Fig 2: CCN2 expression in transformed keratinocytes is α3β1-dependent.(A) GAPDH-normalized relative mRNA expression of CCN2 is significantly decreased in non-stimulated as well as IL-6 and TPA-treated α3β1-depleted keratinocytes. The average of up to four independent measurements of technical duplicates of four RNA samples per group (dots) is presented (mean ± SD, Fisher’s LSD test, *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001). (B) IF (left) and quantification of the mean intensity (right) of CCN2 in non-stimulated, IL-6, and TPA-treated MSCC Itga3 KO and WT keratinocytes. Expression of CCN2 is α3β1 dependent and increases upon IL-6 and TPA treatment (scale bar: 50 μm). 90 cells imaged over three independent experiments were quantified (mean ± SD, Fisher’s LSD test, P < 0.0001). (C) Representative WB confirming α3β1-dependent and IL-6– and TPA-mediated CCN2 expression. Quantification can be found in Fig S5A.Source data are available for this figure.

Fig 3: HB-originating α3β1-depleted keratinocytes show decreased expression of CCN2 during the initiation stage of tumorigenesis.(A) A heat map showing expression of selected epidermal markers, selected from gene list, in GFP-positive HB-originating keratinocytes, isolated from the skin of short-term DMBA/TPA–treated K19 Itga3 KO and WT mice. (B) IHC staining for GFP and CCN2 in the skin of short-term DMBA/TPA–treated K19 Itga3 KO and WT mice (scale bar: 100 μm). (C) Sparse CCN2-positive cells can be observed in epithelia and stroma of all tumors of K19 Itga3 WT mice (arrow heads). Top: quantification of the CCN2-positive area in cross-section of 127 tumors, isolated from six K19 Itga3 WT mice. CCN2 represents less than 1% of total tumor surface in most tumors. Bottom: representative IHC staining for CCN2 (scale bar: 500 μm). (D) Consecutive section of papilloma, isolated from K19 Itga3 WT mouse, stained for GFP and CCN2. No overlap of the two markers can be observed (scale bar: 500 μm).

Fig 4: CCN2 expression in transformed keratinocytes is α3β1-dependent.(A) Quantification of WB from Fig 7C. Bars represent the mean of five independent experiments (mean ± SD, Fisher’s LSD test, *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001). (B) IF images showing co-localization of excreted CCN2 and laminin-332 in the culture of non-treated MSCC Itga3 KO and WT keratinocytes.

Fig 5: Epidermal deletion of α3β1 causes de novo expression of K15 outside of HBs.(A) Overview of the K14 Itga3 KO and WT mouse models. (B) Whole mounts of tail epidermis show the presence of α3β1-depleted K15-positive keratinocytes in upper parts of hair follicles and IFE of K14 Itga3 KO mice (white arrow heads). Remaining α3β1-positive keratinocytes in K14 Itga3 KO mice are preferentially localized to HBs (scale bar: 100 μm). (C) Staining for integrin α3 shows HB localization of α3β1-positive keratinocytes in the back skin of 7-wk-old K14 Itga3 KO mice. α3β1 is found in all basal keratinocytes of K14 Itga3 WT mice of similar age (scale bar: 50 μm). (D) FACS analysis of keratinocytes isolated from back skin epidermis. The chart shows the percentages of α3-positive HB cells (CD34-positive) in the total α3-positive population. Each dot represents a mouse. Gating strategy can be found in Fig S2A (mean ± SD, unpaired t test, *P < 0.05). (E) GAPDH-normalized relative mRNA expression of K15 is increased in the epidermis of back and tail skin of K14 Itga3 KO compared with WT mice. Each dot represents a mouse and is an average of technical duplicate or triplicate (mean ± SD, unpaired t test, *P < 0.05, **P < 0.005).Source data are available for this figure.