S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

CANCER Copyright © 2024 The

Authors, some rights
The role of Piezo1 mechanotransduction in high-­grade reserved; exclusive
licensee American
serous ovarian cancer: Insights from an in vitro model Association for the
Advancement of
of collective detachment Science. No claim to
original U.S.
Government Works.
Hannah M. Micek1, Ning Yang2†, Mayuri Dutta1†, Lauren Rosenstock1, Yicheng Ma1, Distributed under a
Caitlin Hielsberg1, Molly McCord3,4, Jacob Notbohm1,3,4,5, Stephanie McGregor2,5, Creative Commons
Pamela K. Kreeger1,2,5* Attribution
NonCommercial
Slowing peritoneal spread in high-­grade serous ovarian cancer (HGSOC) would improve patient prognosis and License 4.0 (CC BY-­NC).
quality of life. HGSOC spreads when single cells and spheroids detach, float through the peritoneal fluid and take
over new sites, with spheroids thought to be more aggressive than single cells. Using our in vitro model of spher-

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
oid collective detachment, we determine that increased substrate stiffness led to the detachment of more spher-
oids. We identified a mechanism where Piezo1 activity increased MMP-­1/MMP-­10, decreased collagen I and
fibronectin, and increased spheroid detachment. Piezo1 expression was confirmed in omental masses from pa-
tients with stage III/IV HGSOC. Using OV90 and CRISPR-­modified PIEZO1−/− OV90 in a mouse xenograft model, we
determined that while both genotypes efficiently took over the omentum, loss of Piezo1 significantly decreased
ascitic volume, tumor spheroids in the ascites, and the number of macroscopic tumors in the mesentery. These
results support that slowing collective detachment may benefit patients and identify Piezo1 as a potential thera-
peutic target.

INTRODUCTION to demonstrate that many spheroids arose from the same ovary.
High-­grade serous ovarian cancer (HGSOC) has an overall 5-­year These data suggest that tumor cells either aggregate quickly after
survival rate of less than 50%, due to most patients having substan- detaching or are released as a unit through a process of collective
tial metastatic burden at the time of diagnosis. While early-­stage detachment. Unfortunately, animal models are not easily adapted
tumors are restricted to the fallopian tube and ovary, at advanced to address what controls the extent of detachment and whether a
stages, tumors are present through the pelvis and peritoneal cavity tumor cell releases as a single cell or in a spheroid. Therefore, we
(1). Patients diagnosed at stage III/IV frequently have metastases recently developed an in vitro model that isolates tumor cells that
on deposits of visceral adipose (e.g., omentum and small bowel detach as spheroids from cells that detach as single cells (9). In this
mesentery) and accumulation of excess fluid, termed ascites. Iden- model, ovarian cancer cells are cultured on top of a polyacryl-
tifying methods to slow or stop spread and ascites accumulation amide (PAA) gel that rests within a cell strainer (Fig. 1A). As indi-
would improve patient prognosis and quality of life, as metastases vidual cells detach, they pass through the strainer, while spheroids
can ultimately constrict the bladder or bowel and ascites causes that have collectively detached (Sph-­CD) are trapped by the filter
substantial discomfort. However, HGSOC metastasis differs from and can be isolated for further analysis.
that of most solid tumors. Rather than relying on hematogenous or Here, we sought to use our model to begin testing the hypothe-
lymphatic spread, tumor cells detach, float in the peritoneal fluid/ sis that changes in the tumor microenvironment associated with
ascites, and reattach to establish new tumors (2). The processes disease progression increase collective detachment. When design-
regulating transport and reattachment have been studied through ing biomimetic systems, determining how to best mimic the extra-
in vitro (3–6) and in vivo models (7, 8), but analysis of the detach- cellular matrix (ECM) is an essential first step before building in
ment process is much more challenging. higher-­order interactions (13). The role of the ECM in HGSOC has
Tumor cells that have detached can be isolated from ascites as been examined in many of the existing culture systems to under-
both single cells and spheroids of cells on the order of 50 to 100 μm stand tumor cell adhesion, spreading, and proliferation [reviewed
in diameter (9, 10). While most studies of HGSOC metastasis have in (2)], but it is unknown whether it affects cell detachment. While
used single cells, spheroids are believed to disproportionately con- there are many changes in the ECM as HGSOC progresses, the
tribute to metastasis due to their resistance to apoptosis and che- most abundant ECM protein is collagen I (Col I), which increases
motherapy (11). To examine how spheroids are generated, Al 20-­fold by advanced stages (14). Associated with this increase in
Habyan et al. (12) used labeled tumor cells injected into the ovary ECM is an increase in stiffness. This fibrotic response is observed in
numerous solid tumors, including breast cancer, which is known to
use collective processes in metastasis (15). Using PAA gel formula-
1
Department of Biomedical Engineering, University of Wisconsin-­Madison, Madison, tions with stiffnesses reflective of healthy and metastatic omentum
WI 53705, USA. 2Department of Pathology and Laboratory Medicine, University of
Wisconsin School of Medicine and Public Health, Madison, WI 53705, USA. 3Depart-
along with different concentrations of Col I, we were able to test the
ment of Mechanical Engineering, University of Wisconsin-­Madison, Madison, WI individual and combinatorial effects of increased Col I density and
53705, USA. 4Biophysics Program, University of Wisconsin-­Madison, Madison, WI substrate stiffness. We then examined downstream signaling pro-
53705, USA. 5University of Wisconsin Carbone Cancer Center, University of Wisconsin cesses to identify mechanisms regulating collective detachment
School of Medicine and Public Health, Madison, WI 53705, USA.
*Corresponding author. Email: kreeger@​wisc.​edu and analyzed the effect of slowing collective detachment on tumor
†These authors contributed equally to this work. progression in vivo.

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 1 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

during a debulking surgery. On the basis of their assessment, PAA at

~5 kPa was representative of benign tissue (“soft”), and ~44 kPa was
representative of metastatic tissue (“stiff ”; fig. S1A). We then modified
soft and stiff PAA gels with either low (100 μg/ml) or high (2000 μg/
ml) Col I (fig. S1B). Last, we considered the cell types to incorporate
into our model. Our prior analysis of spheroids isolated from patient
ascites suggested that they are composed primarily of epithelial cells
(9). However, it is possible that the originating site has more cellular
diversity. Therefore, we immunostained tumors from patients with
stage III/IV HGSOC for Pax8 (tumor cells), CD68 (monocytes/
macrophages), and αSMA (activated fibroblasts). Large areas of the tu-
mor were composed of only Pax8+ cells, with monocytes/macrophages
and fibroblasts between these regions [fig. S1C; similar findings were
observed with respect to fibroblasts in our prior work (18)]. On the
basis of these findings, we used only tumor cells in our model to ex-

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
amine the impact of stiffness and ECM density.
Using the selected stiffnesses and ECM densities, we observed a
significant increase in Sph-­CD with culture on stiff substrates
across three different cell lines, but no significant differences ob-
served with respect to Col I density (Fig. 1B). While these observa-
tions were based on counting a sampling of the released aggregates,
quantification of total DNA from released OV90 Sph-­CD also
showed that stiffness, but not ECM density, increased collective de-
tachment (fig. S1D). To confirm that the increase in Sph-­CD was
not due to an increase in all forms of cell detachment, we quantified
single cells and spheroids resulting from single cells that aggregated
Fig. 1. Stiffness increases collective detachment of spheroids. (A) Overview of (Sph-­SC) and found that there were no significant differences with
culture system consisting of a PAA gel on top of a filter. At the completion of cul- respect to substrate stiffness or ECM density (fig. S1E). Before
ture, the cover slip and PAA gel were removed for analysis. To isolate the Sph-­CD spheroid detachment, we observed areas of high cell density and
(#1), the filter was inverted and captured spheroids were washed out. To isolate cellular mounding on the surface of the substrate (fig. S1F). These
single cells (#2), the media was collected from the bottom well and passed through areas were termed buds and appear to be the sites where Sph-­CD
a cell strainer. Last, to isolate Sph-­SC (#3), the cell strainer used to isolate single cells detach. As more bud formation would explain the increase in Sph-­
was inverted and washed to remove captured spheroids. All three fractions were
CD, we quantified buds across the substrates and determined there
then transferred to cell culture dishes to image and quantify. (B) Quantification of
were no significant differences (Fig. 1C; note that in remaining fig-
OV90, OVCAR3, and OVCAR8 Sph-­CD from soft (5 kPa) or stiff (44 kPa) PAA gels
functionalized with a low (100 μg/ml) or high (2000 μg/ml) Col I. (C) Quantification
ures, only data for OV90 are shown with data from the other lines
of OV90, OVCAR3, and OVCAR8 buds on soft or stiff PAA gels functionalized with in supplement). However, we note that the average bud size is larg-
Col I (100 μg/ml), five fields of view were averaged per gel. Data in (B) and (C) are the er than the average aggregate size (fig. S1G), suggesting that only a
average ± SD of N = 3 to 4 individual gels. Statistical tests are two-­way analysis of portion of the bud detaches.
variance (ANOVA) with Tukey’s (B) and unpaired t test (C). ns indicates not statisti-
cally significant. *P < 0.05, ***P < 0.001, and ****P < 0.0001. Increased collective detachment was not a result of
well-­established effects of stiffness
While the number of buds was similar across conditions and buds
RESULTS were similar in size on soft and stiff substrates, it is possible that more
Increased substrate stiffness leads to increased collective actively proliferating buds would be less stable and therefore more
detachment of ovarian cancer cells likely to detach. As prior studies have clearly established a link be-
We first sought to determine the approximate stiffness of the omen- tween increased stiffness and cell proliferation in some contexts (19,
tum of a healthy female and a female with stage III/IV HGSOC. It is 20), we examined whether proliferation was involved in collective de-
particularly challenging to collect and process these tissue samples tachment. We first observed that inhibition of proliferation with
through pathology while maintaining physiological temperatures to aphidicolin ablated the stiffness-­dependent increase in Sph-­CD
prevent changes in lipid structure and the resulting mechanics. In the (Fig. 2A and fig. S2, A and B). However, there was no difference in
literature, the moduli of these tissues at high strains (15 to 30%) range overall proliferation between soft and stiff substrates (fig. S2C) or for
from 3 to 32 kPa (healthy) and 15 to 100 kPa (disease) (14, 16, 17). proliferation localized to buds or the monolayer (Fig. 2, B and C). To
This uncertainty results from differences in sample collection (with better understand the role of proliferation in collective detachment,
some studies using tissue subjected to a freeze/thaw cycle) and me- we examined the monolayer from cells treated with vehicle or aphidi-
chanical testing procedures. As an alternative approach to define our colin and observed a significant decrease in the number of buds on
range, we first generated a series of PAA gels of varying stiffness both stiffnesses (fig. S2D), suggesting that proliferation is needed to
(Young’s modulus from 2 to 44 kPa, determined from stress-­strain generate many buds. As the number of buds and proliferation were
curve by dynamic mechanical analyzer) and then asked a blinded not affected by stiffness, we hypothesized that stiffness may instead
gynecologic oncologist to palpate the gels as they would do to tissue affect detachment.

