Kindler syndrome in mice and men

Abstract

Kindler syndrome (KS) in humans is a severe skin blistering disease associated with inflammation and increased risk of epidermal squamous cell carcinoma (SCC). This disease is known to be caused by loss-of-function mutations in Kindlin-1, a focal adhesion β-integrin binding protein. Thus far, it has been unclear what specific signaling events occur in KS keratinocytes to promote tumorigenesis, especially since loss of β-integrins and focal adhesion complexes has been previously shown to prevent or delay tumor formation. In the April issue of Nature Medicine, Rognoni and colleagues generate a transgenic mouse lacking Kindlin-1 in the epidermis to model the key features of KS, and show that Kindlin-1 regulates Wnt and TGFβ signaling independent of β-integrins. These β1-integrin-independent functions of Kindlin-1 may contribute to the increased SCC risk in KS patients.

Kindler syndrome (KS) is a rare autosomal recessive disease characterized by skin blistering, scarring, inflammation, photosensitivity, cutaneous and mucocutaneous squamous cell carcinoma (SCC) formation, and esophageal strictures that require surgical intervention. KS is associated with mutations in the KIND1 (FERMT-1) gene that result in either hypomorphic, or complete loss of Kindlin-1 activity.¹ Kindlin-1 binds directly to several β integrin tails—particularly integrin β1—to enhance their activation from within the cell (“inside-out activation”). Integrin αβ heterodimers are essential for cellular adhesion, proliferation and migration, and are particularly important for skin structure and function.² This can certainly explain the skin blistering phenotypes seen in KS patients, which are phenocopied in integrin β1 skin-specific knockout mouse models.^3,4 However, several additional human symptoms cannot be explained by current murine models. Particularly, a paradox exists regarding the formation of cutaneous SCCs in human KS patients. Current literature has supported a tumor-promoting role for β1 integrins in SCCs. Blockade of β1 integrin using blocking antibodies in a human-based skin model slowed tumor progression, and an activating β1 mutation (T1881A) in a transgenic mouse model showed increased SCC progression to malignancy.^5,6

Given this data, how can the tumor-suppressive function of Kindlin-1 in humans be explained? In a recent issue of Nature Medicine, Rognoni and colleagues demonstrate integrin β1-independent roles for Kindlin-1 using transgenic mouse models.⁷ The authors develop two transgenic mouse lines: one with conditional deletion of FERMT-1 using a K5-Cre promoter (Kind1-K5 mice), and one with integrin β1 T788A/T789A substitutions that prevent Kindlin-1 binding to the integrin β1 tail.⁸ As expected, skin-specific Kindlin-1 loss prevented the severe ulcerative colitis and early death seen in mice that completely lack the KIND1 gene.⁹ Both of the new transgenic Kindlin-1 mutant mouse lines displayed skin blistering at the epidermal-dermal junction in the mouse back skin. Blisters were associated with a brisk local inflammatory response, and increased epidermal back skin proliferation. This is consistent with the K5-Cre model of β1 loss in the epidermis, which also exhibited epidermal thickening.³

The most striking phenotype, which was seen only in the Kind1-K5 mice and not in the TTAA-K5 mice, was extensive hair follicle bulge stem cell expansion, resulting in densely packed hair follicles with increased numbers of hair shafts and bulges per follicle. This stem cell expansion was maintained for approximately one year, and subsequently declined, resulting in hair thinning later in the mouse’s life, which likely reflects stem cell exhaustion. Stem cell niches are known to be maintained through a balance of growth-promoting and growth-suppressive signals. The bulge stem cell niche is maintained specifically through the pro-proliferative effects of Wnt/β-catenin signaling, and the growth-suppressive effects of TGFβ signaling.¹⁰ The authors showed changes in both pathways in the Kind1-K5 mice: enhanced Wnt/β-catenin signaling in the hair follicle bulge and the interfollicular epidermis, with corresponding diminished TGFβ signaling in both compartments. The induction of Wnt/β-catenin signaling is likely due to integrin-independent induction of Wnt5a secretion, though additional experiments are needed to determine the mechanistic basis for this finding. The hair follicle cycling defects were efficiently rescued through blockade of Wnt protein secretion, or stabilization of the β-catenin destruction complex, effectively showing that the hair follicle phenotypes are the result of aberrant Wnt signaling. The diminished TGFβ signaling results from reduced release of active TGFβ1 from the latent form, LAP-TGFβ1, due to decreased activity of αvβ6 integrin signaling. The authors reasoned that this was likely due to loss of Kindlin-1 binding to the tail of integrin β6. It is likely that alterations in both Wnt and TGFβ1 cooperate to drive the hair follicle induction, although the relative importance of these two pathways is not clear. Interestingly, integrin β6-deficient mice have highly proliferative hair follicles due to loss of TGFβ signaling, though this effect seems less robust than that seen in the Kind1-K5 mice.¹¹

