Brief pulses of high-level fluid shear stress enhance metastatic potential and rapidly alter the metabolism of cancer cells

SUMMARY

Circulating tumor cells (CTCs) face challenges to their survival, including mechanical and oxidative stresses that are different from cancer cells in solid primary and metastatic tumors. The impact of adaptations to the fluid microenvironment of the circulation on the outcome of the metastatic cascade is not well understood. Here, we find that cancer cells exposed to brief pulses of high-level fluid shear stress (FSS) exhibit enhanced invasiveness and anchorage-independent proliferation in vitro and enhanced metastatic colonization/tumor formation in vivo. Cancer cells exposed to FSS rapidly alter their metabolism in a manner that promotes survival by providing energy for cytoskeletal remodeling and contractility as well as reducing equivalents to counter oxidative stress associated with cell detachment. Thus, exposure to FSS may provide CTCs with an unexpected survival benefit that promotes metastatic colonization.

In brief

Cancer cells are exposed to elevated fluid shear stress for brief periods during their journey through the circulation. Fluid shear stress exposure triggers rapid transcriptional and metabolic changes that protect cancer cells from oxidative stress and promote productive metastatic colonization.

Graphical abstract

INTRODUCTION

Circulating tumor cells (CTCs) are intermediates in the formation of metastases at sites distant from primary tumors. They exist in the fluid microenvironment of the circulation, where they lack the supportive trophic factors of the primary tumor, adhesion to the extracellular matrix (ECM), and protection from the immune system.^1,2 They are also exposed to a range of hemodynamic forces reviewed in Krog and Henry,³ including fluid shear stress (FSS). Due to the dynamic nature of blood flow throughout the circulatory system, the levels of FSS a CTC may encounter vary greatly (from 1 to 6 dynes/cm² in veins to >1,000 dynes/cm² in the heart). Because most CTCs are larger than the average capillary diameter of ~5 μm, they readily become entrapped after entering the microcirculation and may die, extravasate, or be dislodged to circulate freely again until they encounter the next capillary bed.^4–6 Considering blood flow velocity, the periods of free flow between longer periods of entrapment may be on the order of seconds. Collectively, the death, destruction, and clearance of cells by extravasation or entrapment account for the relatively short (<2 h) half-life of CTCs.^7,8 Thus, although CTCs are exposed to a wide range of FSSs, this exposure is brief and discontinuous.

How hemodynamic forces, including FSS, impact cancer cells is a growing area of interest reviewed in Follain et al.⁹ It remains unclear whether prior findings on adaptive responses in adherent cancer cells and non-transformed cells to exposure to FSS hold true for CTCs as these cancer cells exist in suspension. Compounding this uncertainty is that significant challenges remain in studying the effects of FSS on cancer cells, including improving the accuracy of in vitro modeling of the spatiotemporal dynamics of circulatory flow and in vivo tracking of the fates of individual CTCs.³ Past studies of the effects of FSS on various cancer cell types used cancer cells in suspension to model CTCs and FSS of low levels (1–60 dynes/cm²) or long duration (15 min–10 days). Thus, these studies did not account for either the upper end of the physiological range of FSS or the fact that CTCs are unlikely to flow continuously for such long periods. Nevertheless, they indicated that FSS can induce cellular phenotypes associated with metastasis, including increases in cell survival, adhesion, migration, and invasion;^10–12 expression of markers of epithelial-mesenchymal transition (EMT) and stemness;^13,14 and tumorigenicity.¹⁴

An additional barrier to survival encountered by CTCs is oxidative stress resulting from cell detachment. Oncogenic signaling can promote the survival of detached epithelial cells by enhancing glucose uptake and flux through the pentose phosphate pathway to generate NADPH that combats detachment-induced oxidative stress.¹⁵ Melanoma cells exhibit dependence on the folate pathway to generate NADPH and promote distant metastasis.¹⁶ Indeed, recent work has revealed a variety of metabolic adaptations that CTCs may engage to survive reactive-oxygen species (ROS) toxicity, reviewed in Merteroglu and Santoro.¹⁷ The relationship between mechanical and oxidative stress in CTCs is less clear. Exposure of MDA-MB-231 breast cancer cells to FSS (5–30 dynes/cm²) in a continuous flow closed loop model for up to 6 h increased ROS in a manner that was linked to enhanced cell migration. Thus, ROS may either promote or prevent metastatic behaviors depending on the cellular and or microenvironmental/mechanical context.

In our previous studies, we found that cancer cell lines, including PC-3 prostate cancer cells and MDA-MB-231 breast cancer cells, can resist mechanical destruction from brief, high-level exposure to FSS through engagement of a RhoA-dependent mechanoadaptive process resulting in cortical actin remodeling and actomyosin contractility that protects cells from mechanical damage.¹⁸ Brief, high-level FSS exposure rapidly activates RhoA, consistent with its activation by a variety of mechanical stimuli. Because RhoA is involved in other cellular phenotypes that are associated with metastasis, such as invasion and survival in anchorage-independent conditions, we hypothesized that exposure to FSS in our model may not only protect cancer cells from destruction by FSS but may also simultaneously promote metastatic behaviors in those surviving cells. Here, we sought to determine the effects of brief exposure to higher levels of FSS that CTCs might encounter in circulation on metastatic behavior.

RESULTS

Exposure to FSS enhances metastatic phenotypes in vitro and in vivo

We exposed several cancer cell lines (PC-3, human prostate cancer, Myc-CaP, mouse prostate cancer, MDA-MB-231 human breast cancer) to millisecond pulses of high-level (937–1234 dynes/cm²) FSS in cell suspension over ~10 min^19,20 and evaluated cellular phenotypes associated with metastasis as compared to cell suspensions held under static conditions for a similar duration. Exposure to FSS increased invasion in a Boyden chamber assay (Figures 1A–1C) and increased cell proliferation under anchorage-independent conditions (Figures 1D–1F) for all 3 cell lines. Since we previously demonstrated that in PC-3 cells, exposure to FSS activates RhoA and RhoC but not Rac1,¹ we determined whether the increased invasive potential was dependent on RhoA or RhoC. We showed that in PC-3 cells, the increased invasive potential conferred by FSS was dependent on RhoA, not RhoC (Figure S1A).

Figure 1. Exposure to FSS enhances metastatic potential in vitro and in vivo.

(A–F) Effects of FSS exposure on invasion through a type I collagen matrix by PC-3 (p = 0.0060) (A), MDA-MB-231 (p = 0.0350) (B), and Myc-CaP (p = 0.0303) (C) cells, 18 h post-FSS exposure (paired t test; mean ± SEM). Effects of FSS exposure on the number of PC-3 (p = 0.0343) (D), MDA-MB-231(p = 0.0363) (E), Myc-CaP (p = 0.0416) (F) cells at 24 h after seeding on poly-HEMA-coated wells, expressed as fold change relative to initial cell number (paired t test; mean ± SEM).

(G) Schematic representation of in vivo experimental design made with BioRender.

(H–K) Representative images of pairs of mice injected with either FSS-exposed or static cells, on the day of injection (day 0) and 56 days after injection (day 56). Day 0 PC-3 BLI mean values: static 5.4 × 10⁵ photon/s and FSS 6.5 × 10⁵ photon/s (unpaired t test, p value = 0.289). Day 0 MDA-MB-231 mean values: static 1.2 × 10⁵ photon/s and FSS 1.4 × 10⁵ photon/s (unpaired t test, p value = 0.534). Color scale at day 0 is the same for all images. Time to metastasis, with event time defined as BLI signal > 10⁷ for (I) PC-3 and (J) MDA-MB-231(log rank test, p = 0.0222 and p = 0.0495, respectively). Mice inoculated with (K) Myc-CaP tumor volume measured by calipers (median and 50%/80%/95% nonparametric confidence intervals are shown; repeated measures ANOVA p < 0.0001).

