Fig 2: In vitro cell growth inhibitory activity of ADCA, Cancer cells were treated with humanized GPC1-ADC(MMAE) or human IgG4-ADC (a control-ADC[MMAE]) for 144 h. The humanized GPC1-ADC(MMAE) significantly inhibited the growth of the GPC1-positive cancer cell lines compared with the control-ADC(MMAE). In the GPC1-negative BxPC-3-GPC1 and TE8-GPC1 KO cell lines, neither treatment had any inhibitory effect. B, Induction of G2/M phase cell cycle arrest in BxPC-3, KP-2, PK-8, TE-8, TE-14, and KYSE70 cells treated with humanized GPC1-ADC(MMAE). Cells were treated with either 16 nM control-ADC(MMAE) or 16 nM humanized GPC1-ADC(MMAE). After 24 h, cell cycle analysis was performed by flow cytometry with propidium iodide DNA staining. **P < 0.01, by one-way ANOVA, followed by Dunnett's post hoc test. C, Induction of apoptosis in BxPC-3, KP-2, PK-8, TE-8, TE-14, and KYSE70 cells treated with 16 nM control-ADC(MMAE) or 16 nM humanized GPC1-ADC(MMAE). Caspase 3/7 activation was monitored using the Caspase Glo 3/7 assay system at 48 h after addition of ADCs. Caspase 3/7 activation relative to untreated cells was detected in these cells treated with increasing concentrations of humanized GPC1–ADC. *P < 0.05 and **P < 0.01, by Student's t test. D, Western blot analysis of proteins involved in the apoptosis pathways. BxPC-3 and TE-8 cells were treated with either 16 nM control-ADC(MMAE) or 16 nM humanized GPC1-ADC(MMAE). After 48 h, cells were harvested and lysed. The protein level of poly ADP-ribose polymerase (PARP), cleaved PARP, and Cleaved Caspase-3 was detected by western blot; ß-actin was used as a loading control.

Fig 3: a3(V) and GPC1 are associated with basal and luminal cells, respectively, in normal breast, but co-localize in human tumours.(a) Representative immunofluorescence co-localization of GPC1 (green) with marker K8 (red) in luminal cells, but not with marker K14 (red) or with a3(V) (red) in basal cells, in normal human mammary ducts. Arrowheads and arrows denote luminal and basal cells, respectively. (b,c) Representative co-localization of GPC1 (green) with basal marker K14 (red) and with a3(V) (red) in human triple negative, basal-like tumours (b), and with K8-luminal marker (red) and a3(V) (red) in human luminal A tumours (c). Note that GPC1 appeared to be consistently lower in basal-like than in luminal tumours. White scale bars, 50 µm.

Fig 4: Effects of treatment with recombinant a3(V)-NTD sequences and anti-a3(V)-NTD antibodies on in vitro and in vivo tumour cell growth.Cultured KO/PyMT cells were treated with recombinant a3(V)-NTD sequences prepared in mammalian cells, followed by monitoring of growth by MTT assay (a), or by immunoblotting for cyclin D, CDKs 2 and 6, p-H3Ser10, phospho-Erk1/2 (p-Erk1/2), Erk1/2, phospho-FGFR2 (p-FGFR2), and Ras; and for activated GTP-Ras pulled-down from lysate with Raf-1 RBD domain (b). Cultured WT/PyMT cells were treated with antibodies raised against recombinant a3(V)-NTD sequences prepared in mammalian cells (M) or E. coli (E), followed by monitoring of growth by MTT assay (c), or by immunoblotting for cyclin D, CDKs 2 and 6 and p-H3Ser10, phospho-Erk1/2 (p-Erk1/2), Erk1/2, phospho-FGFR2 (p-FGFR2), and Ras; and for activated GTP-Ras pulled-down from lysate with Raf-1 RBD domain (d). Immunoblotting for a-tubulin was a loading control. (e) tumour volumes were measured in C57BL/6 mice injected with WT/PyMT cells and then, 7 days later, with anti-a3(V)-NTD (M) antibodies or with non-immune control IgG. Cell proliferation assays were performed in triplicate. Mice injected with a3(V)-NTD antibody or IgG were n=8 per group. Data are presented as mean±s.d. P values: *<0.05, **<0.01, ***<0.005. Statistical analysis was via two-tailed Student's t-test, with differences considered significant at P<0.05. Immunoblots were repeated three times, from three independent cell lysates. Representative blots are shown. Quantification of results for significance of differences for b,d is shown in Supplementary Fig. 6. (f) A model in which the acidic PARP and basic variable (VAR) a3(V) subdomains bind basic FGF2 (F2) and acidic GPC1 heparan sulfate (HS) GAG chains, respectively. Via these interactions a3(V) may act as a linker that enhances the efficiency of signalling by HSPG-binding growth factors, which in PyMT cells act via Ras to activate MAPK signalling and the cell cycle.

Fig 5: Structure, characterization, and internalization activity of the humanized GPC1-ADCA, Flow cytometric analysis of GPC1 expression. The shaded histogram profile indicates the isotype control, and the open histogram indicates the humanized anti-GPC1 antibody-staining results. B, Structure of the humanized GPC1-ADC consisting of the humanized anti-GPC1 antibody (clone T2) conjugated to the MMAE payload. C, BxPC-3 cells are incubated with unconjugated humanized anti-GPC1 mAb (blue closed circles) or humanized GPC1-ADC (red closed triangles). The mean fluorescence intensity at various concentrations is shown. D, Time-course analysis of the internalization activity of the antibody or ADC in BxPC-3 cells. The left panel shows unconjugated humanized anti-GPC1 mAb (clone T2); and the right panel, humanized GPC1-ADC. E, GPC1-ADC internalizes and locates in the lysosomes of BxPC-3 cells. Cell surface and intracellular GPC1 visualized on fluorescence microscopy. Green indicates the humanized anti-GPC1 antibody (clone T2) or humanized GPC1-ADC; red, the lysosomal marker LAMP-1; and blue, 4',6-diamidino-2-phenylindole-stained DNA. Scale bar: 10 mm. (Color version of figure is available online)