Upregulation of collagen matrix crosslinking directly increases its ability to relieve stress under the constant strain imposed by solid tumor, a matrix property termed stress relaxation. However, it is unknown how rapid stress relaxation in response to increased strain impacts disease progression in a hypoxic environment. Previously, it has been demonstrated that hypoxia-induced expression of the crosslinker procollagen-lysine, 2-oxoglutarate 5-dioxygenase 2 (PLOD2), in sarcomas has resulted in increased lung metastasis. Here, we show that short stress relaxation times led to increased cell migration along a hypoxic gradient in 3D collagen matrices, and rapid stress relaxation upregulated PLOD2 expression via TGFβ-SMAD2 signaling, forming a feedback loop between hypoxia and the matrix. Inhibition of this pathway led to a decrease in migration along the hypoxic gradients. In vivo, sarcoma primed in a hypoxic matrix with short stress relaxation time enhanced collagen fiber size and tumor density and increased lung metastasis. High expression of PLOD2 correlated with decreased overall survival in patients with sarcoma. Using a patient-derived sarcoma cell line, we developed a predictive platform for future personalized studies and therapeutics. Overall, these data show that the interplay between hypoxia and matrix stress relaxation amplifies PLOD2, which in turn accelerates sarcoma cell motility and metastasis.

Significance:

These findings demonstrate that mechanical (stress relaxation) and chemical (hypoxia) properties of the tumor microenvironment jointly accelerate sarcoma motility and metastasis via increased expression of collagen matrix crosslinker PLOD2.

Cell migration through the extracellular matrix (ECM) is required for cancer metastasis (1). With recent advances in analyzing and mimicking the tumor ECM, the role of ECM properties in cancer cell migration has been more thoroughly investigated, mainly focusing on the increased stiffness of the developing tumor (2). The natural ECM has a hierarchical structure that responds to different stress and strain according to the ECM's viscoelastic properties. Viscoelasticity describes how a material responds to deformation, and is composed of elastic and viscous characteristics. When elastic materials are deformed, a strain is generated simultaneously. Viscous materials, on the other hand, show a time dependency upon deformation. That is, the resulting strain is a function of time. Consequently, when a viscoelastic material is deformed, a time-dependent strain will ensue. Similarly, when the mechanical load is removed, the deformation will gradually return to its original form due to the entropy-driven nature in the viscoelastic material, and the time that is needed to recover a portion of its original form is often called relaxation time (3, 4). It is worth noting that stress relaxation is often related to the application of a constant strain. Relevant to the tumor microenvironment, stress relaxation has been shown to correlate with solid stress (5, 6), which is the mechanical stress that is contained and transmitted by the solid and elastic elements of the ECM and cells (i.e., tumor microenvironment; ref. 7). The increase in solid stress causes short stress relaxation times in the tumor ECM. In addition, the increase in solid stress occurs from an increase in tumor pressure (8), which was attributed to the collagen matrix in the tumor (6). The pressure increases causes a decrease in vascular permeability (5), exacerbating the hypoxic environment (9). Interestingly, different tumors have been shown to have unique stress relaxation profiles (10). Stress relaxation affects numerous cellular behaviors, including cell motility (11), mesenchymal stem cell differentiation (12), and the cell's ability to remodel the ECM (13). However, stress relaxation remains an understudied viscoelastic property of tumor ECM.

Polymer-based hydrogels are currently utilized to recapitulate the three-dimensional (3D) cellular microenvironments (14). In cancer research, hydrogels are designed to mimic the chemical and physical properties of the tumor ECM. Several studies have investigated the effect of stress relaxation on cell behavior (4, 15, 16) particularly in cancer (5, 10). In sarcoma, collagen is one of the most abundant components of the ECM (17); and elevated expression of the collagen crosslinker, procollagen-lysine, 2-oxoglutarate 5-dioxygenase 2 (PLOD2), is associated with a more metastatic phenotype (18, 19). In this study, we investigated the role of stress relaxation in sarcoma cell migration and metastasis using collagen matrix with varying stress relaxation.

Finally, intratumoral hypoxia occurs as the tumor outgrows its blood supply. Hypoxia is a key regulator of cancer development (20) and correlates with poor prognosis for patients with sarcoma (21). Specifically, hypoxia has been shown to regulate PLOD2 expression in sarcoma, leading to a metastatic phenotype (18, 22, 23). This hypoxia-triggered PLOD2 has been shown to modify the ECM. Recently, we have shown that hypoxic gradients, which are present in sarcoma ECM, induce cell migration in the direction of higher oxygen, by modifying the ECM. (16, 18, 24).These modifications can impact mechanical properties of the matrix, one of which is stress relaxation.

To understand how stress relaxation modulates sarcoma progression, it is important to study it independently from other matrix properties (stiffness, porosity, and fiber density) that have been shown to modify cell behavior. By isolating a single factor, this platform can truly be used to explore a stress relaxation change in hypoxia. Here, we decouple stress relaxation from other matrix properties in collagen gels, in both hypoxic and nonhypoxic environments. Using murine and human undifferentiated pleomorphic sarcoma cells (UPS), we show that shorter stress relaxation time in hypoxic gradients leads to faster cell migration along an increasing gradient of oxygen tension causing an increase in PLOD2. The PLOD2 increase is a combinatorial effect of hypoxia and the quicker stress relaxation stimulating PLOD2 upregulation through the TGFβ-SMAD2 signaling axis. This amplification feedback loop was not observed in nonhypoxic conditions. Upon implanting the quicker and slower stress relaxation hydrogels in a murine model, we show increase in collagen fiber size and density in the solid tumor, as well as in metastasis events.

Cell culture and primary tumor cell extraction

KIA was derived from murine model of sarcoma, LSL-KrasG12D/+, Ink4a/Arffl/fl as established previously (18). KIA cells were authenticated as reported previously (18). KIA-GFP was expanded under standard culture conditions in high-glucose DMEM with 10% FPBS and 1% penicillin/streptomycin. hUPS cells were extracted from a primary human tumor sample treated at the Hospital of the University of Pennsylvania (Philadelphia, PA). A clinical pathologist, following obtaining written informed consent from the patients, authenticated the tumor. Studies were conducted in accordance with U.S. Common Rule, and were approved by Institutional review board protocol #822382. hUPS cells were extracted by incubating the primary tumor with a 3 mg/mL collagenase solution in DMEM for 45 minutes at 37°C. The tumor was then spun down and the supernatant was aspirated and replaced with 0.05% trypsin EDTA solution and again incubated for 15 minutes at 37°C. The digestion is then stopped with complete cancer media. The tumors again were spun down, aspirated, and replaced with serum-free media. The tumor was washed two more times and then triturated through a 40-μm cell trainer into a 50-mL conical tube. The cells were then spun down and plated. Cells were selected by continuous culture for up to 8 passages prior to authenticating the line. All cell lines were used before passage 20 and were tested for Mycoplasma every quarter.

Collagen gel preparation and cell encapsulation

Collagen gel was prepared as reported previously (16). Briefly, to make 1 mL of 3 mg/mL collagen solution, 35.96 μL of M199 10× medium (Thermo Fisher Scientific) was mixed with 11.51 μL of 1 mol/L NaOH until it turned dark pink. Next, we used acid-soluble rat tail collagen I at 9.61 mg/mL (Corning #354249). A total of 312 μL of collagen was added to the dark pink solution and mixed thoroughly. A total of 640.35 μL of M199 1× medium (Thermo Fisher Scientific) was then mixed with the solution to create our final collagen gel solution. The solution was then incubated on ice for 2 hours. For cancer cell encapsulation, we first prepared cell pellets of hUPS-GFP and KIA-GFP cells (7.5 × 105 cells) in a 1.5 mL Eppendorf tube. We then mixed the pellet with 0.5 mL of collagen solution by gentle pipetting to give a homogeneous cell suspension. After mixing, the solution was pipetted into a 96-well plate (BD Biosciences) and incubated for 30 minutes at 37°C. Nonhypoxic gels were made with 45 μL of collagen and hypoxic gels were made with 90 μL of collagen.