To test whether exposure to FSS of this same magnitude and duration can influence metastatic colonization or tumor growth, we exposed these cell lines to FSS and then injected equivalent numbers of viable, FSS- and static-exposed cells into the appropriate animal hosts within 45 min of FSS exposure and evaluated metastatic colonization using whole-body bioluminescence imaging (BLI) (Figures 1G and 1H). In both PC-3 and MDA-MB-231 cells injected into the lateral tail vein, exposure to FSS decreased the time necessary to detect metastasis using BLI (Figures 1I, 1J, S1C, and S1D). More rapid development of metastatic colonies following FSS exposure is consistent with more cells surviving their journey through the circulation, potentially because prior FSS exposure increased their resistance to destruction in the circulation (as we showed previously¹⁸) and/or enhanced their ability to generate metastatic colonies, as suggested above by the effects of FSS on cellular phenotypes. To address the former possibility, for PC-3 cells, we found that there was no significant difference in the fraction of cells destroyed immediately following tail vein injection as compared to static controls (Figures S1F and S1G). Prior studies have demonstrated an increase in tumorigenicity of MDA-MB-231 cells exposed to low levels of shear (20 dynes/cm²) for 10 days when implanted subcutaneously in non-obese diabetic severe combined immunodeficiency (NOD-SCID) mice.¹⁷ Therefore, we explored this phenotype in our model of FSS exposure. Myc-CaP cells exposed to FSS exhibited faster tumor growth in syngeneic, immunocompetent hosts compared to static controls (Figures 1K and S1E). Taken together, these findings indicate that exposure to brief pulses of high-level FSS enhances the metastatic phenotypes both in vitro and in vivo.

Exposure to FSS alters the transcriptome and induces cytokine expression and oxidative metabolism

To investigate the mechanisms underlying the effect of FSS exposure on enhanced metastatic behaviors in cancer cells, we exposed PC-3 cancer cells to the same protocol of FSS in the preceding studies and conducted RNA sequencing (RNA-seq) on FSS-exposed and static samples 3, 12, and 24 h post FSS exposure. During this time course, cells were held in poly-HEMA-coated dishes to simulate CTCs, which are not attached to the extracellular matrix (Figure 2A). Principal component analysis revealed that the largest determinant of variation in gene expression overall is the time cells are held in suspension, but FSS exposure contributes to variation at all time points (Figure S2A). Over this time course, progressively more genes become significantly regulated (Figures 2B–2D). Gene Set Enrichment Analysis (GSEA) of the top regulated pathways revealed a significant increase in NF-κB-driven gene expression at 3 h, which was not evident at later time points (Figure 2E). To validate these findings, we conducted an NF-κB reporter assay in PC-3 and MDA-MB-231 cells. This showed that while NF-κB activity increases under both static and FSS-exposed conditions from 0 to 6 h, consistent with a stress response in an attachment-deprived state,²¹ FSS exposure results in significantly higher NF-κB activity in both cell lines (Figures 2F and 2G). We next evaluated the effect of FSS exposure on the secretion of a subset of cytokines controlled by NF-κB by assessing cytokine levels in conditioned media of cells that were held in polyHEMA-coated dishes 24 h after FSS exposure. In PC-3 cells, we found that FSS exposure significantly increased IL-6 and IL-11, and in MDA-MB-231 cells, we found increased IL-6 24 h post-exposure to FSS (Figures 2H and 2I).

Figure 2. Exposure to FSS alters the transcriptome, resulting in changes in cytokine expression and oxidative respiration.

(A) Schematic representation of the experiment using BioRender.

(B and D) Volcano plots demonstrating up- or downregulated gene transcripts across three time points post FSS exposure: 3, 12, or 24 h compared to static controls.

(E and J) GSEA across 3 and 24 h time points of significantly regulated pathways post FSS exposure compared to static controls.

(F and G) NF-κB reporter assays (0–6 h post FSS exposure for PC-3 (p = 0.0013) (F) and MDA-MB-231 cells (p < 0.001) (G).

(H and I) Cytokine secretion 24 h post FSS exposure for PC-3 (VEGF, p = 0.0892; IL-6, p <0.0001; IL-11, p < 0.0001) (H) and MDA-MB-231 cells (IL-6, p < 0.001; TFG-ɑ, p = 0.0725 (I).

(K–M) Oxygen consumption rates (OCR) at basal, ATP-linked (0.0025 μM oligomycin), and maximum respiration (0.0025 μM FCCP) for PC-3 (Basal, p = 0.00188; Oligo, p = 0.1511; FCCP, p < 0.0001) (K), MDA-MB-231 (Basal, p = 0.0061; Oligo, p = 0.2007; FCCP, p = 0.0007) (L), and Myc-CaP (Basal, p = 0.0306; Oligo, p = 0.0160; FCCP, p = 0.0377) (M) (2-way ANOVA; n = 4, mean ± SEM).

At 12 and 24 h post-FSS exposure, we observed a gene expression pattern consistent with a proliferative response, including the upregulation of MYC, E2F targets, and other cell cycle-related genes (Figure 2J), consistent with the FSS-induced anchorage-independent cell proliferation shown above (Figures 1D–1F). By 24 h, there was a notable enrichment in transcripts associated with oxidative metabolism (Figure S2B). To assess the effects of FSS on oxidative metabolism in cells in suspension and to reproduce the state of CTCs, we measured oxygen consumption rate (OCR) in 24 h post-FSS exposure (Figures 2K–2M). We observed a significant increase in basal respiration for all three cell lines. Treatment of cells with oligomycin revealed that there was no difference in ATP-linked respiration or proton leak, while treatment with FCCP showed that FSS exposure increased maximal respiration in all 3 cell lines. Thus, within 24 h of exposure to FSS, cancer cells respond by increasing oxidative metabolism. We validated the transcript levels of AOX1 and COX2 associated with oxidative phosphorylation from the RNA-seq screen and observed an increase in PC-3 and Myc-CaP cell lines 24 h post-FSS exposure. However, this increase was not observed in MDA-MD-231 cells, suggesting alternative mechanisms utilized to enhance oxygen respiration (Figure S2C). To test the functional role of oxidative metabolism during FSS response, we pretreated PC-3, MDA-MB-231, and Myc-CaP cells with Antimycin A to inhibit respiration prior to FSS exposure (Figures S2D–S2F). This resulted in a significant reduction in cell viability during exposure to FSS, without inducing cytotoxicity in static conditions (Figures S2G–S2I), indicating that oxidative metabolism supports cancer cell survival during exposure to FSS.

Exposure to FSS rapidly alters the metabolome and promotes glycolysis

Since numerous studies indicate that alterations in cellular metabolism underlie productive metastasis,^16,22–24 and our analysis of FSS-induced changes in the transcriptome revealed that genes involved in oxidative metabolism were highly regulated, we performed metabolomic profiling of PC-3 and MDA-MB-231 cells immediately after exposure to FSS. Following the 10^th pulse of FSS, or under static conditions held for the same duration, cells were immediately plunged into liquid N₂ to preserve the FSS-responsive and control metabolomes. Numerous metabolites were altered in both cell lines (Figures 3A and 3B). We noted that the energy charge, as evidenced by AMP/ATP and GMP/GTP ratios, trended downward in PC-3 and was significantly lower in MDA-MB-231 cells (Figures 3C and 3D). This finding is consistent with the fact that FSS exposure rapidly induces mechano-adaptation involving actin remodeling and actomyosin contractility, both energetically costly functions.²⁵

Figure 3. FSS exposure rapidly alters the metabolome, resulting in changes in energy utilization and glycolysis.

(A and B) Volcano plots from metabolomic profiling in (A) PC-3 and (B) MDA-MB-231 cells exposed to FSS (n = 4; paired t test without correction).

(C and D) Relative abundance of energy-related metabolites in FSS-exposed and static PC-3 (AMP, p = 0.4378; ATP p = 0.7392; GMP, p = 0.3262; GTP, p = 0.7942) (C) and MDA-MB-231 (AMP, p = 0.0079; ATP p = 0.0001; GMP, p = 0.0206; GTP, p < 0.0001) (D) cells (n = 4; paired 2-way ANOVA with Bonferroni correction; n = 4, mean ± SD).

