Immunoblot analysis of cell lysates from transfected HaCaT cells showed no significant difference in the levels of Flag-K14 expression from all the plasmids (Fig. 2a). Next, we performed immunofluorescence analysis using antibodies against Flag and K5. Flag-K14WT formed network structures that colocalized with endogenous K5 (Fig. 2b), indicating keratin filament dimerization between Flag-K14WT and endogenous K5. In contrast, a part of Flag-K14S281_L284del and Flag-K14N123S did not colocalize with K5. In addition, HaCaT cells transfected with the mutant K14 expression vectors showed numerous small ball-like Flag-tagged protein aggregates, suggesting disruption of the keratin filament network (Fig. 2b). The mean percentages of keratin clumped cells transfected with K14WT, K14S281_L284del and K14N123S were 8%, 25% and 49%, respectively (Fig. 2c). Our findings demonstrate that the heterozygous deletion mutation in K14 caused a loss in its ability to dimerize with endogenous K5; rather, it formed aggregates more frequently than wild-type K14. Sørensen C, et al. 8 previously reported similar functional analysis for mutant K14 proteins associated with the three major subtypes of EBS (EBS-loc, EBS-gen intermed and EBS-gen sev). In their study, the frequency of collapses of keratin filament network significantly increased only in mutant K14 associated with EBS-gen sev, whereas there were no differences among wild-type K14 and mutant K14 associated with EBS-loc and EBS-gen intermed. 8 In contrast, we observed that the numbers of cells in which collapsed keratin filaments were observed correlated with the clinical severity of EBS subtypes. Our study suggests that keratin filament formation assay could be useful for predicting the clinical severity and elucidating the pathological mechanisms of EBS. The mechanisms underlying the disruption of the keratin network should be the focus of further research.