Functional genetics reveals modulators of antimicrotubule drug sensitivity

Abstract

Microtubules play essential roles in diverse cellular processes and are important pharmacological targets for treating human disease. Here, we sought to identify cellular factors that modulate the sensitivity of cells to antimicrotubule drugs. We conducted a genome-wide CRISPR/Cas9-based functional genetics screen in human cells treated with the microtubule-destabilizing drug nocodazole or the microtubule-stabilizing drug paclitaxel. We further conducted a focused secondary screen to test drug sensitivity for ∼1,400 gene targets across two distinct human cell lines and to additionally test sensitivity to the KIF11 inhibitor, STLC. These screens defined gene targets whose loss enhances or suppresses sensitivity to antimicrotubule drugs. In addition to gene targets whose loss sensitized cells to multiple compounds, we observed cases of differential sensitivity to specific compounds and differing requirements between cell lines. Our downstream molecular analysis further revealed additional roles for established microtubule-associated proteins and identified new players in microtubule function.

Figure 1.

Large-scale functional genetics screens reveal modulators of antimicrotubule drug sensitivity. (A) Schematic showing the workflow for the pooled CRISPR screen. (B) Schematic showing the effects of microtubule drugs (nocodazole and paclitaxel) on microtubule dynamics. (C) The curve illustrating the CRISPR score of fitness conferring and growth enhancer genes upon treatment with either nocodazole or paclitaxel. (D) Scatter plot showing the CRISPR scores in untreated versus nocodazole-treated cell pools. (E) Scatter plot showing the CRISPR scores in untreated versus paclitaxel-treated cell pools. (F) Table showing the CRISPR scores of selected iKO in high concentrations of nocodazole or paclitaxel. (G) Autosomal start and end positions (from GRCH38.p12) of genes (black lines) with CRISPR score >1.5, P < 0.05 in 250 nM nocodazole (suppressors) on chromosome 1 were obtained from BioMart and plotted in R (for illustration of all chromosomes see Fig. S1 B); expected versus observed fractions of hits on between the end of chromosome 1 and KIF2C, or between KIF2C and the centromere. Expected fractions were calculated by assuming an even distribution of hits across the chromosome 1 length. (H) Scatter plot illustrating the differential CRISPR scores across all gene targets in the secondary screen. The differential was calculated between nocodazole and untreated and paclitaxel and untreated K562 cell pools. (I) Scatter plot illustrating the differential CRISPR scores across all gene targets in the secondary screen. The differential was calculated between nocodazole and untreated and paclitaxel and untreated HeLa cell pools.

Figure S1.

Visualization of additional screen output. (A) Scatter plot illustrating the differential CRISPR scores across all gene targets in the primary screen. The differential was calculated between nocodazole and untreated and paclitaxel and untreated K562 cell pools. (B) Autosomal start and end positions (from GRCH38.p12) of genes (black lines) with CRISPR score >1.5, P < 0.05 in 250 nM nocodazole (suppressors) were obtained from BioMart and plotted in R. (C) Scatter plot showing the CRISPR scores in untreated versus nocodazole treated HeLa cell pools from the secondary screen. (D) Scatter plot showing the CRISPR scores in untreated versus paclitaxel-treated HeLa cell pools from the secondary screen. (E) Scatter plot showing the CRISPR scores in untreated versus STLC-treated HeLa cell pools from the secondary screen. (F and G) Scatter plots showing the differential CRISPR scores in treated versus untreated HeLa and K562 cells. The differential was calculated between nocodazole and untreated (F) and paclitaxel and untreated cell pools (G). (H) Scatter plot showing the CRISPR scores in untreated versus nocodazole-treated K562 cell pools. (I) Scatter plot showing the CRISPR scores in untreated versus paclitaxel treated K562 cell pools. (J) Scatter plot showing the CRISPR scores in untreated versus STLC treated K562 cell pools.

Figure 2.

Analysis of KIF15 and DLGAP5 iKOs. (A) Secondary screen CRISPR score of KIF15 and DLGAP5 iKOs in HeLa and K562 cells. (B) Representative Z-projected deconvolved immunofluorescence images of mitotic cells of control, KIF15, or DLGAP5 iKO HeLas treated with either nocodazole, paclitaxel, or STLC. Microtubules (DM1α), DNA (DAPI). Scale bar: 9 µm. (C) Percentage of cells with mitotic defects in control or KIF15 iKO, treated with nocodazole (noc), paclitaxel (pac), or STLC. n > 300 cells per condition, across three experimental replicates. (D) Percent of cells with mitotic defects after iKO of DLGAP5, treated with nocodazole, paclitaxel, or STLC. n > 300 cells per condition, across three experimental replicates. (E) Quantification of total spindle tubulin signal in control, KIF15, or DLGAP5 iKO HeLas. n = 71, 60, 96 across three experimental replicates. See Fig. S3 A for representative images. (F) Quantification of total EB1 signal in KIF15 iKO and DLGAP5 iKO HeLa cells. n = 71, 60, 96 across three experimental replicates. Statistical tests performed: Welch’s t test (ns = not significant, ****P = <0.0001). Blue lines indicate the median.

