E‐FABP induces differentiation in normal human keratinocytes and modulates the differentiation process in psoriatic keratinocytes in vitro

Epidermal fatty acid‐binding protein (E‐FABP) is a lipid carrier, originally discovered in human epidermis. We show that E‐FABP is almost exclusively expressed in postmitotic (PM) keratinocytes, corresponding to its localization in the highest suprabasal layers, while it is barely expressed in keratinocyte stem cells (KSC) and transit amplifying (TA) keratinocytes. Transfection of normal human keratinocytes with recombinant (r) E‐FABP induces overexpression of K10 and involucrin. On the other hand, E‐FABP inhibition by siRNA downregulates K10 and involucrin expression in normal keratinocytes through NF‐κB and JNK signalling pathways. E‐FABP is highly expressed in psoriatic epidermis, and it is mainly localized in stratum spinosum. Psoriatic PM keratinocytes overexpress E‐FABP as compared to the same population in normal epidermis. E‐FABP inhibition in psoriatic keratinocytes markedly reduces differentiation, while it upregulates psoriatic markers such as survivin and K16. However, under high‐calcium conditions, E‐FABP silencing downregulates K10 and involucrin, while survivin and K16 expression is completely abolished. These data strongly indicate that E‐FABP plays an important role in keratinocyte differentiation. Moreover, E‐FABP modulates differentiation in psoriatic keratinocytes.


Introduction
Epidermal differentiation is a physiological mechanism characterized by morphological changes and by the expression of a variety of markers. Changes in keratinocyte differentiation are associated with variation in lipid composition of epidermal layers (1). During this process, keratinocytes start to express lipid carriers, including high amounts of epidermal fatty acid-binding protein (E-FABP) (2). E-FABP belongs to the family of FABPs and represents the epidermal subtype. E-FABP is not only exclusively expressed in epidermis (3,4), but it is also detected in adipose tissue, endothelial cells, brain, liver, kidney and mammary tissue (5). E-FABP fulfils different roles, including fatty acid transport, control of fatty acid metabolism and cell migration (6). It also regulates cytokine productions (7), and it is related to all-trans retinoic acid sensitivity in cancer cells (8,9). E-FABP is capable of binding other types of long-chain fatty acids and transporting them from the inner plasma membrane to different cell compartments (10). When fatty acids are transported to the nucleus, binding of FABPs to transcription factors of the PPAR family may activate the differentiation process (10)(11)(12). E-FABP is involved in the differentiation mechanism of several cell types, such as T helper cells, neural cells and mouse keratinocytes (13)(14)(15).
Psoriasis is an immuno-mediated, hyperproliferative disease characterized by abnormal differentiation of epidermis that also displays an altered calcium metabolism and a defective response to extracellular calcium gradients (16). E-FABP is overexpressed in psoriatic keratinocytes as compared to normal cells (3). Yet, a functional role of E-FABP in human keratinocyte differentiation both in normal and psoriatic epidermis remains to be determined.
Here, we show that either silencing or overexpressing E-FABP markedly affects keratinocyte differentiation. We also demonstrate that E-FABP is more expressed in postmitotic (PM) cells from psoriatic epidermis than in the same subpopulation from normal skin. Finally, silencing E-FABP alters differentiation in psoriatic keratinocytes.

Cell culture
Human keratinocytes were obtained from neonatal foreskin, adult and psoriatic skin and cultured in keratinocyte growth medium (KGM) (Lonza, Basel, Switzerland) as described previously (17). To induce differentiation, cells were grown to 30-40% confluency and treated with 1.8 mM calcium chloride.
Keratinocyte subpopulations were obtained as previously shown (18,19). Briefly, total keratinocytes were first allowed to adhere to human type IV collagen-coated dishes for 5 min, to obtain a population enriched in keratinocyte stem cells (KSC). Keratinocytes adhering overnight were considered TA cells, as previously reported (19). Non-adhering cells represent a population of terminally differentiated, postmitotic (PM) keratinocytes.

