Regulation of microtubule dynamics by DIAPH3 influences amoeboid tumor cell mechanics and sensitivity to taxanes

Abstract

Taxanes are widely employed chemotherapies for patients with metastatic prostate and breast cancer. Here, we show that loss of Diaphanous-related formin-3 (DIAPH3), frequently associated with metastatic breast and prostate cancers, correlates with increased sensitivity to taxanes. DIAPH3 interacted with microtubules (MT), and its loss altered several parameters of MT dynamics as well as decreased polarized force generation, contractility, and response to substrate stiffness. Silencing of DIAPH3 increased the cytotoxic response to taxanes in prostate and breast cancer cell lines. Analysis of drug activity for tubulin-targeted agents in the NCI-60 cell line panel revealed a uniform positive correlation between reduced DIAPH3 expression and drug sensitivity. Low DIAPH3 expression correlated with improved relapse-free survival in breast cancer patients treated with chemotherapeutic regimens containing taxanes. Our results suggest that inhibition of MT stability arising from DIAPH3 downregulation enhances susceptibility to MT poisons, and that the DIAPH3 network potentially reports taxane sensitivity in human tumors.

Figure 1. Low DIAPH3 expression is associated with reduced patient survival

. A. Kaplan-Meier analysis of overall survival in PCa patients whose tumors are ‘DIAPH3 low’ (<25^th percentile, mRNA expression) versus ‘DIAPH3 high’ (≥25^th percentile, mRNA expression). Log rank test, p = 0.0377. B. Kaplan-Meier analysis of overall survival in glioblastoma patients whose tumors display ‘DIAPH3 low’ versus ‘DIAPH3 high’ expression. Log rank test, p = 0.0039.

Figure 2. DIAPH3 silencing reduces the population of stable MT and alters MT topology.

A. DIAPH3-silenced and control DU145 cells, plated on collagen I, were stained with Cholera toxin B and Ac-tubulin. Filled arrowheads indicate long MT polymers, open arrowheads indicate short MT polymers/fragments. Scale bar, 10 μm. B. DIAPH3-silenced LNCaP cells or control cells, stained with Ac-tubulin antibodies. Scale bar, 10 μm. C. DIAPH3-silenced HMEC-RasV12 cells or control cells, stained with Ac-tubulin antibodies. Scale bar, 10 μm. D-F. DIAPH3-silenced DU145 cells or controls, plated on glass coverslips, were stained with antibodies against α-tubulin (D), β-tubulin (E), or tubulin βIII (F). Data shown are representative of at least 10 fields per condition, in two independent experiments. G. DU145 cells (control, top, or DIAPH3-silenced, bottom) were infected with 100 μl of CellLight GFP-tubulin, incubated for 1-2 days and imaged by spinning disc confocal microscopy. Scale bar, 25 μm. H. The median length of the longest MT in control vs. DIAPH3-silenced DU145 cells. The freehand line function of ImageJ was used to trace the length of the 2 to 3 longest MT in a single cell expressing emGFP-tubulin. A total of 41 MT in control and 37 MT in DIAPH3-silenced cells were measured. Length was converted from pixels to microns, and median length values for each cell type determined. Similar results were obtained from cells expressing TagRFP-tubulin. Mann Whitney test, p < 0.001. Filled arrowheads indicate long MT polymers, open arrowheads indicate short MT polymers/fragments.

Figure 3. The population of dynamic MT is increased by DIAPH3 loss.

