Fig 2: Group I Paks are activated on centrosomes downstream of Tiam1.(a) MDCK II cells with dox-inducible Tiam1 shRNA were grown for 4 days with (plus dox) or without dox (control) (for representative blot see Supplementary Fig. 4g), fixed and stained for IF. Merge: phospho-Pak1/2 (P*Pak)=green, ?-tubulin=red and DAPI=blue. Scale bars, 5 µm. Graph shows average centrosomal P*Pak levels for plus dox normalized to control in prophase (n=9) and prometaphase (n=5) (>40 centrosomes/rep). (b) MDCK II cell lines expressing dox-inducible Tiam1 shRNA and the indicated RNAi-resistant Tiam1 constructs were treated for 4 days with or without dox as indicated (blots for these cell lines are shown in Fig. 3b), fixed and stained for IF using antibodies against P*Pak1/2, ?-tubulin, HA and DAPI. Prophase cells with positive HA signal were analysed for centrosomal P*Pak as in a (n=4) (>40 centrosomes/rep). (c) MDCK II cells were control treated or treated with Rac1 inhibitor NSC 23766, fixed and stained for IF using antibodies against P*Pak1/2, ?-tubulin and DAPI. Centrosomal P*Pak in prophase cells was analysed as in a (n=3, >40 centrosomes/rep). (d) MDCK II cell lines expressing dox-inducible Tiam1 shRNA and the indicated RNAi-resistant Tiam1 constructs were treated for 4 days with or without dox as indicated (blots for these cell lines are shown in Supplementary Fig. 4d). Cells were fixed, stained and analysed for centrosomal P*Pak in prophase as in b (n=3, >40 centrosomes/rep). (e) HEK293T cells expressing GFP only (EV) or Tiam1-GFP (WT or phospho-mutant forms as indicated) were arrested in mitosis by STLC treatment, then lysates subjected to GFP-trap. Following SDS–PAGE, interaction with Pak2 and Rac was analysed by immunoblotting. All graphs show mean+s.e.m. Paired two-sided t-test: *P<0.05, **P<0.01.

Fig 3: TIAM1 upregulates chemoresistance in CRC xenograft.a The western blot analysis illustrates the protein expression of TIAM1 in the HCT116 cell line stably-transfected by sgRNA against TIAM1 or the corresponding control using CRISPR/Cas9. The relative expression values normalized to reference protein are shown in each image. b Overall workflow of in vivo experiments. c, d Effect of TIAM1 knockdown in HCT116 cells on the xenograft model was assessed by evaluating tumor volume (c) and weight (d) compared to controls. e The bar graphs illustrate mRNA expression of TIAM1, NANOG, OCT-4, and ALDH in CRC tumors transfected with sgRNA for TIAM1 and the corresponding control. Error bars are presented as mean ± SE of three independent experiments, ***P < 0.001 (Mann–Whitney U test)

Fig 4: Functional classification of proteins with Tiam1 GEF*-specific changes in Rac1 binding. (A) Venn diagram comparing proteins with increased Rac1 binding in = 2 SILAC SF-TAP experiments upon expression of indicated GEF constructs. Tiam1 GEF-dead mutant (GEF*)-specific proteins are outlined. (B) Venn diagram comparing proteins with decreased Rac1 binding in = 2 SILAC SF-TAP experiments upon expression of indicated GEF constructs. Tiam1 GEF*-specific proteins are outlined. For A and B the associated tables show clustering of the Tiam1 GEF*-specific proteins with increased or decreased Rac1 binding, respectively, according to their cellular functions based on Ingenuity Integrated Pathway Analysis (IPA). Full protein names and SILAC ratios are outlined in Supplementary File 1. Proteins in highlighted categories were presented as part of a heat map in our recent publication. 18

Fig 5: TMOD3 is a novel P-Rex1-enriched Rac1 interactor that binds to FLII in a P-Rex1-dependent manner. (A) Venn diagram comparing proteins that show increased Rac1 binding under P-Rex1 Wild Type (WT) and P-Rex1 GEF-dead mutant (GEF*) together with decreased Rac1 binding under Tiam1 WT expression in = 2 SILAC SF-TAP experiments. TMOD3 is the only protein that exhibits opposing Rac1 binding patterns upon P-Rex1 WT/GEF* and Tiam1 WT expression. (B) StrepII-FLAG Tandem Affinity Purification (SF-TAP) of SF-Rac1 from NIH3T3 cells treated with ethanol (- dox) or 1 µg ml -1 doxycycline (+ dox) to induce expression of SF-Rac1. Co-precipitated endogenous TMOD3 was detected by protein gel analysis. a-Tubulin was used as a loading control. (C) Representative immunofluorescence images of MCF7 cells subjected to the Duolink in situ Proximity Ligation Assay (PLA) following treatment with 1 µg ml -1 dox (+ dox) for 24 hours to induce expression of indicated GEF constructs. Phalloidin and DAPI were used to visualize the actin cytoskeleton and nuclei, respectively. Scale bar= 20 µm. (D) Quantification of the average Duolink PLA signal from indicated number of MCF7 cells described in C ± s.e.m. Student's t-test was performed to determine statistical significance and p-values are shown on graph. NS= non-significant; ***= p = 0.001. (E) Ingenuity protein-protein network cluster of proteins showing increased Rac1 binding upon expression of P-Rex1 WT in = 2 SILAC SF-TAP experiments. Network shown represents one of the generated networks and displays relationships between proteins identified in the screen (*) with other proteins from the Ingenuity database that have similar functions. Additional relationships (red lines) were incorporated in the network based on information from the literature. Bold lines were used to highlight TMOD3 protein interactions and relationships. (F) Endogenous FLII immunoprecipitation (IP) from NIH3T3 cells transduced with the dox-inducible P-Rex1 WT expression system following treatment with ethanol (- dox) or 1 µg ml -1 dox (+ dox). Co-precipitated endogenous TMOD3 was detected by protein gel blot analysis. a-Tubulin was used as a loading control. Quantification of FLII-bound TMOD3 was assessed using the ImageJ software and normalized to a-Tubulin and total TMOD3 levels in the input. The normalized integrated density detected in the – dox FLII IP sample was set as one highlighting a 2-fold increase in FLII-TMOD3 binding upon expression of P-Rex1 WT as indicated.