Transfection of normal and psoriatic keratinocytes
Human keratinocytes, either normal or psoriatic, were plated in antibiotic-free KGM medium. After 24 h, cells were transfected with 75 nM FABP5 or scrambled siRNA (Thermo Scientific, Denver, CO, USA), on-Target plus smart pool human FABP5 or on-Target plus siControl non-targeting pool, combined with Lipofectamin 2000 and Opti-MEM (both from Invitrogen, Paisley, UK), according to the manufacturer's instructions. Cells were transfected twice and used 48 h later for Western blotting. Psoriatic and normal human keratinocytes transfected with FABP5 siRNA were either treated or untreated with calcium 1.8 mM immediately after transfection. Cells were lysed for Western Blotting, and MTT assay was performed on the cells 48 h after treatment.
The protein delivery was performed on Gene Pulser System (Bio-Rad Laboratories Inc., Hercules, CA, USA). Normal human keratinocytes were grown under subconfluent conditions in KGM medium, detached from the dish and resuspended in 60 ll serumfree DMEM with (+rE-FABP) or without (ÀrE-FABP) 5 lg of rE-FABP or with the same amount of a 18.8-kDa control protein (Cayman Chemicals, Ann Arbor, MI, USA). Both recombinant proteins were purified and delipidated by the manufacturer (Cayman Chemicals) before use. The cell suspensions were transferred into two 0.1-cm sterile cuvettes (Bio-Rad) and left in ice for 10 min. The applied current was 100 V, 250 lFD, 7.0 ms for + rE-FABP sample, 100 V, 250 lFD, 6.9 ms for control protein and 100 V, 250 lFD, 6.4 ms for ÀrE-FABP sample. The treated cells were left in ice for 10 min and either lysed (0 h) or transferred into 3 9 6-cm-well plates. To each plate, 5 ml of KGM medium was added and cells were lysed 48 h later.

MTT assay
Cells were plated in a 96-well tissue culture plate (5000 cells per well), and MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay was performed at 48 and 120 h after transfection. Proliferative cells were detected by incubating with MTT (Sigma-Aldrich, St. Louis, MO, USA) solution at 37°C for 4 h. The formazan dye produced after DMSO solubilization was evaluated by a multiwell scanning spectrophotometer at 540 nm. The results are expressed as viability percentage, as compared to control. Results are calculated as the mean AE SD of three different experiments. Student's t-test was performed for comparison of the means.

Western blotting analysis
Total lysates from lesional and non-lesional psoriatic epidermis and healthy epidermis were obtained as follows. Punch biopsies were washed with PBS and treated with dispase II (Roche, Basel, Switzerland) overnight at +4°C. The day after, epidermal sheets were separated from the dermis and homogenized in RIPA buffer. The study was conducted in accordance with the Declaration of Helsinki Principles, approved by the local ethics committees of the participating institutions. A total of 12 patients with untreated psoriasis and 21 healthy controls were enrolled in the study.

Real-time PCR
Total RNA was extracted from keratinocytes using TRI Reagent method performed as described by the manufacturer (Sigma-Aldrich). One microgram of total RNA extracted was reverse-transcribed as described by the manufacturer (Roche). Nucleotide sequences of the oligomers used (MWG Biotech, Ebersberg, Germany) were as follows: E-FABP-DP 5′-3′: ATGGCCACAGT TCAGCAGCTG, RP 5′-3′: CAGGTGACATTGTTCATGAC; involucrin-DP 5′-3′: GGACTGCCTGAGCAAGAATGTG, RP 5′-3′: TAAG CTGCTGCTCTGGGTTT; K10-DP 5′-3′: CCTTCGAAATGTGTCC ACTGG, RP 5′-3′: CAGGGATTGTTTCAAGGCCA; GAPDH-DP 5′-3′:ACATCGCTCAGACACCATG, RP 5′-3′:TGTAGTTGAGGTC AATGAAGGG. SYBR green Taq-DNA polymerase mixture (Roche, Basel, Switzerland) was used for real-time PCR using an ABI 7500 (Applied Biosystems, Foster City, CA, USA). The differences in cycle number past the threshold (DC t ) is reflective of differences in the initial template concentration in the different samples tested. Data are expressed as fold change relative to GAPDH. PCR was carried out at least three times for each sample, and the experiment was performed in triplicate. Data from each sample were compared with stem cells or healthy PM cells, as calibrators, using the Sequence Detection Software, version 1.2.3, according to the Relative Quantification Study method (Applied Biosystems).

Statistical analysis
The Student's t-test was used to compare the average intensities of Western blot bands, average viabilities and average cell counts. One or two asterisks indicate a significant difference, 0.01 < P < 0.05 and P < 0.01, respectively.

