Marked Synergy by Vertical Inhibition of EGFR signaling in NSCLC Spheroids: SOS1 as a therapeutic target in EGFR-mutated cancer

Drug treatment of 3D cancer spheroids more accurately reflects in vivo therapeutic responses compared to adherent culture studies. In EGFR-mutated lung adenocarcinoma, EGFR-TKIs show enhanced efficacy in spheroid cultures. Simultaneous inhibition of multiple parallel RTKs further enhances EGFR-TKI effectiveness. We show that the common RTK signaling intermediate SOS1 was required for 3D spheroid growth of EGFR-mutated NSCLC cells. Using two distinct measures of pharmacologic synergy, we demonstrated that SOS1 inhibition strongly synergized with EGFR-TKI treatment only in 3D spheroid cultures. Combined EGFR- and SOS1-inhibition markedly inhibited Raf/MEK/ERK and PI3K/AKT signaling. Finally, broad assessment of the pharmacologic landscape of drug-drug interactions downstream of mutated EGFR revealed synergy when combining an EGFR-TKI with inhibitors of proximal signaling intermediates SOS1 and SHP2, but not inhibitors of downstream RAS effector pathways. These data indicate that vertical inhibition of proximal EGFR signaling should be pursued as a potential therapy to treat EGFR-mutated tumors.


Introduction 26
Lung cancer is the leading cause of cancer-related death worldwide; adenocarcinomas are 27 the most common subtype of lung cancer. Oncogenic driver mutations in the RTK/RAS pathway 28 are found in over 75% of lung adenocarcinomas (1). Activating EGFR mutations occur in 10-29 30% of lung adenocarcinomas, and are the major cause of lung cancer in never-smokers. In 30 patients whose tumors harbor either an L858R mutation or an exon 19 deletion (85% of EGFR 31 mutated tumors), first-generation EGFR-tyrosine kinase inhibitors (TKIs) erlotinib and gefitinib 32 enhance progression free survival (2-4). However, resistance to first generation TKI invariably 33 occurs. In most cases, acquired resistance to first generation EGFR-TKIs occurs via either a 34 secondary EGFR 'gatekeeper mutation' (T790M, 50-60% of cases) that renders the receptor 35 insensitive to first generation EGFR-TKIs or oncogenic shift to alternative RTKs (15-30%). To 36 treat patients with T790M-mutated resistant tumors, the third generation EGFR-TKI osimertinib, 37 which selectively targets activating EGFR mutant proteins including T790M but spares wild-38 type EGFR, was developed (5, 6). However, despite further enhancing survival of patients with 39 EGFR-mutant tumors, resistance again emerges. 40 Unlike first-generation EGFR-TKIs, mechanisms driving osimertinib resistance are more 41 variable, including both EGFR-dependent (10-30%) and EGFR-independent mechanisms (7-10). 42 The most common EGFR-independent resistance mechanisms involve reactivation of the 43 RTK/RAS/effector pathway (10), often via enhanced signaling through parallel RTKs (7-9, 11-44 16). Here, combining osimertinib with individual RTK inhibitors can both inhibit the 45 development of resistance through the inhibited RTK and kill cancer cells with resistance driven 46 by the specific RTK being inhibited. However, simultaneous inhibition of multiple RTKs with 47 osimertinib may be required to eliminate oncogenic shift to alternative RTKs (8). Downstream 48 pharmacologic synergy, we demonstrate that SOS1 inhibition using BAY-293 synergizes with 72 osimertinib only under 3D spheroid culture conditions, and in doing so add to the growing 73 evidence that pharmacologic assessment of novel therapeutics designed to treat cancer must be 74 performed under 3D culture conditions (27,31,(34)(35)(36). By assessing the pharmacologic 75 landscape of EGFR/RAS pathway inhibitors, we demonstrate that inhibition of proximal 76 signaling is required to synergize with osimertinib, and that combined EGFR and SOS1 77 inhibition synergizes to inhibit RAS effector signaling in 3D culture. These findings have 78 significant therapeutic implications for the development of combination therapies to treat EGFR-79 mutated lung adenocarcinoma. 80 81 Results 82

