Wnt5a is strongly expressed at the leading edge in non-melanoma skin cancer, forming active gradients, while canonical Wnt signalling is repressed

Abstract

Wnt5a is one of the so-called non-canonical Wnt ligands which do not act through β-catenin. In normal development, Wnt5a is secreted and directs the migration of target cells along concentration gradients. The effect of Wnt5a on target cells is regulated by many factors, including the expression level of inhibitors and receptors. Dysregulated Wnt5a signalling facilitates invasion of multiple tumor types into adjacent tissue. However, the expression and distribution of Wnt5a in cutaneous squamous cell carcinoma (SCC) and basal cell carcinoma (BCC), as well as the effect of Wnt5a on keratinocyte migration has not been studied in detail to date. We here report that Wnt5a is upregulated in SCC and BCC and localised to the leading edge of tumors, as well as tumor-associated fibroblasts. The Wnt5a-triggered bundling of its receptor Fzd3 provides evidence of Wnt5a concentration gradients projecting into the tumor. In vitro migration assays show that Wnt5a concentration gradients determine its effect on keratinoctye migration: While chemotactic migration is inhibited by Wnt5a present in homogenous concentrations, it is enhanced in the presence of a Wnt5a gradient. Expression profiling of the Wnt pathway shows that the upregulation of Wnt5a in SCC is coupled to repression of canonical Wnt signalling. This is confirmed by immunohistochemistry showing lack of nuclear β-catenin, as well as absent accumulation of Axin2. Since both types of Wnt signalling act mutually antogonistically at multiple levels, the concurrent repression of canonical Wnt signalling suggests hyper-active Wnt5a signal transduction. Significantly, this combination of gene dysregulation is not observed in the benign hyperproliferative inflammatory skin disease psoriasis. Collectively, our data strongly suggest that Wnt5a signalling contributes to tissue invasion by non-melanoma skin cancer.

Figure 1. Localization of Wnt5a in non-melanoma skin cancer.

Immunohistochemistry of Wnt5a from SCC (a,b), or BCC (c,d), shown at 40× (a,c), or 200× (b,d) magnification. (e) Three SCC tumors, shown at 10× magnification, illustrating strong Wnt5a – staining at the tumor edge. Figures shown are representative for SCC (n = 12), and BCC (n = 9), respectively. Arrowheads indicate the following structures: black - basal layer of the epidermis, white- tumors, red- tumor associated endothelial cells, blue- fibroblasts, green – hair follicle.

Figure 2. Localization of Fzd3 in non-melanoma skin cancer.

Immunohistochemistry performed as in figure 1 performed on SCC (a–d), or BCC (e–f) shown at 40× (a), 100× (b,d,e), or 400× (c,f) magnification, respectively. Black arrows indicate staining intensity of Fzd3 in the epidermis used to assess staining intensity in tumors (denoted by white arrows). Red arrows denote boundary of tumors, pointing toward stroma.

Figure 3. Localization of Fzd5 in non-melanoma skin cancer.

Immunohistochemistry of SCC (a–d), or BCC (e–h), in each case showing an example of tumors exhibiting high (SCC: a,c; BCC: e,g) or low (SCC: b,d; BCC: f,h) Fzd5 expression. Staining intensities in the tumors can be directly compared to staining intensities of Fzd5 in the granular layer of the epidermis (black arrowheads). Panels a,b,e,f are shown at 40× and c,d, g,h at 200× magnification. The inset on the lower right of (c) shows a tumor-associated fibroblast. Red arrows denote blood vessels.

Figure 4. Spatial relationship of Wnt5a, Fzd3, and Fzd5 in squamous cell carcinoma.

Serial sections of three paraffin embedded SCC tumor samples (top, middle, bottom row, respectively) were stained for Wnt5a, Fzd5, Fzd3, respectively, as described in Methods, and shown at 200× magnification. Red asterisk denotes artificial nuclear staining possibly due to antigen retrieval conditions. Red arrows denote boundaries of tumors, pointing toward stroma.