To alter the stress relaxation, gels were incubated with PBS 24 mg/mL microbial transglutaminase (mTG) for 2 hours at room temperature. After 2 hours, the solutions were aspirated and the gels were cultured in normal culture media. Two-hundred microliters of media was added to the hypoxic gels and 100 μL was added to the nonhypoxic gels.

Control over oxygen levels

The oxygen concentration was controlled as described previously (16). In short, to create a hypoxic hydrogel, the key parameters including diffusion coefficient of the collagen hydrogel material, boundary conditions, and cell oxygen consumption rate, were optimized. Collagen was chosen due to its relatively low oxygen diffusion coefficient (25). In addition, polystyrene Petri dish also has a very low diffusion coefficient (26). Thus, the polystyrene creates an impermeable barrier on the sides and bottom of the hydrogel placed within this common cell culture Petri dish, limiting diffusion of oxygen and nutrients only in one direction—from the culture medium on the top of the hydrogel. Finally, the oxygen consumption rate was modulated to balance the diffusion of oxygen, based on the height of the hydrogel and medium, with the cell consumption rate. This oxygen consumption rate is dependent on the concentration of the encapsulated cells in the hydrogel.

Invasive oxygen measurements

Dissolved oxygen levels were monitored noninvasively and invasively as reported previously (16, 24). A commercially available sensor patch (PreSens) and a Needle-Type Housing Fiber-Optic Oxygen Microsensor (PreSens) were used for the noninvasive and invasive experiments, respectively. For the noninvasive oxygen measurements, the sensor patches were placed at the bottom of a 96-well plate and the gel was formed on top of it. Invasive oxygen sensors were used to measure the oxygen concentration on day 3. The needle was placed into the gel until it hits the bottom and the oxygen was measured every 0.25 mm for the nonhypoxic gel and every 0.5 mm for the hypoxic gel.

Matrix degradation

Analyzing protease-mediated matrix degradation was performed as published previously (16) by incorporating 10 μg/mL of DQ collagen (Invitrogen) into polymer solution prior to cell encapsulation. On day 3, DQ Collagen was analyzed by fluorescence microscopy (Axio Observer Z1 Zeiss) and measured using fluorescence spectrophotometry at a wavelength of 495 nm excitation and 515 nm emissions (Molecular Devices). High-resolution fluorescence and light microscopy images were taken and overlaid to determine the localized protease activity in the cells’ surrounding.

Cell tracking and drug treatment

For cell tracking, a previously developed method (16, 24, 27) was used to assess sarcoma cell migration. KIA-GFP cells were encapsulated in hydrogels with different oxygen levels and mTG treatment. On day 3, cells were tracked with live-cell, 3D confocal microscopy (LSM 780; Carl Zeiss) equipped with an incubator (5% CO2 and 37°C). The time-lapse and z-stack images (>200 μm thickness) were collected every 30 minutes up to 24 hours at five randomly selected positions. Cells that did not start in frame were not included in the analysis to properly optimize the experiment. The images were analyzed with Imaris spot analysis software (Imaris 8.2; Bitplane). The 3D migration analysis was performed with a previously developed strategy (16, 24, 27). In brief, a minimum of 100 individual cells were tracked to generate x, y, and z coordinates at each time point. These were then sorted to only include cells present at time 0. The time that each of the sorted cells were in frame was calculated and the most common time was used to pick cells for tracking analysis, maximizing the sample size of cells in the analysis. Velocity, speed, and mean squared displacements were calculated using code adapted from previously established methods for triplicate tracking trials (n = 3; refs. 16, 24, 27). The statistical analysis was performed using MATLAB (Mathworks, Inc.) to calculate the mean, SD, and SEM. Where appropriate, a t test was performed to determine significance (GraphPad Prism 4.02; GraphPad Software, Inc.). Graphed data are presented as average ± SEM. Significance levels were set at *, P < 0.05.

For minoxidil and TGFβ inhibitor treatment, cells were cultured in hydrogels under the conditions and methods described above for 3 days. On day 3, 0.5 mmol/L minoxidil or 10 μmol/L TGFβ inhibitor dissolved in KIA cell culture media was added to the wells containing the hydrogels and the cells were tracked for 24 hours. Untreated cultures served as controls. Cell tracking and data analysis were performed as described above.

Traction force microscopy

Collagen gel solution was prepared as stated above. One-micron–sized FluoSpheres that are Carboxlyated modified (Thermo Fisher Scientific) were rinsed with PBS three times and then mixed with the cell pellet prior to hydrogel formation. On day 3 of culture, z-stacks of cells and beads were imaged at 1-minute interval for 20 minutes to capture bead movement on a confocal microscope (LSM 780, Zeiss). The tracking analysis was performed as stated above. IMARIS was used to track bead movement through the gel and MATLAB was used for analysis. To calculate the force on the beads, previously published formula for a particle moving through a stress relaxation hydrogel was used (28).

Rheology and stress relaxation analysis

For all rheological experiments, a AR1500EX rheometer (TA Instruments) was used. A time sweep was performed on the mTG and control gels at 1% strain and 10 Hz to monitor the storage (G″) and loss modulus (G′) at 37°C. A solvent trap with DI water or an immersion cup with PBS was used to prevent evaporation. Stress relaxation experiments were performed for strains from 0.5%–1% for 2 minutes. During the stress relaxation test, there was a strain ramp of 5 seconds prior to relaxation. Data recorded after the completion of the strain ramp was analyzed. This strain was chosen to stay within the linear viscoelastic region of the hydrogel, to not permanently deform the hydrogel. As previously reported, a single element Maxwell model was used to fit the stress relaxation constants (16, 28).

In brief, the equation for a single spring in series with a single dashpot that describes viscoelastic fluids was fitted to stress relaxation curves:

formula

In the above equation, |{\sigma _0}$| is the initial stress, |\tau $| is the time constant of stress relaxation, and t is the time. We have previously shown to fit the relaxation of collagen gels using this equation (16). Frequency sweep was performed at 1% strain from 0.5 to 10 Hz to study the change in moduli in response to increasing frequency. Creep and recovery test was also performed. All samples were initially stressed at 10 Pa and relaxed for 2 minutes followed by a strain recovery test. This stress was chosen to stay within the linear viscoelastic region of the hydrogel, to not permanently deform the hydrogel. In the strain recovery test, the applied stress was removed and strain was monitored over time. A four-element Maxwell–Voight model was used to fit the relaxation constant as performed previously (3). A stress–strain curve was generated by getting the corresponding stress value at a given strain.

Reflective microscopy and second harmonic generation

To analyze collagen fibers in the collagen gels, reflective microscopy was used. Images were collected using an LSM 780 (Zeiss) microscope. For reflective microscopy, a 40× oil immersion objective and 405 nm light were used to illuminate and capture collagen fibers. To examine collagen in the murine tumors or metastatic lungs, histologic slides from the center of the tumor were imaged using SHG microscopy as described previously (18). SHG microcopy was carried out with a LSM 510 (Zeiss) microscope equipped with a Chameleon Ultravision II Laser Module (Coherent). A wavelength of 810 nm was used to image hematoxylin and eosin (H&E) slides of tumors and light was collected at 488 nm wavelength.

Collagen fiber morphology, orientation, and cell aspect ratio analysis

Collagen fiber orientation was adapted from a previously developed method using MATLAB (Mathworks; ref. 16) of collagen on mTG and control hydrogels. Collagen fiber width and length were analyzed using a ct-Fire analysis, a previously established method (16, 29) on SHG images of tumors. The software segments the image and calculates intensity gradients within subregions of images and uses them to track the overall directions of fibers. To study cell aspect ratio, code was developed to fit an ellipse around the cell and the aspect ratio is the ratio of the major to minor axis of cells on day 3 in the mTG and control hydrogels.