(E) Effect of FSS exposure on glucose uptake in cell lines (PC-3, p = 0.1048; MDA-MB-231, p = 0.8265; Myc-CaP, p = 0.8289).

(F) Effect of FSS exposure on lactate production in cell lines (PC-3, p = 0.0422; MDA-MB-231, p = 0.0352; Myc-CaP, p = 0.8406).

(G) Effect of 2-deoxyglucose (2-DG, 25 mM, 2 h) treatment on viability of PC-3 and MDA-MB-231 cells exposed to FSS (3-way ANOVA; 2-DG p = 0.0152, interaction 2-DG × FSS, p = 0.0003).

To determine how cell metabolism might adapt to increased energy expenditure following FSS, we focused on the relative abundance of the metabolites of glycolysis and the tricarboxylic acid (TCA) cycle. In PC-3 cells, there were elevated levels of the glycolytic metabolite, dihydroxyacetone phosphate, and reduced levels of early TCA metabolites, citrate, isocitrate, and α-ketoglutarate, suggesting that FSS exposure might promote glycolytic metabolism in these cells (Figures S3A and S3B). However, the differences in glycolytic or TCA metabolites were not observed in MDA-MB-231 cells (Figures S3C and S3D). We next independently measured glucose uptake and lactate production immediately following exposure to FSS in PC-3, MDA-MB-231, and MyC-CAP cells to test if FSS results in changes in glycolysis. There were no significant differences in glucose uptake for all cells tested (Figure 3E); however, there was elevated lactate production in FSS-exposed PC-3 and MDA-MB-231 (Figure 3F). The discrepancy between this result and the lack of increase in lactate observed in the metabolomic profile may stem from a difference in time elapsed after FSS exposure, as the metabolome was assayed immediately after FSS exposure. We then evaluated whether treatment of PC-3 and MDA-MB-231 cells with 2-deoxyglucose (2-DG) for a dose and duration that did not result in reduced clonogenic potential (Figure S3E) would influence the ability of these cells to survive FSS exposure. We found that 2 h treatment with 2-DG sensitized both cell lines to destruction by FSS (Figure 3G).

Exposure to FSS rapidly engages folate metabolism and reduces oxidative stress

In addition to evaluating the effects of FSS exposure on energetic metabolites, we also noted that a subset of amino acids associated with the one-carbon pathway and folate metabolism, serine, glycine, methionine, and threonine, were consistently reduced in both cell lines immediately after exposure to FSS as compared to static conditions (Figures 4A and 4B).²⁶ The folate pathway has previously been demonstrated to promote the survival of melanoma cells in circulation during metastasis. The mechanism highlighted by these authors was that the folate pathway can generate the reductant NADPH, which acts via glutathione to combat ROS. To test this hypothesis, we first evaluated whether FSS exposure alters ROS levels and found that FSS exposure significantly reduced ROS levels in both PC-3 and MDA-MB-231 cells 3 h and 6 h post-exposure to FSS when held in suspension, indicating that FSS exposure leads to decreased levels of ROS in response to cell detachment (Figure 4C). We also evaluated whether exposure to FSS can alter lipid peroxidation, as RhoA has been implicated in both suppressing and promoting ferroptosis,^27,28 and we have demonstrated previously that FSS exposure activates RhoA.¹⁸ Using PC-3 cells with RhoA knockdown, we found that FSS exposure reduced levels of lipid peroxidation in an RhoA-dependent manner (Figure S5A). Surprisingly, we did not observe any significant differences in redox couples (NAD⁺/NADH; NADP⁺/NADPH; GSH/GSSG) in either PC-3 or MDA-MB-231 cells immediately, 3 h, or 6 h post FSS exposure (Figures S4A–S4H).

Figure 4. Exposure to FSS engages folate metabolism to reduce ROS and promote metastatic colonization.

(A and B) Fold change in relative abundance in metabolites that can feed the folate and 1-carbon cycle in PC-3 (Serine, p = 0.0012; Glycine, p = 0.0223; Methionine, p = 0.0009; Threonine, p = 0.0051) (A) and MDA-MB-231 (Serine, p < 0.0001; Glycine, p < 0.0001; Methionine, p < 0.0001; Threonine, p < 0.0001)

(B) cell lines (n = 4; paired 2-way ANOVA with Bonferroni correction; n = 4 mean ± SD).

(C) The effect of FSS exposure (open) relative to static controls (filled) on dihydroethidium (DHE) oxidation in MDA-MB-231 (magenta) and PC-3 (blue) 3 and 6 h post-FSS exposure (paired two-way ANOVA, Bonferroni correction; n = 3, mean ± SD; 3 h, p = 0.0008; 6 h, p = 0.0004).

(D) Effect of MTX (1 μM, 4 h) on FSS resistance for PC-3 (solid line) and MDA-MB-231 (dashed line) cells (n = 3 for all conditions; mean ± SD).

(E) Plot of whole-body BLI signal for individual animals measuring metastatic colonization in vehicle (VHC)- and methotrexate (MTX)-treated animals and cells.

(F) Metastasis-free survival (time to 10⁷ whole body BLI signal) in VHC- and MTX-treated groups (log rank test, p < 0.0001).

While there was no evidence that FSS exposure influenced the ratios of redox couples, this does not rule out the possibility that FSS altered the flux through antioxidant pathways, as we observed that FSS exposure did reduce both cell and lipid ROS (Figures 4C and S5A). We then wanted to test whether perturbation of the folate pathway could alter FSS resistance. We did this by inhibiting the folate pathway with methotrexate (MTX) at an exposure that was not acutely toxic to adherent cells (Figure S5B) and observed that MTX pretreatment sensitized MDA-MB-231, but not PC-3 cells, to FSS exposure (Figure 4D). We note that MDA-MB-231 cells express higher levels of dihydrofolate reductase, the target of MTX (Figure S5C). We then evaluated the effects of pretreatment of MDA-MB-231 cells with MTX prior to injection on metastatic colonization in vivo. We found that MTX pretreatment increased metastasis-free survival (Figure S5D). To extend the MTX treatment window further, during a period in which cells are expected to be in circulation, we injected mice with MTX (10 mg/kg, intraperitoneal) just prior to injecting MTX-pretreated cells. This dose of MTX yields a C_max = 11 μM with t_1/2 = 36.8 min.²⁹ Thus, within 4 h, MTX concentration is in the nM range. This resulted in a greater delay in metastasis formation than pretreatment alone (Figures 4E and 4F). Importantly, MTX in non-toxic MDA-MB-231 cells up to 10 μM for a 24 h exposure (Figure S5B). Moreover, the single in vivo dose of MTX was non-toxic to the mice (Figure S5E). Collectively, these data indicate that exposure to FSS triggers an increase in folate pathway flux and in MDA-MB-231 cells, this protects the cells from rapid destruction by FSS in the short term. Thus, FSS exposure may trigger changes that protect CTCs from oxidative insults during metastasis.

DISCUSSION

The data presented here indicate that cancer cells exposed to brief pulses of high-level FSS rapidly alter their metabolism and gene expression profiles in a manner that promotes their fitness for metastatic colonization. Thus, not only do cancer cells have mechanisms that act to protect them from damage by FSS, but FSS can also be a stimulus for metastatic behavior. Indeed, FSS has potent biological effects on many cell types, most notably endothelial cells, which are constantly exposed to FSS.³⁰ Laminar flow models have been employed to determine the effects of FSS on adherent cancer cells in 2D culture,^31–35 generally at low-level FSS (0.05–5 dynes/cm²). However, the relevance of these findings to CTCs, which are detached, is not clear. Although prior studies indicated that prolonged (15 min–10 days) exposure to lower-level (1–60 dynes/cm²) FSS induced cellular phenotypes in vitro and suggested that this potentially promotes metastatic behavior,^14,36 only one previous study examined the effects of FSS exposure in vivo.³⁶ It tested the metastatic colonization potential of lung cancer cells that had been exposed to FSS in in-suspension (continuous flow loop with maximum wall shear stress of 35–918 dynes/cm²) for 72 h before intracardial injection. Control cells were cultured under standard two-dimensional culture conditions (i.e., attached), and the metastatic colonization by the FSS-exposed cells was higher. Limitations of that study include that CTCs are unlikely to circulate continuously over such a long period of time, and it did not account for the effects of depriving cells of attachment (which our gene expression data indicate are significant; Figure S2A). Comparison of cells exposed briefly to FSS to counterparts held in suspension shows that FSS enhances metastatic potential. Taken together, these data indicate that a broad range of FSS exposures to cancer cells in suspension have important biological effects on those cells, which may model the effects of hemodynamic forces on CTCs in circulation.

