(A) pIC50 plots for PXS-S1A and BAPN against recombinant LOXL2 protein and (B) native human LOX enzyme showing a very similar activity and selectivity profile (C) pIC50 plots for the modified second generation PXS-S2A and BAPN against recombinant LOXL2 protein and (D) native human LOX enzyme showing significantly enhanced LOXL2 selectivity.

(A) Confluency (as % of control) of MDA-MB-231 cells at 96 hours post seeding, in the presence of escalating doses of LOXL2 inhibitors PXS-S1A and PXS-S2A, showing dose dependent inhibition of proliferation. (n = 9 measurements over 3 independent experiments). (B) Absorption at 490nm (MTS Assay) of MDA-MB-231 cells in 1.8mg/mL 3D collagen I matrices at day 8 (measured as % of control), showing dose dependent inhibition of proliferation in response to LOXL2 inhibition (n = 9 measurements over 3 independent experiments) (C) Workflow and representative images for 2D scratch wound closure scoring. Using in house ImageJ scripts, first the wound edge is identified, before a mask is applied and the scratch area calculated. (D) LOXL2 inhibition leads to a dose-dependent decrease in scratch wound closure (plotted as % of control wound closure) (n = 12 wounds over 3 independent experiments) (E) Workflow and representative images for 3D spheroid invasion scoring. Using in house ImageJ scripts, a mask is generated of cells and the invaded areas calculated. (F) LOXL2 inhibition leads to a dose-dependent decrease in 3D spheroid invasion (plotted as % of control spheroid invasion) (n = 12 spheroids over 3 independent experiments).

(A) qRT-PCR confirms transcriptional knockdown of the LOXL2 gene with 2 independent hairpins (shLOXL2 #1 and shLOXL2 #5) compared to control (shCon) (n = 3 independent experimental repeats). (B) Western blotting for LOXL2 protein confirms a reduction in expression levels in hairpins shLOXL2 #1 and shLOXL2 #5 compared to control (shCon). (C) shLOXL2 cells show reduced proliferation (MTS assay) in 2D on plastic compared to control shCon (n = 9 measurements over 3 independent experiments). (D) In 3D collagen I matrices, shLOXL2 cells show reduced proliferation (measured by cell count) over time (n = 9 measurements over 3 independent experiments). (E) 3D spheroid invasion assays; shLOXL2 cells show a significantly reduced invasive index (normalized to initial spheroid size [dotted line]) compared to control (shCon) (n = 9 spheroids from 3 independent experiments).

(A) Treatment of orthotopic MDA-MB-231 human breast cancers in mice with PXS-S1A and (B) PXS-S2B significantly delays tumor growth and reduces tumor burden. (n = 2 independent experiments; 3 mice) (C) Stable silencing of LOXL2 in MDA-MB-231 human breast cancer cells leads to delayed tumor growth and reduced tumor burden in orthotopic models (n = 2 independent experiments; 3 mice).

(A) Immunohistochemistry for Ki67 shows limited effect of the PXS-S1A LOXL2 inhibitor, and a small but significant decrease in Ki67 score by treatment with the PXS-S2B LOXL2 inhibitor in vivo (n = 2 independent experiments) (B) Representative images of Ki67 immunohistochemistry in tumors from control and PXS inhibitor treated mice. (C) Conversely, LOXL2 inhibition does not directly induce apoptosis in cells (PXS-S2A), whereas dual inhibition of LOX and LOXL2 (PXS-S1A) shows a significant increase in apoptosis. (n = 2 independent experiments). (D) Representative images of cleaved caspase-3 immunohistochemistry in tumors from control and PXS inhibitor treated mice. (E) Confirmation of findings in shLOXL2 tumors for Ki67 expression shows a significant decrease in tumors Ki67 score in vivo following LOXL2 silencing (n = 2 independent experiments) (F) Representative images of Ki67 immunohistochemistry in tumors from shControl and shLOXL2 implanted mice. (G) Genetic silencing of LOXL2 leads to higher apoptosis in tumors as evaluated by cleaved-caspase-3 staining (n = 2 independent experiments). (H) Representative images of cleaved caspase-3 immunohistochemistry in tumors from shControl and shLOXL2 implanted mice. Scale = 150 μm.

(A) Inhibition of LOXL2 with PXS-S1A and PXS-S2B leads to a significant decrease in vessel density in primary breast tumors in vivo (measured by endomucin positive cells). (n = 2 independent experiments) (B) Representative images of endomucin immunohistochemistry in tumors from control and PXS inhibitor treated mice (C) The LOXL2 dependent reduction in vessel density was confirmed in our stable shLOXL2 lines which both show a significant decrease in endomucin positive vessels. (D) Representative images of endomucin immunohistochemistry in tumors from shControl and shLOXL2 implanted mice (E) Inhibition of LOXL2 using both the PXS-S1A and PXS-S2B compounds leads to a significant decrease in the number of activated αSMA-positive cells in primary tumors (F) Representative images of αSMA immunohistochemistry in tumors from control and PXS inhibitor treated mice. (G) Silencing of LOXL2 in MDA-MB-231 tumor cells leads to a significant decrease in the number of activated αSMA-positive cells within the primary tumors (H) Representative images of αSMA immunohistochemistry in tumors from shControl and shLOXL2 tumor bearing mice. Scale = 150 μm.

(A) Treatment with PXS-S1A showed a significant reduction in pulmonary metastatic burden, but not in PXS-S2B inhibitor treated mice (n = 2 independent experiments) (B) Representative H&E images of pulmonary metastatic burden in control and PXS inhibitor treated mice (C) Similarly, inhibition of LOXL2 with PXS-S1A showed a significant reduction in hepatic metastatic burden but this was not significant in the PXS-S2B treated mice (n = 2 independent experiments). (D) Representative H&E images of pulmonary hepatic burden in control and PXS inhibitor treated mice. (E) In contrast, genetic silencing of LOXL2 led to a significant reduction in pulmonary metastatic burden in shLOXL2 tumor bearing mice compared to shCon tumor bearing mice (F) Representative H&E images of lungs from shCon and shLOXL2 tumor implanted mice. Scale = 150 μm.