SOS1 deletion inhibits transformation in EGFR-mutated NSCLC cells 83
Previous studies showed that EGFR-mutated NSCLC cell lines show much more robust 84 responsiveness to first generation EGFR-TKIs in 3D (spheroid or organoid) culture compared to 85 2D adherent culture, and further that 3D conditions more readily mirror EGFR-TKI responses 86 seen in vivo (28). To confirm these findings and extend them to third generation EGFR-TKIs, 87 we assessed dose-dependent survival of both first-generation EGFR-TKI sensitive (HCC827, to rest for 48 hours, and then treated with increasing doses of either the first-generation EGFR-92 TKI gefitinib or the third-generation EGFR-TKI osimertinib for four days. HCC827 cells 93 showed responsiveness to both EGFR-TKIs under 2D and 3D culture conditions, however in 94 both cases 3D spheroid cultures showed a > 1-log enhancement in EGFR-TKI efficacy and 95 enhanced overall growth inhibition. While NCI-H1975 cells were not sensitive to gefitinib, 96 osimertinib treatment of H1975 cells showed enhanced efficacy and increased overall growth 97 inhibition in 3D spheroids over 2D adherent cultures. 98 SOS1 and SOS2 are ubiquitously expressed RasGEFs responsible for transmitting EGFR 99 signaling to downstream effector pathways. To determine whether SOS1 or SOS2 were required 100 for 2D anchorage-dependent proliferation or 3D spheroid growth in EGFR-mutated NSCLC 101 cells, SOS1 ((37) and Fig. S1) or SOS2 (31) were deleted in pooled HCC827 and H1975 cells, 102 and both proliferation and spheroid growth were assessed versus NT controls ( Fig. 1B and C). In 103 adherent culture, neither SOS1 nor SOS2 deletion altered proliferation (Fig. 1B). In contrast, 104 SOS1 deletion completely inhibited spheroid growth in both HCC827 and H1975 cells, 105 indicating that SOS1 was required to maintain the transformed phenotype in both cell lines. To 106 determine whether SOS1 was generally required for mutant EGFR-driven transformation, we 107 further deleted SOS1 or SOS2 in both first-generation sensitive NCI-H3255 (L858R) and PC9 108 (ex19) cells and in subcultures of these cell lines that had acquired T790M mutations after 109 continuous EGFR-TKI treatment (H3255-TM (38) and PC9-TM (39)). In all cases, SOS1 110 deletion significantly diminished oncogenic transformation, whereas SOS2 deletion had variable 111 effects on transformation depending on the EGFR mutated cell line examined (Fig. 1D). These 112 data indicate that SOS1 is the major RasGEF responsible for oncogenesis downstream of 113 mutated EGFR. 114 BAY-293 was recently described as a specific inhibitor for SOS1 (33). To determine 115 whether SOS1 inhibition was similarly more effective in 3D spheroids over 2D adherent culture, 116 we assessed dose-dependent survival of H1975 cells after BAY-293 treatment under both 2D and 117 3D culture conditions (Fig. 1E). Similar to what we observed after either EGFR-TKI treatment 118 overall growth inhibition in 3D spheroids over 2D adherent cultures. To confirm the specificity 120 of BAY-293 for SOS1, we further treated 3D spheroid cultured H1975, PC9-TM, and H3255-121 TM cells where either SOS1 or SOS2 had been deleted versus NT controls with increasing doses 122 of BAY-293 ( Fig. 1F and Fig. S2 Previous studies reported that combining osimertinib with an alternative RTK inhibitor 134 may inhibit or treat the development of resistance driven by that specific RTK (7-9), whereas 135 simultaneous inhibition of multiple parallel RTKs with osimertinib may be required to 136 effectively potentiate osimertinib action (8). Further, while many studies show enhanced drug 137 activity in combination therapies versus osimertinib treatment alone, they do not assess whether 138 the effects of the 2-drug combinations are truly synergistic; synergistic interactions between 139 therapeutics allow for maximization of the therapeutic effect while minimizing adverse events 140 and may be required for effective therapeutic combinations with targeted agents (40). 