The relevance of these new mouse models to human disease is further validated by the demonstration of decreased TGFβ signaling and an induced Wnt/β-catenin activity in skin sections from human KS. Despite this similarity, KS patients do not exhibit hair follicle defects, potentially due to the fact that human hair follicle density is much less than that of mice and individual follicles may also be less sensitive to changes in Wnt signaling.

In order to determine whether the Kind1-K5 mice displayed the increased tumor susceptibility observed in human KS patients, authors treated these mice with topical DMBA-TPA to induce epidermal SCC tumor formation. Kind1-K5 mice did exhibit increased tumor incidence, though the tumors that formed were primarily trichofolliculoma-like lesions and basal cell carcinomas, indicating that these tumors likely arose from the expanded hair follicle bulge. These hair follicle-derived lesions are uncommonly seen in KS patients, who typically exhibit cutaneous and mucocutaneous squamous cell carcinomas. It is possible that this same tumor spectrum may arise in these Kind1-K5 mice at a later point during the carcinogen treatment, but that the highly expanded proliferative bulge cell population in the young mice is more easily transformed than interfollicular keratinocytes.

The authors have clearly shown a novel link between Kindlin-1, αv-class integrins, TGFβ signaling, and Wnt/β-catenin signaling in both mice and humans. However, it remains unclear which of these pathways contributes to increased stratified epithelial tumor susceptibility in human KS patients. While not directly shown, it is unlikely that β1 integrins are involved in the tumor suppressive functions of Kindlin-1. To support this, transgenic mice with an integrin β1 Y795F mutation, which partially abrogates Kindlin-1 binding to β1, do not exhibit any differences in tumor susceptibility.¹² It remains to be seen whether β1 mutations that completely abolish Kindlin-1 binding, such as the Y795A or TTAA-K5 mice used in this paper, show differences in tumor susceptibility.¹³ Interestingly, integrin αv deletion in the epidermis was shown to promote cutaneous tumor formation in the absence of p53.¹⁴ In this particular report, however, the increased tumor susceptibility was shown to be independent of TGFβ signaling, and involves increased Akt signaling and an immune suppressive response. Hair follicle-derived trichofolliculoma-like lesions are seen in mice with epidermal expression of constitutively active β-catenin, which become abundant in the skin of 3–4 mo old mice even in the absence of DMBA-TPA treatment.¹⁵ In contrast, loss of TGFβ signaling in the epidermis, through loss of TβRII, leads to no hair follicle defects, but rather spontaneous mucocutaneous squamous cell carcinoma formation in the anal and genital epithelia.¹⁶ This indicates that perhaps the loss of TGFβ signaling, rather than the increase in Wnt signaling, can account for the spectrum of tumors seen in Kindler syndrome patients.

