 10 minutes each time, with PBST. Secondary antibodies (1∶5000 goat-α-rabbit HRP or 1∶10,000 equine-α-mouse HRP) in 3% (w/v) BSA (for Hyal1, Hyal2, and β-actin blots) or 3% (w/v) non-fat dried milk (for RHAMM and CD44 blots) were incubated with the membrane for 1 hour at room temperature with shaking. The membrane was washed 3 times, 10 minutes each time, with PBST. Supersignal® West Dura Extended Duration Substrate (Thermo Scientific) was prepared as directed and applied to the membrane for 5 minutes at room temperature with shaking. The blot was exposed using BioMax Light Film (Kodak, Rochester, NY) and developed in an SRX-101A developer (Konica Minolta Medical & Graphic Inc, Tokyo, Japan). PC-3 cell lysate (50 µg) was used as a positive control for CD44 westerns.
Cells grown on 2D were cultured in an 8 well Lab-Tek II Chambered Coverglass (Nalge Nunc, Naperville, IL). The medium was removed and chambers were washed 2 times with DPBS. Cells were fixed for 10 minutes in 4% (v/v) paraformaldehyde (PFA) (Electron Microscopy Sciences, Hatfield, PA) in ddH2O. Excess PFA was removed and the chambers were washed twice with DPBS. Triton X-100 in DPBS (0.2% (v/v)) was prepared and applied to chambers for 10 minutes. Excess Triton X-100 solution was removed and chambers were washed twice with DPBS. Cultures were blocked in 3% (w/v) BSA in DPBS at 4°C overnight. Cells were incubated with 1∶1000 (v/v) solution of RHAMM or CD44 primary antibodies in 3% BSA at 4°C overnight. A 1∶1000 (v/v) solution of Draq5 (Biostatus Limited, Leicestershire, UK) in DPBS was applied for 10 minutes at room temperature. Chambers again were washed with DPBS and Gel/Mount™ (Biomeda, Foster City, CA) was added to chambers to prevent photobleaching. Cells were visualized using confocal microscopy on a Zeiss LSM 510 VIS (Carl Zeiss, Maple Grove, MN).
Cells in 3D were cultured in 24-well plates as described above. Cell clusters were gently removed from the hydrogel using a micropipette. Clusters were washed gently with 1 mL DPBS 2 times, quickly centrifuging to collect cell clusters each time. Clusters were carefully resuspended in 4% (v/v) PFA and transferred to an 8 well chambered coverglass. Cultures were incubated at room temperature for 10 minutes. Once transferred to the chambered coverglass, the same staining procedures were used as employed for 2D culture. Extreme care was taken to retain as many cell clusters as possible during each liquid removal step.
C4-2 cells were cultured in a T-75 flask to approximately 90% confluency. Cells were released from the flask, suspended in T-medium and counted. Cells were washed three times in ice-cold PBS (pH 8.0, 5 mL). Cells were resuspended in ice-cold PBS (pH 8.0) to a final concentration of 20×106 cells/mL. Freshly prepared, 10 mM sulfo-NHS-SS-biotin (80 µl) was added to the cell mixture and the mixture was incubated at room temperature for 30 minutes with shaking. 50 mM Tris (pH 8.0, 2 mL) was added to deactivate the biotinylation reaction, and the cell pellet was subsequently washed twice with PBS (pH 8.0, 2 mL). The final PBS wash was removed completely from the cell pellet, and RIPA buffer containing PI (500 µl) was applied to the cell pellet. The lysis reaction was incubated on ice for one hour. The lysate was cleared by centrifuging at 12,000 RPM for 10 minutes, after which the supernatant was removed and stored at –20°C.
A BCA assay was performed on the biotinylated lysate as described previously; approximately 1.5 mg of protein was obtained. Avidin agarose (1 mL) was applied to a microcentrifuge tube and the resin was settled by centrifuging 1 minute at 2000 RPM. The resin was washed three times with RIPA buffer (500 µl). The lysate was applied to the resin and the mixture was incubated on a tube rotator at 4°C overnight. The resin was settled by centrifuging 1 minute at 2000 RPM and the unbound lysate was removed. Unbound lysate (10 µl) was mixed with Laemmli Sample Buffer containing 5% (v/v) βME (10 µl). The resin was washed three times with RIPA containing PI (500 µl). Laemmli Sample Buffer containing 5% βME (50 µl) was applied to the beads. Laemmli-containing mixtures were boiled for 5 minutes, then spun down at 13,000 RPM. The biotinylated pulldown and unbound fraction were electrophoresed and transferred to nitrocellulose as described previously. Bone metastatic PCa cell (PC3) lysate (10 µl) mixed with Laemmli sample buffer containing βME (10 µl) was included as a positive control for CD44 expression. Western blotting for CD44, RHAMM, and GAPDH were performed as described under the “Western Blotting” section.