Scanning electron microscopy

Collagen hydrogels, with and without cells, were fixed as described previously and processed for conventional scanning electron microscopy (30). Briefly, dime-sized gels were fixed in 100 mmol/L cacodylate buffer, pH 7.4, containing 3% formaldehyde, 1.5% glutaraldehyde, and 2.5% sucrose for 1 hour, washed, and osmicated at 4°C in Palade fixative containing 1% OsO4. Cells were then dehydrated through a graded series of ethanol; critical point dried using a Tousimis Model 795 critical point dryer, and coated with 4-nm platinum using a Anatech Hummer 6.1 sputter coater. Samples were imaged in an FEI Quanta 200 ESEM using the Everhart–Thornley secondary electron detector at 1.2 keV under high vacuum.

Transmission electron microscopy

Collagen hydrogels, with and without cells, were fixed as described previously (30). Briefly, for conventional electron microscopy (EM), gels were fixed in 100 mmol/L cacodylate buffer, pH 7.4, containing 3% formaldehyde, 1.5% glutaraldehyde, and 2.5% sucrose for 1 hour, washed, and osmicated at 4°C in Palade fixative containing 1% OsO4. Cells were then washed, en bloc stained with Kellenberger uranyl acetate overnight, dehydrated through a graded series of ethanol, and subsequently embedded in epon. Eighty-nanometer–thick sections were cut on a LEICA UCT ultramicrotome, collected onto 300 mesh formvar/carbon-coated grids, and poststained in uranyl acetate and lead citrate. Images were recorded using a Soft Imaging System Megaview III camera mounted on a Tecnai 12 TEM operating at 100 keV and figures were assembled in Adobe Illustrator.

RT-PCR

A quantitative real-time RT-PCR assay was used to assess gene expression. Total RNA was extracted from cells encapsulated in hydrogels with TRIzol (Invitrogen) in accordance with manufacturer's instructions. Hydrogels were placed in 500 μL of TRIzol and homogenized with a microhomogenizer. The suspension was then centrifuged at 12,000 × g for 15 minutes and the supernatant was separated. One-hundred microliters of chloroform was mixed with the solution by manual agitation for 20 seconds and the mixture was centrifuged at 12,000 × g for 10 minutes. The supernatant was isolated and mixed with 250 μL isopropyl alcohol and kept at −4°C for 1 hour. The precipitates were separated by centrifugation at 7,500 × g for 5 minutes and washed with 70% ethyl alcohol. Total RNA was quantified using an UV spectrometer and validated by lack of DNA contamination. One microgram of RNA was transcribed using reverse transcriptase M-MLV and oligo(dT) primers (both from Promega), according to the manufacturer's instructions. TaqMan Universal PCR MasterMix and Gene Expression Assay (Applied Biosystems) was used according to the manufacturer's instructor for PLOD2, MMP2, MMP9, and HPRT1.

The Cancer Genome Atlas analysis

Gene Expression Profiling Interactive Analysis (GEPIA; ref. 31) was used to generate survival curves as well as study trends in RNA-sequencing data, based on querying The Cancer Genome Atlas (TCGA) database. GEPIA also generated and provides all the statistics.

Generation of GpNLuc KIA–stable cell lines

KIA cells (70% confluent) were transfected with pRetroX-Tight-MCS_PGK-GpNLuc (Addgene plasmid #70185) using Lipofecatime 2000 (Thermo Fisher Scientific) in serum-free medium as per the manufacturer's instructions. GpNLUC is a bioluminescent resonance energy transfer (BRET) reporter in which the luciferase bioluminescence excites the green fluorescent protein (32). Mixed populations were then FACS sorted for high GFP signal using a Sony Biotechnology SH-800 “chip” cell sorter at one cell per well in a 96-well plate. Stably expressing clonal populations were then utilized for downstream experiments.

In vivo primary tumor model

The Johns Hopkins University's Institutional Animal Care and Use Committee approved all animal protocols. KIA cells, expressing GFP or GpNLuc that were encapsulated in quick and slow stress relaxation hypoxic hydrogels as reported above, were subcutaneously implanted in 8- to 10-week-old nude mice. Caliper measurements were performed to track subcutaneous tumor growth. The volume of the tumor was calculated using the following equation:

formula

where V is the volume of the tumor, L is the major axis, and W is the minor axis of the tumor (33). In addition to caliper measurements, mice were injected intraperitoneally with 20 μg/mL of furimazine followed by luminescence imaging with IVIS Spectrum (PerkinElmer) within 5 minutes of injections. Average radiance (photons per second per centimeter squared per steradian) was observed at time point and tumor.

In addition, primary tumors were generated as reported previously (18, 24). In short, 1 × 106 cells were injected into each flank of the mouse with either the KIA scrambled control line (KIA Scr) or PLOD2 knockdown line (PLOD2(−)). Tumors were allowed to grow for 20 days prior to extraction.

Tumor rheology

Rheological measurements on the tumors using an AR1500EX rheometer (TA Instruments). A time sweep was performed at 1% strain and 5 Hz to monitor the storage (G’) and loss modulus (G″) at 37°C. An immersion cup with media was used to prevent evaporation. Stress relaxation experiments were performed for 1% strain for 2 minutes with a strain ramp of 5 seconds prior to relaxation. To analyze the stress relaxation, the methodology listed above was used.

All samples were initially stressed at 20 Pa and relaxed for 2 minutes, followed by a strain recovery test. Other studies have previously used a similar to measure mechanical properties in tumors (34, 35). In the strain recovery test, the applied stress was removed and strain was monitored over time. To analyze the stress relaxation, the methodology listed above was used.

In vivo metastatic lesions model

The Johns Hopkins University's Institutional Animal Care and Use Committee approved all animal protocols. KIA cells, expressing GFP or GpNLuc that were encapsulated and cultured for 3 days in slow stress relaxation nonhypoxic gels and quick stress relaxation hypoxic hydrogels. Per side, one quick stress relaxation hypoxic hydrogels or two slow stress relaxation nonhypoxic gels were subcutaneously implanted in 8- 10-week-old nude mice. Tumors were let to grow for 20 days and then resected. Mice were left for additional 15 days after tumor resection, to allow lung metastasis to grow. On day 15, lungs were fixed and processed for histology.

IHC and immunofluorescence staining

For murine tissue, following fixation the tissue was dehydrated with serial ethanol (70%–100%), embedded in paraffin, 5-μm sections were cut, and subsequently treated with H&E stain, Masson trichrome, or PLOD2. Immunofluorescent stains were performed as previously reported on gels (27). SMAD2/3 (ProteinTech), vinculin (Sigma), phaloidin, and YAP (Santa Cruz Biotechnology) were all used at a 1:200 dilution and counterstained with DAPI. An anti-rabbit Alexa Fluor 564 (Life Technologies) was used at 1:500 dilution and imaged with Zeiss LSM 780 or LSM 800 microscope. PLOD2 (ProteinTech) was also used at 1:200 dilution for histology sections.

Quantification of immunostains

IHC stains for PLOD2 were quantified as reported previously (36). In short, background was taken from an empty portion of the slide, the color deconvolution plug-in ImageJ (NIH, Bethesda, MD) was used to extract the DAB color of the image. Finally, the mean intensity was calculated from the DAB intensity over the entirety of the image.

For quantification of SMAD2/3 localization, IMARIS (Bitplane) was used to find the colocalization of SMAD2/3 and DAPI from z-stack images of the immunofluorescence-stained cells. Surface area of the colocalization channel was then calculated using the surface package in IMARIS.

To quantify vinculin spots and actin filaments per cells, the number of each was manually counted and normalized to the number of filaments or spots per cell.

Quantification of pore area using scanning electron microscopy

Scanning electron microscopy images of cells moving through the hydrogel were analyzed using ImageJ. The pore area was measured by drawing a region of interest over the pore.

