Expression of filaggrin in HaCaT keratinocytes
We initially generated recombinant filaggrin monomers containing a linker domain (r-mFLGs1 and 2, Supplementary Fig. 1), which behaved as ~ 45 kDa proteins (Fig. 1A), despite the estimated masses being around 34kDa, which was consistent with a previous report using the equivalent r-mFLG [7]. Polyclonal antibodies against mFLG1 selectively labeled upper layers of the skin epidermis, consistent with filaggrin distribution (Fig. 1B). To investigate the expression profile of filaggrin, we employed HaCaT keratinocytes cultured in conventional medium, since these cells do not undergo apoptosis with abundant expression of filaggrin mRNA [18], retain several keratinocyte characteristics [19], and are readily transfected [12, 20]. The antibodies could predominantly detect the large proteins in HaCaT cells, but not those with genetic ablation of filaggrin, demonstrating that HaCaT cells produce profilaggrin (Pro-FLG). On the other hand, apparent signals for mFLGs (30 ~ 45 kDa) were not detected, and the artificial expression of the gene product of introduced mFLG1 was not feasible in HaCaT cells. These results indicate that HaCaT keratinocytes lost the ability to process Pro-FLG, or mFLGs are unstable in these cells (Fig. 1B and C), albeit they were ascertained for the downregulation of several differentiation markers including Pro-FLG in low Ca2+ medium (data not shown). Given that Pro-FLG-processed products, mFLGs and the N-terminal fragment, prime apoptotic cell death via assemble keratin cytosketeton [21, 22] or intranuclear components [23], these results might explain why HaCaT cells can be maintained in conventional medium with relatively higher Ca2+. Also, it is suggested that the effective processing of Pro-FLG does not occur in cultured keratinocytes [24].
EPM and expression of Pro-FLG
We previously showed that extracellular epimorphin (EPM) prevented cornification in HaCaT cells [25]. To address its possible involvement in the expression of Pro-FLG, these cells were introduced with an expression construct for extracellular EPM tagged with the T7-peptide under the control of dox (Fig. 2A). Upon treatment with extracellular EPM, the amount of Pro-FLG significantly decreased (Fig. 2B), and keratohyalin granule-like structures comprising Pro-FLG were almost eliminated in cells with high EPM expression (Fig. 2B). Intriguingly, filaggrin mRNA was not downregulated in cells with EPM (Fig. 2B), and this was the case for several pro-FLG-processing proteases [26–29] including calpain-1, kallikrein-5 (KLK5), and KLK5- regulator LEKT1 (Supplementary Fig. 2A). Additionally, artificial inhibition of proteasomal or lysosomal degradation had no significant effects, and EPM did not accelerate the decline of Pro-FLG in cells treated with cycloheximide, a translation blocker (Supplementary Fig. 2B and C), suggesting that the decrease in Pro-FLG expression might be attributed to perturbation of protein translation. Indeed, while phosphorylation of the translation initiation factor elF4E was not affected, an active phosphorylation of ribosomal protein S6, which reportedly occurred when translation of mRNA having long ORF is suppressed[30], was seen in cells expressing EPM (Fig. 2C). It is notable, however, that doxycycline reportedly exerts unexpected functions in addition to antibacterial actions in keratinocytes [31–34]. To further ascertain the effect of extracellular EPM on the attenuation of Pro-FLG, parental HaCaT cells were treated with active form of recombinant EPM (r-EPM), in the absence of dox. Consistent with the results using HaCaT transfectants, r-EPM decreased Pro-FLG levels in HaCaT cells (Fig. 2D). Thus, it is conceivable that signals propagated by extracellular EPM hinder the translation of filaggrin mRNA (Fig. 2E)
EPM and processing of mFLG
Next, we tried to investigate possible effects of extracellular EPM on the cleavage of mFLGs that occurs upon epidermal cornification. As seen in transient mFLG expression (Fig. 1C), we failed to detect exogenous mFLG1 in HaCaT cells by a stable expression system using piggyback transposon activity (Fig, 3A). In addition, it has been suggested that doxycycline might regulate enzymatic activity of caspase 14, a keratinocyte-specific protease responsible for mFLG cleavage [35, 36]. Thus, we used another approach where recombinant mFLG1 was fused with a cell-penetrating peptide [37], and added to HaCaT cells treated with r-EPM. To this end, we used a fusion protein system established by Liu laboratory [38], where flag-tagged mFLG1 was fused with positively-charged GFP (+ 36GFP) for cell surface association followed by endocytotic incorporation, and an antibacterial peptide, Aurein, for disruption of the endosomal membrane (r-A/36G/flag-mFLG1) (Fig. 3B). While r-GFP did not adsorb to the surface of HaCaT cells, r-A/36G/flag-mFLG1 was effectively incorporated and accumulated in the cytoplasm (data not shown, see Fig. 3C). Analyses using anti-flag tag antibodies revealed that r-EPM protects the processing/degradation of mFLG1 (Fig. 3C), coincidently with the tendency to prevent cleavage/activation of caspase14, a mFLG-cleavage enzyme for NMF production [6, 7] (Fig. 3D). Concomitant with the inhibitory effect on the expression of Pro-FLG, a source of mFLG (Fig. 2), these results indicate that the signals propagated by extracellular EPM decreases the amount of NMF in fully differentiated keratinocytes.
In conclusion, this study revealed that EPM, which is extruded upon several external stimuli, not only decreases the expression of Pro-FLG but also hinders cleavage of mFLGs for NMF production (Fig. 4). At present, whether these EPM-dependent effects lead to, or are a consequence of, defects in epidermal cornification remains unclear, and the effects of r-EPM on differentiation of FLG-KO cells should be analyzed. Nevertheless, molecular elements for membrane translocation of EPM might be therapeutic targets of keratotic lesions, considering that all of Pro-FLG, mFLG and NMF play distinct but critical roles in epidermal cornification.