The results from our multi-omics study demonstrate that exposure to brief pulses of high-level FSS can enhance metastatic potential by inducing dynamic changes in both gene expression and metabolism that protect cancer cells from immediate destruction by FSS while promoting their survival and proliferation well after FSS exposure. An early change noted in the transcriptomic profile is the transient expression of an NF-κB gene signature at 3 h. This is accompanied by enhanced secretion of certain cytokines, including IL-6. Interestingly, Szczerba and colleagues showed that in heterotypic clusters of CTCs and neutrophils, the latter provided inflammatory cytokines, including IL-6, and acted to maintain proliferation of the CTCs.³⁷ Our data suggest that single CTCs exposed to brief pulses of high-level FSS may achieve similar stimulation in a cell-autonomous fashion that can promote metastatic colonization. Interestingly, our RNA-seq data also suggested that 12 and 24 h after FSS exposure, there is a reduction in response to type I and type II interferons (Figures 2F and 2G). This may result in reduced immunogenicity of CTCs and promote metastasis³⁸). Moreover, the Massagué lab demonstrated in lung cancer cells that inhibition of STING, a known promoter of the production of type I interferons and an interferon-regulated gene,³⁹ in cancer cells, leads to a reduction in cancer cell dormancy.⁴⁰ Collectively, our data suggest that FSS exposure results in multifaceted alterations in cytokine production/response that may act through varying mechanisms to promote proliferation of cancer cells and metastatic progression.

We found that exposure to FSS results in rapid alterations in cellular metabolism. Structural alterations in the actin cytoskeleton and actomyosin contractility, which we previously showed are key to mechanoadaptive FSS resistance in cancer cells, are energetically costly.^18,25 Thus, one would expect that cells exposed to FSS must adjust to this perturbation. We find that there are rapid changes in glycolytic metabolism, and lactate production is increased in both PC-3 and MDA-MB-231 cells. However, while both cell lines are sensitized to FSS exposure by blockade of glucose uptake with 2-DG prior to exposure, FSS exposure does not result in immediate enhanced glucose uptake. It is unclear what drives the early glycolytic activity induced by FSS, but it does not appear to be sustained as cells shift toward oxidative metabolism. Other studies that employed a continuous flow model with up to 4 h FSS exposure (5–20 dynes/cm²) found that glycolysis was increased in colorectal cancer cells under the control of the ATOH8 transcription factor.¹² Perhaps the severity/duration of FSS exposure determines the path of metabolic changes in CTCs. We noted a consistent depletion of amino acids associated with the folate pathway in both PC-3 and MDA-MB-231 cells immediately following exposure to FSS. Cell detachment provokes oxidative stress, and in various cell types, prolonged exposure to FSS while suspended leads to an increase in the levels of ROS.^10,11,15 During metastasis, the folate pathway protects cells from oxidative stress, and folate inhibitors can reduce both CTC numbers and metastatic burden.¹⁶ Our data demonstrate that exposure to FSS reduces ROS of PC-3 and MDA-MB-231 cells 3 and 6 h after exposure and rapidly reduces the abundance of amino acids that are methyl donors to the folate pathway, whereas suppressing this oxidative mechanism significantly reduces metastatic colonization and resistance to FSS. These data are in contrast with a previous study showing that continuous, long-duration (6 h) exposure to low levels of FSS results in elevated ROS; however, CTCs are unlikely to be continuously circulating for this period of time.^10,41 Surprisingly, we did not see changes in redox couples 3 and 6 h after FSS exposure, suggesting that production and utilization of reducing capacity are in balance.

At longer time points following exposure to FSS, we find that cells increase oxidative metabolism. Prior evidence supports a shift toward oxidative metabolism and reduced glycolysis in melanoma cells under the conditions of detachment-induced stress.⁴² Additionally, emerging data suggest that metastatic cancer cells upregulate oxidative phosphorylation and depend upon it in establishing secondary lesions or as CTCs.^43–45 Our data demonstrates that exposure to FSS promotes oxygen consumption, consistent with the upregulation of oxidative phosphorylation in metastatic lesions that others have observed. Suppressing oxidative respiration significantly sensitizes cells to FSS and overall decreases cellular viability. Herein, our results highlight potential routes of reprogramming the metabolome immediately upon FSS exposure, as well as the lasting effects FSS has on regulating the transcriptome and metabolic pathways, which in turn enhance metastatic potential (Figure S6). In summary, this study reveals new mechanisms by which cancer cells adapt to FSS exposure, and these might be targeted in therapies to reduce the productive metastatic colonization of CTCs.

Limitations of the study

Our in vitro model of FSS exposure has limitations. One is that the brief FSS exposure is at a high intensity that CTCs probably encounter only under certain circumstances, such as in turbulent flows around heart valves.^46,47 Thus, in vivo, only a small fraction of CTCs would be expected to be exposed to FSS at the same level as in our model. However, such unusual FSS exposure could have important effects on their metastatic potential. We have not yet evaluated the effect of FSS on metastatic colonization in an immunocompetent mouse model; thus, we cannot yet assess how exposure to FSS might influence anti-tumor immunity. We do not yet know whether all the transcriptomic and metabolic alterations described here depend solely on mechano-adaptation driven by the RhoA-actomyosin axis. RhoA is known to activate YAP/TAZ in response to mechanical stimuli.³⁵ Thus, a RhoA>YAP/TAZ axis may contribute to anchorage-independent survival and proliferation. We found here that RhoA was required for FSS-induced invasion in PC-3 cells, consistent with its role in cancer cell invasion (reviewed in Haga and Ridley⁴⁸). We have not yet established a role for RhoA in FSS-induced cytokine secretion or metabolic alterations. However, RhoA and other RhoGTPases are known to activate NF-κB,^49,50 and RhoA and NF-κB collaborate to regulate mitochondrial glutaminolysis in breast cancer cells.⁵¹ Differences in how FSS influences glycolytic and TCA metabolites as well as the effects of methotrexate on FSS resistance are apparent among the cancer cell lines evaluated in this study. We do not have a mechanistic understanding of these discrepancies at present, but this may reflect metabolic plasticity among these cell lines.

RESOURCE AVAILABILITY

Lead contact

Requests for further information, resources, and reagents should be directed to and will be fulfilled by the lead contact, Dr. Michael D. Henry (mhenry@odu.edu).

Materials availability

This study did not generate new, unique reagents.

Data and code availability

All datasets have been made available for public access through the National Metabolomics Data Repository (NMDR) under NMDR: PR002798; https://doi.org/10.21228/M80C2F. Processed RNA-seq data can be found on Gene Expression Omnibus under GEO: GSE302200 (release date January 28, 2026). All data can be requested through the lead contact.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell lines

PC-3 (male prostate cancer), MDA-MB-231 (female breast cancer), and Myc-CaP (male prostate cancer) cells were obtained from the American Type Culture Collection (ATCC). PC-3 cells were maintained in DMEM:F12 (11320–033; ThermoFisher Scientific) with 10% FBS (S11150; Atlanta Biologicals) and 1x NEAA (11140–050; ThermoFisher Scientific). Both MDA-MB-231 and Myc-CaP cells were maintained in DMEM (11965–092; ThermoFisher Scientific) with 10% FBS and 1x NEAA. All cells were modified to express firefly luciferase through retroviral transduction, as previously described.⁵² PC-3 cells were transduced with a control (SHC016; SigmaAldrich), RhoA (TRCN0000047711, TRCN0000047712; SigmaAldrich), or RhoC (TRCN00000291516, TRCN000047864; SigmaAldrich) short hairpin RNA (shRNA) using lentiviral particles, as described previously.^18,53 Cells were incubated at 37C at 5% CO₂. Parental PC-3 cell line was authenticated by STR analysis (RADIL). All cell lines were tested for mycoplasma contamination by PCR assay (University of Iowa).