141 SOS1 is a common downstream mediator of RTK signaling. We hypothesized that SOS1 142 could be an effective drug target to synergize with EGFR-TKI inhibition to treat EGFR-mutated 143 lung adenocarcinoma. To directly assess synergy between osimertinib and SOS1 inhibition, we 144 use two distinct methods based on the most widely established reference models of drug 145 additivity. The first method, isobologram analysis, assesses changes in the dose-response curves 146 for mixtures of two drugs compared to sham mixtures of each individual drug with itself. The 147 second method, Bliss independence analysis, assesses whether a mixture of two individual drug 148 doses has a greater effect than would be expected if the two drugs acted independently. We will 149 first describe and then use each method in turn to determine the whether SOS1 inhibition using 150 BAY-293 could synergize with the EGFR-TKI osimertinib in EGFR-mutated lung 151 adenocarcinoma cells. 152 Isobologram analysis is a dose-effect analysis based on the principle of Loewe additivity, 153 which states that a drug mixed with itself, and by extension a mixture of two or more similar 154 drugs, will show additive effects. For two drugs (Drug A and Drug B) that have parallel dose-155 response curves so that a constant potency ratio is maintained at all doses of A and B ( Fig. 2A) Bliss independence analysis is an effect-based analysis based on the principle of Bliss 180 additivity, which assumes that two drugs will act independently of each other so that their 181 combined effect can be assessed by assessing the effect of each drug sequentially (Fig. 2F). 182 Unlike isobologram analysis, this method does not require that two drugs being assessed have 183 parallel dose-response curves and can be calculated based as few as three drug treatments, the 184 effect each drug has on its own on the cell population, and the effect of combining the two drug 185 treatments together. By representing the effect of each drug treatment as a probabilistic outcome 186 between 0 (no effect) and 1 (100% effect), we can compare the observed effect of the drug-drug 187 combination to the expected effect if each drug acted independently (Fig. 2E). The ratio of the 188 expected effect to the observed effect is the Bliss Index (BI), where a BI < 1 indicates synergy 189 Overall, the data presented in Fig. 2 indicate that osimertinib and BAY-293 show significant 204 drug-drug synergy in EGFR-mutated H1975 cells, but only in 3D spheroid culture conditions. 205 To determine whether the SOS1 inhibitor BAY-293 could generally synergize with 206 EGFR-TKIs in EGFR-mutated lung adenocarcinoma cells, we extended our assessment of drug-207 drug synergy to isobologram analysis (Fig. 3) and Bliss independence analysis (Fig. 4) in six 208 different EGFR-mutated lung adenocarcinoma cell lines. In cells that were sensitive to first-209 generation EGFR-TKIs (HCC827, PC9, H3255; T790 wild-type), we assess drug-drug synergy 210 between BAY-293 and either a first-generation (gefitinib) or third-generation (osimertinib) 211 EGFR-TKI. In cells that were resistant to first-generation EGFR-TKIs (H1975; PC9-TM, 212 H3255-TM; T790M) we limited our assessment to synergy between BAY-293 and osimertinib. 213 To first determine the individual EC 50 values for gefitinib, osimertinib, and BAY-293 in each 214 cell line, cells were culture as 3D spheroids for 48-72 hours, and then treated with increasing 215 doses of drug for four days followed by assessment of cell viability by CellTiter-Glo (Fig. S3). 216 In five of six cell lines, the individual dose-response curves for BAY-293, osimertinib, and 217 gefitinib (where appropriate) showed similar maximal effects and Hill coefficients, and were 218 thus appropriate for linear isobologram analysis for each 2-drug combination of BAY-293, 219 osimertinib, and gefitinib (41). In contrast, H3255-TM cells were only moderately sensitive to 220 osimertinib, showing at most a 50% reduction in viability at high doses. Therefore, we limited 221 our assessment of drug-drug synergy in H3255-TM cells to Bliss independence analysis. 222 Further, to simplify our assessment of Bliss independence across multiple drugs and cell lines, 223 we limited our drug treatments to 1:2, 1:1, and 2:1 mixtures of each drug combination based on 224 dose equivalence (see Fig. 4A). 225 For each first-generation EGFR-TKI sensitive cell line (HCC827, PC9, H3255), gefitinib 226 and osimertinib did not show any synergy with each other by either isobologram analysis (Fig. 3) 227 or Bliss Independence analysis (Fig. 4), instead showing additive effects (CI and BI ~ 1) as 228 would be expected for two drugs with the same molecular target. In contrast, BAY-293 showed 229 significant synergy with gefitinib and osimertinib by both isobologram analysis (Fig. 3)

