Multiple proteases are involved in mesothelin shedding by cancer cells

Abstract

Mesothelin (MSLN) is a lineage restricted cell surface protein expressed in about 30% of human cancers and high MSLN expression is associated with poor survival in several different cancers. The restricted expression of MSLN in normal tissue and its frequent expression in cancers make MSLN an excellent target for antibody-based therapies. Many clinical trials with agents targeting MSLN have been carried out but to date none of these agents have produced enough responses to obtain FDA approval. MSLN shedding is an important factor that may contribute to the failure of these therapies, because shed MSLN acts as a decoy receptor and allows release of antibodies bound to cell-surface MSLN. We have investigated the mechanism of shedding and show here that members of the ADAM, MMP and BACE families of proteases all participate in shedding, that more than one protease can produce shedding in the same cell, and that inhibition of shedding greatly enhances killing of cells by an immunotoxin targeting MSLN. Our data indicates that controlling MSLN shedding could greatly increase the activity of therapies that target MSLN.

Fig. 1. Identification of C-terminal sequence of MSLN.

a MSLN was immunoprecipitated from cell culture media using mouse monoclonal Ab MN. The areas labeled MSLN were cut out and analyzed by LC-MS. A431 cells do not express MSLN. KB for KB31, KLM for KLM1; RH for RH16; OV for OVCAR8; H9 for A431/H9, an MSLN stably transfected line. IgG H, heavy chain. b C-terminal sequences of MSLN from various cell lines. A431/H9 sequences are from published data. Arrows indicate cut sites.

Fig. 2. Knockdown of ADAM10 and ADAM17 decrease MSLN shedding in KLM1 and OVCAR8 cells, but not in KB31.

a Cells were transfected with siRNAs and lysates were analyzed by western blot using anti-ADAM17 Ab. Actin is a loading control (full blots shown in Supplementary Fig. 1). b Shed MSLN in cell culture media 48–72 h post-transfection from samples in a (P = 0.018 and 0.022, n = 3). c, d ADAM10 siRNA or control were transfected into KLM1 and OVCAR8 cells. ADAM10 RNA was analyzed by real-time PCR after 48 h (c) and shed MSLN was measured 48–72 h post-transfection (d KLM1, P = 0.034; OVCAR8, P = 0.0006, n = 3). e, f Shed MSLN in KLM1 (e P = 0.0146, 0.00037, and 0.00007, n = 3) or OVCAR8 (f P = 0.0014, 0.00013, 0.00006, n = 3) culture media was analyzed 48–72 h post-transfection of control, ADAM10, ADAM17, and double knockdown siRNAs. The shed MSLN levels were normalized by WST-8 assay for cell growth. Statistical significance is noted by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

Fig. 3. Knockdown of MMP15 decreased MSLN shedding in KB31 cells but not in KLM1 cells.

a KB31 cells were transfected with siRNA MMP15. After 48 h, the RNA was analyzed by real-time PCR. b Left, shed MSLN was measured 72 h after siRNA siMMP15_1 transfection in KB31 (**P = 0.009, n = 4). Right, shed MSLN was measured 48 h after siRNA siMMP15_2 (*P = 0.043, n = 5). c, d KLM1 cells were transfected with siRNA of siMMP14 and siMMP15 (n = 5). The relative RNA level (c) and shed MSLN (d) were similarly analyzed as in a and b.

Fig. 4. Knockdown of BACE lowers MLSN shedding in KB31 cells.

a KB31 cells were transfected with two different siRNA oligos targeting BACE2; after 72 h, cell lysates were analyzed by western blotting using anti-BACE2 antibody. b Cells were treated as in a and blotted with anti-MSLN antibody. Anti-actin is a loading control (full blots shown in Supplementary Fig. 7). c Cells were transfected with BACE2 siRNA, 72 h later labeled with anti-MSLN-Alexa647, and analyzed by flow cytometry. d Growth medium collected 48–72 h post-transfection of BACE2 siRNA was assayed for MSLN (n = 3, *P = 0.03). e Real-time PCR analysis of RNA levels 48 h after transfection of siBACE1 and siBACE2. f Shed MSLN level after transfection of BACE1 and BACE2 measured 48–72 h post-transfection (n = 6, ****P < 0.00001).

Fig. 5. Knockdown of BACE2 in MS751, T3M4, and AsPC1 cell lines decreases MSLN shedding.

a Seventy-two hours after transfection of BACE2 siRNA into MS751, T3M4, and AsPC1, cell lysates were analyzed by western blotting using anti-BACE2 antibody (full blots shown in Supplementary Fig. 8). b Shed MSLN was collected 48–72 h after BACE2 siRNA. MS751(n = 3, P < 0.00001), T3M4 (n = 3, P = 0.0031), AsPC1 (n = 3, P = 0.01). Statistical significance compared to a negative control siRNA is noted by asterisks (*P < 0.05, **P < 0.01, ****P < 0.0001).

Fig. 6. Inhibitors of BACE, ADAM, and MMP lower MSLN shedding.

a Twenty micromolar of LY2886721 or Lanabecestat were incubated with KB31, MS751, T3M4, AsPC1, or KLM1 cells in 96-well plates for 48 h. MSLN in culture media was measured (KB31, both P < 0.0001; KLM1, P = 0.77 and 0.38; T3M4, P = 0.0006 and 0.031; OVCAR8, P = 0.118 and 0.987; MS751, P < 0.0001 and 0.009; AsPC1, both P < 0.0001, n = 3). b In all, 10 µM TMI-1 or 10 µM Marimastat were incubated with cells and shed MSLN measured (TMI-1, P = 0.084, <0.0001, 0.002, 0.00014; Marimastat: P = 0.0002, 0.0003, <0.0001, <0.0001, n = 3). c–f KB31 cells (c, d) were treated with 10 µM Lanabecestat or 10 µM Marimastat or both, OVCAR8 (e, f) were treated with TMI-1 or Marimastat or both. Shed MSLN level are shown (c P = 0.0005, <0.0001, <0.0001; e P = 0.002, 0.0023, 0.0027, n = 3). KB31 (d) or OVCAR8 (f) were incubated with inhibitors for 20 h, then SS1P at indicated concentration was added for 72 h. Cell viability was measured by WST-8 assay. Lana Lanabecestat, Mari Marimastat (ns not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).