E-FABP is expressed in normal human postmitotic keratinocytes
As previously shown, E-FABP is expressed in the higher spinous layers with a weak pattern that intensifies moving upward to the outermost layers, consistent with increased cell differentiation (Fig. S1a). To further evaluate E-FABP localization in epidermis, we separated keratinocyte subpopulations as previously shown (14) and analysed E-FABP expression by Western blotting (Fig. 1a) and E-FABP, K10 and involucrin expression by real-time PCR (Fig. 1c). E-FABP was mostly expressed in PM keratinocytes, while it was weakly detectable in TA cells and absent in KSC, both at the mRNA and protein levels. This was also confirmed by immunofluorescence performed on keratinocyte subpopulations, analysed immediately after separation. KSC expressed no E-FABP, involucrin or K10, while only few TA cells blandly expressed E-FABP. On the other hand, almost all PM cells expressed E-FABP and involucrin, while K10 was barely detected (Fig. 1b). Finally, K10 and involucrin were mostly expressed by PM cells and weakly stained in KSC or TA cells (Fig. 1c).

E-FABP modulates differentiation through NF-kB and JNK1 signalling pathways in normal human keratinocytes
Because E-FABP is expressed in PM cells, we asked whether this protein could be involved in keratinocyte differentiation. Increased E-FABP levels in keratinocytes cultured in the presence of high calcium concentration or under confluent conditions have been previously shown (4). However, a direct correlation between RNA and protein levels has not been analysed. We induced differentiation in human keratinocytes either by adding calcium to the medium or by growing cells to confluency. Under calcium-induced keratinocyte differentiation, E-FABP levels increased in a timedependent manner up to 96 h, both at the mRNA (Fig. S1f) and protein levels (Fig. S1b,d), in parallel with K10 and involucrin expression. E-FABP was also overexpressed in confluent keratinocytes from 24 h up to 72 h, as compared to preconfluent cells, both at the mRNA (Fig. S1g) and protein levels (Fig. S1c,e). Similarly, both involucrin and K10 expression increased in confluent keratinocytes (Fig. S1c,e and g).
Although E-FABP is modulated during keratinocyte differentiation, the ability of E-FABP to stimulate human keratinocyte differentiation is not completely understood. We transfected normal human keratinocytes with a recombinant E-FABP (rE-FABP). rE-FABP was successfully delivered to the cells, as confirmed by the 18-kDa band, corresponding to the exogenous protein, at 0 h. At the same time point, a band corresponding to endogenous E-FABP was equally expressed in -rE-FABP and +rE-FABP cells and cells transfected with the control protein (Fig. 2a). On the other hand, at 48 h, no band corresponding to rE-FABP was visible, while a 10-kDa band appeared, possibly indicating a degradation or a modification of the recombinant protein. At this time point, cells transfected with rE-FABP overexpressed K10 and involucrin, while the same cells transfected with a control protein failed to induce K10 and involucrin increase (Fig. 2b).
Because E-FABP induces a modulation of differentiation marker expression, we wanted to confirm this finding by silencing E-FABP. We transfected human keratinocytes with a FABP5-specific siRNA in the presence or absence of calcium chloride. As shown in Figure 2d 120 h after transfection (Fig. S2a). Similarly, confocal microscopy confirmed the reduction in K10 in siRNA-treated cells and revealed that proliferation rate was not affected by E-FABP silencing, as shown by the unchanged expression of Ki67 (Fig. 2c). A semiquantitative analysis of K10 and Ki67 staining is reported in Fig. S2b. To further investigate the role of E-FABP in differentiation, we studied NF-kB/JNK1 signalling pathways in human keratinocytes transfected with scramble siRNA or E-FABP siRNA, in the presence or absence of calcium. As shown in Fig. 2d and S2c, E-FABP silencing reduced the levels of activated NF-jB expression (represented by the phosphorylated form of the protein, P-NF-jB) but not of total NF-jB. The effect on P-NF-jB was even more evident in the presence of calcium. On the other hand, E-FABP downregulation increased IkBa expression, in particular, in the presence of calcium. In addition, E-FABP siRNA reduced active JNK1 expression, while calcium failed to induce JNK1 activation. Altogether, these data suggest that lack of E-FABP decreases differentiation markers through NF-kB and JNK signalling pathways.