Figure 5. Spatial relationship of Wnt5a, Fzd3, and Fzd5 localization in basal cell carcinoma.

Immunohistochemistry of serially cut samples stained for Wnt5a, Fzd5, or Fzd3 as indicated, magnification: 100× (top row), 200× (middle, bottom rows).

Figure 6. Wnt5a inhibits keratinoctye migration when present in homogenous concentration, but acts as chemoattractant when present as gradient.

A. Expression of endogenous and recombinant Wnt5a in whole cell lysates of stably transfected Wnt5a-overexpressing HaCat or control (HaCat-pcDNA) cells verified by western blot. B. Non-Wnt5a overexpressing HaCat-pcDNA cells were seeded in the upper chamber of a Transwell in 0.1% BSA DMEM in the absence or presence of recombinant Wnt5a at 1 µg/ml, as indicated in the figure. The lower chamber was filled with 600 µl DMEM containing 5% FCS as chemoattractant. Results are expressed as percentage of migrating cells when HaCat-pcDNA were seeded in 0.1% BSA DMEM only. The results shown represent mean ± s.d. of two independent experiment, each performed in triplicate, *p≤0.05. C. Comparison of Wnt5a-overexpressing and pcDNA control cell migration. Cells suspended in 0.1% BSA DMEM were seeded in the upper chamber. The lower chamber were filled with 600 µl DMEM containing 5% FCS as chemoattractant. Migration was assessed at 18 h using a colorimetric assay. Results are expressed as percentage of HaCat-pcDNA migrating cells. Results shown represent mean ± s.d. of n = 4 independent experiment, each performed in triplicate, *** p≤0.001. D. Scratch wound assay performed on mitomycin-C treated cells. During migration, HaCat-pcDNA (a, b, c), or Wnt5a-overexpressing cells (d, e, f) were maintained in DMEM containing 10% FCS. Pictures were taken just after the scratch was made (0 hrs) (a and d), as well as 18 h (b and e) and 24 h later (c and f). E. Migration of HaCat-pcDNA control cells in the presence of a Wnt5a concentration gradient. Wnt5a-overexpressing or pcDNA HaCat cells were seeded in the bottom wells of Transwell plates. Immediately before adding the inserts containing HaCat-pcDNA cells in the upper chamber, the media in the bottom wells was replaced to remove pre-secreted Wnt5a. Migration was assessed at 18 h. Results are expressed as percentage of HaCat-pcDNA migrating cells. Results shown represent mean ± s.d. of n = 3 independent experiments, each performed in triplicate, *** p≤0.001.

Figure 7. Specific upregulation of non-canonical Wnt signallling and repression of canonical Wnt signalling in SCC.

(a) Cartoon illustrating functional relationships between Wnt signalling components listed in tables 2 and 3. Red: upregulated, Green: down-regulated. Large dotted lines represent protein binding. (b) Specific dysregulation of SFRP1 and SFRP2 in invasive SCC, but not psoriasis. Fold-dysregulation of transcripts in psoriasis plaques was calculated as described previously and aligned to the SCC data set described in tables 2 and 3. Color coding and bold type set as in table 2. “n.s.”: not significant.

Figure 8. Lack of nuclear β-catenin in SCC and BCC.

Immunohistochemistry of three BCC (a–c) and SCC (d–f) samples stained with an antibody specific for activated β-catenin. Note strong nuclear β-catenin confined to the granular layer of the epidermis in each sample, as well as in a magnified hair follicle immediately below SCC cells (inset in d). All samples shown at 100× magnification, inset at 400×.

Figure 9. Immunohistochemical detection of β-catenin in BCC (b, e), and moderately differentiated SCC (c, f) samples at the ProteinAtlas repository (see main text).

Samples were stained either with an antibody specific for activated non-phosphorylated β-catenin (top) or pan-β-catenin (bottom). Images in (a) and (d) show the β-catenin distribution observed with the respective antibody. Note that strong nuclear β-catenin is confined to the granular layer of the epidermis.