Project Invasive Lobular Breast Cancer (ILC)

E-cadherin is the King of Cell Adhesion

A schematic overview of E-cadherin form and function. Shown are a schematic representation of the homotypic interactions (left), representing the situation between two epithelial cells (middle). The right cartoon shows a detailed representation of the E-cadherin based Adherens Junction.

Loss of E-cadherin is the cause for ILC development and progression.

During his postdoctoral days at the Netherlands Cancer Institute (NKI-AvL, Jonkers and Berns labs), Patrick developed a conditional knockout mouse model for invasive lobular breast cancer (ILC), which was published in Cancer Cell (2006) and Disease Models and Mechanisms (2011).

Based on either Cytokeratin 14 (K14) or Whey Acidic Protein (WAP) induced bi-allelic loss of E-cadherin (Cdh1) and p53 (Trp53), female mice developed metastatic ILC. Both the histological phenotype, as well as the metastatic spectrum, mimicked that of human ILC. However, in contrast to its human counterpart, mouse ILC is relatively high-grade and Oestrogen Receptor (ER) negative. These results were later confirmed in the Jonkers lab through conditional knockout models combining E-cadherin loss with either PTEN loss or PI3K activation. Interestingly, E-cadherin knockout in the MMTV-driven polyoma middle T model of breast cancer, E-cadherin loss does not induce ILC (pers.communication Dr. R. Kemler, 2005).

In short, in the right mouse model, and in combination with a strong oncogenic event, E-cadherin loss leads to metastatic ILC in the mammary gland.

A mouse model for human ILC

A comparison between mouse (left) and human ILC (right). The mouse ILC models is based on a tissue specific condtional knockout of E-cadherin (Cdh1) and p53 (Trp53). K14cre was used as a driver. Shown is an haematoxilin/eosin (HE) staining, magnification = 200x

How p120 controls Anoikis resistance in ILC

A model for the p120 dependent and MRIP mediated control over RhoA/Rock induced anoikis resistance in ILC.

The multifactorial p120 protein is essential for ILC form and function.

It was originally described by Dabbs et al. in 2004 that ILC cells express mostly cytosol p120-catenin (p120). Our group went on to demonstrated that cytosolic and nuclear translocation in ILC is a direct consequence of E-cadherin loss. Because p120 is essential for the stability of classical cadherins on the plasma membrane, this, by definition, means that exclusive cytosolic p120 expression in cells equals the absence of classical cadherins.

Ron Schackmann, a former PhD Student in the lab, showed in a publication in the Journal of Clinical Investigations (2011), that cytosolic p120 promotes Rock-driven actomyosin contraction through inhibition of the Rock inhibitor MRIP, and that this promotes anoikis resistance of ILC cells.

Ron also demonstrated that p120 loss does not predispose to ILC formation but instead leads to high grade metaplastic carcinomas in mice. The resulting paper in Cancer Research (2013) showed that conditional knockout of p120 (Ctnnd1) in combination with p53 inactivation led to aneuploidy and metastatic tumors.

We also performed triple knockout studies showing that p120 is essential for the formation of lobular breast cancer in mice. This paper, published in 2016 in the Journal of Mammary Gland Biology and Neoplasia (together with the jonkers lab) was preceded by studies from Robert van de Ven (PhD student) who could demonstrate an explanation for the induced aneuploidy upon loss of p120 in tumors.

His work was published in Nature Communications (2016) and showed that p120 is central in the spatial control and activation of mitotic cleavage furrow ingression by RhoA, a textbook finding that explained why p120 loss results in multinucleation and aneuploidy. This is not exclusive for breast cancer but occurs in every cancer cell type we looked!

A model for p120 directed cytokinesis

Above a schematic representation is shown of the effect of p120 loss on epithelial cytokinesis. During anaphase, p120 is accumulating at the ingression furrow together with RhoA. When progressing to late anaphase/telophase, p120 controls spatial RhoA activity through its RhoGDI function (See Anastasiadis et al.NCB, 2001), ensuring correct ingression and cleavage into two daughter cells. In the absence of p120 this does not occur correctly, leading to binucleation, and, eventually, cancer aneuploidy.