Chemical reagents

2-deoxyglucose (2-DG) (D8375, Sigma-Aldrich) was dissolved in serum-free DMEM and cells were treated with 25mM 2-DG DMEM or DMEM 2 h prior to FSS exposure. Antimycin A (Cat #J63522.MA, Thermo Scientific) was dissolved in DMSO and cells were pre-treated with 10 μM or 0.1% DMSO for 3 h prior to FSS exposure. Methotrexate hydrate (MTX) (13960, Cayman Chemical) was dissolved in dimethylformamide (DMF), and cells were treated with 1 μM MTX or 1:10,000 DMF for 4 h prior to FSS exposure.

Exposure to fluid shear stress

Cells suspended in DMEM were exposed to 10 pulses of FSS by passage through a 30 gauge ½ inch needle (305106, BD) attached to a 5 mL syringe (309646, BD) at 250 μL/s using a syringe pump (70–3005, Harvard Apparatus), as described previously.^19,20 A modification from published protocols is that the concentration of cells used for the FSS exposure in these experiments was 2×10⁶. Additionally for RNA sequencing and cell proliferations experiments, cells were sheared and subsequently plated in DMEM supplemented with 1x insulin-transferrin-selenium (ITS) (41400–045; ThermoFisher Scientific) and 5 mg/mL of bovine serum albumin (BSA) (03117057001, Roche).

Measurement of cell viability

PC-3 (5,000) or MDA-MB-231 (10,000) cells were seeded in 96-well plates, with duplicates for each drug concentration tested. The day after seeding, the medium was changed to growth medium containing the stated concentration of MTX or DMF. Cells were treated for either 4 or 24 h and the medium was changed to growth medium without MTX. After 24 h, cell viability was measured using the resazurin (R7017, Sigma-Aldrich) dye.

Clonogenic assay

Static control samples were collected from the same wells as the treated samples and held in suspension while the remainder of the sample was exposed to FSS. For each condition, cells were plated in triplicate in a 6-well dish, at 500 cells/well. Cells were plated in growth medium and allowed to grow for 10 (PC-3) or 14 (MDA-MB-231) days before plates were stained with crystal violet solution and colonies (defined as >50 cells) were counted. Data from triplicate cultures were averaged and then normalized to the average of the biological replicates for the untreated, static control.

Model of metastatic colonization

All procedures were approved by the University of Iowa Animal Care and Use Committee (protocols 5121574 and 8111574). Animals were housed in a 12 h day to night cycles and housed 5 animals per cage when available.To determine whether FSS exposure affects metastatic colonization, we injected 1×10⁶ viable trypan-blue negative cells into the lateral tail veins of 8-week old NOD-Prkdc^em26Cd52 Il2rg^em26Cd22/NjuCrl mice (572, Charles River) within 45 min of FSS exposure. Because of the need to minimize the time between FSS exposure and injection into the tail vein, only 2–5 mice were injected in a round. Each was injected with either sheared or static control cells initially collected from 2 15-cm tissue culture dishes. The cycle of FSS exposure followed by injection was repeated until 12 mice representing each condition (sheared and static) had been injected. To adequately power the experiment (α = 0.05 and β = 0.20), 12 mice per condition were used, and these were from 3 independent rounds of FSS exposure. Power analysis was based on a pilot study performed in NOD-Cg-Prkdc^scid Il2rg^tm1Wjl/Szj (NSG) mice, where the mean time to metastasis (whole-body bioluminescence imaging signal (see below) > 10⁷ photons/second) was 71 ± 20 and 94 days for the FSS-exposed and static groups, respectively (Figures 1; S1). PC-3 cells were injected into male mice, whereas MDA-MB-231 cells were injected into female mice, and littermates were randomly assigned to the experimental groups. Sex was chosen to match the origin of the cells used in the tumor models. Animals were pretreated with MTX 10 mg/kg i.p. 60 min before inoculated with pretreated MDA-MB-231 cells.

Bioluminescence imaging (BLI)

After mice were injected with cancer cells, tumor burden was monitored by weekly bioluminescence imaging (BLI). For BLI, we performed intraperitoneal injection of 150 mg/kg of D-luciferin (LUCK, GoldBio), followed 5 min later by the detection of bioluminescence using an AMI HTX (Spectral Instruments Imaging); exposure time was 5 min. The AMIView Imaging Software was used to select an ROI that included the whole body, for quantification of signal intensity. The threshold for classification as metastatic disease was a whole-body BLI signal of ≥10⁷ photons/second.

Subcutaneous tumors

All procedures were approved by the University of Iowa Animal Care and Use Committee (protocols 5121574 and 8111574). To determine whether FSS exposure affects tumor initiation colonization, we injected 5×10⁴ viable trypan-blue negative cells into the lateral tail veins of 8-week old FVB-NJ mice (001800, Jackson Laboratory) within 45 min of FSS exposure. Cells in PBS suspension were mixed with a 1:1 ratio of Matrigel and injected on the bilateral flanks subcutaneously. Tumors were measured using caliber twice weekly. Because of the need to minimize the time between FSS exposure and injection into the tail vein, only 2–5 mice were injected in a round. Each was injected with either sheared or static control cells initially collected from 2 15-cm tissue culture dishes. The cycle of FSS exposure followed by injection was repeated until 15 mice representing each condition (sheared and static) had been injected. To adequately power the experiment (α = 0.05 and β = 0.20), 15 mice per condition were used, and these were from 3 independent rounds of FSS exposure.

METHOD DETAILS

Invasion assay

Invasion assays were performed using 8-μm, 24-well transwell inserts (3428, Corning) coated with 40μL of 0.8mg/mL rat-tail collagen type I (354236, Corning). The collagen matrix was allowed to polymerize for 30 min at 37°C prior to adding the cells. 100 μL of 1×10⁶ cells/mL in DMEM was plated onto the matrix and 600 μL DMEM with 20% FBS was used as the chemoattractant in the lower chamber. Invasion was measured 18 h after seeding by adding D-luciferin to the media in the bottom of the transwell and the transwell insert, for a final concentration of 3 mg/mL and measuring the BLI signal. The medium was then removed from the top of the transwell insert and the non-migratory cells were scraped off before the BLI signal was measured a second time. The ratio of the invaded to total BLI signal was then used to determine the percentage of invading cells. BLI signal was measured for 2 min starting 5 min after luciferin was added at which point photon emission is stabilized. Duplicate samples were evaluated, and the data are represented as the average of the duplicates.

Anchorage-independent cell proliferation

For FSS-exposed and static conditions, trypan-blue negative cells (5×10⁵ cells/mL) were plated on 24-well dishes coated with poly-2-hydroxyethyl methacrylate (polyHEMA; P3932, Sigma-Aldrich) to prevent cell attachment, in DMEM with 1x ITS and 5 mg/mL of BSA. The concentration of these cells was again measured 24 h after FSS exposure and normalized to the original to determine the fold change in growth.

In vivo destruction assay

Prior to injection the PC-3 cells were either held in suspension or exposed to 10 pulses of FSS.¹⁸ We measured the fraction of cells destroyed immediately after lateral tail-vein injection as previously described.¹⁸ Briefly, we injected 5×10⁵ luciferase expressing PC-3 cells in 200 μL of PBS from either condition into NCI BALB/cAnNcr (555, Charles River) mice and within 3 min collected 500 μL of blood via cardiac draw. The blood was then centrifuged at 1,500 ×g for 5 min to separate the plasma from the cellular components. From the blood plasma, 100 μL aliquots were transferred into 96-well black bottom plate before adding 100 μL of assay buffer (200 mM Tris-HCl pH = 7.8, 10mM MgCl₂, 0.5 mM CoA, 0.3 mM ATP, and 0.3 mg/mL luciferin). A standard curve was generated by lysing 5×10⁵ luciferase expressing PC-3 cells in 200 μL of 1% Tween 20 in ddH₂O and then adding 13.3, 6.7, 3.3, 1.3, or 0 μL of lysed cells to 500 μL of blood collected from mice that had not been injected. The spiked blood was then processed the same as the blood from injected mice. The number of cells destroyed was then determined by linear regression.

