The Role of Fibroblast Growth Factor-Binding Protein 1 in Skin Carcinogenesis and Inflammation

Abstract

Fibroblast growth factor-binding protein 1 (FGFBP1) is a secreted chaperone that mobilizes paracrine-acting FGFs, stored in the extracellular matrix, and presents them to their cognate receptors. FGFBP1 enhances FGF signaling including angiogenesis during cancer progression and is upregulated in various cancers. Here we evaluated the contribution of endogenous FGFBP1 to a wide range of organ functions as well as to skin pathologies using Fgfbp1-knockout mice. Relative to wild-type littermates, knockout mice showed no gross pathologies. Still, in knockout mice a significant thickening of the epidermis associated with a decreased transepidermal water loss and increased proinflammatory gene expression in the skin was detected. Also, skin carcinogen challenge by 7,12-dimethylbenz[a]anthracene/12-O-tetradecanoyl-phorbol-13-acetate resulted in delayed and reduced papillomatosis in knockout mice. This was paralleled by delayed healing of skin wounds and reduced angiogenic sprouting in subcutaneous matrigel plugs. Heterozygous green fluorescent protein (GFP)-knock-in mice revealed rapid induction of gene expression during papilloma induction and during wound healing. Examination of wild-type skin grafted onto Fgfbp1 GFP-knock-in reporter hosts and bone marrow transplants from the GFP-reporter model into wild-type hosts revealed that circulating Fgfbp1-expressing cells migrate into healing wounds. We conclude that tissue-resident and circulating Fgfbp1-expressing cells modulate skin carcinogenesis and inflammation.

Figure 1. Generation of Fgfbp1 KO and GFP-knock-in mice

(a) Schematic of the knockout strategy for the Fgfbp1 gene depicts the endogenous locus of the Fgfbp1 gene on chromosome 5 and the targeting construct below. Mice with the targeted insertion of a floxed Fgfbp1 Neo and Gfp cassette in the Fgfbp1 locus (Fgfbp1^loxP-neo-gfp) were crossed with Cre expressing mice resulting in an insertion of a GFP reporter allele (Fgfbp1^gfp), or with FLPase mice resulting in a floxed Fgfbp1 (Fgfbp1^loxp) allele. Fgfbp1^loxP mice were further crossed with cre-mice to generate Fgfbp1-KO mice (Fgfbp^-/-, KO). Numbered arrows depict locations of genotyping primer (Suppl. Table S22). (b-f) PCR analysis of genomic DNA isolated from tail snips. (b) Primers 4 and 7 resulted in an amplicon of 5000 bp only in the allele with the floxed Fgfbp1 Neo and GFP cassette. (c) Primers 1 and 2 resulted in an amplicon of 478 bp with the 5′loxP site in the floxed Fgfbp1 allele and 416 bp in the WT allele. (d) Primers 3 and 6 resulted in an amplicon of 582 bp with the 3′loxP site in the floxed Fgfbp1 allele and 463 bp in the WT allele. (e) Primers 1 and 5 resulted in an amplicon of 484 bp only in the GFP reporter allele. (f) Primers 1 and 6 resulted in an amplicon of 432 bp in the KO allele and 2.5kb in the WT allele.

Figure 2. Fgfbp1-KO mouse epidermis is thicker and has reduced permeability

(a,b) Representative high magnification pictures of Masson's trichrome stained tissue section of skin from WT and KO mice shows a thicker epidermal layer (E) in KO mice. (b) In contrast to the epidermis, dermis, fat layer and muscle are not significantly different. (Mann Whitney test, WT n=5; KO n=6; 20 fields per skin sample). (c) Proliferation of basal keratinocytes in KO epidermis is higher than in WT (PCNA positive nuclei, Student's t test, n=5). (d) Nucleated keratinocyte layers in KO epidermis contain multiple strata whereas WT epidermis is mostly a monolayer (Chi-square test, n=3), (e) Representative images of skin section stained for Claudin 1, Filaggrin, PCNA (scale bar = 100 μm). (f) Transepidermal water loss (TEWL) was decreased in male KO compared to WT males (Student's t test, WT n=14; KO n=15). (g) RNA expression in separated dermis and epidermis (Student's t test, WT n=4; KO n=3).

Figure 3. Effect of topical Aldara treatment

(a) Daily treatment of shaved back skin of Fgfbpl^ggp mice for 4 days. Skin turned red as the inflammation progresses. (b) Myeloperoxidase (MPO) activity (Student's t-test, WT n=4; KO n=5). (c) Induction of Fgfbpl expression in WT skin. (d) GFP fluorescence in Aldara and ctrl (vaseline) treated mice for 5 days. Fluorescence intensity is shown as scaled counts/s in a heatmap from 0 (black) to 0.8 (dark red). (e) Quantification of GFP-fluorescence in panel b (Chi-square, n=2). (f) Masson's trichrome stained tissue sections of WT and KO epidermis (scale bar = 50 μm). (g) Quantification of the thickness of the epidermis (Student's t-test, n=7). (h) Effect on mRNA expression of Krtl6,116,1117a and 1123. (Student's t test, n=7)

