Colocalization of kindlin-1, kindlin-2, and migfilin at keratinocyte focal adhesion and relevance to the pathophysiology of Kindler syndrome

Abstract

Kindler syndrome (KS) results from pathogenic loss-of-function mutations in the KIND1 gene, which encodes kindlin-1, a focal adhesion and actin cytoskeleton-related protein. How and why abnormalities in kindlin-1 disrupt keratinocyte cell biology in KS, however, is not yet known. In this study, we identified two previously unreported binding proteins of kindlin-1: kindlin-2 and migfilin. Co-immunoprecipitation and confocal microscopy studies show that these three proteins bind to each other and colocalize at focal adhesion in HaCaT cells and normal human keratinocytes. Moreover, loss-of-function mutations in KIND1 result in marked variability in kindlin-1 immunolabeling in KS skin, which is mirrored by similar changes in kindlin-2 and migfilin immunoreactivity. Kindlin-1, however, may function independently of kindlin-2 and migfilin, as loss of kindlin-1 expression in HaCaT keratinocytes by RNA interference and in KS keratinocytes does not affect KIND2 or FBLIM1 (migfilin) gene expression or kindlin-2 and migfilin protein localization. In addition to identifying protein-binding partners for kindlin-1, this study also highlights that KIND1 gene expression and kindlin-1 protein labeling are not always reduced in KS, findings that are relevant to the accurate laboratory diagnosis of this genodermatosis by skin immunohistochemistry.

Figure 1. Immunoblotting shows different expression patterns of kindlin-1 (but not kindlin-2 or migfilin) in KS or control keratinocytes

Immunoblotting shows (a) ∼77 and ∼74 kDa protein bands matching the molecular weight of kindlin-1 and its dephosphorylated form in HaCaT cell lysate (lane H). The same ∼77 kDa band was present in lane NHK but absent in lane KS. In addition, an ∼65 kDa band was seen in lane KS. (b) In lanes H, NHK, and KS, an ∼77 kDa band corresponding to the molecular weight of kindlin-2 was seen. (c) An ∼50 kDa band that equates to the expected molecular weight of migfilin was observed in lanes H, NHK, and KS. H, HaCaT cell lysate; NEG, negative control; NHK, normal human keratinocytes; KS, Kindler syndrome.

Figure 2. Co-immunoprecipitation studies show that kindlin-1, kindlin-2, and migfilin can bind to each other

(a) Western blotting on the immunoprecipitates using anti-kindlin-1 antibody shows ∼77 kDa bands in all three lanes, indicating that kindlin-1 forms a complex with kindlin-2 and migfilin. (b) Similarly, western blotting using anti-kindlin-2 antibody reveals ∼77 kDa bands in all three lanes, indicating that kindlin-2 associates with kindlin-1 and migfilin. (c) Immunoblotting performed using rabbit polyclonal anti-collagen VII IgG antibody as irrelevant antibody control shows absence of the ∼77 kDa protein band but persistence of the ∼50 and ∼25 kDa bands.

Figure 3. Confocal microscopy studies show different kindlin-1 localization but preserved kindlin-2 and migfilin localization in NHK and KS keratinocytes, respectively

(a) In NHK, kindlin-1 was expressed both in the cytoplasm and near the cell periphery and colocalized with vinculin at focal adhesions. Similarly, both (b) kindlin-2 and (c) migfilin showed cytoplasmic and peripheral distribution and colocalized with vinculin. (d) In KS keratinocytes, there was cytoplasmic localization of kindlin-1 but kindlin-1 expression at focal adhesions was absent, as evidenced by the lack of colocalization with vinculin. In contrast, (e) kindlin-2 and (f) migfilin subcellular localization was not affected. Bar = 20 μm.

Figure 4. Confocal microscopy shows that kindlin-1, kindlin-2, and migfilin all colocalize at focal adhesions

(a) Kindlin-1 localizes to the ends of actin stress fibers and also demonstrates a perinuclear distribution. (b) Migfilin is distributed both in the cytoplasm and at the cell periphery. (c) Endogenous kindlin-2 also localizes to the ends of actin stress fibers. (d) Merged confocal image shows colocalization of kindlin-1, kindlin-2, and migfilin at focal adhesions as shown by the white staining at the cell periphery (arrows). Bar = 20 μm.