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 2 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

We next investigated the potential role of mechanosensitive tran-

scription factor YAP in stiffness-­induced collective detachment. YAP
is a transcription factor in the Hippo pathway activated in response to
increased stiffness (21) and linked to tumor cell migration and inva-
sion (22). Immunostaining confirmed that the ovarian cancer lines
expressed YAP and that increased stiffness led to increased expression
and nuclear localization (Fig. 2, D and E, and fig. S3A). To test wheth-
er YAP signaling in response to stiffness affects collective detachment,
cultures were treated with the inhibitor verteporfin (23). We observed
an increase in Sph-­CD in response to stiffness with both vehicle and
verteporfin (Fig. 2F and fig. S3B). Neither stiffness nor YAP inhibition
affected single-­cell detachment and aggregation (fig. S3B). Together,
these data suggest that YAP was activated with stiffness in our model
but was not directly involved in the signaling underlying stiffness-­
induced collective detachment.

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
Piezo1 activity increases collective detachment of ovarian
cancer cells
Next, we examined the role of stretch-­activated ion channel Piezo1 in
stiffness-­induced collective detachment. While Piezo1 has been char-
acterized as a mechanosensor for shear stress in the vasculature (24)
and stretch in alveoli (25), it has a key role in cell extrusion initiated by
cell crowding in homeostatic epithelial cell layers (26). Therefore, we
added a Piezo1 antagonist [GsMTx-­4; (27)] into the in vitro model
and measured levels of Sph-­CD from soft and stiff substrates. GsMTx-
­4 treatment led to a loss of stiffness-­induced collective detachment in
OV90 cells (Fig. 3A). GsMTx-­4 can affect both Piezo1 and TRP chan-
nels; therefore, to confirm this result, we used CRISPR (fig. S4A) to
knock out PIEZO1 in OV90, OVCAR3, and OVCAR8 (28). Loss of
Piezo1 was confirmed through quantitative reverse transcription
polymerase chain reaction (qRT-­PCR) and immunofluorescent stain-
ing, and minimal changes in proliferation rate were observed (fig. S4,
B to D). When the knockout cells were cultured on soft and stiff sub-
strates in the in vitro model, the stiffness-­dependent increase in Sph-­
CD was lost (Fig. 3B and fig. S5A). Next, we treated wild-­type cells
seeded on soft and stiff substrates with Yoda1, a small-­molecule ago-
nist of Piezo1 (29). We observed a significant increase in Sph-­CD
from soft substrates with Yoda1 treatment, suggesting that Piezo1 ac-
tivity is sufficient to facilitate spheroid collective detachment in the
absence of other stiffness-­dependent effects (Fig. 3C and fig. S5B).
Despite the reported role of Piezo1 in cell extrusion from monolayers,
we did not observe a change in single cells or Sph-­SC in response to
either PIEZO1 KO or Yoda1 (fig. S5, A and B).
Fig. 2. Stiffness-­induced proliferation and YAP signaling are not responsible We next explored whether the dependence on Piezo1 resulted
for increased collective detachment. (A) Quantification of OV90 Sph-­CD from from changes in expression or activity on stiff substrates. PIEZO1 ex-
soft or stiff gels treated with vehicle [dimethyl sulfoxide (DMSO)] or aphidicolin pression in cells in the monolayer showed no difference between soft
(4 μg/ml) to inhibit proliferation. (B) Representative images from Click-­iT EdU stain- and stiff substrates (Fig. 3D). Likewise, immunostaining of the
ing of OV90 cells seeded on soft or stiff PAA gels. Spheroid buds are circled in white. monolayer demonstrated similar levels of Piezo1 at the protein level
(C) Quantification of proliferating OV90 cells within buds on the gel or in the mono-
(Fig. 3, E and F). This suggests that increased stiffness increases the
layer (bulk) for soft and stiff gels, three fields of view were averaged per gel.
(D) Representative images of YAP immunostaining (green) for OV90 on soft and stiff
activity of Piezo1, a force-­dependent ion channel that is responsive to
gels. (E) Total YAP signal and percentage of YAP localized to the nucleus quantified changes in membrane tension (30). While typically associated with
from immunofluorescent images of OV90 cells seeded on soft or stiff substrates, activation by an external force such as shear, myosin-­II–mediated
three fields of view were averaged per gel. (F) Quantification of OV90 Sph-­CD from tractions can lead to Piezo1 activity (31). As substrate stiffness can
soft or stiff PAA gels treated with vehicle (DMSO) or 2 μM verteporfin to inhibit affect tractions, we used traction force microscopy and determined
YAP. For (A), (C), (E), and (F), data shown are the average ± SD of N = 3 to 6 individ- that tractions were significantly higher in cells on the stiffer sub-
ual gels per condition. Statistical tests are two-­way ANOVA with Tukey’s [(A), (C), strates (Fig. 3G). When OV90 were treated with ML-­7 to inhibit
and (F)], unpaired t test (E). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. myosin light chain kinase, the number of Sph-­CD was significantly
Scale bars, 100 μm (B) and 10 μm (D). DAPI, 4′,6-­diamidino-­2-­phenylindole; A.U., decreased, consistent with a role for contractility (Fig. 3H). The num-
arbitrary units.
ber of buds was not affected by ML-­7 treatment (fig. S5C), further

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 3 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

expression) and in the stroma in stage III/IV HGSOC, while there was
minimal Piezo1 detected in the absence of tumors (Fig. 3J).

Piezo1 activity increases MMP levels, resulting in remodeled

ECM and increased collective detachment
The data indicate that the increase in Sph-­CD is a result of Piezo1 ac-
tivity specifically regulating spheroid release, as bud formation and
proliferation were not affected by stiffness (Figs. 1C and 2, B and C)
and Sph-­CD were smaller than buds (fig. S1G). Spheroid release re-
quires a loss of cell-­cell and cell-­ECM interactions between the spher-
oid and the substrate/monolayer. As Piezo1 activity has been shown
to induce expression of matrix metalloproteinases (MMPs) (32, 33),
we evaluated whether MMP activity affected stiffness-­induced collec-
tive detachment. Batimastat, a pan-­MMP inhibitor, was incorporated
into the model and found to negate the stiffness-­induced increase in