Western blotting

40 μg of protein was loaded onto 12% SDS-polyacrylamide gels and transferred to PVDF membranes. Membranes were blocked for 1 h using blocking buffer (927–90001, LI-COR) before being probed with primary antibodies (RhoA, ARH04, Cytoskeleton, INC; RhoC, D40E4, Cell Signaling Technology; αTubulin, 12G10, Developmental Studies Hybridoma Bank) overnight. Membranes were washed 3x before being probed with secondary antibody (IRDye 800 Goat α rabbit, 926–32211, LI-COR; IRDye 680 Goat α mouse, 926–68070, LI-COR) and signal was assessed using Odyssey (LI-COR) system.

Metabolomic profiling

Sample preparation-

Metabolomic profiling was performed on cells that were processed and snap frozen (using liquid nitrogen) immediately after they were exposed to 10 pulses of FSS or held in suspension as described above. Specifically, on the final FSS pulse, the cells were transferred into 10 mL of ice-cold PBS, the number of viable cells was determined, and the cells were centrifuged at 500g for 3 min at 4°C. Cell debris was removed from pellets by resuspension in ice-cold PBS, 2×10⁶ cells were transferred into a new tube, the cells were pelleted again, and these washed cell pellets were snap frozen in liquid nitrogen. LC and GC-MS datasets are made available through National Metabolomics Data Repository (NMDR) under project PR002798; https://doi.org/10.21228/M80C2F.

Metabolite Extraction-

Frozen cell pellets were extracted with 1 mL of ice-cold 2:2:1 methanol/acetonitrile/water containing a mixture of 9 internal standards (d₄-Citric Acid, ¹³C₅-Glutamine, ¹³C₅-Glutamic Acid, ¹³C₆-Lysine, ¹³C₅-Methionine, ¹³C₃-Serine, d₄-Succinic Acid, ¹³C₁₁-Tryptophan, d₈-Valine; Cambridge Isotope Laboratories) at a concentration of 1 μg/mL each. Samples were then frozen in liquid nitrogen followed by a 10-min sonication and incubation while gently rotating at −20°C for 1 h. After incubation, samples were centrifuged for 10-min at maximum speed (15,000g). Supernatants were then transferred to fresh 1.5 mL microcentrifuge tubes. A pooled quality control (QC) sample was prepared by adding an equal volume of each sample to a fresh 1.5 mL microcentrifuge tube. Processing blanks were utilized by adding extraction solvent to microcentrifuge tubes. Then, a 300μL aliquot of all samples, pooled QCs, and processing blanks were evaporated in 1.5 mL microcentrifuge tubes using a speed-vac at room temperature.

GC-MS Method-

The resulting dried extracts were derivatized using methyoxyamine hydrochloride (MOX) and N,O-Bis(trimethylsilyl)trifluoroacetamide (TMS) [both purchased from Sigma]. Briefly, dried extracts were reconstituted in 30 μL of 11.4 mg/mL MOC in anhydrous pyridine (VWR), vortexed for 10 min, and heated for 1 h at 60°C. Next, 20 μL TMS was added to each sample, and samples were vortexed for 1 min before heating for 30 min at 60°C. The derivatized samples, blanks and pooled QCs were then immediately analyzed using GC/MS. GC chromatographic separation was conducted on a Thermo Trace 1300 GC with a TraceGold TG-5SilMS column (0.25 μm film thickness; 0.25mm ID; 30 m length). The injection volume of 1 μL was used for all samples, blanks, and QCs. The GC was operated in split mode with the following settings: 20:1 split ratio; split flow: 24 μL/min, purge flow: 5 mL/min, Carrier mode: Constant Flow, Carrier flow rate: 1.2 mL/min). The GC inlet temperature was 250°C. The GC oven temperature gradient was as follows: 80°C for 3 min, ramped at 20°C/min to a maximum temperature of 280°C, which was held for 8 min. The injection syringe was washed 3 times with pyridine between each sample. Metabolites were detected using a Thermo ISQ single quadrupole mass spectrometer. Data were acquired from 3.90 to 21.00 min in EI mode (70eV) by single ion monitoring (SIM). Using TraceFinder 4.1 or 5.1, metabolites were identified in extracted experimental, blank, and pooled QC samples, by matching spectra to an in-house generated library of target and confirming peaks and retention times and then relatively quantified by peak integration. After TraceFinder analysis, instrument drift over the run was corrected by local regression analysis using the NOREVA tool as described by Li et al.⁵⁴ After correction by NOREVA, metabolite values were log2 transformed and then within each run normalized to the average value of the static experimental group on a per metabolite basis.

LC-MS Method-

Dried extracts were reconstituted in 30 μL acetonitrile/water (1:1 v/v), vortexed well, and rotated on a rotator in a −20C freezer overnight. In the morning, the resuspended samples were vortexed and centrifuged for 10 min at maximum speed, and the supernatant was transferred to LC-MS autosampler vials for analysis. LC-MS data were acquired on a Thermo Q Exactive hybrid quadrupole Orbitrap mass spectrometer with a Vanquish Flex UHPLC system or Vanquish Horizon UHPLC system. The LC column used was a Millipore SeQuant ZIC-pHILIC (2.1 × 150 mm, 5 μm particle size) with a ZIC-pHILIC guard column (20 × 2.1 mm). The injection volume was 2 μL. Two mobile phases were used, solvent “A” containing 20 mM ammonium carbonate [(NH₄)2CO₃] and 0.1% Ammonium Hydroxide (v/v) [NH₄OH], pH is ~9.1, and solvent “B” containing 100% Acetonitrile. The method was run at a flow rate of 0.150 mL/min. The gradient started at 80% B and decreased to 20% B over 20 min before returning to 80% B in 0.5 min and being held there for 7 min.⁵⁵ There is a 2-min pre-equilibration time prior to sample injection and during re-equilibration, from 24.5 to 26.5 min, flow is increased to 0.3 mL/min. The pooled QC samples were injected at the beginning, end, and every ~8 samples throughout the run. The mass spectrometer was operated in full-scan, polarity-switching mode from 1 to 20 min, with the spray voltage set to 3.0 kV, the heated capillary held at 275°C, and the HESI probe held at 350°C. The sheath gas flow was set to 40 units, the auxiliary gas flow was set to 15 units, and the sweep gas flow was set to 1 unit. MS data acquisition was performed in a range of m/z 70–1,000, with the resolution set at 70,000, the AGC target at 1 × 106, and the maximum injection time at 200 ms.⁵⁵ Using TraceFinder 4.1 or 4 5.1, metabolites were identified in experimental, blank, and pooled QC samples, by matching spectra to an in-house generated library of high mass accuracy target peaks x retention times and then relatively quantified by peak integration. After TraceFinder analysis, instrument drift over the run was corrected by local regression analysis using the NOREVA tool as described by Li et al.⁵⁴ After correction by NOREVA, metabolite values were log2 transformed and then within each run normalized to the average value of the static experimental group on a per metabolite basis.

RNA isolation

RNA was isolated from PC-3 cells that had been exposed to 10 pulses of FSS or held in suspension and then plated at 1×10⁶ cells/mL in 6-well plates that had been coated with polyHEMA to prevent cell attachment. For RNA extraction, the cytosolic contents were isolated by lysis in RLN buffer (50mM Tris-HCl pH = 8, 140m NaCl, 1.5mM MgCl₂, 0.5% NP-40) for 5 min, follwed by centrifugation at 12,000g for 5 min. Trizol was then added to the supernatant and the Direct-zol RNA Purification Kit (R2050, ZYMO Research) was used according to the manufacturer’s guidelines.

