Extended Data Fig. 10. Effects of CCM2 and TLNRD1 knockdown relative to MAP3K3 knockdown and laminar flow.

a-d. CRISPRi TeloHAEC with control non-targeting guides or with guides to CCM2 or TLNRD1 (2 guides apiece) were nucleofected with Cas9 particles containing control non-targeting guides or with 3 guides targeting exon 3 of MAP3K3 (Synthego). Cells were treated with 2μg/ml doxycycline for a total of 5 days (3 prior to nucleofection & 2 afterwards), to induce the CRISPRi machinery. RNA was harvested 48 hours later (after phase contrast imaging), and RNA-seq libraries sequenced to a depth of 10-12 million reads. The CCM2 guides reduced target gene expression, on average, by 3.4-3.6 fold (p.<2e-9), while TLNRD1 guides reduced target gene expression by 9.4-9.9 fold (p. <2e-43), consistent with the effects of these guides in our other bulk RNAseq data (Fig. 3c). MAP3K3 transcript levels were not significantly reduced, but genome-mappable reads for the targeted exon (#3) were greatly reduced, and most of the remaining reads showed multiple mismatches, indicating efficient introduction of Cas9-targeted deletions. N=2 per condition (from one experiment, 1 RNAseq library for each of 2 CRISPRi guides per target - CCM2, TLNRD1 or non-targeting controls). Correlation coefficient (R), and p.values given in each panel are from a two-sided Pearson correlation test.

a. The difference between the effect of CCM2 knockdown in cells with MAP3K3 knockdown and the effect of CCM2 knockdown in control cells ([CCM2kd_with_MAP3K3kd/Control_with_MAP3K3kd] / [CCM2kd/Control], Y-axis) was plotted against the effect of CCM2 knockdown in control cells ([CCM2kd/Control], X-axis, plotting all genes regulated at p. <5e-4 in either contrast). Labeled genes are the top 5 up- or down-regulated genes by log2 fold change, on each axis. Diagonal line: slope -1 reference.

b. As in (a), but for the difference between the effect of TLNRD1 knockdown in cells with MAP3K3 knockdown and the effect of TLNRD1 knockdown in control cells (Y-axis), versus the effect of TLNRD1 knockdown in control cells (X axis). The negative correlations in (a) & (b) indicate that MAP3K3 perturbation partially reverses the transcriptomic effects of CCM2 or TLNRD1 knock down, consistent with a role of MEKK3/MAP3K3 signaling in regulating transcription downstream of both CCM2 & TLNRD1.

c. The difference between the effect of MAP3K3 knockdown in cells with CCM2 knockdown and the effect of MAP3K3 knockdown in control cells (Y-axis) versus the effect of MAP3K3 knockdown alone (X axis, plotting all genes regulated p. < 5e-4 in either contrast). Diagonal line: slope -1 reference.

d. As in (c), but for the difference between the effect of MAP3K3 knockdown in TLNRD1 knockdown cells and the effect of MAP3K3 knockdown in control cells (Y-axis) versus the effect of MAP3K3 knockdown in control cells (X axis). The negative correlations in both (c) & (d) indicate that perturbation of CCM2 or TLNRD1 partially reverses the transcriptional effects of MAP3K3 knockdown, consistent with the expectation that decreased expression of upstream inhibitors can compensate for decreased expression of MEKK3.

e-h. CRISPRi TeloHAEC with control non-targeting guides or with guides to CCM2 or TLNRD1 (2 guides apiece) were grown in static culture or subjected to flow in an Ibidi flow chamber for 48 hours. In each case, cells were treated with 2μg/ml doxycycline to induce the CRISPRi machinery for 5 days (3 days prior & 2 days after introduction of laminar flow). After phase contrast imaging, RNAseq libraries were prepared and sequenced to a depth of 10–12 million reads. N=2 per condition (one experiment, with 1 RNAseq library for each of 2 CRISPRi guides per target - CCM2, TLNRD1 or non-targeting controls). R and p.values for panels (e-h) as per (a-d).

e. The effects of CCM2 knockdown in static culture (Y-axis) compared to the effect of flow in control cells (X-axis, showing all genes regulated at p. < 5e-4 in either contrast). Labeled genes are the top 5 up- or down-regulated genes by log2 fold change, on each axis. Diagonal line: slope = +1 reference.

f. As in (e), but for the effects of TLNRD1 knockdown in static culture. The positive correlations indicate that TLNRD1 or CCM2 knockdown in static culture is similar to the effect of flow, consistent with the observation that TLNRD1 or CCM2 knockdown increases the number and parallelness of actin stress fibers (Fig. 6b-e), a characteristic of flow response in unperturbed ECs ¹³.

g. The difference between the effect of flow in cells with CCM2 knockdown and the effect of flow in control cells ([Flow_CCM2kd/Static_CCM2kd] / [Flow_Ctrl/Static_Ctrl], Y-axis) versus the effect of flow in control cells ([Flow_Ctrl/Static_Ctrl], X-axis). Diagonal line: slope = -1 reference.

h. As in (g), but showing the difference between the effect of flow in cells with TLNRD1 knock down and the effect of flow in control cells (Y-axis) versus the effect of flow in control cells (X-axis). The negative correlations in (g & h) indicate that CCM2 or TLNRD1 knockdown cells have a weaker transcriptional response to flow than control cells. The negative correlations are also consistent with the observations in (e & f) that CCM2 or TLNRD1 knockdown in static culture are similar to the effects of flow, such that lesser fold-changes in gene expression would be required for CCM2 or TLNRD1 knockdown cells to achieve the full normal transcriptional response to flow.

i. Representative images of CRISPRi teloHAEC with control non-targeting guides or with guides to CCM2 or TLNRD1, that were subjected to flow in an Ibidi flow chamber for 48 hours. Cells were imaged by phase contrast microscopy using a 20x objective. N=2 per condition from one (one experiment, with 2 CRIPSRi guides to CCM2, TLNRD1 or controls), and with 4 images per guide.

j. The normal alignment to flow in control teloHAEC (measured as the angle, relative to flow, of the long axis of each cell) is significantly abrogated in both CCM2 & TLNRD1 KD cells (increased average angle relative to flow). Average values for all cells in each of 4 images for each of 2 guides per target were calculated (35 to 103 cells per image). Significance was assessed by two sided T-test on these average values. N=8. Boxplot center line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range; points, outliers. Note that alignment to flow is not completely blocked in CCM2 or TLNRD1 KD cells, since the average angle relative to flow does not reach the 45% value expected if orientation were entirely random.

k. As per (j), but measuring the ratio of the long vs. short axis lengths for each cell (“length/width”) from the fit ellipse function in FiJi.