1. Supplementary Fig S1. Cell growth rate influences reported drug sensitivity in fixed duration assays. The influence of cancer genetics on drug response can be inferred from profiles of anti-proliferative activity across large panels of cancer cell lines. In these assays, cell lines are each exposed to drug for a fixed duration (typically 72 hours). In principle, cytotoxic effects might be distinguished from cytostatic effects in these assays by comparing cell number at 72 hours to the baseline read corresponding to the initial seeding density. In practice, however, fast growing cell lines must be plated at low seeding density to maintain logarithmic growth over the 72 hour period and, consequently, the baseline read cannot be reliably distinguished from a cell-free media control given the noise typical of assays run in 384-well microtiter plates for high-throughput screens. The widely differing growth rates of cancer cell lines in standard culture conditions therefore confound the detection of synthetic lethal relationships for drugs that have a growth arrest effect in most cell lines but a cytotoxic effect in a defined subset. This is illustrated in the Figure which shows theoretical dose-response curves obtained at 0.5, 1, 2, …, 5 doubling times (DT) for a cytostatic compound1 to simulate the effect of doubling time on measured response at 72 hours for growth rates typical of cancer cell lines in standard culture. At three or more DT, both the maximal effect and IC50 start to become indistinguishable from a true cytotoxic effect (red curve) given the noise inherent to a conventional high throughput proliferation assay.
To overcome this artifact, we established the growth rates in vehicle control conditions for hundreds of genomically characterized cancer cell lines and developed a pharmacogenomic screening assay normalized for growth rate by running each assay for 2DT (blue curve) rather than for a fixed duration. We have found that this method, which we call CCSP (Cancer Cell Sensitivity Profiling), reliably distinguishes cytotoxic from cytostatic effects based on the constant 25% signal window separating the dose response curve plateaus. In addition to showing superior efficacy (curve plateau at maximal inhibition), cytotoxic compounds also manifest left-shifted curves giving rise to lower absolute IC50 values relative to cytostatic compounds of similar potency in 2DT assays. By combining both efficacy and potency, absolute IC50 from a 2DT profiling assay serves as a suitable single measure to identify the most sensitive cancer subtypes and is biased to favor cytotoxic effects.
1 defined as a compound that prevents cell growth and proliferation but does not kill cells.; hence, at maximal effect, as many cells remain in the well as were seeded at the experiment outset.

2. Supplementary Fig S2. Inhibition of AurA kinase is equivalent in Rb+ and Rb- cells. Inhibition of phosphoThr288 signal in an MSD assay after 1 hour incubation with various concentrations of alisertib (A) or MK5108 (B) in Rb-null MDA-MB-468 cells or Rb+ MDA-MB-231 breast cancer cells. IC50 values (nM) from each assay are shown in blue and red for the two cell lines respectively. A B
5 nM
7 nM
3 nM
3 nM
Log [alisertib] mM
Log [MK5108] mM

3. A B Rb GAPDH Untreated RB1 siRNA Control siRNA
Supplementary Fig S3. Modulating RB1 expression in cancer cells modifies sensitivity to AurAi. A-B, Calu-6 Rb+ NSCLC cells were treated with siRNA directed at RB1 and 24 hours after transfection their response to alisertib (A) or MK5108 (B) was tested using the xCELLigence Real-Time Cell Analysis (RTCA) system for 5 days. A B
alisertib
MK5108
Untreated
Scramble siRNA
RB1 siRNA
Untreated
Scramble siRNA
RB1 siRNA
20mM
20mM
Cell Index
Cell Index
6.6mM
6.6mM
Cell Index
Cell Index
2.2mM
2.2mM
Cell Index
Cell Index
0.73mM
Cell Index
0.73mM
Cell Index
0.24mM
Cell Index
0.24mM
Cell Index
0.08mM
Cell Index
0.08mM
Cell Index
0.027mM
0.027mM
Cell Index
Cell Index
0.009mM
0.009mM
Cell Index
Cell Index
DMSO
DMSO
Cell Index
Cell Index
Hours
Hours Rb
GAPDH
Untreated
RB1 siRNA
Control siRNA