Quantification of transmission electron microscopy

ImageJ was used to measure the bands in transmission electron microscopy (TEM) images of the collagen fibers. A line was drawn over the length of the light or dark segments to analyze the length of the banding.

Quantification of metastatic lesion

The number of metastatic nodes was counted manually from H&E histologic sections of the lungs. In addition, using ImageJ, the area of metastasis was calculated and normalized to the area of the lung to determine percent area covered by metastasis.

Statistical analysis

Statistical analysis was performed using MATLAB (Mathworks, Inc) or Prism 4 (GraphPad Software, Inc.) to calculate the mean, SD, and SEM. t test and one-way ANOVA were performed where appropriate to determine significance (GraphPad). Biological and technical replicates are indicated throughout the figure captions. All graphical data are reported as mean ± SEM (*, P < 0.05; **, P < 0.01; ***, P < 0.001; and ***, P < 0.0001).

Design and characterization of stress relaxation collagen gel platform

To generate collagen gels with varying stress relaxation times, we fabricated a collagen hydrogel with defined fiber density by standardizing the ice incubation time prior to gelation as described previously (16). Once the hydrogel structure was established, microbial transglutaminase (mTG) was added on top of the completely formed hydrogel, to further crosslink the collagen fibers and modulate the stress relaxation time of the hydrogel. We hypothesize that mTG crosslinks the existing structure at primary ammines in between the already formed collagen fibrils. The stress relaxation time is the time it takes for a material to return to normal after a strain is applied (Fig. 1A). The stress relaxation time is determined by fitting the Maxwell equation to the stress relaxation curve. Mathematically, the Maxwell model is used to understand viscoelastic materials, such as the ECM. In brief, the equation is derived for a single spring, elastic component, in series with a single dashpot, viscous component. It is derived from adding the stresses from the elastic (Hookean) modulus with the Newtonian viscosity. This equation relates the stress change with time at a constant strain (see Materials and Methods). It has been used previously to model biological tissues (4). The stress relaxation time is determined by fitting the Maxwell equation to the stress relaxation curve. Low concentration of microbial–transglutaminase (mTG; 24 mg/mL) was chosen to allow for crosslinking of the collagen fibers as a means to change stress relaxation without altering the stiffness of the collagen gel (Fig. 1B–D). Consequently, the stress relaxation time decreased by half an order of magnitude (Fig. 1B–D). In Fig. 1D, we can see that the modulus, tensile elasticity G′, plateaus faster in the mTG-treated gel, corresponding with a quicker stress relaxation. It should be noted that tests of the mTG crosslinked gels and nontreated controls using higher stains show a difference in the stress relaxation with no difference in the modulus between the two hydrogels (Supplementary Fig. S1). The stress relaxation time was also verified using a creep and strain recovery test (Supplementary Fig. S2A and S2B); and a similar decrease in stress relaxation was observed. Furthermore, hydrogel response to frequency was evaluated to explore the difference in slope when frequency is plotted versus G′. The greater the rate of change in G′, as frequency increases, confirms a quicker stress relaxation material (Supplementary Fig. S2C and S2D; ref. 12).

To further characterize the hydrogel scaffold, reflection microscopy and scanning electron microscopy were performed on the mTG-treated hydrogel and the nontreated control. Interestingly, we could not observe any differences between the two hydrogel structures at the length scale of cells (Fig. 1E). In addition, using TEM, we could not detect a significant difference in the length of the light and dark segments of the collagen fibers, thus revealing no difference in the fibril organization of the collagen fibers (Fig. 1E). These light and dark segments are indicative of the how collagen alpha helixes are arranged in the superstructure of collagen. Overall, using mTG to modify the stress relaxation of the collagen gels through additional crosslinks to the fibers, from the above reflective imaging, scanning electron microscopy, and TEM analyses, we could not detect change in the hydrogel structure (Supplementary Fig. S3). We therefore conclude that there is no significant structural difference, at the cellular length scale, between the mTG hydrogel and the control.

Morphology and migration of sarcoma cells in hydrogels with controlled stress relaxation profile

Having established the hydrogel platform, the influence of stress relaxation on sarcoma cell migration was explored in both hypoxic and nonhypoxic gradients. To investigate the effect of stress relaxation on cell migration, we used murine (KIA) autochthonous genetically engineered mouse model (16, 24, 37). The KIA cells were chosen due to their highly migratory and invasive behavior as well as the ability to metastasize from subcutaneous tumors (16, 18, 24, 38). To establish oxygen gradients, we used our previously reported approach, in which we balanced the diffusion coefficient of the material, boundary conditions, and cell concentration (16). We measured oxygen tension throughout the hydrogels’ depth in all conditions and confirmed the presence of hypoxic and nonhypoxic gradients in all conditions after 72 hours (Fig. 2A). In addition, we compared the morphology of the cells in the nonhypoxic region of the hypoxic hydrogels to the cells in nonhyoxic hydrogels and found no difference in cell aspect ratios in the two different gels with similar oxygen tension (Supplementary Fig. S4). We further confirmed no change in the stress relaxation of the collagen hydrogels after 72 hours in both hypoxic and nonhypoxic conditions, showing that the stress relaxation time is consistent over the course of the experiment (Supplementary Fig. S5). KIA cells were found to migrate more in both hypoxic and mTG-treated gels compared with untreated gels (Fig. 2B and C).

We have previously demonstrated that a large-cell aspect ratio is consistent with a more migratory phenotype (16). Thus, we examined the motility of KIA cells in the three-dimensional space on the third day after encapsulation (16), and found there were minimal differences in the instantaneous cell velocities between conditions (Supplementary Fig. S6). Next, we examined the migration persistence measured by the mean square displacement (MSD), and found no significant increase in the nonhypoxic conditions (Fig. 2D). However, in the hypoxic gradient, we noticed a slight increase in the cell migration persistence in the x and y directions, but a stark increase in the z direction in the mTG-treated gels (Fig. 2E). This demonstrates that the cells in quicker stress relaxation gel migrate faster.

Finally, to examine how the cells are interacting with the matrix, we performed traction force microscopy. For this, 1-μm carboxylated polystyrene beads were incorporated into the matrix with cells and were tracked in 3D space as cells deform the matrix. The beads were large enough to neglect Brownian displacements and are effectively tethered to the collagen matrix fibers due to the reactive carboxyl groups (39). Thus, any movement of the beads through the matrix is a result of cell-mediated matrix deformation. We plotted the trajectories of the beads in 3D space and found that their displacement decreased in the quicker stress relaxation hydrogel (Fig. 2F, i and ii). The amount of force that the cells are applying on the matrix was then calculated from the displacement of the beads (28). We found that cells in the quicker stress relaxation hydrogels apply less force onto the matrix compared with control gels (Fig. 2F, iii). Small forces have been previously reported for migrating cancer cells (40, 41) as well as dividing cells (42, 43), further supporting the biological relevance of these measurements.

PLOD2 increase in mTG-treated gels

We next analyzed why shorter stress relaxation time induced an increase in sarcoma cell migration. All analyses below were performed after cells were cultured for 3 days in the hydrogel. To study the whether cell–material interactions are enhanced in mTG-treated gels, we looked for (i) vinculin spots, which are indicative of focal adhesions (44); (ii) actin fiber rearrangement and localization (45), and (iii), YAP nuclear localization (46). We could not detect significant differences among the KIA cells due to changes in stress relaxation or hypoxia (Supplementary Fig. S7). As a result, we conclude that cellular responses to changes in the stress relaxation of the collagen gels examined here are not YAP mediated.