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
collective detachment (Fig. 4A and fig. S6A). This effect was specific
to collective detachment, as there were no significant effects on
single-­cell detachment. To identify the specific MMPs contributing to
stiffness-­induced collective detachment, media from cells cultured on
soft or stiff substrates was assayed for a set of eight MMPs (Fig. 4B and
fig. S6B). Of the MMPs included in the screen, MMP-­1 and MMP-­10
were the most consistently expressed. MMP-­1 was significantly up-­
regulated on stiff substrates in OV90 cells; while the level of MMP-­1
was not affected by stiffness in OVCAR3 or OVCAR8 cells, we noted
that the concentration in those conditions was comparable to the el-
evated level of OV90 on stiff substrates. MMP-­10 was also significant-
ly elevated on stiff substrates in OV90 cells and was increased but
not statistically significant in response to stiffness for OVCAR3 and
OVCAR8. As there are no specific inhibitors for MMP-­1 or MMP-­10
to isolate which MMP is responsible for increased detachment, we
Fig. 3. Piezo1 activity is needed for stiffness-­induced collective detachment.
(A) Quantification of Sph-­CD from OV90 cells on soft or stiff gels treated with vehi- examined our Piezo1 knockout cells cultured on stiff substrates to de-
cle (DMSO) or 10 μM GsMTx-­4 to inhibit Piezo1. (B) Quantification of Sph-­CD from termine which MMP was consistent with the Piezo1 mechanism. For
wild-­type (WT) OV90 or OV90 with PIEZO1 knocked out (KO) by CRISPR on soft or all three cell lines, we observed a significant decrease in MMP-­1 with
stiff gels. (C) Quantification of Sph-­CD from OV90 cells on soft or stiff gels treated loss of Piezo1 (Fig. 4C and fig. S6B).
with vehicle (DMSO) or 10 μM Yoda1 to activate Piezo1. (D) qRT-­PCR for PIEZO1 for While the media levels of MMP-­1 and MMP-­10 were not entirely
OV90 cells on soft and stiff gels at 72 hours. Data are average ΔΔCt ± SD relative to consistent with the prior results for collective detachment, it is impor-
GAPDH and soft gels, technical replicates. (E) Representative Piezo1 staining of tant to consider that MMPs will act locally on the ECM and that these
OV90 on soft and stiff gels. (F) Piezo1 signal from OV90 cells seeded on soft or stiff effects may be different when bulk concentrations are only modestly
gels, at least five fields of view were averaged per gel. (G) Root mean square (RMS)
different. MMP-­1 is a collagenase that degrades Col I, and MMP-­10 is
traction of OV-­90 cells on soft and stiff substrates at 24 hours, two to five fields of
view were averaged per gel. (H) Quantification of Sph-­CD from OV90 cells seeded
a stromelysin that acts on several ECM components, including fibro-
on soft or stiff gels treated with vehicle (DMSO) or 0.5 μM ML-­7 to inhibit myosin nectin (34, 35). To determine whether the MMPs were affecting ECM
light chain kinase. (I) Representative images of immunofluorescent staining of be- organization of the substrate/buds, the PAA gels were stained with
nign and HGSOC omenta for Piezo1. Pan-­cytokeratin (PanCK) was used to mark collagen-­binding adhesion protein 35 (CNA35) and an antifibronec-
tumor cells. (J) Integrated intensity of Piezo1 staining within the tissue, three fields tin antibody. CNA35–enhanced green fluorescent protein (EGFP) is a
of view were averaged per patient. For (A) to (H), data shown are the average ± SD of fluorescently tagged collagen-­binding protein that preferentially la-
N = 3 to 5 individual gels per condition. In (J), data shown are the average ± SD of bels fibrillar collagen (36), resulting in minimal background staining
N = 7 to 10 patients per group. Statistical tests are two-­way ANOVA with Tukey’s [(A) of the Col I covalently cross-­linked to the PAA gel (fig. S7A). Cells on
to (C) and (H)], unpaired t test [(D) and (F)], unpaired t test with Welch’s correction stiff substrates showed a marked reduction in Col I signal, consistent
(G), Mann-­Whitney test (J). *P < 0.05, **P < 0.01, and ****P < 0.0001. Scale bars,
with the observation of elevated MMP-­1 (Fig. 4D and fig. S7B).
100 μm.
Fibronectin was also decreased on stiff substrates but not as strongly.
qRT-­PCR confirmed that transcription of COL1A1 and FN1 was not
supporting that proliferation-­induced bud formation is independent affected by stiffness (fig. S7C), supporting that these differences result
of stiffness. from MMP activity. Combined, these data support that MMP-­1 is el-
To confirm that Piezo1 has potential clinical relevance in HGSOC evated by Piezo1 activity, resulting in remodeling of the ECM in the
metastasis, we stained omenta from patients with stage III/IV HGSOC monolayer and increased collective detachment.
compared to patients without omental malignancies. As expected,
malignancies in the omentum were associated with a stark increase in Loss of Piezo1 slows tumor metastasis in vivo
cellularity and Pan-­CK+ tumor cells (Fig. 3I). Piezo1 was expressed To evaluate the impact of Piezo1-­ induced collective detachment
both in the tumor bulk (defined as areas with consistent Pan-­CK in vivo, we used a xenograft model widely used to study metastasis in

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 4 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
Fig. 4. Stiffness increases MMPs, leading to decreased ECM. (A) Quantification of Sph-­CD from OV90 cells on soft or stiff gels treated with vehicle (DMSO) or 10 μM
batimastat. (B) MMP concentrations for OV90 cells on soft or stiff gels. (C) MMP-­1 and MMP-­10 concentrations for WT OV90 [same data in (B)] or OV90 with PIEZO1 knocked
out by CRISPR on stiff gels. (D) Representative images and quantification of immunofluorescent staining of OV90 cells seeded on soft or stiff gels for 72 hours stained with
CNA35-­EGFP (to visualize deposited Col I) or antifibronectin. For (A) to (D), data shown are the average ± SD of N = 3 to 5 individual gels per condition. Statistical tests are
two-­way ANOVA with Tukey’s (A), Sidak’s multiple comparisons [(B) and (C)], and unpaired t test (D). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Scale bars,
50 μm. FN, fibronectin.

HGSOC (Fig. 5A) (37). As OV90 cells showed the most robust re- were similar in size; fig. S8A). Histological analysis of the mesentery at
sponse to stiffness and Piezo1, we tested OV90 and OV90-­​PIEZO1−/− day 13 showed microscopic tumors on the mesentery in both mice
cells. While prior reports stated that ascites does not develop with (fig. S8B), and the percentage of tissue covered by tumor cells was
OV90 in nude mice (38), we observed a small volume of ascites by day lower but not significantly different (Fig. 5E). Overall, the data sup-
13 that was measurable at day 35 in Scid mice (Fig. 5B). Ascites vol- port that loss of Piezo1 has no impact on tumor initiation but may
ume was significantly larger in mice injected with wild-­type OV90 slow metastasis to the mesentery through a reduction in spheroid-­
compared to OV90-­​PIEZO1−/− cells. Nearly all mice had macroscopic based spread through the ascites.
tumors in the omentum by day 13, and all mice had macroscopic tu-
mors in the omentum at day 35, regardless of Piezo1 status (Fig. 5C).
Omental tumors increased in volume with time and did not show a DISCUSSION
difference in size with respect to Piezo1 expression (Fig. 5D). Simi- Using an in vitro culture system to study the role of the microenviron-
larly, nearly all mice injected with wild-­type OV90 had visible tumors ment on collective detachment of HGSOC cells, we determined that
in the mesentery by day 13, and this increased to all mice at day 35. In increased stiffness resulted in an increased number of ovarian cancer
contrast to the omentum, significantly fewer mice injected with spheroids. Single-­cell detachment was unaffected, suggesting that the
OV90-­​PIEZO1−/− cells had macroscopic tumors in the mesentery at response to stiffness affected a step in the collective detachment pro-
both time points (Fig. 5C). Tumors on the mesentery were substan- cess. Mechanistic investigations revealed that increased stiffness led to
tially smaller than omental masses; because of the limited number of Piezo1 activation, which led to increased MMP levels and remodeling
tumors in the OV90-­​PIEZO1−/− mice, it was not possible to compare of Col I and fibronectin that promoted detachment. Interrupting this
tumor sizes, but a trend of smaller tumors with OV90-­​PIEZO1−/− was cascade by knockout of Piezo1 slowed progression in an in vivo mod-
observed (Fig. 5D). A preference of ovarian cancer cells to colonize el of HGSOC metastasis, with less ascites accumulation and lower
the omentum has been described in patients (39), and large tumors tumor burden on the mesentery.
are present on the omentum after only a few weeks in the mouse mod- While the process of collective detachment has been proposed as a
el (6). Therefore, we hypothesized that the decrease in tumors in the mechanism of HGSOC metastasis (12), there are many open ques-
mesentery resulted from a reduction of tumor spheroids released tions surrounding how the process occurs and what factors regulate
from the omental mass due to loss of Piezo1. Consistent with this hy- the release of single cells versus spheroids. We previously developed
pothesis, we observed significantly fewer spheroids in ascites from an in vitro model using a filter to separate Sph-­CD from single cells
mice injected with OV90-­​PIEZO1−/− cells (Fig. 5B; the spheroids and any spheroids that form by aggregation (Sph-­SC). However, the

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 5 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
Fig. 5. Loss of PIEZO1 slows metastasis to the mesentery. (A) Tumors were initiated by intraperitoneal injection of WT and PIEZO−/− OV90 cells. After 13 or 35 days, the
omentum and mesentery were examined for tumor burden. (B) By 35 days, all mice developed ascites, with the volume significantly less in mice injected with OV90 with
PIEZO1 knocked out by CRISPR. Quantification of spheroids isolated from ascites fluid, data shown are average spheroids per field of view (FOV). N = 6 to 8 mice per group,
with at least three fields of view averaged per mouse. ***P < 0.001 by unpaired t test. (C) Mice were categorized by the presence or absence of macroscopic tumors. N = 12
to 13 mice per group per time point. P values are for Fisher’s exact test. (D) The total volume of all macroscopic tumors on the two tissue sites. Because of the low number
of macroscopic tumors in the mesentery, statistics were not possible. (E) Hematoxylin and eosin (H&E)–stained sections of mesentery were imaged, and the percentage
of tissue that was tumor was quantified. N = 9 to 11 mice per group, not significant by Mann-­Whitney test. Data from day 13 were collected from a single trial. The day 35
data are a concatenation of two separate experiments; ascites collection was only conducted on the second trial.

