Homophilic and heterophilic cadherin bond rupture forces in homo- or hetero-cellular systems measured by AFM based SCFS

Cadherins enable intercellular adherens junctions to withstand tensile forces in tissues, e.g. generated by intracellular actomyosin contraction. Single molecule force spectroscopy experiments in in-vitro experiments can reveal the cadherin-cadherin extracellular region binding dynamics such as bond formation and strength. However, characterization of cadherin homophilic and heterophilic binding in their native conformational and functional state in living cells has rarely been done. Here, we used Atomic Force Microscopy (AFM) based Single cell force Spectroscopy (SCFS) to measure rupture forces of homophilic and heterophilic bond formation of N-, OB- and E-cadherins in living fibroblast and epithelial cells in homo- and hetero-cellular arrangements, i.e. between same type of cells and between cells of different type. In addition, we used indirect immunofluorescence labelling to study and correlate the expression of these cadherins in intercellular adherens junctions. We showed that N/N and E/E cadherin homophilic bindings are stronger than N/OB, E/N and E/OB heterophilic bindings. Disassembly of intracellular actin filaments reduces the cadherin bond rupture forces suggesting a contribution of actin filaments in cadherin extracellular binding. Inactivation of myosin did not affect the cadherin rupture force in both homo- and hetero-cellular arrangements. Whereas, myosin inactivation particularly strengthened the N/OB heterophilic bond and reinforced the other cadherins homophilic bonds.


Introduction
Cell adhesion to neighbouring cells or the extracellular matrix (ECM) environment is a very important process in regulating crucial biological activities such as embryonic development, tissue assembly and dynamics, wound healing and cancer metastasis. Generally, cells communicate with other cells through adherens, gap or mechanosensitive junctions (1). Cadherins from adherens junctions are a class of calcium dependent cell adhesion molecules (CAMs) which comprise three different domains: (i) an intracellular or cytoplasmic domain which binds to the actin cytoskeleton through adaptor proteins such as α-catenin, β-catenin and p120 catenin, (ii) a transmembrane domain and (iii) an extracellular domain. The extracellular domain consists of five extracellular cadherin (EC) repeats. A dimer of EC1-EC5 of one cell interacts with the corresponding cadherin dimer of a neighbouring cell through homophilic or heterophilic interaction (2,3).
Several assays have been developed to investigate cell-cell interactions in the last two decades, such as dual micropipette assay (4), flipping assay (5), FRET (6) and AFM based SCFS (7,8). Comparing all assays, the AFM based SCFS assay provides a wide range of forces (10 pN to 10 6 pN) (9) and a controlled force application (loading rate) on the cell-cell adhesion cadherin bond by retracting the AFM cantilever at a well-defined velocity (10). In SCFS, cell adhesion force measurements are performed in near physiological conditions. Being a multifunctional toolbox in nanobiotechnology (11), AFM provides a functionalized cantilever to pick up a live cell guided by optical microscopy. It allows probing the rupture force between cadherin molecules present in two cells, by separating the two cells. The rupture force can be quantified and reveals differences in the specific type of cadherins secreted by different cell types.
According to the presence or absence of the HAV (His-Ala-Val) cell recognition sequence in the EC1 domain, classical cadherins are classified into type I (E-, N-and others) and type II cadherins (OBand others) (2,3). The most commonly expressed cadherin found in fibroblasts is N-cadherin (cad-2) (12). Primary rat fibroblasts differentiate into myofibroblasts in vitro using transforming growth factor-β1 (TGF-β1). TGF-β1 induces the expression of alpha-smooth muscle actin (α-sma), an increased expression of OB-cadherins (cad-11) and a decreased expression of N-cadherin (13). This TGF-β1 induced cadherin switch from N-cadherin to OB-cadherin increases the intercellular adhesion strength between myofibroblasts by strengthening individual OB-cadherin bonds. Single molecule force spectroscopy (SMFS) measurements on OB-and N-cadherins showed that the rupture force between OBcadherins homophilic interaction is larger than between N-cadherins (14). A biochemical analysis of N-and OB-cadherins expression in human dermal fibroblast and Dupuytren's myofibroblast shows increased OB-cadherin and decreased N-cadherin expression in myofibroblasts compared to dermal fibroblasts (1). The E (epithelial)-cadherin (cad-1) is the dominant cadherin expressed in most epithelial cell lines like MDCK (Madine-Darby Canine Kidney) cells (15). The more motile, trypsin sensitive subpopulation of MDCK cells shows a low level of N-cadherin expression (16).
Hetero-cellular interactions between different cell types occur in tissue and organ morphogenesis. Involvement of specific cadherins in these interactions plays a pivotal role in cancer cell metastasis (17) whereas heterophilic interactions between cell specific cadherins mediate cancer cell invasion (18).