Previous studies linked an increase in expression of collagen crosslinkers, such as PLOD2, to TGFβ-driven SMAD2/3 nuclear localization (47, 48). Specifically, in breast cancer, TGFβ and SMAD2/3 have been linked to cell mechanosensing through PDZ-binding motif (TAZ) signaling, where TAZ specifically is required for nuclear accumulation of SMAD2/3 (49). Moreover, in stiffer matrices, cells have more nuclear SMAD2/3 (50). In pancreatic cancer, increased TGFβ expression leads to more migratory cells (51), while in prostate cancer, an increase in TGFβ RNA expression, correlates to an increase in patient mortality (52). We thus hypothesized that a quicker stress relaxation time upregulates TGFβ expression enabling nuclear localization of SMAD2/3, which in turn further amplifies PLOD2 expression. To test this hypothesis, we first examined the expression pattern of SMAD2/3 in the different conditions. We found a significant increase in SMAD2/3 nuclear localization in the mTG hydrogels (Fig. 3A and B). We then examined PLOD2 expression in the different conditions with and without inhibition of TGFβ. Without inhibition of TGFβ, we found higher PLOD2 expression in mTG-treated gels, regardless of oxygen levels, compared with untreated controls. KIA cells embedded in mTG-treated hypoxic gel exhibited the highest PLOD2 expression, suggesting an additive effect of these two cues. When TGFβ inhibitor was added, PLOD2 expression levels dropped (Fig. 3C). Overall, we conclude that these results link TGFβ to PLOD2 expression.

To explore the consequences of increased PLOD2 expression, we performed scanning electron microscopy and analyzed how cells interact with the collagen matrix, and found that the pores around the cells are larger in mTG-treated hydrogels on day 3 (Fig. 3D). We sought to determine whether the size of cell-associated pores is due to cell-mediated matrix degradation. Using DQ collagen assays, we did not detect differences in protease activity either overall or locally next to cells (Supplementary Fig. S8A and S8B). Moreover, we did not observe significant differences in matrix metalloprotease (MMP) 2 and 9 expression (Supplementary Fig. S8C) or in the G’ of the different gels on day 1 and day 3 of culture (Supplementary Fig. S8D). Taken together, these results suggest that the change in the size of cell-associated pores was not protease mediated.

To explore changes in cellular migration due to increased PLOD2 expression, we performed tracking studies of KIA-GFP cells treated with a TGFβ inhibitor or with a PLOD2 inhibitor (18, 24), and fluorescently labeled PLOD2-deficient cells. When KIA cells were treated with TGFβ inhibitor, we observed a decrease in migration of KIA cells in mTG hypoxic gradients to a level similar to the KIA cells in hypoxic gradients in the z direction (Fig. 3E, i), further corroborating the impact of decreased PLOD2 expression with TGFβ inhibition. In addition, when PLOD2 was inhibited with minoxidil, or knocked down (Fig. 3E, ii and iii); we found a decrease in MSD in the z direction, in both the mTG hypoxic gel and hypoxic gel, to a level below the untreated controls. The PLOD2-specific inhibitor and the knockdown cell line are impeding the hypoxic and TGFβ contributions to PLOD2 expression. We thus conclude that the increase in migration, in the quicker stress relaxation gels, is due to an increase in PLOD2 expression induced by TGFβ stimulation through SMAD2/3 activation. We also confirm a combinatory effect of hypoxia in upregulating that migration via PLOD2 expression.

Quicker stress relaxation in hydrogels leads to increased collagen fiber size/density and metastasis

To examine the effect of matrix stress relaxation on sarcoma growth and metastasis in vivo, hypoxic control and mTG-treated constructs were implanted subcutaneously into nude mice (Fig. 4A). Sarcoma cell migration in response to an oxygen gradient was previously established in vivo (16, 18, 24). In this model, we examine the coregulation of stress relaxation with hypoxia. For the in vivo assays, we hypothesize that quicker stress relaxation is caused by PLOD2 upregulation in hypoxic conditions, resulting in more cell migration and metastasis. To test this hypothesis, we use hypoxic hydrogel-encapsulated cells as a substitute for subcutaneous KIA tumors, which are generally hypoxic when grown beyond 300 mm2 (24). Specifically, the hydrogel model allows us to directly manipulate stress relaxation of the material. KIA cells, transfected with a BRET reporter, were used for all in vivo experimentation (32). Tumors were allowed to grow for up to 20 days and monitored using IVIS imaging, before the animals were euthanized. mTG-treated gels displayed enhanced fluorescent signal over time compared with control hypoxic gels (Fig. 4B). This observation correlated with an apparent increase in tumor volume (Fig. 4C). However, tumor volume, from the mTG hydrogel, was not significantly greater at any of the examined time points.

We next analyzed PLOD2 staining and found it to be significantly increased in the mTG tumors compared with controls. (Fig. 4D). This confirmed the phenotype that we observed in vitro. Finally, we analyzed the collagen fiber density and architecture in the day 20 tumors using Masson trichrome and second harmonic generation (SHG) microscopy and image analysis with the ct-FIRE code suite (29). Masson trichrome stain showed more collagen fibers in the tumors generated from the mTG-treated hypoxic hydrogels (Fig. 4E). By analyzing the SHG images, we further found higher collagen fiber width, length, and density in the mTG tumors (Fig. 4F and G; Supplementary Fig. S9). Together, these data demonstrate that quicker stress relaxation time increases collagen content and fiber architecture through upregulation in PLOD2 expression.

In addition, to further confirm the effect of PLOD2 on stress relaxation, subcutaneous tumors were generated in nude mice using the KIA scramble control cells (KIA Scr) and KIA cells with a PLOD2 knockdown (PLOD2(−)). Cells was injected subcutaneously to form the tumors, and the tumors where then allowed to grow for 20 days (Fig. 4H). The growth of these tumors was similar to what was seen with the hydrogel implantation, where we saw no difference in the tumor volume over time between the KIA Scr and PLOD2(−) tumors (Fig. 4I). At day 20, the storage modulus (G’), and stress relaxation time was measured (Figs. 4J–L). From these experiments, we observed that the tumors had the same G’, but different stress relaxation times. The stress relaxation time was also confirmed using a creep and recovery test (Supplementary Fig. S10). Overall, when PLOD2 crosslinking is knocked down, the tumor has a slower stress relaxation time. This confirms the hypothesis about the importance of PLOD2 crosslinking in generating a quicker stress relaxation.

Quicker stress relaxation in hydrogels augments metastasis

As described previously, loss of PLOD2 expression disrupts pulmonary metastasis and because lungs are the main metastatic sight for sarcoma (18), we next examined the impact of stress relaxation on lung metastasis. We prepared nonhypoxic and mTG-treated hypoxic constructs and implanted them after 3 days in culture. This time in culture was to allow transcriptional and translational changes responsible for the phenotype to take place prior to implantation. We speculated that while hypoxic conditions are rapidly generated in subcutaneous tumors, priming the cells to the quicker stress relaxation time would enhance their metastatic potential. After 20 days, tumors were resected and mice were monitored for an additional 15 days (Fig. 5A). This model, with extended time for tumor growth, allows for KIA cells to migrate out of the hypoxic hydrogel into the vasculature and disseminate to the lungs and form metastasis. After 15 days, metastatic lesions were found in the lungs in both conditions, but were more extensive in the mice that were implanted with the mTG-treated hypoxic constructs (Fig. 5B) including the number of metastatic lesions per lung as well as the area covered by metastasis (Fig. 5C). In addition, SHG imaging was performed on lung regions showing that there was more collagen deposited in the metastasis originated from the mTG-treated hypoxic constructs (Fig. 5D). This is indicative of more mature tumors. We observed no difference in the collagen fiber length and width in the lungs (Supplementary Fig. S11).