physical setup of the model with a solid substrate under the cells and in vitro models; a further challenge is that most biomaterial ap-
means that it is possible that some Sph-­CD form from single cells that proaches that do so cannot recreate the range of stiffnesses observed
quickly aggregated in solution before reaching the filter (9). Our re- in metastatic HGSOC. Therefore, we used PAA gels, where the elastic
sults here support that most of the spheroids captured in the top filter modulus of the substrate can be varied independently of the density of
are produced by collective detachment. First, if the Sph-­CD arise from adhesive ligands conjugated to the surface (42, 43). While this system
single cells that aggregate quickly, then we would expect to see a con- cannot approximate a three-­dimensional environment, it is an appro-
comitant increase in single cells when Sph-­CD increase; however, priate mimic for collective detachment in HGSOC as tumors are lo-
stiffness only affected Sph-­CD numbers. Second, we observed bud-­ cated along the tissue/peritoneal fluid boundary. While most studies
like structures on the monolayers that could be the source of Sph-­ use the PAA system to vary stiffness at a set ECM density, those exam-
CD. We note that Sph-­CD were smaller in size than these buds ining combinations of stiffness and density have demonstrated effects
(fig. S1G), consistent with our stiffness-­dependent mechanism where by both on adhesion (43) and cell spreading (44). Similarly, in a three-­
MMP cleavage of ECM could release a group of cells. Elucidating the dimensional model that can modulate Col I fiber density and bulk
full mechanism of this process is beyond the scope of the current stiffness independently, both variables were found to affect breast can-
work but will likely identify additional approaches to limit collective cer cell invasion (45). However, our results demonstrate an impact of
detachment in metastasis. In addition, while we have focused to date stiffness that is independent of Col I density (Fig. 1), and others have
on a monoculture to model the tumor, macrophages and fibroblasts shown that breast cancer cell response to prolactin stimulation was
are present in the surrounding tumor microenvironment. Their pres- affected by stiffness alone (46). This reinforces that the influence of
ence may be particularly important for ECM-­based mechanisms as stiffness and ECM density should be considered both alone and in
stromal fibroblasts are important to develop a robust fibronectin net- combination. Of course, the ECM of tumors includes components be-
work (40) and ECM receptors such as DDR2 play roles in the stromal yond Col I (14), so additional studies are needed to determine wheth-
compartment in HGSOC (41). er changes in ECM composition associated with HGSOC progression
Here, we sought to expand our model to test the hypothesis that complement or counteract the impact of increased stiffness ob-
changes in the tumor microenvironment associated with disease pro- served here.
gression will increase collective detachment. While there are many To identify the mechanism responsible for the stiffness-­induced in-
changes in the microenvironment as tumors progress, it is well estab- crease in collective detachment, we considered the potential roles of
lished that in HGSOC, the metastatic sites have increased Col I depo- proliferation (47), YAP (48), and Piezo1 (49). Consistent with our ob-
sition and are substantially stiffer (14, 18). Separating the effects of servation that Sph-­CD may result from the bud-­like structures on the
increased Col I and stiffness is a nontrivial problem in many in vivo cell monolayer, proliferation was necessary for robust bud formation

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 6 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

and increased Sph-­CD but was not affected by the increase in stiffness. predict that the tumor bulk would remain at a pathophysiological
Likewise, YAP/TAZ was activated by stiffness but not responsible for stiffness.
the observed increase in Sph-­CD (Fig. 2). Piezo1 is a stretch-­activated On the basis of our in vitro findings, we next asked how loss of
ion channel that converts mechanical forces into cellular signaling PIEZO1 would affect tumor progression in vivo. A recent study found
through an influx of Ca2+ (49). Similar to others (50), we found Piezo1 that PIEZO1 knockdown in A1847 ovarian cancer cells slowed tumor
expressed in HGSOC tumors but did not observe changes in Piezo1 growth and metastasis; however, this study used a subcutaneous xe-
expression in response to stiffness (Fig. 3). Piezo1 is activated by nograft (50) rather than an intraperitoneal xenograft that better mim-
changes in membrane tension that open and close the ion channel, in- ics the patient’s tumor microenvironment (37). We focused our
cluding shear stress (24) and substrate stiffness (51, 52). Through analysis on the visceral fat deposits commonly involved in advanced
chemical inhibition, activation, and CRISPR knockout, we confirmed HGSOC—the omentum (~80% of patients with stage III/IV HGSOC)
that Piezo1 activity is necessary for the stiffness-­dependent increase in and the mesentery (~30% of patients with stage III/IV HGSOC) (59).
Sph-­CD (Fig. 3). Cellular tractions were increased on stiffer substrates, The high prevalence of tumors on the omentum suggests that it is an
and inhibiting actomyosin contractility decreased Sph-­CD release, early site of distal metastasis. Our results demonstrated no effect of
providing a potential mechanism by which substrate stiffness increases PIEZO1 loss on tumor initiation or omental burden. As HGSOC pro-
Piezo1 activity (31). Relevant to our context, Piezo1 was shown to reg- gresses, tumor burden increases throughout the peritoneum, and

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
ulate homeostatic non-­apoptotic cell extrusion in epithelial monolay- there is an onset of ascites accumulation with larger volumes correlat-
ers, and inhibition of Piezo1 led to reduction of cell extrusion (26). ing to poor patient outcomes (60). With PIEZO1 knocked out, we ob-
However, we did not observe an increase in single-­cell detachment served decreases in ascites fluid, number of tumor spheroids in this
with stiffness, suggesting that the findings here represent additional fluid, and macroscopic tumor numbers on the mesentery consistent
pathways by which Piezo1 can regulate epithelial cell organization. with a reduction in overall metastatic burden.
Piezo1 has been linked to other collective processes such as reepitheli- These results are intriguing when we consider the differences be-
alization during wound healing, with loss of Piezo1 accelerating wound tween the untreated animal model and standard of care for patients.
healing in mice (53). First, the finding of large omental disease is consistent with patient ob-
Given the data suggesting that Sph-­CD detach from buds, we con- servations, where omental masses are dealt with by omentectomy dur-
sidered the potential for alterations in the ECM within the bud or ing debulking surgery. The decrease in metastases on the mesentery
monolayer. Stiffness did not affect expression of COL1A1 or FN1, two suggests that PIEZO1 inhibition could be used to reduce reseeding of
of the most abundant ECM proteins in metastatic HGSOC (14). We the visceral adipose after debulking; this is particularly important as
observed changes in MMP-­1 and/or MMP-­10 in a subset of condi- disease on the mesentery is often not amenable to surgical removal
tions, as well as decreases in Col I and fibronectin for all cell lines on (61) and the challenges associated with recovery usually prohibit a sec-
stiffer substrates. This finding may seem in conflict with our prior ond debulking surgery (62). In addition, the decrease in spheroids sug-
study demonstrating a role for fibronectin in Sph-­CD for reattach- gests that PIEZO1 inhibition could be used to improve response to
ment; however, in that study, we demonstrated that the ECM was pri- adjuvant chemotherapy as spheroids are more resistant to anoikis and
marily produced by cells after they detached rather than taken from chemotherapy (11). Future studies incorporating animal models of
the substrate (9). Consistent with the proposed MMP-­based mecha- standard of care (63) could further clarify the impact of inhibiting
nism, Piezo1 is linked to increased MMP-­2 in endothelial cells (54) PIEZO1 on metastatic ovarian cancer and determine whether it is a
and human trabecular meshwork cells (55). promising target for therapy. Given the pleiotropic effects of Piezo1
The finding of a MMP-­based mechanism raises the question of throughout the body, this would likely be best done in combination
whether the increase in MMPs on stiff matrices would be sufficient to with targeted delivery approaches (64).
soften the tumor microenvironment and perhaps enough to lessen
spheroid production. The model system used in this work is a PAA
gel; therefore, its stiffness was not susceptible to MMP degradation MATERIALS AND METHODS
and softening. It was recently shown that a compressed collagen gel Unless noted, all materials were purchased from Thermo Fisher
(elastic modulus, 4 kPa) treated with recombinant MMP-­1 (800 ng/ Scientific.
ml) softened from 4 to 1 kPa over 5 hours (56). While this would sug-
gest we could see substantial softening, we note that the highest con- Cell lines
centration of MMP-­1 observed in our cultures was on the order of All experiments used ovarian cancer tumor cell lines OV90, OVCAR3,
10 ng/ml (80-­fold less), which suggests that this effect would be slower and OVCAR8 (American Type Culture Collection). Cell lines were
or lessened—particularly as the stiffness in metastatic HGSOC is an authenticated by human short tandem repeat analysis at the TRIP
order of magnitude higher. Correlating in vitro concentrations to Lab at the University of Wisconsin (UW)–Madison. Cells were cul-
in vivo settings is challenging, but we would expect that dilution in tured in 1:1 medium 199 (with Earle’s salts and l-­glutamine; Sigma-­
the ascites would result in lower concentrations. Our prior analysis of Aldrich, St. Louis, MO) and MCDB 105 medium (Sigma-­Aldrich).
14 patients undergoing primary debulking surgery found a maxi- Experiments were conducted in serum-­free medium, and mainte-
mum concentration MMP-­1 (5 ng/ml) in ascites (57). It is possible nance culture included 15% heat-­inactivated fetal bovine serum. All
that this question could be assessed using a polyethylene glycol– media included 1% penicillin-­streptomycin.
based system with MMP-­degradable linkers (58); however, this will
likely overestimate any softening relative to the in vivo setting, where PAA gel preparation
surrounding fibroblasts would be depositing matrix at the same time The stiffness of a benign omentum (soft) and an omentum with meta-
as MMPs are active. Therefore, while we acknowledge that there may static disease (stiff) were estimated by a blinded gynecologic oncolo-
be local softening that could turn down the release of spheroids, we gist who manually palpated PAA gels over a range of stiffnesses

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 7 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