C4-2 cells were maintained in antibiotic-free medium for three days prior to transfection. Mixed 6 pmol/cm2 Stealth RNAi™ siRHAMM duplex (GGCGUCUCCUCUAUGAAGAACUAUA and UAUAGUUCUUCAUAGAGGAGACGCC) and 0.5 µl/cm2 Lipofectamine™ RNAiMAX in Optimem® I for RHAMM knockdown or 6 pmol/cm2 low GC Stealth RNAi™ negative control duplex and 0.5 µl/cm2 Lipofectamine in Optimem for transfection control. Transfection mixtures were incubated for 15 minutes at room temperature before applying to cells. Transfection reactions were incubated for six hours at 37°C and plates were swirled gently every hour to mix transfection reagents.
The transfection mixture was aspirated from the cells, and cells were washed with DPBS. Antibiotic free medium was applied to the cells and transfected cells were incubated at 37°C overnight before using for experiments. Efficacy of RHAMM knockdown was assessed through Western blotting for RHAMM as described under the “Western Blotting” section.
HAase Activity was examined via substrate gel electrophoresis
A solution of 125 mM DSC was prepared in 40% (v/v) DMSO and filtered in a Steriflip filter (Millipore) to sterilize. DSC (3.33 or 10 µl of 125 mM) was added to 800 µl or 2.4 mL T-medium in 24 or 6 well plates, respectively, to a final concentration of 500 µM DSC, the IC-50 value of this inhibitor
The toxicity of DSC on PCa cells was determined using a trypan blue exclusion assay. 600,000 cells were plated into 6 well plates with T medium, 5% (v/v) FBS, 1% (v/v) PS (4 mL). After 24 hours of culture, the cultures were treated with DSC or vehicle control as described previously. At three and six day timepoints, medium was aspirated and cells were released with trypsin (250 µl). Cells were pelleted, then resuspended in 500 uL T medium. 50 uL of a 0.4% (w/v) solution of trypan blue in buffered isotonic salt solution was added to each sample. The total number of white (live) and blue (dead) cells was determined by counting on a hemocytometer. The percentage of live cells was determined from this data.
Error bars on all figures display standard error of the mean (SEM). Significance was determined using Student’s two sample two-tailed t-tests with p<0.05 considered significant.
In initial experiments, we evaluated C4-2 cell morphology in the HA hydrogel cell culture system to determine if differences could be seen in response to FBS used as a motogenic factor to stimulate motility. Control cultures were treated with TCM™, a plant-based serum replacement to maintain cell growth and viability, without the ability to promote invasion and migration. While TCM™ treated C4-2 cells showed a lower metabolic output compared to FBS, assessed by WST assay, there was no significant difference in estimated cell counts for cells growing in either condition at either three or six day timepoints (see
As shown in
Images of C4-2 cells cultured for three or six days with either 2% TCM™ (control) or 5% FBS (motogenic factor) in the HA hydrogel invasion assay. Arrow and inset show a magnified view of the image to better display invadopodia. Black scale bars represent 200 µm, white scale bar on inset represents 50 µm (A). Quantification for average number of invadopodia (B) or percent merging clusters (C) in each imaged field. Error bars = SEM, n = 3, ***p<0.001.
Quantification of the number of invadopodia revealed that the number of these structures was significantly higher (p<0.0001) for FBS treatment than for TCM™ treatment at both three and six day timepoints (
Next, we evaluated the parameters of invadopodia number and merging cluster percentage in the HA hydrogel for PCa cells representing increasing disease progression. For this study we compared highly aggressive, metastatic C4-2 cells to their less aggressive LNCaP parent cell line. Similar metabolic activities and cell counts were observed when comparing LNCaP and C4-2 cultures at three and six day timepoints (see
At both three and six days of culture in the HA hydrogel, LNCaP cells grew in well-defined, irregularly shaped, grape-like clusters (
Images of LNCaP or C4-2 cells cultured for three or six days with 5% FBS in the HA hydrogel invasion assay. Scale bars represent 200 µm (A). Quantification for average number of invadopodia (B) or percent merging clusters (C) for each imaged field. Error bars = SEM, n = 3, *p<0.05, ***p<0.001.