4. Supplementary Fig S4. A. 2D structure of isoquinoline AurAi (compound 1). b, Crystal structure of 1 bound to AurA showing details of the binding site interactions between 1 and AurA to illustrate the close interaction with Thr217. A B

5. LY3295668 Abs IC50 (µM) Doubling Time (DT, hr)
Supplementary Fig S5. Variation in cell line growth rate across the panel does not appear to influence response to LY using the fixed doubling time (2 DT) assay format. Abs IC50 profile of LY across the cancer cell line panel plotted against the average population doubling time for each cell line. 20 18 16 14 12 10 8 6 4 2
LY Abs IC50 (µM) 20 30 40 50 60 70 80 90
100
110
120
130
Doubling Time (DT, hr)

6. Supplementary Fig S6. The median and mean Abs IC50 values for LY in cell lines from different histologic subtypes. Number of RB1mut v RB1WT cells in each histologic class as well as the p-value (ANOVA test) for the difference of the mean values is shown.

7. RB1 WT RB1 mut RB1 WT RB1 mut RB1 WT RB1 mut
Supplementary Fig S7. LY shows a cytotoxic profile in RB1mut but not RB1WT lung cancer cells. The effect of LY on a panel of 24 lung cancer cell lines after treating for 2DT was tested using either the CellTiter-Glo® viability assay (CTG, Promega) or propidium iodide (PI) staining to measure DNA replication. RB1mut, but not RB1WT, lung cancer cells show high Max inhibition and low Abs IC50 values in response to 2DT treatment with LY using CTG. DNA replication during these short term assays , determined using propidium iodide (PI) staining, is potently inhibited in both RB1mut and RB1WT cells consistent with enhanced apoptosis rather than cytostasis accounting for the different responses. p-value calculated using ANOVA shown in blue.
Viability Assay (CTG)
Potency (Abs IC50)
Viability Assay (CTG)
Efficacy (Max Inh)
DNA replication (PI)
Potency (Abs IC50)
p =
p =
p = 0.520 20 40 60 80
100
0.001
0.01
0.1 1 10
0.001
0.01
0.1 1 10
Abs IC50 (µM)
Max inhibition (%)
Abs IC50 (µM)
RB1 WT
RB1 mut n 17 7
Median Abs IC50 (µM)
11.110
0.012
RB1 WT
RB1 mut n 17 7
Median % Inh
42.894
90.973
RB1 WT
RB1 mut n 17 7
Median Abs IC50 (µM)
0.019
0.078

8. Supplementary Fig S8. Knockout of the RB1 gene from MDA-MB-361 cells confers enhanced resistance to palbociclib and sensitivity to the cytotoxic effects of LY The anti-proliferative effect of palbociclib (A) or LY (B) in a 2DT CTG assay in mock treated cells (solid circles) or MDA-MB-361 cells artificially depleted of RB1 by CRISPR/Cas9-mediated gene knockout (solid triangles). n = 3 independent experiments. (C) MDA-MB-361 mock treated cells and MDA-MB-361 RB1 k/o cells were treated with 1mM LY for the indicated times and PARP and caspase 3 cleavage was determined by immunoblot. A
MDA-MB-361
MDA-MB-361-Rb1 k/o B
% Inhibition
% Inhibition
Log [palbociclib] µM
Log [LY ] µM C
MDA-MB-361
MDA-MB-361-
Rb k/o
Cleaved
PARP
Cleaved
Caspase-3
GAPDH
16 hours
24 hours
16 hours
24 hours
48 hours
No treatment
48 hours
No treatment