(Young’s modulus from 2 to 44 kPa, values determined from stress-­ dimethyl sulfoxide (DMSO) as a negative control. Six hours later
strain curve by dynamic mechanical analyzer, TA Instruments RA III). (48 hours in total), cells were fixed, permeabilized, subjected to the
The “benign” tissue corresponded to 5 kPa (soft; 10% acrylamide and Click-­iT reaction, and imaged. Proliferating cells in the bulk of the
0.03% bis-­acrylamide), and the “metastatic” tissue corresponded to monolayer and within buds were counted in FIJI, and the percentage
44 kPa (stiff; 10% acrylamide and 0.45% bis-­acrylamide; fig. S1A). For of proliferating cells in the bulk and within buds was calculated.
both gel conditions, 0.5% Irgacure-­2959 was added as a photoinitiator
(Advanced Biomatrix, Carlsbad, CA). Silanized glass coverslips (9 mm Interventions
by 9 mm; Electron Microscopy Sciences, Hatfield, PA) were placed on Aphidicolin (4 μg/ml), vertoporfin (2 μM), GsMTx-­4 (10 μM),
top 19-­μl drops of prepolymer solution and cross-­linked under ultra- Yoda1 (10 μM), ML-­7 (0.5 μM), or batimastat (10 μM) was incorpo-
violet (UV) light at 254 nm for 15 min, producing gels with an ap- rated into the media (Tocris, Bristol, United Kingdom) after cells had
proximate thickness of 235 μm (enough to ensure that cells cannot attached. An equivalent volume of DMSO was incorporated to ac-
sense the coverslip stiffness) (65). Phosphate-­buffered saline (PBS) was count for vehicle effects. Fresh inhibitor/activator/vehicle was spiked
added, and the gels were swelled overnight. Sulfo-­SANPAH (0.5 mg/ into the medium every 24 hours.
ml) dissolved in 50 mM Hepes (pH 8.0) was added to gels, and UV
light was applied for 25 min. Gels were washed twice in 50 mM Hepes Immunostaining

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
and 2× in PBS on a shaker. A low (100 μg/ml) or high (2000 μg/ml) Cells were fixed in 4% paraformaldehyde for 15 to 25 min and blocked
concentration of PureCol Collagen I (Advanced Biomatrix) diluted in for 45 to 60 min in 10% goat serum or 10% donkey serum. Piezo1
PBS was added to the gels and incubated overnight [for experiments primary antibody (1:200; Proteintech, San Diego, CA) and YAP pri-
in Figs. 2 to 4, Col I (100 μg/ml) was used]. The Col I solution was mary antibody (1:500; Cell Signaling Technologies, Danvers, MA)
aspirated, and 50 mM tris-­HCl was added to the gels and incubated were applied overnight at 4°C. Secondary antibody [Alexa Fluor 488
at room temperature for 15 min to quench any remaining sulfo-­ goat anti-­rabbit (1:500) or Alexa Fluor 647 donkey anti-­rabbit (1:250)]
SANPAH. Gels were washed three times with PBS, and UV light was was applied for 1 hour at room temperature. Cells were counter-­stained
applied for 30 min to sterilize. Conjugation was confirmed by immu- with Hoechst 33258 (1:500) for 5 min at room temperature. To visual-
nostaining with Col I primary antibody (1:500; Abcam, Waltham, MA; ize deposited Col I and fibronectin, paraformaldehyde-­fixed samples
ab34710) and Alexa Fluor 488 goat anti-­rabbit (1:500). were incubated with 50 μM CNA35-­EGFP (36) in 5% bovine serum
albumin (BSA) at 37°C overnight. After washing (0.1% BSA in PBS) at
In vitro model least 3 times for 30 min each, fibronectin primary antibody in 5% BSA
The in vitro model used in these experiments was described previ- (1:100; Abcam, ab2413) was applied overnight at 4°C. Secondary anti-
ously (9). Briefly, HGSOC cell lines (OV90, OVCAR3, or OVCAR8) body (1:200; Alexa Fluor 647 goat anti-­rabbit) was applied for 1 hour at
were seeded onto PAA gels functionalized with Col I at a density of room temperature, and cells were counterstained with Hoechst 33342
926,000 cells/cm2 in media with 15% serum for 4 hours. These gels (1:500) for 15 min at room temperature.
were placed into a 40-­μm cell strainer (Sigma-­Aldrich) fitted into a Formalin-­fixed, paraffin-­embedded samples from patients who
six-­well plate filled with 10 ml of serum-­free media. After 72 hours, had undergone surgical debulking for HGSOC or other conditions
the Sph-­CD were collected by removing the PAA gel from the filter, were obtained from archived pathology samples through a protocol
inverting the filter on top of a 50-­ml conical tube and washing with approved by the UW-­Madison Institutional Review Board (IRB) (ta-
4 ml of serum-­free media. Sph-­SC were collected by filtering the me- ble S1). Sections (5 μm) were prepared by the TRIP lab (UW Carbone
dium from the culture plate through a separate 40-­μm cell strainer Cancer Center), deparaffinized, subjected to antigen retrieval, and
and collecting the spheroids in the same manner as the Sph-­CD blocked with 10% goat serum. Primary antibodies for Piezo1 (1:500;
spheroids. Single cells that had detached but not aggregated were col- Novus Biologicals, Centennial, CO) and pan-­cytokeratin (1:500;
lected from the media that had passed through the strainers. To quan- OriGene, Rockville, MD) were applied overnight at 4°C. Secondary
tify the number of Sph-­CD, Sph-­SC, or single cells that detached from antibodies [Alexa Fluor 488 goat anti-­rabbit (1:500) and Alexa Fluor
each gel, the media was distributed in 0.5 ml of aliquots in a 48-­well 647 goat anti-­guinea pig (1:500)] were applied at room temperature
plate. Wells were imaged and analyzed in FIJI using a macro script for 1 hour. Slides were mounted using ProLong Diamond Antifade
(66). Briefly, background was subtracted, images were converted to with 4′,6-­diamidino-­2-­phenylindole.
binary, and “Fill Holes” used to fill holes were created by binary con- For all staining, a no primary control was included to account for
version. “Analyze Particles” was used to measure the area of objects nonspecific signal. All imaging was done on a Zeiss Axio Observer.Z1
and count them as single cells or spheroids. Particles were considered inverted microscope with an AxioCam 506 mono camera and a Plan-­
a spheroid if more than three cells could be visually identified and the Apochromat 10× 0.4–numerical aperture (NA) or 20× 0.8-­NA air
diameter was greater than 20 μm; single cells had a 10-­to 20-­μm di- objective. Coverslips with PAA gels were inverted on coverslips to im-
ameter. Spheroid buds were identified and counted manually; buds prove focus while imaging. For quantification of staining, the mean
had a higher cell density and at least three cells with a raised z-­position fluorescent intensity was measured in FIJI, and the background was
adjacent to one another (fig. S1). subtracted on the basis of measurement of controls without primary
antibodies. For each individual gel, the mean value of at least three
Click-­iT EdU analyzed images is reported.
To visualize proliferating cells in the monolayer/buds, Click-­iT
5-­ethynyl 2′-­deoxyuridine (EdU) Alexa Fluor 488 imaging was per- Quantitative reverse transcription polymerase
formed according to the manufacturer’s instructions. OV90 cells chain reaction
were seeded on soft and stiff PAA gels and, after 42 hours of culture, RNA was isolated from cells on the monolayer of soft and stiff sub-
changed to serum-­free media supplemented with 10 μM EdU or strates at 72 hours using the Monarch Total RNA Miniprep Kit (New

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 8 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

England Biolabs, Ipswich, MA). The quality and amount of isolated for each gene target; an equal volume of virus constructs with
RNA were determined using a NanoDrop One, and cDNA was syn- PIEZO1 gRNA 1, 2, and 3 were used to increase the probability of
thesized by SuperScript III First-­Strand Synthesis. Quantitect primers an effective gene edit. The viral supernatant was added to each cell
for PIEZO1, COL1A1, and FN1 (Qiagen, Germantown, MD) were line at a 2:3 virus to ovarian cancer media volume ratio. Polybrene
used in combination with SsoAdvanced Universal SYBR Green Su- was added at 8 μg/ml, and the plate was centrifuged at 165g for
permix (Bio-­Rad, Hercules, CA) on a CFX Connect Real-­Time PCR 2 hours to aid transduction. After incubation for 48 hours, puro-
Detection System (Bio-­Rad). mycin (4 μg/ml) was added for an additional 48 hours for selec-
tion. Knockout was verified by qRT-­PCR for PIEZO1 (Qiagen,
Traction force microscopy Hilden, Germany) and immunostaining.
For traction force microscopy, PAA gels with a thickness of 100 μm
were prepared using the formulations above with fluorescent MMP quantification
particles (0.5 μm in diameter, red, Life Technologies). Glass cov- Media was collected 72 hours after seeding from cells on soft and stiff
erslips (18 mm in diameter) were hydrophobically treated to pre- gels, soft gels were treated with Yoda1, and stiff gels were seeded with
vent sticking to the gel by dipping each coverslip in a solution of PIEZO1 knockout cells. Media was centrifuged, and the supernatant
97% hexane, 0.5% acetic acid, and 2.5% (tridecafluoro-­1,1,2,2-­ was snap-­frozen in liquid nitrogen and quantified by Eve Technolo-