and Bliss 230
Independence analysis (Fig. 4), suggesting that SOS1 inhibition can act as a secondary treatment 231 for all EGFR-TKIs. Further, in all three T790M mutated cell lines (H1975, PC9-TM, H3255-232 TM), BAY-293 again showed synergy with osimertinib. These data suggest that combined 233 SOS1 and EGFR inhibition is a robust therapeutic combination that synergize to inhibit EGFR-234 mutated lung adenocarcinoma cell growth. 235 236

Synergy between BAY-293 and osimertinib is independent of SOS2 237
We showed that SOS2 deletion sensitized NCI-H1975 cells to the SOS1 inhibitor BAY-238 293 ( Fig 1F). We wanted to determine whether the synergy we observed between EGFR-and 239 SOS1-inhibition ( Fig. 3 and 4) was enhanced by SOS2 deletion in EGFR-mutated NSCLC cell

Assessment of Inhibitor Landscape in EGFR-mutated cells lines shows synergy upon 294 inhibition of upstream pathway effectors 295
Since the most common EGFR-independent resistance mechanisms involve reactivation of 296 RTK/RAS/effector pathways (7-10), we wanted to assess whether inhibition of different proteins 297 within the EGFR/RAS signaling pathway could synergize to inhibit 3D survival of EGFR 298 (T790M) mutated cancer cells. To determine drug-drug synergies after inhibition of EGFR-RAS 299 pathway signaling at different levels, we assessed synergy between osimertinib, inhibitors of 300 EGFR signaling intermediates of RAS (BAY-293 for SOS1 and RMC-4450 for SHP2), and 301 inhibitors of the Raf/MEK/ERK (trametinib) and PI3K/AKT (buparlisib) pathways (Fig. 7A). 302 H1975 and PC9-TM cells were treated with each individual inhibitor or 1:1 DEQ mixtures of 303 every drug-drug combination, and the combination index was calculated to assess drug-drug 304 synergy. Since 3255-TM cells are not suitable for isobologram analysis, these cells were treated 305 with full-dose mixtures based on dose equivalence and the Bliss Index was calculated for each 306 drug-drug combination (Fig. 7B). Intriguingly, all three cell lines showed drug-drug synergy 307 with any combination of EGFR, SOS1, and SHP2 inhibition. In contrast, inhibition of 308 downstream Raf/MEK/ERK or PI3K/AKT pathways failed to consistently synergize with either 309 osimertinib or any other inhibitor (Fig. 7B, top). These data support the premise that combined 310 vertical inhibition of proximal EGFR signaling may constitute an effective strategy to treat 311 EGFR-mutated lung adenocarcinomas. 312 SHP2 is important for the stabilization of the GRB2:SOS1/2 complexes on EGFR (42), 313 and the mechanism of allosteric SHP2 inhibitors depends on SOS1 (43), although the 314 contribution of SOS2 to SHP2 inhibitors was not assessed. To determine whether SOS2 deletion 315 altered the spectrum of drug-drug synergies in EGFR-mutated cells, parallel studies were 316 performed in EGFR-mutated cells where SOS2 was deleted (Fig. 7B, bottom). Unlike what we 317 observed for synergy between EGFR-and SOS1 inhibition, synergy between SOS1 and SHP2 318 inhibition was enhanced by SOS2 deletion. These data suggest that SOS2 plays a role in SHP2-319 dependent signaling. SOS1 inhibition also synergized with MEK inhibition in SOS2 KO cells. 320 Given the strong synergy between SOS1 inhibition and SOS2 deletion in inhibiting 321 Raf/MEK/ERK signaling (Fig. 6), these data suggest that deep inhibition of MEK signaling is 322 sufficient to inhibit survival in EGFR-mutated cells. 323 To further evaluate synergy between inhibitors of proximal EGFR signaling, we 324 examined combinations of EGFR-SOS1-and SPH2 inhibition both by expanded evaluation of 325 each 2-drug combination and by assessing whether combined inhibition of EGFR, SOS1, and 326 SHP2 would be more effective than two drug combinations of these inhibitors. To assess each 327 two-drug combination, H1975 cells cultured under 3D spheroid conditions were treated with 328 dose-equivalent combinations of osimertinib, BAY-293, and RMC-4550, assessed for cell 329 viability, and subjected to isobologram analysis to assess drug-drug synergy. Each two-drug 330 combination showed synergy at three different DEQ ratios (Fig. 7C), suggesting that inhibition 331 of any two proximal signaling proteins may be an effective therapeutic regimen to treat EGFR-332 mutated cancer. To assess whether adding a third proximal inhibitor to each two-drug 333 combination would further enhance synergistic inhibition of spheroid survival, each 2-drug 334 combination was mixed at 1:1 ratio, and then a third proximal pathway inhibitor was added to 335 give the indicated 3-drug mixtures (Fig. 7D). Isobologram analysis of these three drug mixtures 336 revealed that addition of a third proximal pathway inhibitor to any 2-drug combination of 337 osimertinib, BAY-293, and RMC-4550 further enhanced synergy above what was observed for 338 each 2-drug combination (Fig. 7D) Finally, comparing the combination index for the three-drug 339 combination at a 1:1:1 ratio when each drug is treated independently versus the two-drug 340 combinations showed marked synergy for the three drug combination, but that this synergy was 341 not significantly enhanced compared to the combination of osimertinib and BAY-293 (Fig. 7E). 342 These data indicate that vertical inhibition of proximal EGFR signaling with the combination of 343 osimertinib and a SOS1 inhibitor may be the most the most effective therapeutic combination to 344 treat EGFR-mutated NSCLC. 345 346