Figure S2.

Analysis of identified hits. (A) Representative max intensity projection of undeconvolved immunofluorescence images of iKO HeLa cells for control, KIF15, or DLGAP5. Stained for microtubules or DNA. Quantified in Fig. 2 E. Scale bar: 5 µm. (B) Quantification of total tubulin immunofluorescence signal in the KIF15 and DLGAP5 iKO HeLa cells treated with nocodazole (noc). n = 73, 70, 61 across three experiments. (C) Quantification of total spindle tubulin immunofluorescence signal in iKO HeLa cells for control, KIF15 or DLGAP5 treated with paclitaxel (pac). n = 66, 73, 66 across three experiments. (D) Mitotic EB3 speed quantification in control, KIF15 or DLGAP5 iKO HeLa cells expressing td-Tomato EB3. n = 82, 148, 111 kymographs, n = 19, 25, 22 cells across three experiments. (E) Interphase EB3 speed quantification in control, KIF15 or DLGAP5 iKO HeLa cells expressing td-Tomato EB3. n = 82, 148, 111, n = 30, 30, 30 cells across three experiments. (F) Interphase EB3 speed quantification in control, HMMR, or SAMHD1 iKO HeLa cells expressing td-Tomato EB3. n = 86, 171, 142 kymographs, n = 30, 30, 32 cells across three experiments. (G) Mitotic EB3 speed quantification in control, HMMR, or SAMHD1 iKO HeLa cells expressing td-Tomato EB3. n = 82, 110, 130 kymographs, n = 20, 19, 23 cells across three experiments. (H) Representative max intensity projection of undeconvolved immunofluorescence images of iKO HeLa cells for control, HMMR, or SAMHD1. Stained for microtubules or DNA. Quantified in Fig. 3 F. Scale bar: 5 µm. (I) Quantification of total mitotic spindle tubulin immunofluorescence signal in the HMMR and SAMHD1 iKO HeLa cells treated with nocodazole. n = 73, 69, 67 across three experimental replicates. (J) Quantification of total mitotic spindle tubulin immunofluorescence in the HMMR and SAMHD1 iKO HeLa cells treated with paclitaxel. n = 60, 61, 60 across three experimental replicates. (K) Representative confocal image of an interphase HeLa cells showing the localization of GFP-tagged SAMHD1. Scale bar: 5 µm. Statistical tests performed: Welch’s t test (***P = <0.001, ****P = <0.0001). Blue lines indicate the median.

Figure 3.

Analysis of: HMMR and SAMHD1 iKOs. (A) Table showing the secondary screen CRISPR score of HMMR and SAMHD1 iKOs in HeLa and K562 cells. (B) Representative Z-projected deconvolved immunofluorescence images of mitotic cells of control, HMMR, or SAMHD1 iKO in HeLas treated with nocodazole, paclitaxel, or STLC. Microtubules (DM1α), DNA (DAPI). Scale bar: 9 µm. (C) Percentage of mitotic cells with mitotic defects in control or after iKO of HMMR, treated with either nocodazole (noc), paclitaxel (pac), or STLC, n > 300 cells per condition, across three experimental replicates. (D) Percentage of mitotic cells with mitotic defects after iKO of SAMHD1, treated with nocodazole, paclitaxel, or STLC. n > 300 cells per condition, across three experimental replicates. (E) Quantification of total EB1 immunofluorescence signal in control, HMMR, or SAMHD1 iKO HeLa cells. n = 94, 86, 51 across three experimental replicates. (F) Quantification of total spindle tubulin immunofluorescence in the HMMR and SAMHD1 iKO HeLa cells. n = 61, 69, 62 across three experimental replicates. See Fig. S2 H for representative images. Statistical tests performed: Welch’s t test (***P = <0.001, ****P = <0.0001). Blue lines indicate the median.

Figure 4.

Analysis of HN1 and HN1L double knock-out cells. (A) Table showing the primary and secondary screen CRISPR score of HN1 or HN1L iKOs in K562 cells. (B) Representative confocal immunofluorescence images of mitotic metaphase and interphase HeLa cells showing the localization of GFP-tagged HN1 and HN1L proteins, Scale bar: 5 µm. (C) Percentage of mitotic cells with mitotic defects in control and after HN1+HN1L double iKO, treated with nocodazole (noc), paclitaxel (pac), or STLC. n > 100 cells per condition, across three experimental replicates. (D) Representative Z-projected deconvolved immunofluorescence images of mitotic cells of HN1+HN1L double iKO in HeLas treated with nocodazole, paclitaxel, or STLC. Microtubules (DM1α), DNA (DAPI). Scale bar: 10 µm. (E) Left: Live confocal stills of td-Tomato EB3 expressing control or HN1+HN1L double iKO HeLa cells; middle: max projection over 30 s middle; right: representative kymographs generated from region highlighted by red arrows. Scale bars: 5 µm/2 µm (kymograph). (F) EB3 speed quantification in control, HN1, HN1L, or HN1/HN1L double iKO HeLa cells. n = 93, 103, 104, 127 kymographs, n > 31 cells across three experimental replicates. (G) Quantification of total spindle tubulin signal in the control or HN1+HN1L double iKO HeLa cells. n = 70, 59 across three experimental replicates. (H) Quantification of total EB1 immunofluorescence in control or HN1+HN1L double iKO HeLa cells. N = 76, 75 across three experimental replicates. Statistical tests performed: Welch’s t test (ns = not significant, ***P = <0.001, ****P = <0.0001). Blue lines indicate the median.