We suspected that the invadopodia and merging cluster formation observed in the HA hydrogels was due in part to the chemical make-up of the hydrogel, not simply its mechanical properties or the process of being encapsulated in 3D. To test this, we compared C4-2 cell growth in HA hydrogels to that in non-physiologically relevant soft agar gels.
The rheological properties of the HA hydrogels were compared to that of 0.3% agar in
In soft agar, C4-2 cells formed small cell clusters that appeared to grow in size slightly between two and five days of culture (
Images of C4-2 cells cultured for zero, two, or five days with 5% FBS in the HA hydrogel or in 0.3% LMP agar. Scale bars represent 200 µm (A). Average elastic (G’) and loss (G”) moduli across one hour time for HA hydrogel (B) or 0.3% LMP agar (C). Error bars = SEM, n = 3.
While both HA and agar gels are soft and elastic, it is important to note that these gels differ in that the HA gel crosslinks covalently while the agar gel crosslinks physically. The crosslinking method, along with possible differences in porosity and stiffness, represent additional variables between the two gels. While we cannot know if any of these variables may account in part for the observed cell behavior, it is clear that the HA gels provide context-specific signals to the cells that the agar gels do not.
To understand how PCa cells interact with the HA hydrogel to form merging clusters and invadopodia within the invasion assay, we investigated the expression of several classes of HA-interacting proteins, starting with the HA receptors. Western blotting showed that both common isoforms of RHAMM were expressed across the LNCaP series, with lower expression by LNCaP cells compared to the more aggressive cell lines (
Western blotting and densitometry for RHAMM (A) and western blotting for CD44 (B) across the LNCaP series of cell lines with β-actin used as a load control. Cell surface biotinylation for RHAMM and CD44 with GAPDH utilized as a cytoplasmic control (C). PC-3 lysate was used as a positive control for CD44, Error bars = SEM, n = 3, *p<0.05 compared to LNCaP.
Because RHAMM is reportedly found to be bound to CD44 on the cell surface, we were interested to see if RHAMM was present on the cell surface where it could interact with the HA hydrogel in the absence of CD44. To test this, we performed a cell-surface biotinylation assay under non-permeabilizing conditions where only proteins accessible on the cell surface would be labeled (
Next, we investigated the effects of 3D HA hydrogel culture on RHAMM expression and localization. As shown by immunofluorescence, C4-2 cells cultured on 2D (
C4-2 cells with 5% FBS on 2D (A) or in 3D HA hydrogel for 3 days (B) immunostained for RHAMM (green) and nuclei (blue). Arrows indicate differences in nuclear RHAMM between the two culture conditions. A quantification method was utilized that measured RHAMM staining intensity across a single line both inside and outside of the nucleus (C). Arrows indicate marked points to average staining intensity between and outside of as depicted in the chart to the right. Results of the quantification method show nuclear and cytoplasmic or membrane (cyto/mem)-bound RHAMM for both culture conditions (D). Western blotting for RHAMM expression after five days of 2D or 3D culture with β-actin used as a load control (E).
Because we found that the C4-2 cells were differentially expressing and localizing RHAMM in response to 3D HA hydrogel culture and that LNCaP cells expressed less RHAMM than C4-2 cells, we hypothesized that RHAMM expression was necessary for invadopodia and/or merging cluster formation in the HA hydrogel invasion assay. To test this hypothesis, we utilized a RHAMM knockdown approach on C4-2 cells.
RHAMM Western blotting for siRHAMM or low GC control transfected C4-2 cells across six days of culture with β-actin used as a load control (A). Fluorescent staining of siRHAMM or low GC control transfected cells at day 2 on 2D for F-actin (green) and nuclei (blue) (B). Quantification of average process length for all cellular processes within the imaged field (C). Error bars = SEM, n = 3, *p<0.05.