Direct interactions between fibroblast and epithelial cells may play an important role in the epithelial
to mesenchymal transition (EMT) process (19). Hetero-cellular interactions between normal fibroblasts and gastric cancer cells induce E-cadherin loss and increase metastasis in gastric cancer via EMT (20). The investigation of hetero-cellular interactions between fibroblast and epithelial cells using biophysical techniques such as SCFS will help to better understand the role of classical cadherin interactions both in EMT and Mesenchymal to Epithelial cell transition (MET) processes.
Actin filaments associated with myosin are the major contractile component responsible for intracellular force generation. Generally, these forces are generated by the myosin assembly and motility on the actin filaments. Myosin light chain is phosphorylated by the myosin light chain kinase (MLCK) and this activates the myosin cross linking to the actin filaments with actomyosin contractile force generation. Intracellular forces are then transmitted to the neighbouring cells and to the extracellular environment through cadherins and integrins, respectively that are connected to actin filaments. Disassembling actin filament rich stress fibres by treating fibroblasts with Cytochalasin D results in decreased cell stiffness (21). Addition of Cytochalasin D reduces the cadherin mediated binding forces between myofibroblasts, as measured by SCFS, and shows that cadherins are linked structurally and possibly functionally to the intracellular actin network (14). Inactivating myosin-II activity by treating fibroblasts with ML-7 inhibits the MLCK, which further prevents myosin mediated actomyosin contractility which results in actin cytoskeleton softening and thus decreased cell stiffness (22).
In the present study, we have studied the expression of N-and OB-cadherins in three types of fibroblasts extracted from the same patient with Dupuytren's disease using fluorescence microscopy: (1) normal fibroblasts-NFs from normal healthy skin, (2) scar fibroblasts-SFs from cutaneous scar tissue and (3) Dupuytren's myofibroblast-DFs from the nodules of the palmar fascial strands. Using AFM-SCFS, we measured the rupture forces between fibroblasts grown in a confluent monolayer and fibroblasts attached to the AFM cantilever (NF-NF, SF-SF and DF-DF). Loading rate dependent rupture force measurements showed that NF and SF exhibit larger rupture forces than DFs. These results correlated with the cadherin types present in the adherens junctions of respective fibroblast types. Heterocellular interaction forces were also measured between fibroblasts grown in monolayers and epithelial cells attached to the cantilever. Regarding the epithelial cell, we used epithelial cell line called MDCK cells to study the hetero-cellular interactions between MDCK and fibroblasts mediated by cadherins expression and binding dynamics. Immunofluorescence studies of MDCK and fibroblast co-cultures showed the presence of N-cadherins at the fibroblast-MDCK junction and E-cadherin loss in MDCK.
Cytochalasin D treatment decreases the interaction forces in both homo-cellular and hetero-cellular interactions. In ML-7 treatment, no change in interaction forces observed in homo-cellular and heterocellular interactions except for DF-DF interaction. Contrarily, there is an increase in DF-DF rupture forces after ML-7 treatment and reveals that OB-and N-cadherin heterophilic bond strengthens the cell-cell interaction when there is no intracellular contractile force.