Figure 5. Kindlin-2 and migfilin expression is not altered after RNAi-mediated knockdown of kindlin-1 in HaCaT cells

(a) Gradual reduction of kindlin-1 expression was noted from day 2 post-transfection. (b) Expression of kindlin-2 does not seem to be altered by kindlin-1 knockdown. (c) However, there was an increase in migfilin expression from day 6 post-transfection in the scrambled control and kindlin-1-transfected HaCaT cells. Oligofectamine-only (OLI)-treated HaCaT keratinocytes lysate was used as control and showed no reduction in kindlin-1 expression. In contrast, no reduction in kindlin-1 expression was seen when the cells were treated with scrambled siRNA.

Figure 6. Kindlin-2 and migfilin localization remains unchanged in kindlin-1 siRNA-transfected HaCaT keratinocytes

(a) In HaCaT keratinocytes transfected with kindlin-1 siRNA, there was no localization of kindlin-1 at focal adhesions (arrows). However, (b) kindlin-2 and (c) migfilin still colocalized with vinculin at focal adhesions. In contrast, in the scrambled control-transfected HaCaT cells, there was no change in (d) kindlin-1, (e) kindlin-2, and (f) migfilin subcellular localization. Bar = 20 μm.

Figure 7. Immunofluorescence microscopy shows two patterns of labeling for kindlin-1, kindlin-2, and migfilin in KS skin

(a) Immunolabeling with anti-kindlin-1 antibody on normal skin sections shows bright epidermal staining predominantly at the cell periphery (but also in the cytoplasm) mostly within the basal keratinocyte layer (arrows) but with less intense pan-epidermal labeling. (b) In the KS patient with the homozygous nonsense mutation p.E516X/p.E516X there is barely detectable staining for kindlin-1. (c) In contrast, in the KS patient with the homozygous mutation p.R288X/p.R288X there is similar kindlin-1 immunolabeling intensity to control skin (cf. Figure 7a). (d) Anti-kindlin-2 immunostaining of normal skin sections shows pan-epidermal membranous labeling with the absence of staining being noted along the lower pole of basal keratinocytes. (e) Marked reduction in anti-kindlin-2 antibody immunoreactivity is seen in the patient's skin harboring the homozygous nonsense mutation p.E516X/p.E516X. (f) Kindlin-2 antibody labeling in the patient with the homozygous nonsense mutation p.R288X/p.R288X is similar to that in normal skin. (g) Anti-migfilin staining of normal skin sections shows membranous and cytoplasmic labeling of the basal keratinocytes. (h) Anti-migfilin antibody labeling in the patient with homozygous nonsense mutation p.E516X/p.E516X shows barely detectable labeling. (i) Anti-migfilin antibody labeling in the patient with the homozygous nonsense mutation p.R288X/p.R288X is similar to that in normal skin. White dashed lines indicate the position of the dermal–epidermal junction in (b), (e), and (h). Bar = 50 μm.

Figure 8. Semi-quantitative RT-PCR shows correlation between KIND1 mRNA expression and the corresponding kindlin-1 labeling patterns

(a) Compared with the three normal controls (lanes 1–3), patients with reduced anti-kindlin-1 immunostaining (lanes 4–6) had variably reduced KIND1 mRNA levels, whereas the patient with normal immunolabeling of anti-kindlin-1 (lane 7) had a normal or near-normal level of KIND1 mRNA. (b) Patients with reduced kindlin-2 labeling (lanes 4–6), however, had no reduction in KIND2 mRNA expression compared with normal controls (lanes 1–3) or the patient with normal kindlin-2 immunostaining (lane 7). (c) Similarly, there was no difference in FBLIM1 mRNA expression in any of the patients compared with normal controls. (d) GAPDH mRNA expression was used as loading control in these experiments.