Ascribing gain, loss, or lack of function to the spontaneous PCSK5 M452I mutation of MCF10DCIS cells, a common engineered model of breast ductal carcinoma in situ.

Introduction:: The MCF10A breast epithelial cell line and its transformed derivatives are widely used in the field of cancer bioengineering. The MCF10DCIS derivative is a particularly valuable experimental model of ductal carcinoma in situ (DCIS) for its ability to form lesions in mice that resemble clinical DCIS [1-2]. This derivative arose by serial in vivo passaging and selection for DCIS-like phenotypes that coincided with the acquisition of a heterozygous mutation in the proprotein convertase, PCSK5 [3]. PCSK5 is a serine endoprotease that activates by autocatalytic cleavage and processes proproteins to their mature active forms before secretion. The most recognized PCSK5 substrate is GDF11, a TGFβ-family ligand that promotes pro-epithelial and anti-invasive signaling in basal-like breast cancer [4], a subtype with similarities to MCF10A. The PCSK5 peptidase domain contains three active-site residues that together form a catalytic triad charge-relay system. The MCF10DCIS mutation, PCSK5M452I, is located at the end of the peptidase domain and therefore could influence enzymatic activity positively or negatively. PCSK5M452I is not recorded in any clinical database, and computational predictions are ambiguous about its effect, if any, on function. Here, we describe an ongoing detailed investigation of the PCSK5M452I allele and its associated DCIS-like phenotypes with the objective of determining whether mutation and phenotype are linked. The results of this work have implications for the interpretation of hundreds of studies in the literature using MCF10DCIS.


Materials and Methods:: Plasmids—Site-directed mutagenesis was used to generate PCSK5M452I or PCSK5T288P variants that were recombined into pLX304 for C-terminal V5 epitope tagging and constitutive expression. sgRNA-resistant V5-tagged PCSK5 was similarly prepared by introducing three silent mutations and recombining into pSLIK for doxycycline-inducible expression.

Antibodies—Rabbit polyclonal antisera was raised against PCSK5 amino acids 604-619, affinity purified, and validated by Western blot of cells overexpressing V5-tagged PCSK5.

PCSK5 Activity Assay—pLX302 GDF11-V5 and pLX304 PCSK5-V5 variants were transfected into 293T cells and conditioned medium and protein lysate were collected after 24 hours. GDF11 concentration was measured by ELISA and PCSK5 was measured by Western blot.

PCSK5 targeting—sgRNA designed to target upstream of the first site of the PCSK5 catalytic triad was cloned into EDCPV [5]. MCF10DCIS cells were tranduced and cultured in 2% Matrigel for 12 days supplemented with 10 ng/ml GDF11 and Shield to stabilize Cas9. Single Venus-positive cells were isolated, expanded, and genotyped after TOPO cloning and sequencing the sgRNA-targeted exon.

PCSK5 Guide-Resistant Add-Back—PCSK5–/– MCF10DCIS cells were transduced with sgRNA-resistant PCSK5M452I or PCSK5T288P at limiting viral titer and selected with hygromycin until control plates were clear.

Results, Conclusions, and Discussions:: We confirmed expression of PCSK5WT and PCSK5M452I alleles in MCF10DCIS cells by RT-PCR. To evaluate the enzymatic activity of PCSK5 variants, we co-expressed GDF11 with either PCSK5WT, PCSK5M452I, or a cancer-associated PCSK5T288P variant that is catalytically null [4]. Conditioned medium was collected to measure mature secreted GDF11 by ELISA (Figure 1A). GDF11 was undetectable in cells transfected only with PCSK5WT, and PCSK5WT increased GDF11 secretion fivefold when co-transfected, indicating negligible contributions of endogenous GDF11 or PCSK5 in these cells. Cells expressing PCSK5T288P secreted GDF11 comparably to cells lacking ectopic PCSK5, confirming specificity of the assay. Compared to PCSK5WT, secreted GDF11 was slightly but significantly reduced in cells expressing PCSK5M452I, suggesting that the M452I mutation is mildly hypomorphic. Results were similar when cells were co-transfected with oncogenic HRAS or stimulated with oxidative, electrophilic, or proteotoxic stressors. The differences in GDF11 release were consistent with trends in relative abundance of the autocatalytically mature form of PCSK5 compared to its unglycosylated and glycosylated proforms measured by immunoblotting (Figure 1B).

Using a custom antibody, we found that PCSK5 protein in MCF10DCIS cells was below the detection limit of 16,000 copies per cell, although PCSK5 became detectable when cells were cultured with basement membrane ECM. To assess function of individual PCSK5 variants in MCF10DCIS cells, we are adopting a knock-out and allele-specific addback approach. PCSK5-targeting sgRNA was designed to cause a premature stop upstream of the first catalytic site, and PCSK5–/– clones were isolated with GDF11 supplementation to minimize compensatory adaptations. When grown as 3D spheroids without GDF11, PCSK5–/– cells were less compact and more invasive (Figure 2), reminiscent of when GDF11 is acutely knocked down [4]. We optimized a sgRNA-resistant PCSK5 (Figure 3) by introducing silent mutations into a lentivector delivering different inducible epitope-tagged PCSK5 alleles and transduced two PCSK5–/– MCF10DCIS clones at a limiting titer favoring single integration events. These stable lines will be tested for 3D phenotypes in vitro and colonization/growth in vivo after intraductal injection [6]. Our results thus far suggest that PCSK5M452I is not a gain-of-function allele, assuaging concern that the mutation may confer unusual properties to MCF10DCIS cells.

Acknowledgements (Optional): : We thank all members of the Janes lab for their support and feedback. We are especially grateful to Cheryl Borgman for all her hard work to keep the lab running smoothly.

References (Optional): : [1] J Natl Cancer Inst. 2000 Jul 19;92(14):1185-6. doi: 10.1093/jnci/92.14.1185a.

[2] Cancer Cell. 2008 May;13(5):394-406. doi: 10.1016/j.ccr.2008.03.007.

[3] J Pathol. 2016 Nov;240(3):315-328. doi: 10.1002/path.4778.

[4] Dev Cell. 2017 Nov 20;43(4):418-435.e13. doi: 10.1016/j.devcel.2017.10.027.

[5] Nat Commun. 2017 Feb 22;8:14370. doi: 10.1038/ncomms14370.

[6] Breast Cancer Res. 2009;11(5):R66. doi: 10.1186/bcr2358.