Discussion 347
Activating EGFR mutations are found in 10-15% of lung adenocarcinomas and are the major 348 cause of lung cancer in never smokers. The third-generation EGFR-TKI osimertinib enhances 349 both progression-free (44) and overall survival (45) compared to first generation EGFR-TKIs 350 and is now considered first-line treatment in EGFR-mutated NSCLC. Osimertinib resistance 351 often develops via activation of parallel RTK pathways (7-9), and broad inhibition RTK 352 signaling may enhance osimertinib efficacy and delay therapeutic resistance. Here, we 353 demonstrate that inhibition of the common RTK signaling intermediate SOS1 using  showed marked synergy with osimertinib in 3D spheroid-cultured EGFR-mutated NSCLC cells. 355 Our observations that (i) osimertinib-BAY-293 synergy was only observed in 3D spheroids but 356 not in adherent (2D) cultures and (ii) synergy between RTK-signaling intermediates and 357 osimertinib was not broadly applicable EGFR downstream signaling components but was limited 358 to proteins upstream of RAS reveal novel insights into pharmacologic studies assessing 359 therapeutics designed to treat NSCLC. 360 While most studies designed to identify or test therapeutic targets to treat cancer are done 361 in 2D adherent culture, a growing body of evidence suggests that pharmacologic assessment of 362 novel therapeutics must be performed in 3D culture systems (34). This is particularly true for 363 NSCLC, where multiple studies have now revealed the importance of 3D culture systems in 364 order to recapitulate in vivo findings. EGFR-mutated cells show differential RTK expression and 365 phosphorylation in 3D versus 2D conditions (27) and respond more robustly to EGFR-TKIs in 366 3D cultures compared to 2D settings ( Fig. 1 and (28)); KRAS-mutated cell lines deemed 367 "KRAS-independent" in 2D culture (46-50) still require KRAS for anchorage-independent 368 growth (51-54), and some KRAS G12C -mutated NSCLC cell lines respond to KRAS(G12C) 369 inhibitors in 3D culture and in vivo but not in 2D adherent culture (35). The relevance of 3D 370 culture systems extends to the identification of novel therapeutic targets and therapeutic 371 combinations. We recently showed that SOS2 is specifically required for PI3K-dependent 372 protection from anoikis in KRAS-mutated NSCLC cells (32) and SOS2 deletion synergizes with 373 MEK inhibition to kill KRAS mutated cells only under 3D culture conditions (31). Here, we 374 show marked synergy between vertical inhibition of EGFR and SOS1 in EGFR mutated cancer 375 cells, but only under 3D culture conditions (Fig. 2). CRISPR screens performed in spheroid 376 cultures of KRAS-and EGFR-mutated NSCLC cell lines more accurately reproduce in vivo 377 findings and identify drivers of oncogenic growth compared to screens performed in 2D cultures 378 (55). Intriguingly, in this study SOS1 was essential for 3D spheroid survival but not 2D spheroid 379 growth of both EGFR-and KRAS-mutated cells. These data are in complete agreement with our 380 data from Fig. 1 showing the requirement for SOS1 in 3D transformation but not 2D 381 proliferation, and support our conclusion that SOS1 is an important therapeutic target in EGFR-382 mutated NSCLC. These data suggest that future studies assessing novel therapeutics to treat 383 lung adenocarcinomas must be performed in a 3D setting. 384 Osimertinib resistant can occur via oncogenic shift to alternative RTKs including c-MET 385 (11), HER2 and/or HER3 (7-9), IGF1R (12), and AXL (13-16). The variety of RTK bypass 386 pathways that can lead to osimertinib resistance suggests that broad inhibition of RTK signaling 387 may be a more effective therapeutic strategy than any individual RTK inhibitor to limit 388 osimertinib resistance, whereas once resistance via oncogenic shift to an alternative RTK occurs Here, we show that osimertinib does not broadly synergize with inhibitors of downstream 407 EGFR/RAS/RAS effector signaling. Instead, we found that synergy was limited to combinations 408 of osimertinib with inhibitors of proximal EGFR signaling intermediates SOS1 and SHP2 (Fig.  409   7). Further, SOS1 inhibition significantly enhanced osimertinib-dependent inhibition of both 410 Raf/MEK/ERK and PI3K/AKT signaling (Fig. 6), whereas inhibition of individual downstream 411 Raf/MEK/ERK or PI3K/AKT effector pathways did not synergize with osimertinib (Fig. 7) to 412 inhibit 3D spheroid growth. We hypothesize that these two findings are inexorably linked, so 413 that any potential therapeutic must synergize with osimertinib to inhibit all downstream RAS 414 effector signaling to show drug-drug synergy in 3D culture. In support of this idea, previous 415 studies showed inhibition of SRC family kinases (SFK) potentiated osimertinib to a much greater 416 extent than either MEK or PI3K inhibition (20), and that SFK inhibition synergized with 417 osimertinib to inhibit both Raf/MEK/ERK and PI3K/AKT signaling (20, 62). 418 Overall, our data suggest that inhibitors of proximal signaling may be the most 419 efficacious therapeutics to combine with osimertinib to treat EGFR-mutated tumors. Toward  As an example, we will show data for one trial analyzing the combination of osimertinib 552 and BAY-293 in 3D spheroid cultured H1975 cells in Fig. 2B Fig. 2, wells were 572 treated with a full dose of each individual drug or drug combination in a 10  10 matrix of dose 573 combinations for osimertinib and BAY-293 on a 1/3-log scale. Data were normalized to the 574 maximum luminescence reading of untreated cells, and a heat-map depicting cell viability was 575 generated using Prism 8. The Bliss index was calculated by first converting viability (on a scale 576 of 0 to 1) for each treatment to the effect of each drug or drug combination, where 0 represents 577 no effect and 1 represents 100% effect (no viable cells). 578