Figure S3.

Analysis of identified hits 2. (A) Clustal analysis of HN1, HN1L homologs across species. (B) Quantification of total spindle tubulin immunofluorescence signal in control, HN1, or HN1L iKO HeLa cells. n = 64, 70, 62 across three experimental replicates. (C) EB3 speed quantification in control or HN1+HN1L double iKO HeLa cells treated with nocodazole (noc) or paclitaxel (pac). n = 171, 188, 170, 162 kymographs, n = 31, 33, 32, 33 cells across three experiments. (D) Quantification of total mitotic spindle tubulin signal in the control or HN1+HN1L double iKO HeLa cells treated with nocodazole. n = 71, 72 across three experiments. (E) Quantification of total spindle tubulin signal in the HN1+HN1L double iKO HeLa cells treated with paclitaxel. n = 71, 72 across three experiments. (F) Live confocal image of td-Tomato EB3 in HeLa cells expressing control or HN1+HN1L double iKO. Stills were max projected over 30 s. Scale bar: 2 µm (G) Quantification of EB3 speed of cells in F n = 60, 63 kymographs, n = 24, 25 cells across three experiments. (H) Western blot of control, CARMT1 knockout, or CARNMT1 knock-out K562 cells expressing hardened CARNMT1 rescue construct probed for CARNMT1 or β-actin. (I) Percentage of cells appearing as monopolar in control, CARNMT1 knockout, or CARNMT1 knock-out K562 cells expressing hardened CARNMT1 rescue construct after STLC treatment. n > 100 cells each in three experiments. (J) Immunofluorescence images of CARNMT1 knock-out K562 cells expressing hardened CARNMT1 rescue construct stained for CARNMT1, tubulin, and DNA. Scale bar: 5 µm. (K) Representative max intensity projection of undeconvolved immunofluorescence images of iKO K562 s for control or CARNMT1. Stained for microtubules or DNA. Quantified in Fig. 5 F. (L) Quantification of mitotic spindle tubulin signal in control or CARNMT1 iKO HeLas. n = 58, 64 across three experiments. (M) Quantification of total EB1 immunofluorescence in control or CARNMT1 iKO K562 s. n > 60 across three experimental replicates. (N) Interphase EB3 speed quantification in control or CARNMT1 iKO K562 s expressing td-Tomato EB3. n = 130, 111 kymographs, n = 34, 35 cells across three experiments. Statistical tests performed: Welch’s t test (**P = <0.01, ***P = <0.001, ****p = <0.0001). Blue lines indicate the median. Source data are available for this figure: SourceData FS3.

Figure 5.

Analysis of CARNMT1 knock-out cells. (A) Table showing the secondary screen CRISPR score of CARNMT1 iKO in HeLa and K562 cells. (B) Representative confocal images of mitotic and interphase HeLa cells showing the localization of GFP-tagged CARNMT1. Scale bar: 2 µm. (C) Representative Z-projected deconvolved images of control or CARNMT1 iKO K562 cells with nocodazole, paclitaxel, or STLC. Microtubules (DM1α), DNA (DAPI). Scale bar: 9 µm. (D) Percentage of mitotic cells without mitotic defects in control or CARNMT1 iKO, treated with nocodazole (noc) or paclitaxel (pac). n > 300 cells per condition, across three experiments. (E) Percentage of cells displaying monopolar spindle in control or CARNMT1 iKO in STLC. n > 300 cells per condition, across three experiments. (F) Quantification of total spindle tubulin immunofluorescence signal in control or CARNMT1 iKO in K562 s. n = 58, 64 across three experimental replicates. For representative images, see Fig. S3 K. (G) Quantification of total spindle tubulin immunofluorescence in the CARNMT1 iKO K562 cells treated with nocodazole. n = 62, 63 across three experimental replicates. (H) Quantification of total spindle tubulin immunofluorescence in the CARNMT1 iKO K562 cells treated with paclitaxel. n = 62, 61 across three experimental replicates. Statistical tests performed: Welch’s t test (ns = not significant, ****p = <0.0001). Blue lines indicate the median.