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
tetrahydrooctyl) triethoxysilane and air drying overnight. No. 1.5 gies (Calgary, Canada) using a bead-­based multiplex immunoassay
glass-­bottom dishes (Cellvis, Mountain View, CA) were treated as for MMP-­1, MMP-­2, MMP-­3, MMP-­7, MMP-­8, MMP-­9, MMP-­10,
described previously (67), 20 μl of the gel solution was pipetted MMP-­12, and MMP-­13. MMP-­1 readings from some samples were
onto each dish, and a coverslip was placed on top. The dishes were near the saturation limit; therefore, all samples were assayed at up to a
placed upside-­down in a centrifuge and spun for 3 min at 15 rela- 1:50 dilution using an MMP-­1–specific enzyme-­linked immunosor-
tive centrifugal force to localize the fluorescent particles to the top bent assay (Abcam).
of each gel. Dishes were carefully removed, exposed to UV light at
254 nm for 15 min, and then swelled overnight in Dulbecco’s In vivo studies
Phosphate-­Buffered Saline. Gels were conjugated with ECM and Female BALB/c scid (CBySmn.Cg-­​Prkdcscid/J) mice (approximately
seeded as above; after 24 hours, images were collected of the cells 10 weeks of age) were procured from the Jackson Laboratory
(phase contrast) and particles (fluorescence) using the Zeiss (Bar Harbor, ME). All animal protocols were approved by the In-
microscope at 37°C. The cells were removed from the substrates stitutional Animal Care and Use Committee (IACUC) at the
by adding trypsin, which allowed the PAA substrates to recover UW-­Madison School of Medicine and Public Health. Mice were in-
to an undeformed reference state, and the fluorescent particles traperitoneally injected with 2 million OV90 or OV90-­​PIEZO1−/−
were imaged again. Fast Iterative Digital Image Correlation was cells in 100 μl and euthanized by CO2 asphyxiation after 13 or
used to measure cell-­generated substrate displacements with re- 35 days. Ascites was removed from the mice at day 35 using 22 ½ gauge
spect to the undeformed reference images (68). Tractions were needle, and the approximate volumes of ascites were recorded. The
calculated using unconstrained Fourier transform traction cy- peritoneal cavity was then washed by injection of 4 ml of PBS, and
tometry (69) with a correction for finite substrate thickness (70, the collected ascites/PBS was filtered through a 40-­μm filter to iso-
71). The results yielded the traction vector field applied by the late spheroids for quantification as above (Fig. 1A). Mice were nec-
cells to the substrate. ropsied, and macroscopic tumor numbers on the omentum and
mesentery were counted and measured by a digital caliper. The vol-
PIEZO1 CRISPR ume (V) of the macroscopic tumors was approximated using the
CRISPR-­Cas9 was used to knock out PIEZO1 in OV90, OVCAR3, formula V = 1/2 × tumor width (W) × tumor width (W) × tumor
and OVCAR8. Three PIEZO1 guide RNA (gRNA) sequence oligos length (L) (72). Tissue samples were fixed, paraffin-­embedded, sec-
(sequences 1 to 3; table S2) designed and validated from the Zhang tioned, and stained with hematoxylin and eosin (H&E). Images of
lab at the Broad Institute were selected and synthesized (Integrat- the whole H&E-­stained tissue sections were acquired with a Zeiss
ed DNA Technologies, Coralville, IA) (28). lentiCRISPR v2 was a Axio Observer.Z1 inverted microscope using the Tiles tool. Micro-
gift from F. Zhang (Addgene plasmid #52961; http://n2t.net/add- scopic tumor lesions were outlined on the images and analyzed by
gene:52961; RRID:Addgene_52961); this one vector system deliv- ImageJ. Microscopic tumor burden was defined as the area of a tis-
ers hSpCas9 and chimeric gRNA expression cassettes (28). The sue section occupied by tumors divided by the total area of the tis-
lentiCRISPRv2 vector was dephosphorylated and digested with sue section.
Bsm BI restriction enzyme and subsequently gel-­purified. The for-
ward and reverse gRNA oligos were phosphorylated, annealed, Statistics
and ligated into the cut vector. Whole plasmid sequencing (Plas- Statistical tests are detailed in figure legends and were performed
midsaurus, Eugene, OR) was used to confirm PIEZO1 gRNA in- in GraphPad Prism, and P < 0.05 was considered significant. All
sertion into the lentiCRISPRv2 construct. in vitro experiments were performed at least twice with one repre-
One Shot Stbl3 chemically competent Escherichia coli were sentative experiment shown.
transformed with the PIEZO1 lentiCRISPRv2 plasmid and grown
in LB with ampicillin (100 μg/ml) for selection and expansion. Research standards
Human embryonic kidney 293T cells were cotransfected with Animal protocols were approved by the IACUC at the UW-­Madison
packaging plasmids from the Lenti-­vpak Lentiviral Packaging Kit School of Medicine and Public Health. Tissue samples from human
(Origene) and the PIEZO1 lentiCRISPRv2 plasmid. After 48 and patients were obtained with a consent waiver due to de-­identification
72 hours of incubation, viral batches were collected and combined procedures approved by the UW-­Madison IRB.

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 9 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

Supplementary Materials 21. M. Jang, J. An, S. W. Oh, J. Y. Lim, J. Kim, J. K. Choi, J. H. Cheong, P. Kim, Matrix stiffness
This PDF file includes: epigenetically regulates the oncogenic activation of the Yes-­associated protein in gastric
Figs. S1 to S8 cancer. Nat. Biomed. Eng. 5, 114–123 (2021).
Tables S1 and S2 22. S. W. Chan, C. J. Lim, K. Guo, C. P. Ng, I. Lee, W. Hunziker, Q. Zeng, W. Hong, A role for TAZ in
migration, invasion, and tumorigenesis of breast cancer cells. Cancer Res. 68, 2592–2598
(2008).
REFERENCES AND NOTES 23. C. Wang, X. Zhu, W. Feng, Y. Yu, K. Jeong, W. Guo, Y. Lu, G. B. Mills, Verteporfin inhibits YAP
1. A. S. O'Shea, Clinical staging of ovarian cancer. Methods Mol. Biol. 2424, 3–10 (2022). function through up-­regulating 14-­3-­3σ sequestering YAP in the cytoplasm. Am. J. Cancer
2. H. M. Micek, M. R. Visetsouk, A. J. Fleszar, P. K. Kreeger, The many microenvironments of Res. 6, 27–37 (2016).
ovarian cancer. Adv. Exp. Med. Biol. 1296, 199–213 (2020). 24. S. S. Ranade, Z. Qiu, S. H. Woo, S. S. Hur, S. E. Murthy, S. M. Cahalan, J. Xu, J. Mathur,
3. H. A. Kenny, C. Y. Chiang, E. A. White, E. M. Schryver, M. Habis, I. L. Romero, A. Ladanyi, M. Bandell, B. Coste, Y. S. Li, S. Chien, A. Patapoutian, Piezo1, a mechanically activated ion
C. V. Penicka, J. George, K. Matlin, A. Montag, K. Wroblewski, S. D. Yamada, A. P. Mazar, channel, is required for vascular development in mice. Proc. Natl. Acad. Sci. U.S.A. 111,
D. Bowtell, E. Lengyel, Mesothelial cells promote early ovarian cancer metastasis through 10347–10352 (2014).
fibronectin secretion. J. Clin. Invest. 124, 4614–4628 (2014). 25. K. Diem, M. Fauler, G. Fois, A. Hellmann, N. Winokurow, S. Schumacher, C. Kranz, M. Frick,
4. C. M. Novak, E. N. Horst, E. Lin, G. Mehta, Compressive stimulation enhances ovarian Mechanical stretch activates piezo1 in caveolae of alveolar type I cells to trigger ATP
cancer proliferation, invasion, chemoresistance, and mechanotransduction via CDC42 in release and paracrine stimulation of surfactant secretion from alveolar type II cells. FASEB
a 3D bioreactor. Cancers 12, 1521 (2020). J. 34, 12785–12804 (2020).
5. A. Russo, Z. Yang, G. M. Heyrman, B. P. Cain, A. Lopez Carrero, B. C. Isenberg, M. J. Dean, 26. G. T. Eisenhoffer, P. D. Loftus, M. Yoshigi, H. Otsuna, C. B. Chien, P. A. Morcos, J. Rosenblatt,