A. Montage of TagRFP-tubulin visualized by spinning disc confocal microscopy. Note relative persistence of example MT in control cell (inset, top), yet rapid disappearance of example MT in DIAPH3-silenced cell (inset, bottom), indicating increased MT dynamics. B. Quantitation of the maximum MT length change in DU145 cells. MT length was measured over 10 frames (30 s period), and the maximal length change calculated as described in the Methods. The median of the maximum MT length change in control or DIAPH3-silenced cells is shown as a Tukey plot, and was analyzed with a Mann-Whitney test. C. TFM measurements of RMS value of traction in control or DIAPH3-silenced cells, with and without incubation with 2 μM nocodazole. Note increased contractility in response to nocodazole in both cell lines, but to a greater extent in DIAPH3-deficient cells, especially at 11 and 26 kPa stiff substrates, at which tractions were of comparable magnitude between control and DIAPH3-silenced cells. * = p < 0.0001, Student’s t-test. D. Schematic summarizing effects of DIAPH3 silencing or nocodazole on contractile force. Dynamic MT inhibit contractility. MT depolymerization with nocodazole disrupts MT, and increases traction (left). DIAPH3 loss, by increasing dynamic MT, is predicted to decrease contractility (right). E. Representative contraction maps, demonstrating the location and magnitude of traction exerted by control and DIAPH3-silenced DU145 cells. Note asymmetry of forces in control cells versus lack of force polarity in DIAPH3-deficient cells. F. Representation of traction orientation and polarity. For each ellipse, representing a single cell, semi-axes are determined by the eigenvalues of the matrix comprised by the first-order moment of the traction (M) and orientation determined by the corresponding eigenvectors of M. Note the greater circular contour (symmetry) of DIAPH3-deficient cells, indicating reduced traction polarity relative to more elliptical (polarized asymmetry of) control cells. G. Traction polarity, obtained from the ratio of the eigenvalues of M in control or DIAPH3-silenced DU145 cells at the indicated substrate stiffnesses. Note increased traction polarity with increasing substrate stiffness in control cells, yet relative insensitivity to substrate stiffness in DIAPH3-deficient cells.

Figure 4. DIAPH3 interacts with MT.

A. MT-relevant interactome of DIAPH3 in U87 cells. Cells stably expressing GFP- or GFP-DIAPH3 were immunoprecipitated with antibodies against GFP, and subjected to proteomic analysis by tandem LC-MS/MS. The fold-enrichment of interaction with GFP-DIAPH3 over GFP in each protein is shown. Note prevalence of tubulins as DIAPH3-interacting proteins. The pie chart illustrates the fraction of each subtype of MT-relevant proteins to the total number of MT-relevant proteins detected. B. Co-localization of DIAPH3 and α-tubulin in DU145 cells stably expressing GFP-DIAPH3. Scale bar, 10 μm. C. Co-localization of DIAPH3 and Ac-tubulin in DU145 cells stably expressing GFP-DIAPH3. Scale bar, 10 μm. D. Enrichment of DIAPH3 interaction with MT-relevant proteins, when MT are intact (25 ^oC) vs. depolymerized (4 ^oC), in DU145 cells. After immunoprecipitation from cells stably expressing GFP- or GFP-DIAPH3, interacting proteins were analyzed by LC-MS/MS. The fold-enrichment at which each protein interacted with GFP-DIAPH3 or GFP at 25 ^oC vs. 4 ^oC was determined. Bottom, pie chart illustrating the fraction of each subtype of MT-relevant proteins to the total number of MT-relevant proteins detected. E-F. Interaction of DIAPH3 with α-tubulin. Reciprocal co-immunoprecipitations of GFP-DIAPH3 or GFP with α-tubulin in DU145 (E) or U87 (F) cells was performed at 25 ^oC or 4 ^oC. Note greater extent of binding of GFP-DIAPH3 to α-tubulin, at 25 ^oC vs. 4 ^oC, despite equivalent precipitation of bait at these two temperatures.

Figure 5. DIAPH3 silencing increases MT responsiveness to MT stabilizing agents.