The TCGA analysis and patient relevance of high PLOD2 expression

To study the clinical significance of PLOD2 expression in sarcoma, TCGA database was queried to analyze survival rates, as well as correlations from patient RNA-seq data. A total of 261 patients were queried, and the results were pooled using GEPIA to generate RNA-seq correlations, disease-free survival, and overall survival (31). We explored the correlation between upregulation of SMAD2, TGFβ, and hypoxia-inducible factor 1-alpa (HIF1α) and PLOD2 with patient disease-free survival and overall survival. There is no significant difference in the overall survival, or disease-free survival, of patients who are high expressing for SMAD2 and TGFβ1 (Fig. 6A and B). In addition, while we do not see an increase in overall survival in patients who are high expressing for HIF1α, we detect an increase in their disease-free survival (Fig. 6C). Interestingly, we could detect a small but significant decrease in disease-free survival for high PLOD2-expressing patients; and a significant increase in mortality rates, from 40% for low-expressing PLOD2 patients to 20% for high-expressing PLOD2 patients (Fig. 6D). To confirm the TCGA data, we explored the ability of the mTG hydrogel platform to test the response of primary patient UPS cells, recently derived from human tumors (hUPS). The oxygen profiles in the gels on day 3 were similar to what we observed for encapsulated KIA cells (Fig. 6E). We observed distinct phenotypes in the different hypoxic and nonhypoxic gels treated with or without mTG (Fig. 6F); and an increase in aspect ratio in the more migratory cells (Fig. 6G), which is also similar to what was observed with KIA cells. Finally, there was no difference in cell migration between the nonhypoxic conditions (Supplementary Fig. S12A), but there was a significant increase in migration in the mTG hypoxic gel condition compared with the control hypoxic gel in the z direction (Fig. 6H; Supplementary Fig. S12B). There was no difference in the velocity of hUPS cells in the hydrogels (Supplementary Fig. S13). Overall, hUPS and KIA cells show similar trends.

Proposed mechanism of response to quicker stress relaxation

We propose a novel mechanism for cellular response to quicker collagen stress relaxation time in the hypoxic sarcoma microenvironment (Fig. 7). As the tumor grows beyond its blood supply, hypoxic regions develop, which induces PLOD2 expression via HIF1α upregulation, thereby leading to metastasis (18, 24). This enhanced crosslinking quickens the stress relaxation time of the collagen matrix, which in turn stimulates TGFβ shuttling of SMAD2 to the nucleus, further increasing PLOD2 expression. We propose a feedback loop between hypoxia and matrix properties, resulting in an amplification of cell migration and metastasis in sarcoma.

This study demonstrates the importance of stress relaxation time as a new mechanical property that is impacted by the hypoxic environment and in turn enhancing sarcoma motility and metastasis. Stress relaxation is a key factor that links the viscous and elastic components of the ECM. This is the first reported direct correlation between hypoxia and matrix properties that jointly regulate sarcoma fate.

The collagen hypoxic hydrogel platform was used to explore this mechanical property and biological impact. Using collagen gel and mTG, we can control for faster stress relaxation without modulating the hydrogels’ stiffness and micron-level structure. With this system, by controlling cell concentration and hydrogel thickness, we are able to change oxygen tension to achieve hypoxic (1%–12%) and nonhypoxic (15%–20%) gradient hydrogels. Overall, this platform allows us to modulate oxygen and stress relaxation without altering the micro- and nanoscale collagen fiber architecture; thereby allowing us to create a platform that more accurately mimics in vivo ECM to study finite properties of the tumor microenvironment. From our studies, we observed an increase in hypoxic sarcoma cell migration along the oxygen gradient from low to high tensions. This is expected as we have previously shown that sarcoma cells persistently migrate in the direction of higher oxygen concentrations (16, 24). Importantly, the cells in the hypoxic hydrogels with the quickest stress relaxation time, exhibited the most robust migration within the oxygen gradient as well as an elongated phenotype quantified by aspect ratio. Using traction force microscopy, we further found that less displacement and less force is applied in the quicker stress relaxation hydrogel. We speculate that cells need less force to move through the quicker stress relaxation gels, thus able to migrate faster and to longer distances compared with controls.

To explore the molecular mechanisms behind the increase in cell migration, we begun by analyzing YAP and could not detect significant differences among the KIA cells in the different conditions. Unlike epithelial cancer cells where YAP localization changes in stiffer gels during EMT (53), KIA cells, which derived from a genetic model of murine UPS, are already mesenchymal therefore they do not go under EMT. We speculate that KIA cells do not respond to changes in stress relaxation via YAP. As a result, we conclude that cellular responses to changes in the stress relaxation of the collagen gels examined here are not YAP mediated. It is important to note, that our studies did not explore the response of KIA cells to changes matrix stiffness and that our gels are soft (100 Pa), and thus we cannot exclude that KIA maybe YAP-sensitive in stiffer matrix conditions. We were able to deduce that the quicker stress relaxation increased PLOD2 expression in cells via TGFβ and SMAD2. This increase in PLOD2 via TGFβ and SMAD2 had an additive effect when coupled with hypoxia. Using scanning electron microscopy, we found that the pores around the cells are larger in quicker relaxing hydrogels on day 3. This ability to create larger pores is due to the lower polymer-to-crosslinking ratio, which requires less force to move the fibers. Previous studies have shown that an increase in crosslinking leads to an increase in cell migration (54) as well as it being a critical factor in tumor progression (18, 55). As we further found that the change in the size of cell-associated-pores was not protease mediated, we hypothesize that pore size and cell movement in mTG-treated hydrogels is most likely due to cells physically deforming the collagen fibers.

We next explore the impact of stress relaxation on tumor progression and metastasis. Tumors generated from hypoxic hydrogels with quicker stress relaxation times exhibited an increase in collagen fiber size and density as well as PLOD2 expression. Moreover, when generating tumors from PLOD2 knockdown KIA cells, we could not observe change in tumor volume and stiffness, but found slower stress relaxation compared with control tumors. This confirms that loss of PLOD2 crosslinking affect stress relaxation in sarcoma, similarly to what we observed in the in vitro experiment, and with tumors generated from hydrogel implantation. Finally, to examine impact on metastasis, we performed a survival assay where fully grown primary tumors were removed, allowing mice to live for additional 15 days, examining pulmonary metastasis. We found more extensive metastasis with more collagen deposition in the mice that were implanted with the quicker stress relaxed hydrogels, further corroborating our migration observations in vitro.

Finally, the clinical relevance of these observations was determined by querying publicly available TCGA global expression data. We found a dramatic decrease in patient survival in those who had high expression levels of PLOD2. Using this new hypoxic collagen gel platform, the response of primary patient hUPS cells was found to be similar to what was predicted using the KIA cell line, demonstrating the ability of this platform to screen future therapeutics to inhibit migration and subsequently minimize cancer progression.

This is one of the first studies to demonstrate the effect of stress relaxation on tumor growth and metastasis in a physiologically mimicking in vitro system. We therefore propose that our platform has substantial predictive potential and will enable a more thorough investigation of migratory phenotypes in primary human samples. This platform can further be adapted to explore stress relaxation in oxygen gradients as they affect other cancers.

K. Weber is a vice president at AAOS (volunteer national organization for orthopedics) and has done expert testimony for Wolters Kluwer. No potential conflicts of interest were disclosed by the other authors.