N/OB heterophilic binding is weaker than N-and OB-homophilic binding
Investigation of cell-cell interactions using AFM becomes more possible using a simple cell force spectroscopic setup. AFM based SCFS setup is explained with the simple schematics shown in Fig   1A. A tipless cantilever, functionalized with concanavalin A (conA), was placed on a cell, which makes initial adhesion to the substrate and is appropriately round in shape. The cantilever was approached towards that cell until a certain loading force has been reached. After a dwell time of 5 sec the cell has adhered sufficiently and stays attached to the cantilever when the cantilever is retracted from the support as shown in Fig. 1B. The force curve obtained during cell capture is shown in Fig.   1C. After a recovery time of 10 minutes, the cantilever with the attached cell was approached towards and retracted from another cell attached to the Petri dish. Cell-cell interactions and rupture forces between cells were probed. Fig. 1D shows a cell-cell (NF-NF) interaction force curve. The force curve contains approach (red arrows) and retract curve (blue arrows). The cell capturing and cell-cell interaction events are visible in the retract curve. In case of cell-cell interactions, two distinct features can be seen in the retract curve: rupture (continuous line arrows) and tether events (discontinuous line arrows). The adhesion molecules that are well anchored to the intracellular actin filaments interact with their counterparts on the other cell, and the breakage of the adhesion molecules mechanical bonds can be seen as a rupture event. This rupture event can be due to a single bond breakage or to multiple bond breakages. The rupture force was calculated from the height of the rupture event. When adhesion molecules are not anchored to actin filaments membrane, tethers can be pulled over large distances, which eventually will also break (tether events). The rupture and tether events observed during cell capture were due to the interaction and bond breakage of either specific adhesion molecules or other nonspecific interactions, which were not characterized here. Here we determined rupture forces between three types of fibroblasts isolated from primary human cells using SCFS and assessed the specific cadherins at the interaction site using fluorescence microscopy. The cell-cell rupture force was measured using an approach and retraction velocity of 3 µm/sec, a maximum loading force of 3 nN and the contact time of 2 s. The histogram plot of measured rupture forces versus the number of rupture events shows the force distribution for each fibroblast type (NF-NF Fig Table 1 which lists the corresponding 25, 50 and 75 percentile values). To verify that these rupture forces were due to the cadherin-cadherin bond breakages, the rupture events were recorded in the presence of EGTA (ethylene glycol tetraacetic acid, a calcium chelating agent) in the SCFS setup, effectively removing all free calcium from the extracellular space. Addition of EGTA completely inhibited the cadherin mediated cell-cell interaction with reduced numbers of rupture events ( Fig. 2A, B and C, blue bar). Under normal conditions, force curves showed multiple rupture events due to interactions of multiple cadherins ( Supplementary Fig. 1A), whereas in the absence of Ca 2+ , i.e. in the presence of EGTA, such rupture events were not seen in force curves ( Supplementary Fig. 1B). To understand the cadherincadherin binding strength, we exerted varying force (loading) rates on the bonds by approaching and retracting the AFM cantilever at different velocities (3, 5, 7.5 and 10 µm/sec), which named "pulling rate" in force spectroscopy. For all three fibroblast types, the corresponding rupture forces showed a linear increase depending on the pulling rate applied (Fig. 2D). The median rupture force values for respective pulling rates for all three fibroblast types were listed in Table 1  In order to assess the presence of specific cadherin types in cell-cell interaction sites, all three fibroblast types were immunostained for N-and OB-cadherin. Dual immunostaining for N-(red) and OB-(green) cadherins showed that NF and SF express exclusively N-cadherin whereas DF express both Nand OB-cadherin at the interaction site between cells (Fig. 2E). In the overlay (orange), heterophilic interactions between the N-and OB-cadherins are visible. Single immunostaining for N-cadherin (red) showed that all three fibroblast types express N-cadherin at the interaction site (Supplementary

E-, N-and OB-cadherin at the fibroblast-epithelial hetero-cellular adherens junctions
The significance of studying hetero-cellular interactions may lead to sorting out different cell types by their expression and assembly of cell specific cadherins at the interaction site. The investigation of cadherin homophilic and heterophilic interactions may pave the way for a better understanding of cadherin mediated intracellular signalling. Heterophilic cadherin rupture forces were measured between epithelial cells and fibroblasts. A monolayer of fibroblasts was grown in a Petri dish and MDCK epi-thelial cells were attached to a tipless cantilever functionalized with conA. Fibroblast-MDCK interactions were studied by approaching a cantilever with attached MDCK cells towards the fibroblast cell monolayer at 3 µm/sec velocity with 3 nN maximum contact force and 2s contact time. In a similar fashion, MDCK-MDCK interactions were studied and the resulting median rupture force value was 75.51 pN. Regarding fibroblast-MDCK interaction, the median rupture force values were 55.85 pN for NF-MDCK, 39.68 pN for SF-MDCK and 46.70 pN for DF-MDCK (See Table 1