A. DU145 cells were treated with varying concentrations of paclitaxel for 30 min prior to lysis and immunoblotting with Ac-tubulin antibodies. B. Top, cells were treated with 500 nM of each MSA for 30 min at 37 ^oC, and assessed for Ac-tubulin levels. Bottom, data were quantified after normalization of Ac-tubulin to β-tubulin intensities, followed by ratiometric comparison between MSA-treated and untreated conditions. Note greater fold-change in Ac-tubulin levels in DIAPH3-deficient cells following treatment with MSA relative to baseline (untreated) conditions. Data shown are average ± SD from 3 combined, independent trials. C. Quantitation of Ac-tubulin fluorescence, from at least 25 cells per condition. Note that the reduced Ac-tubulin fluorescence in cells silenced for DIAPH3, relative to controls, persists in the presence of taxol. D. Quantitation of perinuclear Ac-tubulin fluorescence, from at least 25 cells per condition. Note the greater fold-increase in fluorescence by MSA treatment (relative to untreated conditions) in DIAPH3-deficient cells. A (*) indicates p < 0.0001. E. Intracellular accumulation of OG-PTX in cells expressing or silenced for DIAPH3 was monitored spectrophotometrically, and fluorescence normalized to protein concentration in each well. n = 2 independent trials.

Figure 6. Greater cytotoxicity of MT-stabilizing agents in cells silenced for DIAPH3.

A-C. DU145 cells expressing or silenced for DIAPH3 were incubated with varying concentrations of each MSA for 4 d, and cell survival monitored as crystal violet absorbance. Values are normalized to vehicle (DMSO) treatment. A (*) indicates a p-value of <0.02 (Student’s t-test). D. LNCaP cells expressing or stably silenced for DIAPH3 were treated with the indicated concentrations of docetaxel for 4 d, and survival assessed as in (A-C). Inset, silencing of DIAPH3. E. HMEC-RasV12 cells expressing or stably silenced for DIAPH3 were treated with the indicated concentrations of docetaxel and assessed as in (D). Inset, silencing of DIAPH3. n = 2 independent trials.

Figure 7. Low DIAPH3 expression is associated with improved clinical outcomes in breast cancer patients after taxane-containing chemotherapy.

A and B. DIAPH3 levels differ in responders and non-responders to neo-adjuvant T/FAC chemotherapy, as predicted by expression of the DLDA-30 signature. A. The median expression level of DIAPH3 is shown as a box and whisker plot. B. Fisher’s exact test demonstrates increased chemotherapeutic response in patients with low median DIAPH3 expression. C. DIAPH3 levels are lower in cancers with a complete pathologic response (pCR) based on histopathological inspection. Contingency table demonstrating the frequency of clinically-achieved pCRs in patients treated with T/FAC therapy. Cox proportional hazards analysis revealed a significant association of low DIAPH3 expression with therapeutic responsiveness. D-F. Kaplan-Meier curves demonstrating that low DIAPH3 expression is associated with a longer time to recurrence (D, and E,29) and with greater overall survival (F,29) in patients treated with taxane-containing neoadjuvant chemotherapy. G. Cox proportional hazards regression analyses, derived from datasets used in panels D-F, demonstrating increased hazards ratios (improved response) for recurrence-free and overall survival status in patients with low DIAPH3 expression.

Figure 8. Greater pathologic complete response (pCR) after treatment with neoadjuvant taxane-containing therapy in TNBC expressing low DIAPH3.

A. Prediction of a pCR in 2 cohorts of TNBC patients whose cancers display low DIAPH3 expression. B. DIAPH3 expression is lower in TNBC patients predicted to respond to T/FAC therapy than those predicted not to respond. C. Achievement of a histologically-defined pCR in cancers with low DIAPH3 levels. Cox proportional hazards regression analysis revealed significant association between low DIAPH3 expression and pCR. D. Lower median DIAPH3 expression in patients achieving a pCR. E. Kaplan-Meier analyses demonstrating a trend toward extended time to recurrence in patients with low DIAPH3 expression.

Figure 9. Model for the association of DIAPH3 loss with amoeboid motility and taxane sensitivity.

Low DIAPH3 expression reduces the extent of MT stability, and in turn increases dynamic MT content. These transitions may lead to either 1) amoeboid behavior and worse prognosis (in untreated patients) or 2) taxane sensitivity and better prognosis (in taxane-treated patients).