The movies in the left panel show H2B-GFP tagged p120::iKD cells undergoing division in the presence of Doxycycline. Note the failure in cleavage furrow ingression (left movie) and successive failures in cytokinesis leading to multi nucleation (middel and right movies).

 

Growth Factor Receptor Activation in ILC

Using a combination of reverse-phase protein array (RPPA) analyses, mRNA sequencing, conditioned medium growth assays and CRISPR/Cas9-based knock-out experiments, we demonstrated in an NWO-VENI sponsored study that E-cadherin loss causes increased responsiveness to autocrine growth factor receptor (GFR)-dependent activation of phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)/AKT signalling. Most important, autocrine activation of GFR signalling and its downstream PI3K/AKT hub was independent of oncogenic mutations in PIK3CA, AKT1 or PTEN. Pharmacological inhibition of Akt resulted in robust inhibition of cell growth and survival of ILC cells, and impedes tumour growth in a mouse ILC model. These findings were published in Scientific Reports (2018).

Because E-cadherin loss evokes hypersensitisation of PI3K/AKT activation independent of oncogenic mutations in this pathway, we proposed clinical intervention of PI3K/Akt in ILC based on functional E-cadherin inactivation, irrespective of activating pathway mutations. We are currently pushing hard, together with our ELBCC clinical partners, to get this into a clinical setting.

An upcoming study called NeoAKT (NCT05919108), headed by Dr. Jason Mouabbi (MD Anderson), will investigate the neoadjuvant combination of capivasertib (Truqap) and letrozole in postmenopausal patients with stage II to III, estrogen receptor–positive, HER2-negative invasive lobular carcinoma.

E-cadherin loss induces GFR hyperactivity

Western blot showing the impact of E-cadherin loss on IGF1 induced AKT activation in ILC cells. Note the 8.8x and 4.4x increase in phosphorylation on the different epitopes, indicating a >30 fold increase in signaling upon E-cadherin loss. E-cadherin loss also inhibits FOXO and BMF transcription and  induces a 20x sensitivity towards AKT inhibitors such as MK2206 and Capivasertib (not shown).

The cartoon depicts a schematic overview of the molecular impact of E-cadherin loss in ILC.

 

Loss of α-catenin leads to E-cadherin positive ILC phenotypes

Shown are the effect of α-catenin loss on E-cadherin expressing ductal breast cancer cells. Left panel shows a western blot after inducible knock-down and restoration of expression respectively. Right panels show the impact on epithelial morphology and E-cadherin localisation/expression.

Other E-cadherin associated molecules that cause Lobular disease

Although E-cadherin loss is the event leading to ILC, approximately 10% of all ILCs retain E-cadherin expression on the plasma membrane. Moreover, within the 90% of ILC that have lost E-cadherin expression, “only” 60-80% show a mutational event leading to protein loss.

We have therefore spent a substantial amount of effort and resources in finding the other hits that would lead to E-cadherin deficiency or loss of function. Eva Vlug and Jolien de Groot (former PhD candidates in the lab) identified α-catenin (CTNNA1) as a prominent candidate. Upon functional assessment of the impact of CNTTA1 loss of function in cell lines, we found that inactivation of α-catenin leads to cell motility and anoikis resistance that is RhoA/Rock/actomyosin dependent (just like in E-cadherin KO cells). Surprisingly, this phenotype is accompanied by retention of E-cadherin on the cell membrane. 

 

Using a transplantation-based mouse model of conditional p53 KO stem cells, Jolien could show that loss of α-catenin leads to tumor formation with ILC-like phenotypes. These data were published in the Journal of Pathology in 2018.

We concluded that α-catenin loss is sufficient to disturb the E-cadherin to F-actin linkage, resulting an ILC phenotype. This conclusion was incorporated in a comprehensive review in Cold Spring Harbor Perspective in Biology (2018), where we wrote that ILC is an “Actin”-disease. In other words, we postulated that disturbance of the E-cadherin to F-actin is sufficient and necessary to for ILC formation and progression.

