Diverse microtubule-destabilizing drugs induce equivalent molecular pathway responses in endothelial cells

Abstract

Drugs that modulate microtubule (MT) dynamics are well-characterized at the molecular level, yet the mechanisms linking these molecular effects to their distinct clinical outcomes remain unclear. Several MT-destabilizing drugs, including vinblastine, combretastatin A4, and plinabulin, are widely used, or are under evaluation for cancer treatment. Although all three depolymerize MTs, they do so through distinct biochemical mechanisms. Furthermore, their clinical profiles and therapeutic uses differ considerably. To investigate whether differential modulation of molecular pathways might account for clinical differences, we compared gene expression and signaling pathway responses in human pulmonary microvascular endothelial cells (HPMECs), alongside the MT-stabilizing drug docetaxel and the pro-inflammatory cytokine TNF-α. RNA-sequencing and phosphoproteomics revealed that all three MT destabilizers triggered equivalent molecular responses. The substantial changes in gene expression caused by MT destabilization were completely dependent on Rho family GTPase activation. These findings suggest that the distinct clinical profiles of the destabilizing drugs depend on differences in pharmacokinetics (PK) and tissue distribution rather than molecular actions. The washout rate of the three drugs differed, which likely translates to PK differences. Our data provide insights into how MT destabilization triggers signaling changes, potentially explaining how these drugs induce cell cycle re-entry in quiescent cells and how plinabulin ameliorates chemotherapy-induced neutropenia.

Figure 1.

Microtubule destabilizers elicit similar transcriptional responses in HPMECs. Cells were treated for 6 hours with each condition. (A) Principal component analysis of RNA-seq data from HPMECs treated with MT destabilizers. Donors are shown as shapes, treatments as colors. MT destabilizers are colored in shades of blue/green. Percentages indicate proportion of variance explained by each component. (B) Number of differentially expressed genes (DEGs) in HPMECs. Conditions were compared to vehicle (0.1% DMSO), and MT depolymerizers were compared to each other. Genes were classified as differentially expressed if | log₂FC | > 1 and p-adjusted < 0.05. Directionality reflects changes in the first comparator relative to the second. (C) Top 25 DEGs across all conditions ranked by absolute log₂FC. DEGs were selected based on the highest absolute log2FC across conditions and displayed in descending order of log2FC for the CA4 condition. Genes not differentially expressed in a given condition are shown in gray. (D) Volcano plot of vinblastine-treated HPMECs compared to baseline. Genes with the highest adjusted p-values (black) or log2FC (blue) are labeled. (E) Reactome pathway analysis of DEGs. The top 3 pathways (by adjusted p-value) were selected for MT depolymerizers, and the top 1 for TNF-α and DTXL. Darker colors indicate more significant adjusted p-values.

Figure 2.

Transcriptional changes are downstream of Rho activation. (A) Change in VB-activated actin stress fibers after Rho and ROCK inhibition. HPMECs were pre-treated with vehicle (H₂O), 1 μg/ml C3 transferase (Rho inhibitor, RHOi) for 4 hours, or 10 μM Y-27632 (ROCK inhibitor, ROCKi) for 1 hour. After pre-treatment, cells were treated with 0.1% DMSO or 100 nM VB for 30 min. Actin stress fibers were stained with fluorescent phalloidin and imaged on a confocal microscope. Representative images were selected and scaled by the brightest image. Insets show selected regions (red boxes) at higher magnification. Scale bar = 50 μm. (B) Changes in DEGs after Rho and ROCK inhibition. Following pre-treatment, cells were treated with 0.1% DMSO or 100 nM VB for 6 hours, with a booster of 0.5 μg/ml C3 transferase for RHOi. DEGs are genes with | log₂FC | > 1 and p-adjusted < 0.05. Comparison between perturbation+DMSO and H2O+VB indicates changes in gene expression are not due to the perturbation having a similar effect as MT depolymerization. (C) Model for MT loss-dependent changes in gene expression. MT destabilization activates Rho GTPases, which signal partially through ROCK as well as other effectors. ROCKi leads to a reduction in ROCK-dependent downstream gene expression, while RHOi inhibits changes through all arms of the pathway, completely abolishing changes in gene expression induced by MT destabilization.

Figure 3.

Differentially modified phosphosites (DMPs) are similar across MT destabilizers. (A) Principal components of phosphoproteomic data for short-term MT destabilization in HPMECs. Donors are represented by different shapes, and treatments correspond to different colors. (B) Number of DMPs after 30 min of treatment. Each condition was compared to vehicle (0.1% DMSO), and MT destabilizers were compared to each other. Modified peptides were classified as significantly differentially phosphorylated if | log₂FC | > 1 and p-adjusted < 0.05. (C-D) Functional pathway analysis of differentially modified proteins. After DMPs were identified, the names of modified proteins were analyzed using Reactome and PANTHER GO Slim. For MT depolymerizers, the top 5 pathways from Reactome and the top 7 pathways from PANTHER were selected, based on adjusted p-value. For TNF-α, the top 2 pathways were included, while for DTXL, only the top pathway was included. (E) Predicted kinase activity in HPMECs treated with VB or TNF- α for 30 min. Centered phosphosites were submitted to the Kinase Library’s updated library to predict kinase activity based on substrate modifications. VB was selected as a representative depolymerizer. Kinases with phosphorylated substrates are shown in green, dephosphorylated substrates in purple, and those with both phosphorylated and dephosphorylated modifications in an open black circle.

Figure 4.

Plinabulin washes out faster from cells than other microtubule destabilizers. HPMECs were incubated with 100 nM of each drug at time 0, followed by four sequential washes at 3, 4, 6, and 9 hours. Cells were harvested at 0, 1, and 6 hours post-initial wash, and intracellular drug concentrations were measured using LC-MS. Drug concentrations are presented as absolute intracellular concentrations (left) and normalized to their initial values (right).