Role of actin assembly in homo-and hetero-cellular adherens junctions
Cytochalasin D disrupts the actin assembly and results in cell softening (21). Here, we used 5 µM cy-   Table 2 and plotted in Supplementary Fig. 5.    Table 3 and plotted in Supplementary Fig. 8.  and SF were shown to express α-sma, but no large stress fibres. Thus both cell types were considered as fibroblast phenotype (29). When SF was seeded in a physiological environment such as a decellularized dermal matrix, cells expressed large stress fibres and thus showed a proto-myofibroblast or myofibroblast phenotype (30). In contrast, DF showed α-sma positive large stress fibres and thus were considered as myofibroblast phenotype (29). In comparison to earlier reports (1,14), N-cadherin expression was seen in the normal fibroblast phenotype (in this study: NF and SF) and OB-cadherin expression in the myofibroblast phenotype (DF). In contrast to rat fibroblasts, which show a transition in expression from N-to OB-cadherin triggered by TGF-β1 (14), we found in our study that DF expressed both N-and OB-cadherins and N/OB heterophilic binding. Fibroblasts extracted from the palmar region (cords and nodules) of patients with Dupuytren's disease express stress fibres and thus exhibiting a myofibroblastic phenotype (29) without any mechanical or biochemical stimulation such as TGF-β1 (14). Biochemical expression of N-cadherin was observed in Dupuytren's myofibroblast and results from a collagen gel contraction study showed that myofibroblasts displayed reduced contraction in the presence of N-cadherin blocking peptide (1). Obviously, N-cadherin has an important function in myofibroblast intercellular adherens junctions. By immunofluorescence, N-cadherin and

OB-cadherin expression and their homophilic (in NF and SF) and heterophilic (in DF) pair formation
were observed in all three fibroblast phenotypes. In this study, we found that the presence of different cadherins was strongly correlated with the rupture forces measured by SCFS.
In our study, the E/E-cadherin homophilic interactions in MDCK showed rupture force values closely related to previous studies (28). This confirms the initial adhesion in MDCK homo-cellular arrangements could be largely dominated by E-cadherin homophilic binding that displays the larger rupture This mechanically active heterotypic contact between N-cadherin expressing cancer associated fibroblasts and an E-cadherin expressing epithelial (A431) cancer cell line (A431) enables fibroblasts to steer cancer cell invasion (18). Loss of E-cadherin was observed in co-cultures of fibroblast with epithelial cells, whereas normal fibroblasts can induce E-cadherin loss to promote EMT in gastric cancer (20). In chronic inflammatory conditions, epithelial cell-fibroblast interactions stimulate EMT in human bronchial epithelial cells from chronic obstructive pulmonary patients (19). Accordingly, we found reduced E-cadherin and increased N-cadherin in our multi-cell cultures with immunofluorescence which might imply the initiation of an EMT process. Furthermore, N/N homophilic adhesion (NF-MDCK and SF-MDCK) and N/OB heterophilic adhesion (DF-MDCK) were present at the inter-action sites between epithelial cells and fibroblasts.
In AFM based SCFS, a varying cantilever pulling rate allowed for characterizing the cadherin binding strength. Rupture forces generally increase with increasing pulling rate, which leads to increased loading rates (35). In this study, E-cadherin and N-cadherin homophilic and OB-cadherin heterophilic binding rupture forces showed a linear relationship related to the pulling rate. In the fibroblast homocellular arrangement, N-cadherin homophilic binding was stronger in NF and SF compared to N/OBcadherin heterophilic binding in DF. Similarly, in fibroblast-epithelial cell hetero-cellular arrangement, all three fibroblast types interacting with MDCK show similar rupture forces. In general, Ecadherin homophilic binding in MDCK homo-cellular arrangement displayed the strongest binding strength which reflects previous findings (14,27,28).
Differences in force peak values can be found when results are compared to other studies. Due to the stochastic process of cadherin, protein binding forces can be distributed differentially. Rupturing of molecular bonds is always effected by thermal fluctuations, leading to varying rupture forces and thus cadherin binding events are stochastic (36). Even the VE-, N-and OB-cadherin SMFS and SCFS measurements showed three different interaction forces, as the three force peaks were present in rupture force histograms (14,24). However, cadherin pairs (VE-, E-and N-) exhibited single force states as well which correlates well to results found in earlier SCFS studies (27). Similarly, we observed one single force peak in the histograms which correspond to a single rupture force of cadherin bond unbinding.
Cell-cell adhesion is mediated by cadherins in adherens junctions. Cadherins are linked with their cytoplasmic domain to the intracellular actin cytoskeleton through adaptor proteins such as αand βcatenin (37). Disruption of actin filaments by cytochalasin D affected the cadherin extracellular domain homophilic and heterophilic binding dynamics in our study. It seems that the inactivation of actin filaments with cytochalasin D has a direct effect on the cadherin extracellular binding activity by altering the cadherin cytoplasmic link to the actin filaments (14). However, this phenomenon was found exclusively for OB-cadherin homophilic binding (14). In our study we could show a similar effect for both homophilic (N/N and E/E) and heterophilic (E/N, N/OB and E/OB) adhesion in homoand hetero-cellular arrangements.
ML-7 inhibits the activity of MLCK by interacting with the phosphorylation event of myosin light chain (MLC). Thus the binding of myosin to actin filaments and ATPase driven contractile force generation are inhibited (38,39,40). In the current study, disabling actin-myosin contraction using ML-7 showed no effect on the cadherin extracellular binding dynamics except for N-/OB-cadherin heterophilic binding. Myosin inactivation particularly strengthened the N-/OB-cadherin extracellular binding activity demonstrated by the change of rupture forces. A hypothetical biophysical mechanistic pathway that could explain the observed N/OB-cadherin reinforcement effect is stated in Fig. 6. Myosin II acts as an actin crosslinker (41) whereas myosin VI acts as a mediator protein, which binds cadherin to actin filaments (42). Loss of myosin II selectively inhibits myofibroblast differentiation in fibroblasts of fibrotic lung when compared to healthy phenotype (43). From our current findings and previous results from others, we speculate that (Fig. 6): (1) There is no influence of actomyosin contraction or inactivated myosin on homophilic or heterophilic cadherin extracellular binding dynamics