RNA sequencing

Samples were collected at 3, 12, and 24 h after FSS exposure. RNA sequencing and library preparation were performed by Novogene Co. LTD. Paired-end 150-bp sequences were generated using the Illumina platform. Sequences were aligned, and fragments counted, using the STAR aligner.⁵⁶ Differential gene expression analysis of the dataset was performed using DEseq2⁵⁷ in Rstudio. The differential gene expression output was used to make a pre-rank list (log2 fold change * −log10 pvalue) of genes. Gene set enrichment analysis was then done on that pre-ranked gene list using fast gene set enrichment analysis (ref: https://www.biorxiv.org/content/10.1101/060012v3). Processed RNA-seq data can be found on Gene Expression Omnibus under GEO accession: GSE302200.

Glutathione measurement

PC-3 and MDA-MB-231 cells were exposed to FSS or held in suspension as outline above and 400μL samples of static and sheared conditions were taken 3 and 6 h after FSS exposure. Cells were pelleted by centrifuging at 500g for 3 min and the supernatant was discarded before the cell pellets were flash frozen on liquid nitrogen. To measure glutathione levels, samples were first deproteinized by adding 3 volumes of 5% 5-Sulfosalicylic Acid Solution to cell pellets, frozen and thawed twice and incubated on ice for 5 min. Samples were then centrifuged for 10 min at 10,000 ×g to remove the precipitated protein. Supernatant was collected in a new tube and used to determine total glutathione content as described previously.⁵⁸ The measurement of GSH uses a kinetic assay in which catalytic amounts of GSH cause a continuous reduction of 5,5-dithiobis (2-nitrobenzoic acid) (DTNB) to TNB and the GSSG formed is recycled by glutathione reductase and NADPH so the GSSG present will also react to give a positive value in this reaction. The yellow color of 5-thio-2-nitrobenzoic acid (TNB) generated is measured spectrophotometrically at 412 nm. The rate at which color accumulates is proportional to the amount of total glutathione. Reduced and oxidized glutathione were distinguished by the addition of 10μL 2-VP mixed 1:1 (v/v) with ethanol to 50μL of sample. The assay uses a standard curve of reduced glutathione to determine the amount of glutathione in the biological sample. Glutathione levels were normalized to the protein content.

NADP(H) measurement

PC-3 and MDA-MB-231 cells were exposed to FSS or held in suspension as outline above and 400μL samples of static and sheared conditions were taken 3 and 6 h after FSS exposure. Cells were pelleted by centrifuging at 500g for 3 min and the supernatant was discarded before the cell pellets were flash frozen on liquid nitrogen. Cell pellets were extracted using 200μL of 1% (DTAB) (D8638, Sigma) in 0.2N NaOH before adding 200μL of PBS. The measurement from the extracted cells was then performed using the NADPH/NADP Glo-Assay (Promega; Cat#G9081) following manufactures instructions. Luminescence was measured using AMI X imager with Aura software (Spectral Instruments). The concentration of NADP and NADPH were then converted to the total amount and normalized to protein concentration.

Reactive oxygen species (ROS) measurement

For both PC-3 and MDA-MB-231 cell lines, cells were subjected to either 10 pulses of FSS at 2×10⁶ cells/mL or held in suspension in phenol red free RPMI with 1x ITS and 5mg/mL of BSA media. Cells were then plated on polyHEMA coated 24-well dishes at 5×10⁵ cells/mL in the phenol red free RPMI with ITS and BSA. ROS were measured by incubating 1mL of cells with 10μM of dihydroethidium (DHE; Thermo Fischer; Cat#D11347) for 30 min at 37C. ROS levels were measured immediately after FSS exposure, as well as 3 and 6 h later by flow cytometry using the data from 488nm excitation laser and comparing the ratio of the geometric mean to the initial static control.

Lipid peroxide (LPO) measurement

The cells were subjected to either 10 pulses of FSS at 2 × 10⁶ cells/mL or held in suspension in phenol red free RPMI with 1x ITS and 5mg/mL of BSA media. Cells were then immediately stained with Bodipy-C11 (5 μM, 20 min incubation at 37C) at 5 × 10⁵ cells/mL in the phenol red free RPMI with ITS and BSA (Bodipy-C11; Thermo Fischer; Cat# D3861). Ratiometric analysis was conducted on the emission peak shift from unoxidized red (590 nm) to oxidized green (510 nm) fluorescence. LPO ratios were measured immediately after FSS exposure and comparing the ratio of the geometric means of the ratios to the initial static control.

Real-time qPCR

cDNA was prepared from 1μg of extracted RNA using the iSCRIPT cDNA synthesis kit (1708891, Bio-RAD) and following the manufacturer’s guidelines. Real-time PCR was performed using the Power-Up SYBR reagent (A25776, Applied Biosystem), on duplicate samples, using the following primers: AOX1 Human (Forward: GGCCACAGTGATGTTGTAATG; Reverse: ATC CTC TAA GCC CAC AGA AAG), AOX1 murine (Forward: CCGACCCAA GAGCTGATA TTT; Reverse: CACGGTATGTCTGTGTCTTCTC), COX2 Human (Forward: CCGAGGTGTATGTATGAGTGT G; Reverse: CAGGAGGAAGGGCTCTAGTATAA), COX2 murine (Forward: GTCATTGGTGGAGAGGTGTATC; Reverse: CAGGAGGATGGAGTTGTTGTAG). Average values for the triplicates were then used to determine the ΔC_t, by comparing the C_t for AOX1 and COX2 to that for TATA Box binding-protein within the sample. The ΔC_t for each gene was then compared to the average ΔC_t of the control (either the static sample at 24 h timepoint, or the non-FSS-treated sample for the time point) to obtain the ΔΔC_t, which is same as log₂[fold change in expression]. The corresponding statistics were performed on the ΔΔC_t values.

Lactate production

The rate of lactate production was measured using Lactate-Glo assay kit (Promega, Cat#J5021). Our cancer cell lines were exposed to FSS or held in static conditions, washed with PBS, then immediately seeded on a 96-well plate in DMEM media. Lactate production was stopped using 5 μL of inactivation solution, provided by the kit, and placed on a plate-shaker for 3 min, followed by 5 μL of neutralization solution then on the plate-shaker for 1 min 50 μL of lactate detection reagent was added then incubated for an 1 h at room temperature then luminescence was recorded. Lactate standard was diluted from 10 mM stock with a range from 0.1 mM to 0.00625 mM. The amount of lactate in each well was calculated from the standard curve. The rate of lactate production was calculated as:

$$\text{Rate}\text{of}\text{lactate}\text{production}=(\text{lactate}\text{in}\text{cell}\text{culture})/((\text{cell}\text{number}{\text{well}}^{-1})\times 24\times 60\times 60)).$$

Glucose uptake

The rate of glucose uptake was measured using Glucose-Glo assay kit (Promega, Cat#J6021). Our cancer cell lines were exposed to FSS or held in static conditions, washed with PBS, then immediately seeded on a 96-well plate in PBS. 50 μL of 2DG at 1mM was added to the samples then placed on a plate-shaker for 10 min 25 μL of stopping buffer, provided by the kit, was added, followed by a quick shake, followed by 25 μL neutralization buffer added. 100μL of 2DG6P Detection Reagent was added then incubated for 1 h at room temperature. Luminescence was recorded. Glucose standard was diluted from 10 mM stock with a range starting at 0.1 mM–0.00625 mM. The rate of glucose uptake was calculated as:

$$\text{Rate}\text{of}\text{glucose}\text{consumption}=(\text{glucose}\text{in}\text{medium}\text{control-glucose}\text{in}\text{cell}\text{culture})/((\text{cell}\text{number}{\text{well}}^{-1})\times 24\times 60\times 60)).$$

Oxygen consumption rate (OCR) analysis

We exposed cell lines to FSS or held them in suspension and seeded them on polyHEMA coated plates for 24 h before measuring OCR. We used Oroboros Instruments Oxygraph-2k to measure basal, ATP-linked/proton leak, and maximal respiration at 7.5×10⁵ cells per replicate. ATP-link respiration/proton leak was induced by 0.0025 μM oligomycin and maximal respiration using 0.0025 μM FFCP. All measurements were recorded for 10 min intervals then averaged and reported in units of pmol min⁻¹ cell number⁻¹.