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
J. Coppeta, J. E. Burdette, Versican secreted by the ovary links ovulation and migration in Crowding induces live cell extrusion to maintain homeostatic cell numbers in epithelia.
fallopian tube derived serous cancer. Cancer Lett. 543, 215779 (2022). Nature 484, 546–549 (2012).
6. E. I. Harper, T. S. Hilliard, In Vivo and ex vivo analysis of omental adhesion in ovarian 27. C. Bae, F. Sachs, P. A. Gottlieb, The mechanosensitive ion channel Piezo1 is inhibited by
cancer. Methods Mol. Biol. 2424, 199–216 (2022). the peptide GsMTx4. Biochemistry 50, 6295–6300 (2011).
7. S. Mittal, P. Gupta, P. Chaluvally-­Raghavan, S. Pradeep, Establishment of In Vivo ovarian 28. N. E. Sanjana, O. Shalem, F. Zhang, Improved vectors and genome-­wide libraries for
cancer mouse models using intraperitoneal tumor cell injection. Methods Mol. Biol. 2424, CRISPR screening. Nat. Methods 11, 783–784 (2014).
247–254 (2022). 29. W. M. Botello-­Smith, W. Jiang, H. Zhang, A. D. Ozkan, Y. C. Lin, C. N. Pham, J. J. Lacroix,
8. M. Asem, A. Young, C. Oyama, A. ClaureDeLaZerda, Y. Liu, M. J. Ravosa, V. Gupta, A. Jewell, Y. Luo, A mechanism for the activation of the mechanosensitive Piezo1 channel by the
D. Khabele, M. S. Stack, Ascites-­induced compression alters the peritoneal small molecule Yoda1. Nat. Commun. 10, 4503 (2019).
microenvironment and promotes metastatic success in ovarian cancer. Sci. Rep. 10, 11913 30. A. H. Lewis, J. Grandl, Mechanical sensitivity of Piezo1 ion channels can be tuned by
(2020). cellular membrane tension. eLife 4, (2015).
9. H. M. Micek, L. Rosenstock, Y. Ma, C. Hielsberg, L. Montemorano, M. K. Gari, S. M. Ponik, 31. K. L. Ellefsen, J. R. Holt, A. C. Chang, J. L. Nourse, J. Arulmoli, A. H. Mekhdjian, H. Abuwarda,
P. K. Kreeger, Model of collective detachment in high-­grade serous ovarian cancer F. Tombola, L. A. Flanagan, A. R. Dunn, I. Parker, M. M. Pathak, Myosin-­II mediated traction
demonstrates that tumor spheroids produce ECM to support metastatic processes. APL forces evoke localized Piezo1-­dependent Ca(2+) flickers. Commun. Biol. 2, 298 (2019).
Bioeng 7, 016111 (2023). 32. W. Morozumi, K. Aoshima, S. Inagaki, Y. Iwata, S. Nakamura, H. Hara, M. Shimazawa, Piezo
10. S. Miceska, E. Skof, G. Gasljevic, V. Kloboves-­Prevodnik, Morphological and 1 is involved in intraocular pressure regulation. J. Pharmacol. Sci. 147, 211–221 (2021).
immunocytochemical characterization of tumor spheroids in ascites from high-­grade 33. Y. Yu, X. Wu, S. Liu, H. Zhao, B. Li, H. Zhao, X. Feng, Piezo1 regulates migration and invasion
serous carcinoma. Cells 12, 2390 (2023). of breast cancer cells via modulating cell mechanobiological properties. Acta Biochim.
11. S. Makhija, D. D. Taylor, R. K. Gibb, C. Gercel-­Taylor, Taxol-­induced bcl-­2 Biophys. Sin. (Shanghai) 53, 10–18 (2021).
phosphorylation in ovarian cancer cell monolayer and spheroids. Int. J. Oncol. 14, 34. P. A. Arakaki, M. R. Marques, M. C. Santos, MMP-­1 polymorphism and its relationship to
515–521 (1999). pathological processes. J. Biosci. 34, 313–320 (2009).
12. S. Al Habyan, C. Kalos, J. Szymborski, L. McCaffrey, Multicellular detachment generates 35. S. Bord, A. Horner, R. M. Hembry, J. E. Compston, Stromelysin-­1 (MMP-­3) and
metastatic spheroids during intra-­abdominal dissemination in epithelial ovarian cancer. stromelysin-­2 (MMP-­10) expression in developing human bone: Potential roles in skeletal
Oncogene 37, 5127–5135 (2018). development. Bone 23, 7–12 (1998).
13. H. M. Micek, M. R. Visetsouk, K. S. Masters, P. K. Kreeger, Engineering the extracellular 36. S. J. Aper, A. C. van Spreeuwel, M. C. van Turnhout, A. J. van der Linden, P. A. Pieters,
matrix to model the evolving tumor microenvironment. iScience 23, 101742 (2020). N. L. van der Zon, S. L. de la Rambelje, C. V. Bouten, M. Merkx, Colorful protein-­based
14. O. M. T. Pearce, R. M. Delaine-­Smith, E. Maniati, S. Nichols, J. Wang, S. Bohm, V. Rajeeve, fluorescent probes for collagen imaging. PLOS ONE 9, e114983 (2014).
D. Ullah, P. Chakravarty, R. R. Jones, A. Montfort, T. Dowe, J. Gribben, J. L. Jones, 37. A. K. Mitra, D. A. Davis, S. Tomar, L. Roy, H. Gurler, J. Xie, D. D. Lantvit, H. Cardenas, F. Fang,
H. M. Kocher, J. S. Serody, B. G. Vincent, J. Connelly, J. D. Brenton, C. Chelala, P. R. Cutillas, Y. Liu, E. Loughran, J. Yang, M. Sharon Stack, R. E. Emerson, K. D. Cowden Dahl,
M. Lockley, C. Bessant, M. M. Knight, F. R. Balkwill, Deconstruction of a metastatic tumor M. V. Barbolina, K. P. Nephew, D. Matei, J. E. Burdette, In vivo tumor growth of high-­grade
microenvironment reveals a common matrix response in human cancers. Cancer Discov. serous ovarian cancer cell lines. Gynecol. Oncol. 138, 372–377 (2015).
8, 304–319 (2018). 38. L. Hernandez, M. K. Kim, L. T. Lyle, K. P. Bunch, C. D. House, F. Ning, A. M. Noonan,
15. K. J. Cheung, A. J. Ewald, A collective route to metastasis: Seeding by tumor cell clusters. C. M. Annunziata, Characterization of ovarian cancer cell lines as in vivo models for
Science 352, 167–169 (2016). preclinical studies. Gynecol. Oncol. 142, 332–340 (2016).
16. N. Alkhouli, J. Mansfield, E. Green, J. Bell, B. Knight, N. Liversedge, J. C. Tham, R. Welbourn, 39. K. M. Nieman, H. A. Kenny, C. V. Penicka, A. Ladanyi, R. Buell-­Gutbrod, M. R. Zillhardt,
A. C. Shore, K. Kos, C. P. Winlove, The mechanical properties of human adipose tissues and I. L. Romero, M. S. Carey, G. B. Mills, G. S. Hotamisligil, S. D. Yamada, M. E. Peter, K. Gwin,
their relationships to the structure and composition of the extracellular matrix. Am. J. E. Lengyel, Adipocytes promote ovarian cancer metastasis and provide energy for rapid
Physiol. Endocrinol. Metab. 305, E1427–E1435 (2013). tumor growth. Nat. Med. 17, 1498–1503 (2011).
17. R. M. Delaine-­Smith, S. Burney, F. R. Balkwill, M. M. Knight, Experimental validation of a 40. S. Libring, A. Shinde, M. K. Chanda, M. Nuru, H. George, A. M. Saleh, A. Abdullah,
flat punch indentation methodology calibrated against unconfined compression tests T. L. Kinzer-­Ursem, S. Calve, M. K. Wendt, L. Solorio, The dynamic relationship of breast
for determination of soft tissue biomechanics. J. Mech. Behav. Biomed. Mater. 60, 401–415 cancer cells and fibroblasts in fibronectin accumulation at primary and metastatic tumor
(2016). sites. Cancers (Basel) 12, 1270 (2020).
18. K. C. Fogg, C. M. Renner, H. Christian, A. Walker, L. Marty-­Santos, A. Khan, W. R. Olson, 41. A. M. Schab, M. M. Greenwade, E. Stock, E. Lomonosova, K. Cho, W. R. Grither, H. Noia,
C. Parent, A. O'Shea, D. M. Wellik, P. S. Weisman, P. K. Kreeger, Ovarian cells have increased D. Wilke, M. M. Mullen, A. R. Hagemann, I. S. Hagemann, P. H. Thaker, L. M. Kuroki,
proliferation in response to heparin-­binding epidermal growth factor as collagen density C. K. McCourt, D. Khabele, M. A. Powell, D. G. Mutch, P. Zhao, L. P. Shriver, G. J. Patti,
increases. Tissue Eng. Part A 26, 747–758 (2020). G. D. Longmore, K. C. Fuh, Stromal DDR2 promotes ovarian cancer metastasis through
19. J. J. Northey, A. S. Barrett, I. Acerbi, M. K. Hayward, S. Talamantes, I. S. Dean, J. K. Mouw, regulation of metabolism and secretion of extracellular matrix proteins. Mol. Cancer Res.
S. M. Ponik, J. N. Lakins, P. J. Huang, J. Wu, Q. Shi, S. Samson, P. J. Keely, R. A. Mukhtar, 21, 1234–1248 (2023).
J. T. Liphardt, J. A. Shepherd, E. S. Hwang, Y. Y. Chen, K. C. Hansen, L. E. Littlepage, 42. R. J. Pelham Jr., Y. Wang, Cell locomotion and focal adhesions are regulated by substrate
V. M. Weaver, Stiff stroma increases breast cancer risk by inducing the oncogene ZNF217. flexibility. Proc. Natl. Acad. Sci. U.S.A. 94, 13661–13665 (1997).
J. Clin. Invest. 130, 5721–5737 (2020). 43. J. D. Mih, A. S. Sharif, F. Liu, A. Marinkovic, M. M. Symer, D. J. Tschumperlin, A multiwell
20. T. A. Ulrich, E. M. de Juan Pardo, S. Kumar, The mechanical rigidity of the extracellular platform for studying stiffness-­dependent cell biology. PLOS ONE 6, e19929 (2011).
matrix regulates the structure, motility, and proliferation of glioma cells. Cancer Res. 69, 44. A. Engler, L. Bacakova, C. Newman, A. Hategan, M. Griffin, D. Discher, Substrate
4167–4174 (2009). compliance versus ligand density in cell on gel responses. Biophys. J. 86, 617–628 (2004).