The role of Wnt signaling in cutaneous tumor formation is unclear. Until recently, it was difficult to define a role for Wnt signaling in the interfollicular epidermis (IFE) using transgenic mouse models, due to weak reporter signal and the profound hair follicle phenotypes upon manipulation of Wnt pathway components. A recent paper demonstrated that Axin2-Cre-ERT2 mice can be used for manipulation of expression of Wnt pathway genes specifically in the IFE and hair follicle (HF) infundibulum, without affecting the HF bulge, secondary hair germ (SHG), or dermal papilla (DP).¹⁷ This transgenic model showed that, despite low level Wnt signaling in the IFE, β-catenin is essential for IFE proliferation. However, determining whether Kindlin-1 and IFE Wnt activation play a role in epidermal SCC will require additional mouse genetic models. Alternatively, the use of human skin organotypic skin reconstructs may also be valuable for deciphering the roles of TGFβ, αv-class integrins, and Wnt signaling in epidermis. There are key architectural differences between mouse and human skin that could account for some differences in the tumor types seen in the new Kindlin-1 epidermal knockout mice.^18,19 Conducting functional experiments directly in native human skin human which contains far fewer hair follicles, may avoid the rapid formation of hair follicle-derived tumors, and facilitate study of these pathways in potentially more medically-relevant setting.

What about the other skin phenotypes seen in KS patients? Interestingly, while the Kind1-K5 mice exhibited the hyperpigmentation seen in KS patients, the TTAA-K5 mice did not exhibit this abnormality. This result is surprising, considering that the pigmentation defect was initially thought to be due to the robust blister-associated inflammatory response in the skin. Despite similar blistering, there may be differences in the inflammatory response between the two transgenic mouse lines. Integrin αvβ6 could be influencing the inflammatory milieu of the skin in these mice, since tumors that form from αv-null epidermis exhibit an absence of neutrophils, macrophages and natural killer cells.¹⁴ αv-integrins in leukocytes themselves are essential for T cell infiltration into inflamed tissues.²⁰ This could play important roles in KS patients, where Kindlin-1 expression is lost in all tissues, not just epidermis. It remains to be seen whether Kindlin-1 loss in leukocytes leads to the same loss of αv-integrin function, and whether this plays a role in skin inflammation seen in KS patients.

It is also unclear what role the blistering-induced inflammation plays in SCC formation in KS patients. An inflammatory milieu could predispose these mice and men to SCC formation resulting from the associated hyper-proliferative wound healing response, which may make the proliferating keratinocytes more susceptible to accumulating mutations in driver oncogenes. This is commonly seen in patients with epidermolysis bullosa, another family of genetic epidermal blistering disorders associated with a robust inflammatory response, chronic ulceration, and an increased risk of developing SCCs.²¹ Reduction of inflammation using pharmacological inhibitors, or other genetically modified mice, may help clarify the role of inflammation in tumor development in the setting of KS.

Finally, further investigation will be required to define the specific mechanism by which Kindlin-1 loss promotes Wnt signaling in keratinocytes. How does Kindlin-1 loss lead to induction of Wnt5a secretion? Does Kindlin-1 interact directly with cell-cell adhesion complexes? One study showed co-localization of Kindlin-1 with both focal adhesion and cell-cell adhesions in tissue sections of the human intestinal epithelium, implicating a potential role for Kindlin-1 in restraining Wnt signaling through direct interaction with β-catenin.²² However, an independent study showed no change in β-catenin localization or expression upon loss of Kindlin-1 in cultured keratinocytes.²³ Whether this is also the case in native human skin in unknown. It will thus be a fruitful area of research to determine the specific mechanism, and relative importance of any cross-talk between Kindlin-1 at focal adhesions interacting with the extracellular matrix, and downstream signaling from cell–cell adherens junctions.

Glossary

Abbreviations:

DP

dermal papilla

DMBA-TPA

7,12-Dimethylbenz(a)anthracene- 12-O-Tetradecanoylphorbol-13-acetate

HF

hair follicle

IFE

interfollicular epidermis

K5

keratin 5

K14

keratin 14

KS

Kindler syndrome

SHG

secondary hair germ

SCC

squamous cell carcinoma

TGFβ

transforming growth factor β

TβRII

transforming growth factor β receptor II

Rognoni E, Widmaier M, Jakobson M, Ruppert R, Ussar S, Katsougkri D, Böttcher RT, Lai-Cheong JE, Rifkin DB, McGrath JA, et al. Kindlin-1 controls Wnt and TGF-β availability to regulate cutaneous stem cell proliferation. Nat Med. 2014;20:350–9. doi: 10.1038/nm.3490.