Cell culture
Cell culture was performed as described previously (29). Fibroblasts were harvested from tissues of patients undergoing hand surgery (approved by the local Ethics Committee-Ärztekammer Bremen, #336/2012) and isolated as described previously (29). Cells were grown until the passage-9 for fibroblasts and 13 for MDCK in DMEM medium and incubated at 37°C in a humidified atmosphere of 95% air and 5% CO 2 . Cell culture was established for two days before proceeding with further SCFS measurements. Medium was supplemented with 10% fetal bovine serum (FBS) and 2% penicillinstreptomycin.

Cantilever functionalization
The silicon-nitride tipless cantilevers (Nanoprobe SPM Tips, NP-OW 9861) were washed with 1 % SDS (sodium dodecyl sulphate), Helizyme (B. Braun Vet Care GmbH) and distilled water solution each for overnight. The cantilevers were then treated with plasma (Ar) at high power for 5 min. In order to functionalize the plasma treated cantilevers with concanavalin A (conA) (C2010, Sigma-Aldrich), the cantilevers were placed in a phosphate buffered saline (PBS) solution containing conA (2 mg/ml) for 2 h at room temperature. The conA coated cantilevers were stored in PBS at 4°C (44).

Cell attachment to the cantilever
Prior to cell-cell adhesion measurements, cells that were used for attachment to the cantilever were released from the culture flask by treatment with trypsin for 2 min and trypsin was neutralized by centrifugation and replenishment with new medium. The trypsinized cells were transferred into the Petri dish containing firmly attached cell monolayers that are grown for two days. After 5 min incubation at 37°C, the Petri dish was used for the single-cell force spectroscopy-AFM setup. The conA functionalized cantilever was then placed over a suitable cell with round morphology which initiated its attachment to the cell monolayer. Then, the conA coated cantilever was approached towards the cell with a 3.5 nN maximum loading force for 5 sec at a velocity 5 µm/sec until the cell was captured. The cantilever with attached cell was taken few µm away from the cell monolayer and the whole setup was left undisturbed for 10 min in order to establish firm cell adhesion to the cantilever.

AFM cell adhesion force measurements and data analysis
Single-cell experiments were conducted using a MFP3D AFM (Asylum Research, Santa Barbara, CA, USA