Figure 4. DMBA/TPA (D/T) effects are delayed in Fgfbp1-KO mice

(a) Treatment scheme: A single topical treatment with DMBA was followed by biweekly TPA treatments for 180 days. (b) In vivo fluorescence of Fgfbpl ^+/gfp mouse skin after 180 days of treatment shows GFP activity in papillomas. Visible (left panel) and green fluorescence channels (right panel) are shown. Arrows indicate papillomas (scale bar = 5 mm). (c) Immunostaining of representative tissue sections with anti-GFP antibody shows staining of the epidermis (E) excluding the basal layer of keratinocytes. (Dermis (D), scale bar =200 μm (top panel), 100 μm (bottom panel), magnified in Suppl. Fig. S5b). (d) GFP RNA expression in Fgfbpl^+/gfp mice after 180 days of treatment (n=3). (e) The thickness of the epidermal layer progressively increased upon treatment. Biopsies were taken and the epidermal thickness was measured. On days 57 and 124 only papilloma-free skin sections were measured. (f) Papilloma on day 161. (g) Kaplan-Meier plot of papilloma occurrence (p<0.01; Mantel-Cox test, WT n=9; KO n=8). (h) Papilloma frequency (2-way ANOVA, WT n=9; KO n=8). (i) Vascular leakiness in papilloma indicated by extravasated erythrocytes (student's t-test, n=5, 35 fields per sample). (j) Representative anti CD31-stained tissue section; E=epidermis, V=vessel; interrupted white line surrounds area with extravasated erythrocytes (scale bar = 100 μm). (k) Illumina bead array analysis of DMBA/TPA treated skin shows induction of Fgfbp1 (left panel) and increased expression of S100a8, Krt16, Sprr2d and Klk6. Baseline expression is higher in Fgfbp1-KO skin but induced upon DMBA/TPA treatment (Student's t-test, WT n=4; KO n=3).

Figure 5. Wound healing is delayed in Fgfbp1-KO mice

(a) The back skin of WT and KO mice was wounded on day 0 and wound sizes were measured daily for 7 days (scale bar = 5 mm). (b) Quantification of wound sizes (2-way Anova, WT n=20; KO n=12). (c) Activation of the Fgfbp1 expression during wound healing indicated by fluorescence in Fgfbp1^+/gfp mouse skin. Heat map images are shown (scale bar = 2.5 mm). (d) GFP quantitation after wounding. GFP fluorescence peaks on day 4 (student's t-test for each time point, n=10). (e) Immunostaining of representative tissue sections with anti-GFP antibody (scale bar = 1 mm, magnified scale bar 100 μm, higher magnification in Suppl. Fig. S8b). (f) Representative images of hematoxylin/eosin stained wounds (scale bar = 1 mm), magnified sections show extravasated red blood cells (eR), single erythrocytes (arrow heads); (scale bar = 100 μm, granulation tissue (G), fat (F), scab (S), dermis (D), muscle (M), hair follicles/glands (H), epidermis (E) (higher magnification in Suppl. Fig S8c). (g) Relative opening of wounds, calculated by the ratio between wound opening divided by wound diameter (distance between collagen-containing dermis times 100, n=5). (h) Quantification of extravasated erythrocytes (Student's t test; WT n=5; KO n=6 wounds; means of at least 10 fields per wound). (i) Quantification of microvessels and capillaries (Student's t test, WT n=5; KO n=6 wounds; at least 10 fields per wound).

Figure 6. Fgfbp1-expressing cells migrate into wounds

(a to c) Host Fgfbp1 expressing cells enter wound in transplanted skin. (a) Time line (in days) and schematic of experiments. Skin from a WT mouse was transplanted onto the back of a Fgfbp1^+/gfp mouse, wounded after 7 days and fluorescence monitored. (b) Left panels: Bright field images and GFP fluorescence during wound healing (scale bar = 5 mm). Right panels: Immunostaining of a representative tissue section with an anti-GFP antibody shows staining of inflammatory cells that entered transplanted skin wound (scale bar = 250 μm). (c) Quantification of GFP signal in intact and wounded skin transplant (2-way Anova, n=7). (d to f) Wound healing with concurrent bone marrow transplant from the Fgfbp1^+/gfp reporter model. (d) Schematic of the experiment. (e) Bright field and GFP fluorescence images of wounds over 4 days (scale bar = 5 mm). Immunostaining of representative tissue sections with anti-GFP antibody confirmed GFP-positive cells in granulation tissue of the wound (Scale bars = 500 μm top right panel, 100 μm bottom right panel; higher magnification in Suppl. Fig. S9e). (f) Quantification of GFP activity; 2-way Anova, n=8. (g) Matrigel plug neoangiogenesis assay. H&E stained tissue sections of subcutaneously injected matrigel after 7 days in WT and KO mice. Scale bar, 100 μm. (h) Quantification of invasive cells per field in different regions of the matrigel. Regions 1 and 5 represent the edges of the plugs, regions 2-4 the center; WT n=12; KO n=20.