E-FABP expression in psoriatic keratinocyte subpopulations
Psoriasis is characterized by altered keratinocyte differentiation, and E-FABP expression appears to be increased in psoriatic epidermis. E-FABP was more expressed in psoriatic skin as compared to healthy skin in vivo (Fig. S3a,b). This was also confirmed by Western blotting, showing that lesional psoriatic epidermis expresses higher levels of E-FABP, as compared to non-lesional psoriatic or healthy epidermis, partially mimicking the expression of involucrin (Fig. 3a,b). Consistently, in psoriatic sections, E-FABP and involucrin were mostly detected in high spinous layers (Fig. S3b). Moreover, S100A7 and K16 were exclusively expressed in psoriatic epidermis, their expression partially overlapping E-FABP localization in the high spinous layers (Fig. S3b). In addition, cytoplasmic survivin, a marker of KSC, was expressed in the basal layer of normal epidermis, while it was expressed also in suprabasal psoriatic keratinocytes with an almost exclusively nuclear pattern (Fig. S3a,b).
To further evaluate the differential expression of E-FABP in normal and psoriatic epidermis, we separated keratinocyte subpopulations and evaluated E-FABP expression by real-time PCR and Western blotting. At the mRNA and protein levels, E-FABP was mostly expressed in PM psoriatic keratinocytes, while it was almost undetectable in KSC and TA cells (Figs 3c,f and S3c, respectively). This was in line with increased mRNA levels of K10 and involucrin in psoriatic PM keratinocytes as compared to KSC and TA cells (Fig. 3d,e). In addition, by comparing equal amounts of PM cell lysates from healthy and psoriatic skin, we found that psoriatic PM cells expressed significantly higher levels of E-FABP as compared to normal PM cells both at the protein (Figs 3g and S3d) and mRNA levels (Fig. 3h).

E-FABP modulates differentiation and psoriatic markers in psoriasis
Although psoriatic keratinocytes are difficult to culture (21), low-calcium and serum-free conditions allow to culture psoriatic cells for multiple passages (16,22). Under these conditions, low-passage psoriatic keratinocytes expressed higher levels of K10, K16 and S100A7 than normal keratinocytes (Fig. 4a). To further evaluate E-FABP's role in psoriasis, we treated psoriatic keratinocytes with a siRNA specific for E-FABP, with or without high calcium concentration. In absence of calcium, E-FABP silencing induced a slight decrease in K10 and involucrin expression, while psoriatic markers survivin and K16 were upregulated. Upon high-calcium conditions, in E-FABP siRNA-treated psoriatic keratinocytes, K10 and involucrin expression was further downregulated and survivin was markedly reduced, while K16 expression was completely abolished (Fig. 4b,c).