Conception and design: D.M. Lewis, H. Pruitt, J.M. McCaffery, S. Gerecht

Development of methodology: D.M. Lewis, H. Pruitt, N. Jain, J.M. McCaffery, Z. Xia

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D.M. Lewis, H. Pruitt, N. Jain, M. Ciccaglione, J.M. McCaffery, K. Weber

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.M. Lewis, N. Jain, M. Ciccaglione, Z. Xia, S. Gerecht

Writing, review, and/or revision of the manuscript: D.M. Lewis, H. Pruitt, N. Jain, M. Ciccaglione, Z. Xia, K. Weber, T.S.K. Eisinger-Mathason, S. Gerecht

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.M. Lewis, N. Jain

Study supervision: D.M. Lewis, S. Gerecht

We would like to thank Dr. Luo Gu and Dr. Denis Wirtz for helpful discussions about collagen hydrogel design. This work was supported by the shared resources from the Sydney Kimmel Cancer Center, Johns Hopkins University (P30 CA006973), fellowship from the 2T32CA153952-06, Nanotechnology Cancer Research training grant (to D.M. Lewis and H. Pruitt), The NCI Physical Sciences-Oncology Center (U54CA210173), and the President's Frontier Award (both to S. Gerecht).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Friedl
P
,
Wolf
K
. 
Tumour-cell invasion and migration: diversity and escape mechanisms
.
Nat Rev Cancer
2003
;
3
:
362
.
2.
Swaminathan
V
,
Mythreye
K
,
O'Brien
ET
,
Berchuck
A
,
Blobe
GC
,
Superfine
R
. 
Mechanical stiffness grades metastatic potential in patient tumor cells and in cancer cell lines
.
Cancer Res
2011
;
71
:
5075
80
.
3.
Xia
Z
,
Patchan
M
,
Maranchi
J
,
Trexler
M
. 
Structure and relaxation in cellulose hydrogels
.
J Appl Polym Sci
2015
;
132
.
4.
Chaudhuri
O
,
Gu
L
,
Klumpers
D
,
Darnell
M
,
Bencherif
SA
,
Weaver
JC
, et al
Hydrogels with tunable stress relaxation regulate stem cell fate and activity
.
Nat Mater
2015
;
15
:
326
34
.
5.
Stylianopoulos
T
,
Martin
JD
,
Chauhan
VP
,
Jain
SR
,
Diop-Frimpong
B
,
Bardeesy
N
, et al
Causes, consequences, and remedies for growth-induced solid stress in murine and human tumors
.
Proc Natl Acad Sci
2012
;
109
:
15101
8
.
6.
Nia
HT
,
Liu
H
,
Seano
G
,
Datta
M
,
Jones
D
,
Rahbari
N
, et al
Solid stress and elastic energy as measures of tumour mechanopathology
.
Nat Biomed Eng
2017
;
1
:
0004
.
7.
Nia
HT
,
Datta
M
,
Seano
G
,
Huang
P
,
Munn
LL
,
Jain
RK
. 
Quantifying solid stress and elastic energy from excised or in situ tumors
.
Nat Protoc
2018
;
13
:
1091
.
8.
Stylianopoulos
T
,
Martin
JD
,
Snuderl
M
,
Mpekris
F
,
Jain
SR
,
Jain
RK
. 
Coevolution of solid stress and interstitial fluid pressure in tumors during progression: implications for vascular collapse
.
Cancer Res
2013
;
73
:
3833
41
.
9.
Jain
RK
. 
Antiangiogenesis strategies revisited: from starving tumors to alleviating hypoxia
.
Cancer Cell
2014
;
26
:
605
22
.
10.
Netti
PA
,
Berk
DA
,
Swartz
MA
,
Grodzinsky
AJ
,
Jain
RK
. 
Role of extracellular matrix assembly in interstitial transport in solid tumors
.
Cancer Res
2000
;
60
:
2497
503
.
11.
Lo
C-M
,
Wang
H-B
,
Dembo
M
,
Wang
Y-l
. 
Cell movement is guided by the rigidity of the substrate
.
Biophys J
2000
;
79
:
144
52
.
12.
Chaudhuri
O
,
Gu
L
,
Klumpers
D
,
Darnell
M
,
Bencherif
SA
,
Weaver
JC
, et al
Hydrogels with tunable stress relaxation regulate stem cell fate and activity
.
Nat Mater
2016
;
15
:
326
34
.
13.
Carey
SP
,
Goldblatt
ZE
,
Martin
KE
,
Romero
B
,
Williams
RM
,
Reinhart-King
CA
. 
Local extracellular matrix alignment directs cellular protrusion dynamics and migration through Rac1 and FAK
.
Integr Biol
2016
;
8
:
821
35
.
14.
Tibbitt
MW
,
Anseth
KS
. 
Hydrogels as extracellular matrix mimics for 3D cell culture
.
Biotechnol Bioeng
2009
;
103
:
655
63
.
15.
Lou
J
,
Stowers
R
,
Nam
S
,
Xia
Y
,
Chaudhuri
O
. 
Stress relaxing hyaluronic acid-collagen hydrogels promote cell spreading, fiber remodeling, and focal adhesion formation in 3D cell culture
.
Biomaterials
2017
;
154
:
213
22
.
16.
Lewis
D
,
Tang
V
,
Jain
N
,
Isser
A
,
Xia
Z
,
Gerecht
S
. 
Collagen fiber architecture regulate hypoxic sarcoma cell migration
.
ACS Biomater Sci Eng
2018
;
4
:
400
9
.
17.
Frantz
C
,
Stewart
KM
,
Weaver
VM
. 
The extracellular matrix at a glance
.
J Cell Sci
2010
;
123
:
4195
200
.
18.
Eisinger-Mathason
TK
,
Zhang
M
,
Qiu
Q
,
Skuli
N
,
Nakazawa
MS
,
Karakasheva
T
, et al
Hypoxia-dependent modification of collagen networks promotes sarcoma metastasis
.
Cancer Discov
2013
;
3
:
1190
205
.
19.
Noda
T
,
Yamamoto
H
,
Takemasa
I
,
Yamada
D
,
Uemura
M
,
Wada
H
, et al
PLOD2 induced under hypoxia is a novel prognostic factor for hepatocellular carcinoma after curative resection
.
Liver Int
2012
;
32
:
110
8
.
20.
Semenza
GL
. 
Targeting HIF-1 for cancer therapy
.
Nat Rev Cancer
2003
;
3
:
721
.
21.
Francis
P
,
Namløs
HM
,
Müller
C
,
Edén
P
,
Fernebro
J
,
Berner
J-M
, et al
Diagnostic and prognostic gene expression signatures in 177 soft tissue sarcomas: hypoxia-induced transcription profile signifies metastatic potential
.
BMC Genomics
2007
;
8
:
73
.
22.
Du
H
,
Pang
M
,
Hou
X
,
Yuan
S
,
Sun
L
. 
PLOD2 in cancer research
.
Biomed Pharmacother
2017
;
90
:
670
6
.
23.
Gilkes
DM
,
Bajpai
S
,
Wong
CC
,
Chaturvedi
P
,
Hubbi
ME
,
Wirtz
D
, et al
Procollagen lysyl hydroxylase 2 is essential for hypoxia-induced breast cancer metastasis
.
Mol Cancer Res
2013
;
11
:
456
66
.
24.
Lewis
DM
,
Park
KM
,
Tang
V
,
Xu
Y
,
Pak
K
,
Eisinger-Mathason
TK
, et al
Intratumoral oxygen gradients mediate sarcoma cell invasion
.
Proc Natl Acad Sci U S A
2016
;
113
:
201605317
.
25.
Abaci
HE
,
Truitt
R
,
Tan
S
,
Gerecht
S
. 
Unforeseen decreases in dissolved oxygen levels affect tube formation kinetics in collagen gels
.
Am J Physiol Cell Physiol
2011
;
301
:
C431
40
.
26.
Gao
Y
,
Baca
AM
,
Wang
B
,
Ogilby
PR
. 
Activation barriers for oxygen diffusion in polystyrene and polycarbonate glasses: effects of low molecular weight additives
.
Macromolecules
1994
;
27
:
7041
8
.
27.
Lewis
DM
,
Blatchley
MR