Later that year, Max Rätze, joined the lab as a PhD candidate to work on a KWF funded project to find additional tumor suppressors for ILC development. In a collaboration with Prof. Edwin Cuppen and Prof. René Bernards, we exome-sequenced well-characterized ILC samples and searched for cooperating mutations in ILC CDH1 wild type samples. Out of the 300+ mutations we picked a frameshift mutation in Afadin (AFDN) because of its obvious ties with junctional formation and F-actin linkage. With the help of Dr. Noburu Ishiyama (former postdoc group Takai) we could show that Afadin is indeed a novel invasion suppression for ILC. We also functionally mapped the Coiled-Coil domain of Afadin to be necessary and sufficient to prevent tumor invasion and ILC formation. Whether this is due to direct F-actin binding or indirectly via α-catenin remains unclear. These results were published in 2025 in the Journal of Pathology.

During the course of this project, we also identified a new inactivation mutation in CDH1. This mutation (S180Y) was originally picked up in an ILC organoid (P008) in the lab of Prof. Clare Isacke (ICR London). With the help of Dr. Pere Roca-Cusachs we could show that CDH1(S180Y) cannot form in trans X-dimers, which is the underlying cause for its inactivity in ILC. These findings were published in Oncogene (2022).

Our assumption is that cooperating hits result in ILCs that retain E-cadherin expression, which should therefore be considered as biochemically non-classical ILCs

The CDH1 S180Y mutation inhibits trans dimer formation

Modeling of the new CDH1S180Y mutation in the ILC patient derived organoid P008.

Representation of the main molecular interactions between the first two residues of the adhesion arm (D155 and W156) of a cadherin and the acceptor pocket of an opposite cadherin. At the left, acceptor pocket corresponding to the E-cadherin wild-type model. At the right, acceptor pocket corresponding to the S180Y mutant model. 
Afadin is an ILC tumor suppressor

Afadin loss of function induces lobular-type morphological features. Shown are representative H&E stained examples of primary tumors that developed in control mice  (wild type) or mice transplanted with ∆AFDN cells, cells lacking the coiled coil domain (∆CC) or the Afadin C terminal (∆Cterm). Note the acquisition of invasive or Indian files (arrows).  Size bar = 100µm.

The p120-Kaiso link and consequences for gene transcription in ILC

Upon E-cadherin loss, translocated p120 interacts with a bimodal transcription factor/repressor called Kaiso (ZBTB33). Kaiso (named after a Calypso dance style in Barbados) can bind a canonical Kaiso Binding Sequence (cKBS), and a presumed methyl (CG) binding site (CG-KBS). Binding to p120 induces relief of transcriptional repression on the cKBS, leading to expression of distinct genes such as MYC and WNT11. Robert van de Ven in our lab) defined a set of candidate target genes called the Anoikis Resistance Transcriptome (ART) of ILC. This set, published in the Journal Disease Models and Mechanisms (2015), confirmed the target Wnt11 as a driver of RhoA-dependent actomyosin contraction in ILC, and set the stage for later studies by another PhD student Max Rätze.

Max, together with Thijs Koorman (assistant professor in the lab), could show that ID2 is a bona fide transcriptional p120/Kaiso target gene upon E-cadherin loss. More importantly, they unravelled a mechanism whereby cytosolic ID2 binds non-phosphorylated Rb, which prevents phosphorylation by CDK4/6 and subsequent cell cycle progression. We postulated that this mechanism, published in Oncogene (2022), is the basis for the low-grade characteristics in ILC. Moreover, when ILC cells are transferred to anchorage independent conditions, this machinery fully inhibits cellular division in suspension, thereby preventing DNA damage and subsequent anoikis. We are now studying the impact of these ILC specific mechanisms on cell proliferation and susceptibility to cyclin dependent kinase inhibitors.

In 2024 we coupled the ART of ILC to a Kaiso specific regulation and called it KART. The resulting paper in the Journal of Pathology shows that the 33 KART genes are prognostic in breast cancer and strongly associate with the ILC histotype. KART was reciprocal to the LobSig signature (low KART correlates with favorable prognosis) but with a comparable prognostic power.

Prognostic power of the KART signature

Shown are the associations of normalized KART (left) and LobSig (right) expression signatures with IDC-NST and ILC in TCGA dataset.
All points are confirmed ER-POS and HER2-NEG breast cancers.
IDC-NST n = 227; ILC n = 98.