NF-κB activation assay

PC-3 and MDA-MB-231 cells were transduced with lentiviral particles containing pHAGE NFκB-TA-LUC-UBC-GFP-W vector (Cat#49343, AddGene, a gift from Darrell Kotton⁵⁹) and flow sorted for GFP expression. To evaluate NF-κB activation, cells were suspended in DMEM with 1x ITS and 5mg/mL BSA at 1 × 10⁶ cells/mL after exposing cells to FSS or hold them in suspension and 100μL was seeded in polyHEMA coated 96-well black bottom plate. Luciferase activity was assessed immediately and 6 and 24 h after seeding. Luciferase activity was normalized to the static 0 h sample.

Cytokine measurement

PC-3 and MDA-MB-231 were held in suspension or exposed to FSS as outlined above and plated at 1 × 10⁶ cells/mL on polyHEMA coated 24-well. 24 h after FSS exposure, 500μL static and sheared samples were collected and centrifuged at 1500g for 15 min. The supernatant was collected and stored at −80C until the assay was performed. Cytokines were measured using a custom human 8-plex LEGENDplex panel (900001023, BioLegend). The assay was performed per manufactures instructions and flow cytometry on the beads was performed on a Cytek Aurora. Data analysis was performed using the LEGENDplex software from BioLegend. Data is presented as an average of duplicate reads for each biological replicate.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistics

Within a cell line, differences in invasion by, and the proliferation of FSS-exposed vs. static cells were analyzed for significance using the Students t test. For the invasion data from the RhoA and RhoC knockdown experiments, the metabolomics data, and real-time PCR, repeated measures two-way ANOVA with Bonferroni’s correction for multiple comparisons was used. Data for invasion are plotted as mean ± standard error of the mean; those shown in other graphs are plotted as mean ± standard deviation. Time to metastasis is presented as Kaplan-Meier curves and was analyzed using the log rank test. For the volcano plots in Figure 3 paired t-tests without correction were used as an initial screen to identify significantly regulated metabolites which were subsequently validated in the two-way ANOVA analysis above. The statistics described above were analyzed using GraphPad Prism 9. Principal component analysis was performed on metabolomics data using R v4.0.2. Differential expression analysis for RNAseq data within timepoints, as well as the effect size estimate for PCA analysis, were done using DEseq2 in R. For all experiments biological replicates n = 3, unless otherwise stated in the figure legend or respective method section.

Additional resources

All datasets have been made available for public access through National Metabolomics Data Repository (NMDR) under project PR002798; https://doi.org/10.21228/M80C2F and Processed RNA-seq data can be found on Gene Expression Omnibus under GEO accession: GSE302200. All data can be requested through the lead contact.

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116908.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-RhoA	Cytoskeleton	Cat#ARH04; RRID: AB_2728698
Rabbit monoclonal anti-RhoC	Cell Signaling Technology	Cat#3430; RRID: AB_2179246
Mouse monoclonal anti-α-tubulin	Developmental Studies Hybridoma Bank	Clone ID: 12G10 anti-alpha-tubulin; Antibody Registry ID: AB_1157911
Goat polyclonal anti-Rabbit IRDye 800 CW	LI-COR	Cat#926-32211
Goat anti-Mouse IgG2a IRDye 680LT	LI-COR	Cat#926-68051
Bacterial and virus strains
DH5 alpha	Life Technologies	18265–017
Chemicals, peptides, and recombinant proteins
2-deoxyglucose	Sigma-Aldrich	Cat#D8375; CAS#154-17-6
Antimycin A	Thermo Scientific Chemicals	Cat#J63522.MA; CAS#1397-94-0
Methotrexate	Caymen Chemical	Cat#13960; CAS#154-17-6
K+ D-Luciferin	GoldBio	LUCK-1G
Rat-Tail Collagen Type I	Corning	354236
Poly(2-hydroxyethylmethacrylat)	Sigma-Aldrich	P3932; CAS#25249-16-5
LI-COR Blocking Buffer	LI-COR	927–90001
Insulin-Transferrin-Selenium (ITS)	ThermoFisher Scientific	41400–045
Bovine Serum Albumin (BSA)	Roche	03117057001
Matrigel Matrix	Corning	Cat#CB-40234
Dihydroethidium (DHE)	ThermoFischer Scientific	Cat#D11347
BODIPY 581/591 C11	ThermoFischer Scientific	Cat# D3861
Critical commercial assays
Cell Titer Blue	Promega	G8080
Resazurin	Sigma-Aldrich	R7017
Direct-zol RNA Miniprep Kits	Zymo Research	R2050
NADP/NADPH-Glo™ Assay	Promega	Cat#G9081
iScript cDNA Synthesis Kit	Bio-Rad	Cat#1708891
PowerUp™ SYBR™ Green Master Mix	Applied Biosystems	Cat#A25776
Lactate-Glo Assay	Promega	Cat#J5021
Glucose-Glo Assay	Promega	Cat#J6021
8-plex LEGENDplex panel	BioLegend	900001023
Deposited data
Processed RNA-seq data	GEO	GSE302200
Total Ratiometric Raw metabolomic data	Metabolomics Workbench	PR002798 https://doi.org/10.21228/M80C2F
Experimental models: Cell lines
PC-3	ATCC	CRL-1435
MDA-MB-231	ATCC	HTB-26
Myc-CaP	ATCC	CRL-3255
Experimental models: Organisms/strains
NOD-Prkdcem26Cd52Il2rgem26Cd22/N\uCrl	Charles River	572
FVB/NJ	Jackson Lab	001800
NCI BALB/cAnNcr	Charles River	555
Oligonucleotides
AOX1 Human Forward primer: GGCCACAGTGATGTTGTAATG	This paper	N/A
AOX1 Human Reverse primer: ATC CTC TAA GCC CAC AGA AAG	This paper	N/A
AOX1 Murine Forward primer: CCGACCCAA GAGCTGATA TTT	This paper	N/A
AOX1 Murine Reverse primer: CACGGTATGTCTGTGTCTTCTC	This paper	N/A
COX2 Human Forward primer: CCGAGGTGTATGTATGAGTGTG	This paper	N/A
COX2 Human Reverse primer: CAGGAGGAAGGGCTCTAGTATAA	This paper	N/A
COX2 Murine Forward primer: GTCATTGGTGGAGAGGTGTATC	This paper	N/A
COX2 Murine Reverse: CAGGAGGATGGAGTTGTTGTAG	This paper	N/A
Recombinant DNA
pLKO.1_RhoA shRNA	Sigma-Aldrich	TRCN0000047712
pLKO.1_RhoA shRNA	Sigma-Aldrich	TRCN0000047711
pLKO.1_SCR shRNA	Sigma-Aldrich	SHC001
pQCXIN-luc	Our Lab	Schafer15
pHAGE NFkB-TA-LUC-UBC-GFP-W	Darrell Kotton, Addgene	Cat#49343
Software and algorithms
GraphPad Prism		v.9
R		v4.0.2
AMIView Imaging Software	Spectral Instruments Imaging	N/A
AURA Software	Spectral Instruments Imaging	N/A
Other
PHD1000 Syringe Pump	Harvard Apparatus	703006
Oxygraph-2K Respirometer	Oroboros Instruments	N/A

Highlights.

• Brief pulses of fluid shear stress induce phenotypes associated with increased metastasis

• Fluid shear stress induces a switch to oxidative metabolism

• Cancer cells exposed to fluid shear stress show reduced levels of reactive oxygen species

• Fluid shear stress rapidly activates folate metabolism, which protects cancer cells