,
Park
KM
,
Gerecht
S
. 
O 2-controllable hydrogels for studying cellular responses to hypoxic gradients in three dimensions in vitro and in vivo
.
Nat Protoc
2017
;
12
:
1620
.
28.
McKinnon
D
,
Domaille
D
,
Brown
T
,
Kyburz
K
,
Kiyotake
E
,
Cha
J
, et al
Measuring cellular forces using bis-aliphatic hydrazone crosslinked stress-relaxing hydrogels
.
Soft Matter
2014
;
10
:
9230
6
.
29.
Bredfeldt
JS
,
Liu
Y
,
Pehlke
CA
,
Conklin
MW
,
Szulczewski
JM
,
Inman
DR
, et al
Computational segmentation of collagen fibers from second-harmonic generation images of breast cancer
.
J Biomed Opt
2014
;
19
:
016007
.
30.
McCaffery
JM
,
Farquhar
MG
. 
Chapter 29: Localization of GTPases by indirect immunofluorescence and immunoelectron microscopy
.
Methods in Enzymology.
Philadelphia: Elsevier
; 
1995
. p
259
79
.
31.
Tang
Z
,
Li
C
,
Kang
B
,
Gao
G
,
Li
C
,
Zhang
Z
. 
GEPIA: a web server for cancer and normal gene expression profiling and interactive analyses
.
Nucleic Acids Res
2017
;
45
:
W98
W102
.
32.
Schaub
FX
,
Reza
MS
,
Flaveny
CA
,
Li
W
,
Musicant
AM
,
Hoxha
S
, et al
Fluorophore-NanoLuc BRET reporters enable sensitive in vivo optical imaging and flow cytometry for monitoring tumorigenesis
.
Cancer Res
2015
;
75
:
5023
33
.
33.
Tomayko
MM
,
Reynolds
CP
. 
Determination of subcutaneous tumor size in athymic (nude) mice
.
Cancer Chemother Pharmacol
1989
;
24
:
148
54
.
34.
Insana
MF
,
Pellot-Barakat
C
,
Sridhar
M
,
Lindfors
KK
. 
Viscoelastic imaging of breast tumor microenvironment with ultrasound
.
J Mammary Gland Biol Neoplasia
2004
;
9
:
393
404
.
35.
Wisdom
KM
,
Adebowale
K
,
Chang
J
,
Lee
JY
,
Nam
S
,
Desai
R
, et al
Matrix mechanical plasticity regulates cancer cell migration through confining microenvironments
.
Nat Commun
2018
;
9
:
4144
.
36.
Lessey
BA
,
Savaris
RF
. 
Comparison of HSCORE assessment of endometrial β3 integrin subunit expression with digital HSCORE using computerized image analysis (ImageJ)
.
Anal Quantitat Cytopathol Histopathol
2013
;
35
:
210
.
37.
Kirsch
DG
,
Dinulescu
DM
,
Miller
JB
,
Grimm
J
,
Santiago
PM
,
Young
NP
, et al
A spatially and temporally restricted mouse model of soft tissue sarcoma
.
Nat Med
2007
;
13
:
992
.
38.
Yoon
SS
,
Stangenberg
L
,
Lee
Y-J
,
Rothrock
C
,
Dreyfuss
JM
,
Baek
K-H
, et al
Efficacy of sunitinib and radiotherapy in genetically engineered mouse model of soft-tissue sarcoma
.
Int J Radiat Oncol Biol Phys
2009
;
74
:
1207
16
.
39.
Bloom
RJ
,
George
JP
,
Celedon
A
,
Sun
SX
,
Wirtz
D
. 
Mapping local matrix remodeling induced by a migrating tumor cell using three-dimensional multiple-particle tracking
.
Biophys J
2008
;
95
:
4077
88
.
40.
Wolf
K
,
Te Lindert
M
,
Krause
M
,
Alexander
S
,
Te Riet
J
,
Willis
AL
, et al
Physical limits of cell migration: control by ECM space and nuclear deformation and tuning by proteolysis and traction force
.
J Cell Biol
2013
;
201
:
1069
84
.
41.
Franck
C
,
Maskarinec
SA
,
Tirrell
DA
,
Ravichandran
G
. 
Three-dimensional traction force microscopy: a new tool for quantifying cell-matrix interactions
.
PLoS One
2011
;
6
:
e17833
.
42.
He
L
,
Sneider
A
,
Chen
W
,
Karl
M
,
Prasath
V
,
Wu
P-H
, et al
Mammalian cell division in 3D matrices via quantitative confocal reflection microscopy
.
J Vis Exp
2017
:
e56364
.
doi: 10.3791/56364
.
43.
Ng
MR
,
Besser
A
,
Brugge
JS
,
Danuser
G
. 
Mapping the dynamics of force transduction at cell–cell junctions of epithelial clusters
.
Elife
2014
;
3
:
e03282
.
doi: 10.7554/eLife.03282
.
44.
Humphries
JD
,
Wang
P
,
Streuli
C
,
Geiger
B
,
Humphries
MJ
,
Ballestrem
C
. 
Vinculin controls focal adhesion formation by direct interactions with talin and actin
.
J Cell Biol
2007
;
179
:
1043
57
.
45.
DuFort
CC
,
Paszek
MJ
,
Weaver
VM
. 
Balancing forces: architectural control of mechanotransduction
.
Nat Rev Mol Cell Biol
2011
;
12
:
308
.
46.
Dupont
S
,
Morsut
L
,
Aragona
M
,
Enzo
E
,
Giulitti
S
,
Cordenonsi
M
, et al
Role of YAP/TAZ in mechanotransduction
.
Nature
2011
;
474
:
179
.
47.
Laklai
H
,
Miroshnikova
YA
,
Pickup
MW
,
Collisson
EA
,
Kim
GE
,
Barrett
AS
, et al
Genotype tunes pancreatic ductal adenocarcinoma tissue tension to induce matricellular fibrosis and tumor progression
.
Nat Med
2016
;
22
:
497
505
.
48.
Remst
DF
,
Davidson
ENB
,
Vitters
EL
,
Bank
RA
,
van den Berg
WB
,
van der Kraan
PM
. 
TGF-ss induces Lysyl hydroxylase 2b in human synovial osteoarthritic fibroblasts through ALK5 signaling
.
Cell Tissue Res
2014
;
355
:
163
71
.
49.
Varelas
X
,
Sakuma
R
,
Samavarchi-Tehrani
P
,
Peerani
R
,
Rao
BM
,
Dembowy
J
, et al
TAZ controls Smad nucleocytoplasmic shuttling and regulates human embryonic stem-cell self-renewal
.
Nat Cell Biol
2008
;
10
:
837
.
50.
Taylor
MA
,
Amin
JD
,
Kirschmann
DA
,
Schiemann
WP
. 
Lysyl oxidase contributes to mechanotransduction-mediated regulation of transforming growth factor-β signaling in breast cancer cells
.
Neoplasia
2011
;
13
:
406
IN2
.
51.
Chen
Y
,
Yu
G
,
Yu
D
,
Zhu
M
. 
PKCα-induced drug resistance in pancreatic cancer cells is associated with transforming growth factor-β1
.
J Exp Clin Cancer Res
2010
;
29
:
104
.
52.
Reis
STd
,
Pontes-Júnior
J
,
Antunes
AA
,
Sousa-Canavez
JMd
,
Abe
DK
,
Cruz
JASd
, et al
Tgf-β1 expression as a biomarker of poor prognosis in prostate cancer
.
Clinics
2011
;
66
:
1143
7
.
53.
Zanconato
F
,
Forcato
M
,
Battilana
G
,
Azzolin
L
,
Quaranta
E
,
Bodega
B
, et al
Genome-wide association between YAP/TAZ/TEAD and AP-1 at enhancers drives oncogenic growth
.
Nat Cell Biol
2015
;
17
:
1218
.
54.
Anguiano
M
,
Castilla
C
,
Maška
M
,
Ederra
C
,
Peláez
R
,
Morales
X
, et al
Characterization of three-dimensional cancer cell migration in mixed collagen-Matrigel scaffolds using microfluidics and image analysis
.
PLoS One
2017
;
12
:
e0171417
.
55.
Levental
KR
,
Yu
H
,
Kass
L
,
Lakins
JN
,
Egeblad
M
,
Erler
JT
, et al
Matrix crosslinking forces tumor progression by enhancing integrin signaling
.
Cell
2009
;
139
:
891
906
.