Discussion
In this work, we show that E-FABP is abundantly expressed in PM keratinocytes both in healthy and psoriatic epidermis. Moreover, decreased E-FABP expression deranges keratinocyte differentiation in normal and psoriatic cells. E-FABP is localized in suprabasal layers of healthy epidermis, with increased expression in the spinous and granular layers. The lipid composition of cell membranes changes during keratinocyte differentiation, increasing from suprabasal layers to the more differentiated stratum granulosum and corneum. It has been previously suggested that E-FABP is a potential marker of TA keratinocytes in human epidermis (5). However, the method by O'Shaughnessy et al. does not distinguish between TA and PM subpopulations, including both cell subsets in the non-adherent pool of keratinocytes. By separating KSC from TA and PM cells based on b1-integrin expression levels, as previously reported (18,19), we show that E-FABP is highly expressed in PM cells, while it is barely detected in TA cells and KSC. Indeed, E-FABP is expressed in the highest epidermal layers, and PM keratinocytes are terminally differentiated cells that do not proliferate in vitro (18). This result indicates the possibility of a tight correlation between E-FABP function and the state of keratinocyte differentiation. Keratinocyte differentiation is induced  by high-calcium conditions and in confluent cells. We show that, when keratinocytes are induced to differentiate, E-FABP mRNA and protein increase, along with a higher expression of involucrin and K10 protein, suggesting that not only E-FABP is mainly localized in differentiated cells in vivo, but it also increases when differentiation is promoted in culture. It remained to be determined whether E-FABP is able to induce differentiation in human epidermis (23). In the neural system, E-FABP stimulates PC12 cells differentiation by inducing neurite extension (14). There is also evidence that E-FABP is actively involved in the differentiation programme of murine keratinocytes (4,24). E-FABP-deficient mice display defects in water barrier functions, albeit having normal skin morphology and no major defects in other organs (23,25). This seems to be in contrast with a possible role for E-FABP in keratinocyte differentiation. However, keratinocytes obtained from these mice show reduced expression of differentiation markers such as K1, involucrin and loricrin, in the absence of exogenous ligand, and decreased susceptibility to calcium-induced differentiation, as compared to E-FABP +/+ cells (15). These findings suggest the presence of still unknown compensatory mechanisms in mice lacking E-FABP expression, which lead to normal development of epidermis. In line with E-FABP involvement in the differentiation process of keratinocytes, it has been recently shown that, once bound to a specific ligand, E-FABP is able to physically interact with the transcription factor PPARb, thus activating the differentiation programme in mouse keratinocytes (13). However, to date, no data on human keratinocytes indicate an active role of E-FABP in keratinocyte differentiation. In this study, we show that administration of rE-FABP to human keratinocytes induces an increased expression of K10 and involucrin. The fact that E-FABP is able to act in the absence of ligand could be accounted for by some residual fatty acids in the media. Treatment of keratinocytes with rE-FABP also results in the appearance of a smaller 10-kDa E-FABP. This may be due to the degradation of the recombinant protein itself (26) or the product of a post-translational modification of rE-FABP. The role of E-FABP in keratinocyte differentiation was confirmed by silencing with specific E-FABP siRNA. As expected, E-FABP inhibition induced a downregulation of both K10 and involucrin, indicating that E-FABP stimulates human keratinocyte differentiation. E-FABP depletion does not influence keratinocyte viability that appears to be slightly affected by decreased differentiation, in line with previous findings (15). In agreement with recent reports in mice showing that reduced NF-kB activity is responsible for decreased K10 expression in E-FABP À/À mouse keratinocytes (15), we report that repression of differentiation induced by E-FABP silencing is associated with a reduction in activated NF-kB expression in normal human keratinocytes. However, calcium treatment did not influence NF-kB expression, as previously suggested (27).
In mouse epidermis, activation of PPAR-b/c signalling, which is mediated also by E-FABP binding, reproduces a psoriasis-like phenotype (26). Psoriasis is a skin disease with defects in cell differentiation and proliferation. Here we show that E-FABP is abundantly expressed in psoriatic epidermis as compared to normal skin, as previously reported (4,28). In normal human skin, E-FABP is mainly expressed in granular layers, with a decreased staining from upper to lower spinous layers. By contrast, in psoriatic skin, E-FABP is highest in spinous layers, thus behaving as other differentiation markers, such as involucrin and K10, suggesting a possible involvement of E-FABP in the altered differentiation process observed in psoriasis. It has been shown that increasing the proliferation time period of TA keratinocytes generates a psoriatic phenotype, implying that psoriasis possibly originates by alteration of TA cells (29). On the other hand, in the present study, psoriatic PM keratinocytes express higher levels of E-FABP than normal PM cells. This suggests that E-FABP identifies keratinocytes that have concluded the differentiation programme and that do not contribute to the proliferative compartment of the skin. Recent works show that cultured psoriatic keratinocytes retain some characteristics of the disease under optimized culture conditions (30). We show that psoriatic keratinocytes at early passages, under low-calcium and serum-free conditions, express K16 and S100A7. Downregulation of E-FABP in these cells reduces involucrin and K10 expression, thus showing that E-FABP mediates differentiation also in psoriatic cells. Not only E-FABP mediates psoriatic keratinocyte differentiation, but E-FABP depletion also upregulates both K16 and survivin, which are overexpressed in psoriatic cells in vivo. Thus, E-FABP seems to play an important role in the dysregulated epidermal homoeostasis observed in psoriasis. This effect is completely abrogated when cells are treated with high calcium. Psoriatic keratinocytes in culture display an inborn defect in calcium metabolism, thus being not responsive to high extracellular calcium (16). E-FABP downregulation further reduces psoriatic keratinocytes sensitivity to calcium, leading to decreased differentiation and to reduced expression of psoriatic markers. Altogether, these data suggest that E-FABP modulation may alter calcium metabolism and uptake in psoriasis.
We can conclude that E-FABP is able to induce normal human keratinocyte differentiation and that its upregulation could partially be involved in the altered differentiation mechanism in psoriasis.