9. A B Cell Line Abs IC50 (µM) Cancer type NCI-H1436 0.05 Lung SCLC
Supplementary Fig S9. Sensitivity of RB1 mutant cancer cells to LY does not require amplification of Myc family genes (MYC, MYCN, MYCL). A. Box plot of LY Abs IC50 for SCLC cell lines with RB1 mutation and/or Myc-family gene amplification. P-value for the difference in mean IC50 in RB1 mutant SCLC cells with or without amplification of a Myc-family gene is shown (ANOVA). B. List of RB1 mutant cancer cell lines without co-occurring amplification of Myc-family genes (MYC, MYCN, MYCL) that are sensitive to LY (Abs IC50 < 1 µM). A B
p=0.7
Cell Line
Abs IC50 (µM)
Cancer type
NCI-H1436
0.05
Lung SCLC
DU-4475
0.06
Breast TNBC
U-266
0.08
Multiple myeloma
NCI-H209
0.10
NCI-H1734
Lung Adenocarcinoma
NCI-H2227
0.14
WERI-Rb-1
0.17
Retinoblastoma
M059J
Glioma
NCI-H1618
0.20
Lung_SCLC
NCI-H2081
0.22
NCI-H1048
0.33 H9
0.34
Lymphoma
HLE
0.47
Liver
MDA-MB-468
0.71 10 1
0.1
Abs IC50 (CTG µM)
0.01
MYC-amp & RBmut
RBmut only
MYC-amp only
GeoMean
0.12
0.38
0.99
Count 10 8 2

10. Supplementary Fig S10. Sensitivity of RB1 mutant cancer cells to LY does not require high expression levels of Myc family genes (MYC, MYCN, MYCL). A. LY Abs IC50 among RB1 mutant cancer cell lines of all tumor types plotted against the highest expression value among the three Myc family genes (MYC, MYCN, MYCL) B. LY Abs IC50 among cancer cell lines with amplification of MYC, MYCN, or MYCL, plotted against RB1 expression. Among these cell lines, if RB1 log2 RLM < 5: Abs IC50 < 0.4 uM for 91% (10/11) cells lines. Conversely, if RB1 log2 RLM > 5: Abs IC50 < 0.4 uM for 28% (17/60) cells lines. A B
Abs IC50 (µM)
Abs IC50 (µM)
r2=0.094
r2=0.079
max(MYC, MYCN, MYCL) (Log2 of RNAseq RLM Signal)
RB1 expression (RNAseq Log2 of RLM Signal)

11. A B RB1mut RB1WT RB1mut RB1WT Median=11.5 Median=11.3 Median=9.5
Supplementary Fig S11. Co-factor expression is similar in RB1mut and RB1WT cells. A. TPX2 and B. TACC3 RNA expression levels among RB1 mutant and wild-type cancer cell lines. The difference in expression is not statistically different in the two RB1 groups with either gene. A B
Median=11.5
Median=11.3
Median=9.5
Median=9.3 7 8 9 10 11 9 10 11 12 13
TPX2 RNA Expression
TACC3 RNA Expression
RB1mut
RB1WT
RB1mut
RB1WT

12. Supplementary Fig S12. SYK inhibition cannot account for the anti-proliferative activity of Aurora-A inhibitors in RB1 mutant cancer cells. Geometric mean Ki nM (SEM, number of experiments) for inhibitors of Aurora (described here) and SYK (from Zhang et al. 2012) in enzyme assays. In some experiments Ki values were undefined (below or above the threshold of detection) and for these cases geometric mean and SEM values are listed as a minimum or maximum values.
Compound
AurA Ki (nM)
AurB Ki (nM)
SYK Ki (nM)
SYK/AurA ratio
MK5108
<0.81 (>0.037, n=2)
4.7
(1.03, n=2)
2421
(327, n=2)
>3000
Alisertib
<1.1
(>0.21, n=5)
5.8
(0.80, n=6)
10,558
(35, n=4)
>9,600
Barasertib 61
(19, n=3)
2.8
(0.96, n=3)
>11,700
(n=2)
 >190
Compound 1
<0.9
(>0.073, n=7) 76
(13, n=6)
>60,000
>68,000 LY
0.8
(0.066, n=4)
1038
(265, n=5)
>74,000
SYK (BAY )
>13,400
(>1210, n=2)
1801
(670, n=3) 12
(5.8, n=2)
<0.0009
SYK (R-406)  66
(7.3, n=2)
327
(94, n=2) 29
(12, n=3)
0.44