When cultured on 2D and stained with phalloidin to visualize the cytoskeleton, siRHAMM C4-2 cells showed a different morphology compared to control cells (
To determine if the siRHAMM phenotype observed in 2D culture translated to differences in 3D culture, we cultured C4-2 cells in the 3D HA hydrogel invasion assay. Compared to control, siRHAMM treated cells showed no visible difference in morphology in the HA hydrogel invasion assay at either timepoint (
Images of siRHAMM or low GC control transfected C4-2 cells cultured for two or five days with 5% FBS in the HA hydrogel invasion assay. Scale bars represent 200 µm (A). Quantification for average number of invadopodia (B) or percent merging clusters (C) in each imaged field. Error bars = SEM and n = 3.
After determining that RHAMM knockdown did not affect invadopodia and cluster formation of C4-2 cells in the HA hydrogel invasion assay, we investigated if another family of HA interacting proteins, HAases, may be responsible. Western blot analysis showed that cells of the LNCaP series express both Hyal1 and Hyal2 (
Western blotting for Hyal1, Hyal2, and β-actin as a load control across the LNCaP cell series and for LNCaP and C4-2 conditioned medium (CM) (A). Hyaluronidase substrate gel activity assay for LNCaP series cell lysates and 50 µg BTH (B). Alcian blue stains HA (clear areas indicate HAase activity), Coomassie stain shows total protein as a load control. Clear arrows indicate major clear zones, black arrows indicate protein bands suitable for load controls, MW = molecular weight marker in kDa.
To determine if HAase expression was necessary for invadopodia and/or cluster formation in the HA hydrogel invasion assay, we utilized a HAase inhibitor, DSC. When applied to C4-2 cells in the HA hydrogel invasion assay at the IC-50 value of 500 µM, DSC showed noticeably smaller, unbranched cell clusters on day three compared to vehicle control, as well as a lack of invadopodia (
Images of C4-2 cells cultured for three or six days with 5% FBS and either 500 µM DSC or vehicle control in the HA hydrogel. Scale bars represent 200 µm (A). Quantification for average number of invadopodia (B) or percent merging clusters (C) in each imaged field. Trypan blue exclusion assay to test C4-2 cell viability at three and six days with either 500 µM DSC or vehicle control (D). Error bars = SEM, n = 3, *p<0.05, **p<0.01.
To ensure that the observed differences in invadopodia and merging cluster formation were not attributable to DSC toxicity to the cells, we performed a trypan blue exclusion assay. This assay tested C4-2 cell viability with 500 µM DSC at the three and six day timepoints used for the HA hydrogel invasion assay. The trypan blue assay showed similar levels of high cell viability (>95%) compared to vehicle control for cells observed at both timepoints (
In this report, we provided data showing that the 3D HA hydrogel invasion assay provides an easily quantified, physiologically relevant method to study cancer invasion and the functions of individual HA interacting proteins likely to be involved in migration through HA-rich connective tissues. The quantifiable parameters of invadopodia and merging cluster formation were shown to be independent of cell number, and yielded low error values allowing for the use of statistical tests of significance. We suggest that this method could easily be modified to fit the unique experimental needs for examination of a variety of invasive cancer cells. While there is no doubt that there remain differences between soft, HA-rich connective tissues and the 3D HA hydrogels we use, this method offers a useful alternative to existing methods for mechanistic studies of cell invasion. As the field progresses, a deeper understanding of the mechanical properties and composition of these tissues will inform the development of more physiologically relevant hydrogels. Additionally, while understanding of how the chemically modified HA compares to native HA is somewhat limited, our data demonstrates that cells can interact with the crosslinked HA composing the hydrogel by degrading and binding it. How these interactions compare to interactions with native HA is unknown, but remains an exciting area of discovery.