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 10 of 11

S c i e n c e A d v a n c e s | R e s e ar c h A r t i c l e

45. A. J. Berger, K. M. Linsmeier, P. K. Kreeger, K. S. Masters, Decoupling the effects of stiffness 62. M. Pergolotti, A. Bailliard, L. McCarthy, E. Farley, K. R. Covington, K. M. Doll, Women's
and fiber density on cellular behaviors via an interpenetrating network of gelatin-­ experiences after ovarian cancer surgery: Distress, uncertainty, and the need for
methacrylate and collagen. Biomaterials 141, 125–135 (2017). occupational therapy. Am. J. Occup. Ther. 74, 7403205140p1–7403205140p9 (2020).
46. C. E. Barcus, P. J. Keely, K. W. Eliceiri, L. A. Schuler, Stiff collagen matrices increase tumorigenic 63. M. Popa, V. Fosse, K. Kleinmanns, L. Bjorge, E. McCormack, Xenograft models of ovarian
prolactin signaling in breast cancer cells. J. Biol. Chem. 288, 12722–12732 (2013). cancer for therapy evaluation. Methods Mol. Biol. 2424, 275–293 (2022).
47. J. Schrader, T. T. Gordon-­Walker, R. L. Aucott, M. van Deemter, A. Quaas, S. Walsh, 64. S. Correa, N. Boehnke, A. E. Barberio, E. Deiss-­Yehiely, A. Shi, B. Oberlton, S. G. Smith,
D. Benten, S. J. Forbes, R. G. Wells, J. P. Iredale, Matrix stiffness modulates proliferation, I. Zervantonakis, E. C. Dreaden, P. T. Hammond, Tuning nanoparticle interactions with
chemotherapeutic response, and dormancy in hepatocellular carcinoma cells. ovarian cancer through layer-­by-­layer modification of surface chemistry. ACS Nano 14,
Hepatology 53, 1192–1205 (2011). 2224–2237 (2020).
48. S. Dupont, L. Morsut, M. Aragona, E. Enzo, S. Giulitti, M. Cordenonsi, F. Zanconato, 65. S. Sen, A. J. Engler, D. E. Discher, Matrix strains induced by cells: Computing how far cells
J. Le Digabel, M. Forcato, S. Bicciato, N. Elvassore, S. Piccolo, Role of YAP/TAZ in can feel. Cell. Mol. Bioeng. 2, 39–48 (2009).
mechanotransduction. Nature 474, 179–183 (2011). 66. J. Schindelin, I. Arganda-­Carreras, E. Frise, V. Kaynig, M. Longair, T. Pietzsch, S. Preibisch,
49. B. Coste, J. Mathur, M. Schmidt, T. J. Earley, S. Ranade, M. J. Petrus, A. E. Dubin, C. Rueden, S. Saalfeld, B. Schmid, J. Y. Tinevez, D. J. White, V. Hartenstein, K. Eliceiri,
A. Patapoutian, Piezo1 and Piezo2 are essential components of distinct mechanically P. Tomancak, A. Cardona, Fiji: An open-­source platform for biological-­image analysis. Nat.
activated cation channels. Science 330, 55–60 (2010). Methods 9, 676–682 (2012).
50. Y. Xiong, L. Dong, Y. Bai, H. Tang, S. Li, D. Luo, F. Liu, J. Bai, S. Yang, X. Song, Piezo1 67. K. Vazquez, A. Saraswathibhatla, J. Notbohm, Effect of substrate stiffness on friction in
activation facilitates ovarian cancer metastasis via Hippo/YAP signaling axis. Channels collective cell migration. Sci. Rep. 12, 2474 (2022).
(Austin) 16, 159–166 (2022). 68. E. Bar-­Kochba, J. Toyjanova, E. Andrews, K.-­S. Kim, C. Franck, A fast iterative digital volume

Downloaded from https://www.science.org at National Sun Yat-Sen University on September 18, 2024
51. M. Lopez-­Cavestany, S. B. Hahn, J. M. Hope, N. T. Reckhorn, J. D. Greenlee, S. C. Schwager, correlation algorithm for large deformations. Exp. Mech. 55, 261–274 (2015).
J. A. VanderBurgh, C. A. Reinhart-­King, M. R. King, Matrix stiffness induces epithelial-­to-­ 69. J. P. Butler, I. M. Tolic-­Norrelykke, B. Fabry, J. J. Fredberg, Traction fields, moments, and
mesenchymal transition via Piezo1-­regulated calcium flux in prostate cancer cells. strain energy that cells exert on their surroundings. Am. J. Physiol. Cell Physiol. 282,
iScience 26, 106275 (2023). C595–C605 (2002).
52. M. M. Pathak, J. L. Nourse, T. Tran, J. Hwe, J. Arulmoli, D. T. Le, E. Bernardis, L. A. Flanagan, 70. X. Trepat, M. R. Wasserman, T. E. Angelini, E. Millet, D. A. Weitz, J. P. Butler, J. J. Fredberg,
F. Tombola, Stretch-­activated ion channel Piezo1 directs lineage choice in human neural Physical forces during collective cell migration. Nat. Phys. 5, 426–430 (2009).
stem cells. Proc. Natl. Acad. Sci. U.S.A. 111, 16148–16153 (2014). 71. J. C. del Alamo, R. Meili, B. Alonso-­Latorre, J. Rodriguez-­Rodriguez, A. Aliseda, R. A. Firtel,
53. J. R. Holt, W.-­Z. Zeng, E. L. Evans, S.-­H. Woo, S. Ma, H. Abuwarda, M. Loud, A. Patapoutian, J. C. Lasheras, Spatio-­temporal analysis of eukaryotic cell motility by improved force
M. M. Pathak. (BioRxiv, 2020). cytometry. Proc. Natl. Acad. Sci. U.S.A. 104, 13343–13348 (2007).
54. H. Kang, Z. Hong, M. Zhong, J. Klomp, K. J. Bayless, D. Mehta, A. V. Karginov, G. Hu, 72. A. Faustino-­Rocha, P. A. Oliveira, J. Pinho-­Oliveira, C. Teixeira-­Guedes, R. Soares-­Maia,
A. B. Malik, Piezo1 mediates angiogenesis through activation of MT1-­MMP signaling. Am. R. G. da Costa, B. Colaco, M. J. Pires, J. Colaco, R. Ferreira, M. Ginja, Estimation of rat
J. Physiol. Cell Physiol. 316, C92–C103 (2019). mammary tumor volume using caliper and ultrasonography measurements. Lab. Anim.
55. W. Morozumi, K. Aoshima, S. Inagaki, Y. Iwata, Y. Takagi, S. Nakamura, H. Hara, (NY) 42, 217–224 (2013).
M. Shimazawa, Piezo1 activation induces fibronectin reduction and PGF2α secretion via
arachidonic acid cascade. Exp. Eye Res. 215, 108917 (2022). Acknowledgments: We thank A. O’Shea for assistance with experimental design and S. Ponik
56. A. Micalet, J. Pape, D. Bakkalci, Y. Javanmardi, C. Hall, U. Cheema, E. Moeendarbary, and K. Masters for helpful discussions as this project evolved. Funding: This work was
Evaluating the impact of a biomimetic mechanical environment on cancer invasion and supported by the National Institutes for Health research grants R01CA240965 (P.K.K.),
matrix remodeling. Adv. Healthc. Mater. 12, e2201749 (2023). R01CA232517 (P.K.K.), and NSF CMMI-­2205141 (J.N. and P.K.K.). Core facilities were supported
57. M. J. Carroll, K. Kaipio, J. Hynninen, O. Carpen, S. Hautaniemi, D. Page, P. K. Kreeger, A by NIH P30CA014520 to the Carbone Cancer Center. The funders had no role in study design,
subset of secreted proteins in ascites can predict platinum-­free interval in ovarian cancer. data collection, or analysis. Author contributions: Conceptualization: H.M.M. and P.K.K.
Cancers (Basel) 14, 4291 (2022). Methodology: H.M.M., N.Y., M.D., and P.K.K. Investigation: H.M.M., N.Y., M.D., L.R., Y.M., C.H.,
58. J. L. Leight, D. L. Alge, A. J. Maier, K. S. Anseth, Direct measurement of matrix M.M., and J.N. Formal analysis: H.M.M., N.Y., M.D., and P.K.K. Data curation: S.M. Resources: S.M.
metalloproteinase activity in 3D cellular microenvironments using a fluorogenic peptide Writing—original draft: H.M.M. and P.K.K. Writing—review and editing: H.M.M., M.M., J.N., and
substrate. Biomaterials 34, 7344–7352 (2013). P.K.K. Competing interests: The authors declare that they have no competing interests. Data
59. J. Prat; FIGO Committee on Gynecologic Oncology, Staging classification for cancer of the and materials availability: All data needed to evaluate the conclusions in the paper are
ovary, fallopian tube, and peritoneum. Int. J. Gynaecol. Obstet. 124, 1–5 (2014). present in the paper and/or the Supplementary Materials.
60. J. B. Szender, T. Emmons, S. Belliotti, D. Dickson, A. Khan, K. Morrell, A. Khan, K. L. Singel,
P. C. Mayor, K. B. Moysich, K. Odunsi, B. H. Segal, K. H. Eng, Impact of ascites volume on Submitted 18 October 2023
clinical outcomes in ovarian cancer: A cohort study. Gynecol. Oncol. 146, 491–497 (2017). Accepted 22 March 2024
61. A. H. Brand, Ovarian cancer debulking surgery: A survey of practice in Australia and New Published 26 April 2024
Zealand. Int. J. Gynecol. Cancer 21, 230–235 (2011). 10.1126/sciadv.adl4463

Micek et al., Sci. Adv. 10, eadl4463 (2024) 26 April 2024 11 of 11