13. A HeLa B H446 C H2228 E D 24hr 48hr H2228 H446 DMS-53
Supplementary Fig S13. RB1 mutant cancer cells show a delayed mitosis without AurB kinase inhibition and with subsequent apoptosis in response to AurAi. Histone H3 Ser10 and Ser28 are known substrates for AurB kinase and their phosphorylation increases at mitosis in the absence of AurB inhibition. Rb null HeLa (A), H446 (B) and H2228 (C) cells were treated with either or MK5108 at the indicated concentrations for 24 hours and lysates were immunoblotted with the indicated antibodies. The concentration dependent increase in cyclin B1 and phosphoH3 is consistent with mitotic block and inconsistent with AurB inhibition at these concentrations since AurB inhibitors cause a decrease in phosphoH3 in equivalent conditions. (D) Cyclin B1 levels after the indicated doses and duration of LY treatment in the indicated RB1mut and RB1WT lung cancer cell lines. (E) Apoptosis predominantly occurs during mitotic arrest of RB1mut cells in response to LY High content imaging analysis was used to analyze the fraction (% over DMSO control) of cells expressing cyclinB1 and with both caspase 3/7 activation and, either, > 2N or ≤ 2N DNA content after 24 hr treatment with 100 nM LY A
HeLa
[LY ] nM
[MK5108] nM
DMSO
500
100 20 5 1
DMSO
500
100 20 5 1
Cyclin B1
H3 ser10
H3 ser28
Actin B
H446
[LY ] nM
[MK5108] nM
DMSO
500
100 20 5 1
DMSO
500
100 20 5 1
Cyclin B1
H3 ser10
H3 ser28
Actin C
H2228
[LY ] nM
[MK5108] nM
DMSO
500
100 20 5 1
DMSO
500
100 20 5 1
Cyclin B1
H3 ser10
H3 ser28
Actin E D
24hr
48hr
H446
H2228
DMS53
RB1 mut RB1 WT
% cells
DMSO
0.25mM
0.1mM
DMSO
0.25mM
0.1mM
H2228
CyclinB1
Actin
H446
CyclinB1
Actin
DMS-53
CyclinB1
Actin

14. Blue = DNA, Green = a-tubulin, Red = pericentrin
Supplementary Fig S14. Despite prolonged mitotic block and increased apoptosis, the spindle morphology of Rb null cells is not notably different to Rb wild-type cells following LY treatment. Representative mitotic phenotypes after 24h treatment with 100 nM Aurora A inhibitor showing typical mitotic spindle/centrosome abnormalities. Maximum intensity projections of Z stacks taken with a Zeiss LSM880 confocal microscope. No obvious differences in drug-induced changes to spindle morphology were noted between Rb-positive DMS-53 and MDA-MB-231 cells when compared to Rb-negative H228 and HeLa cells (the HPV-positive cervical cancer cell line HeLa, which works very well for imaging mitotic spreads, does not express Rb due to viral E7). These experiments are somewhat confounded, however, by the rapid apoptosis apparent in the Rb-null cells treated with LY and by the fact that the Rb null cells, without drug exposure, frequently showed large or supernumerary centrosomes.
Blue = DNA, Green = a-tubulin, Red = pericentrin

15. _____________ MB-361 MCF-7 P PR P PR Myc N-Myc MYCL1 β-Actin
Supplementary Fig S15. ER+ breast cancer cell lines selected for resistance to the CDK4/6i palbociclib do not show increased expression of Myc proteins (Myc, N-Myc, MYCL1). Immunoblots for Myc family proteins from palbociclib-resistant (PR) variants of MDA-MB-31 and MCF7 versus parent cells (P) described in the text.
_____________
MB MCF-7
P PR P PR
Myc
N-Myc
MYCL1
β-Actin