effect = 1 -viability 579
From the effect data, the expected effect for each drug combination is calculated: 580

Expected effect = E A + E B -E A * E B 582
The Bliss Index is the ratio of the expected effect / actual effect: 583 Bliss Index = (expected effect) / (actual effect) 584

Bliss Index = (E A + E B -E A * E B ) / (E A+B MIX ) 585
A Bliss Index of 1 indicates that the actual and expected effects are equivalent, and the effects of 586 the two drugs are additive. Bliss Index < 1 indicates increasing synergy, whereas Bliss Index >1 587 indicates antagonism. 588 Excess over Bliss is calculated by determining how much greater the actual effect of the drug 589 combination is versus the expected effect, and is calculated as: -10 -9.67 -9.33 -9 -8.67 -8.33 -8 BAY-293 -8 -7.67 -7.33 -7 -6.67 -6.33 -6 -5.67 -5.33 -5 RMC-4550 -9 -8.67 -8.33 -8 -7.67 -7.33 -7 -6.67 -6.33 -6 610 Each two-drug combination was set as a single "drug mixture" at a 1:1 ratio, and the third drug 611 was combined with this drug mixture at 2:1, 1:1, and 1:2 drug ratios. To generate the proper two 612 and three-drug mixtures for analysis, 21 total dose response curves were generated. The five 613 dose-response curves on the right represent the mixtures used to generate the isobologram plots 614 in Fig. 7D. The other two two-drug mixtures in bold (2-drug 2:1 and 1:2 mixtures) were used to 615 generate the isobologram plots in Fig. 7CCombination

632
To calculate the three-drug combination index where each drug was considered independently 633 (Fig. 7E)