Using this HA hydrogel method, we tested our hypothesis that the formation of invadopodia along with merging cluster formation that we observed in the HA hydrogels correlated with the ability of the PCa cells to interact with the hydrogel matrix by virtue of their expression of functional HA interacting proteins. Although it has been examined in several studies, the expression and function of the HA receptor, CD44, in prostate cancer invasion and in the LNCaP series of cell lines has failed to lead to general consensus. Some groups have reported a lack of CD44 mRNA and/or protein expression in these cell lines
The alternative HA receptor, RHAMM/CD168, was an intriguing candidate because of its foundational role in cell motility through G protein dependent pathways
The RHAMM expression data, combined with the data showing lower amounts of invadopodia and merging cluster formation in LNCaP cells compared with C4-2, led us to wonder if RHAMM expression may be a driving factor behind motility in the LNCaP series. However, for RHAMM to interact with the HA hydrogel matrix, it would need to be on the cell surface, and RHAMM has been thought to be sequestered on the cell surface by binding to the surface receptor for HA, CD44
To begin investigating if expression of RHAMM allows the C4-2 cells to functionally interact with the components of the HA hydrogel, we examined RHAMM expression and localization in 3D vs. 2D culture. We found that RHAMM was localized predominantly in the nucleus of cells grown in 2D, but a larger pool was found in the cytoplasm or membrane in 3D culture. This redistribution may reflect RHAMM translocalization from the nucleus to the cytoplasm or membrane in response to the HA present in the cells’ microenvironment or it may reflect a direct transport of newly synthesized RHAMM to the cell surface in cells grown in HA-rich 3D hydrogels. Supporting this, RHAMM expression in 3D culture was significantly higher (p = 0.013) than in 2D culture at day 5. These increases in RHAMM expression and cell surface and/or cytoplasmic localization also may lead to increased activation of downstream signaling pathways associated with motility
With the previous suggestion that RHAMM expression is related to PCa aggression
Next, we investigated what role, if any, HAases play in PCa stromal invasion. HAases were expressed, secreted, and activated at similar levels at all stages of progression the cell lines represented. This suggests that even locally advanced prostate cancers, as represented by LNCaP cells, already have activated HAases that can assist in migration through HA-rich connective tissues. This data indicates that HA degradation by HAases may be as important in the development of early metastatic potential of a tumor as it is in the later stages of metastasis. Previous studies have shown that HAase secretion is 10 fold higher in PCa cells compared to both normal adult prostate and benign prostatic hyperplasia
To test the need for active HAase in prostate cancer cell motility in 3D HA-rich hydrogels, we utilized a HAase inhibitor, DSC, and C4-2 cells treated at its IC-50 concentration of 500 µM
Invadopodia are formed through a combination of both production of degradative enzymes and activation of motility promoting proteins
Compelling arguments can be made suggesting Hyal1, Hyal2, or a combination of the two enzymes, act as driving force(s) behind PCa stromal invasion. Hyal1 has been most strongly implicated in the metastasis of several cancer types, including PCa
Taken together, our results suggest that DSC, and other HAase inhibitors, may be effective in preventing stromal invasion by PCa cells, even early in the course of locally invasive disease. In combination with other interventions, DSC or other HAase inhibitors, such as ascorbic acid derivatives or alkyl gallates, could be incorporated into treatment regima for PCa patients to help prevent tissue invasion and metastasis. DSC is currently FDA approved for asthma and allergy treatment, making it an attractive candidate for further pre-clinical study
Here, we describe the development and use of a novel 3D HA hydrogel system that can be used to study aspects of invasion, including formation of invadopodia. This assay provides a more typical microenvironment for PCa cells that have metastasized to the bone marrow or other HA-rich connective tissues. Invasion in this assay can be quantified by counting the number of invadopodia and merging cell clusters. Invasion also depends on the ability of the PCa cells to interact with the HA within the hydrogel matrix. Knockdown of RHAMM expression did not affect invadopodia or merging cluster formation, while inhibition of HAase activity decreased both invadopodia and merging cluster numbers. HAase expression and activity are necessary but not sufficient for PCa cell invasion in the HA hydrogel. HAases may play an essential role in PCa invasion through HA-rich connective tissues, such as the bone marrow. Therefore, HAase inhibitors are attractive candidates for drugs to prevent PCa invasion and metastasis even early in disease progression.
Mean Cell Counts of LNCaP and C4-2 Cells Growing in HA Hydrogels. Cell numbers were determined by counting cells in clusters as described in Materials and Methods.
(DOCX)
We thank Drs. Kirk Czymmek and Jeffrey Caplan along with the staff of the Bioimaging Center at the Delaware Biotechnology Institute for their help with microcopy and quantification methods. We thank the members of the Farach-Carson laboratory, especially Drs. Swati Pradhan Bhatt and Daniel Harrington and Ms. Eliza Fong, for many helpful discussions. We thank Genzyme for providing HA. We thank Drs. Nicholas Petrelli and Robert Sikes, co-directors of the Center for Translational Cancer Research (CTCR), for use of the facility.