Introduction

The lipid mediator platelet activating factor (PAF, 1-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine) is found in excess in psoriatic scale and plasma [15] as well as in other inflammatory skin conditions [34]. Bayerl et al. [6] also reveal increased expression of PAF receptors in psoriatic skin.

Phospholipase A2 enzymes are central mediators of inflammatory responses, including psoriasis [30]. The PLA2 superfamily of enzymes is often divided into three broader categories: (1) extracellular secretory PLA2s (sPLA2), (2) cytosolic calcium-dependent PLA2s (cPLA2) and (3) cytosolic calcium-independent PLA2s (iPLA2). Recent evidence suggests that all classes of PLA2 enzymes may participate in agonist-induced arachidonic acid (AA) release [30, 36]. In HaCaT keratinocytes, we previously detected expression of sPLA2 isoenzymes group IIa, IId, V and X, as well as the ubiquitous group IVα (cPLA2) and VI (iPLA2) [4]. PAF is known to activate cPLA2 in several cell types [14], but its role in activating other PLA2 isoenzymes is less well characterized.

The two synthetic fatty acids tetradecylthioacetic acid (TTA) and tetradecylselenoacetic acid (TSA) [22, 33] have previously been reported to have anti-inflammatory properties [35]. To further investigate PLA2-involvement in inflammatory skin conditions this paper examines and compares the activation and participation of PLA2 isoenzymes in PAF and calcium ionophore (A23187)-induced AA-release in keratinocytes, and if TTA and TSA interfere with this.

Results and discussion

Keratinocytes in the outer layer of the epidermis are differentiated, as opposed to proliferating keratinocytes attached to basal laminae. In order to obtain a cell state resembling mature skin, HaCaT keratinocytes [7] used in this study (kindly provided by Prof. N. Fusenig, Heidelberg, Germany) were differentiated in culture for 3 days prior to experimentation (protocol as described in refs. [3, 4]). The differentiation state of the cells was documented by examining the expression of three markers of suprabasal keratinocytes: Keratin 10 [8], S100A7 (psoriasin) [9] and sPLA2 IIA [2]. They all show an upregulation of more than fivefold compared to HaCaT undifferentiated state (Fig. 1a). [RNA was isolated using a Qiagen RNeasy Mini kit according to kit protocol, and RT2-PCR performed using the MX3000 system from Stratagene (primer sequences shown in Table 1)]. Other studies of HaCaT cells also demonstrate that upregulation of these genes are associated with a highly differentiated state [16, 29], thus the data suggest that the cultivation protocol used, indeed, differentiates the keratinocytes.

Fig. 1
figure 1

PAF induces AA and OA release in differentiated keratinocytes. a Initial cell state. Fold induction of differentiation specific marker genes in proliferating vs post-confluent HaCaT cells. (N = 2, result shown from one representative experiment). b Dose–response relationship for PAF stimulation of HaCaT cells. The response is measured as fold induction of [3H] arachidonic acid and [14C] oleic acids compared to the unstimulated control. PAF exposure time was 60 min. (All dose–response data in cell culture have been statistically validated using one way ANOVA at the 95% confidence level, and the results shown are representative of at least three consecutive experiments, using at least three parallel samples in each experiment)

Full size image
Table 1 Primers for keratinocyte differentiation markers
Full size table

Previous studies of PAF-stimulated AA-release in keratinocytes have used actively proliferating cells [10, 27]. In order to establish the pro-inflammatory potential of PAF in differentiated keratinocytes, HaCaT cells were stimulated with PAF-16 (purchased from Calbiochem), and the subsequent release of [3H]AA and [14C] oleic acid (OA) (NEN, Perkin Elmer) was measured. Changes in the levels of these free fatty acids may indicate the involvement of AA-selective or non-selective phospholipases, respectively [30]. (Protocol as described previously [3, 4], except that bovine serum albumin was excluded due to evidence that it may inhibit PAF activity [13].) At 60 min exposure and 20 μM concentration, PAF typically induces an AA induction of about twofold in a dose-dependent manner (Fig. 1b). The fold induction of OA is nearly as large as for AA, which is especially notable, since OA release is typically a minor fraction of the fold induction of AA-release [4, 25], and only rarely equal to it [36]. We may conclude that PAF has a pro-inflammatory effect in fully differentiated HaCaT, by induction of AA-release. Furthermore, AA-non-selective phospholipase A2s, as indicated by the observed OA release, contribute substantially to this effect.

The physiological relevance of the PAF concentration used deserves to be further commented upon. HaCaT cells were 100% viable at this PAF concentration as determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenoltetrazolium bromide (MTT) dye (Sigma) uptake assay [21] (data not shown). This study uses natural PAF, a readily degradable lipid—it has a half life of 0.5–2 s in serum [24]. HaCaT cells express both sPLA2 GX [4] and PAF-AH II [19], which are both known to degrade PAF [12, 19]. Additionally, recent developments in new quantitative methods for PAF suggest that the PAF concentration in serum in healthy humans may be as high as 0.2–0.3 μM [26], which is at least 100 times higher than previously thought. Thus, the PAF concentration used in this study is non-toxic and physiologically plausible.

Group IVα is the only truly AA-selective PLA2 isoenzyme [30]. In order to examine activation of group IVα PLA2 in PAF-mediated AA-release, HaCaT cells were treated with PAF in the presence or absence of the group IVα inhibitor methyl arachidonyl fluorophosphonate (MAFP) [11] (obtained from Cayman Chemicals), followed by lysis and analysis by in vitro group IVα PLA2 enzyme activity assay (for method description see [3, 18]). [All inhibitors used, their vehicles and the calcium ionophore do not affect viability in the concentrations applied in this study, as verified by MTT assays (not shown).] PAF-induced activation of group IVα PLA2 enzyme in HaCaT cells, and this response is reduced by 50% by MAFP (Fig. 2a). This in vitro assay is highly specific and the result thus strongly indicates that group IVα PLA2 participates in PAF-induced AA-release [18].

Fig. 2
figure 2

Different PLA2 inhibitors reduce PAF-induced AA and OA release. a In vitro cPLA2 activity assay on lysate from HaCaT cells pre-treated with 25 μM MAFP (90 min) and stimulated by 20 μM PAF (60 min) prior to lysis. [95% significance indicated by asterisk (compared to unstimulated control) and Δ (compared to PAF-stimulated sample) as determined by Student’s t test]. Results shown are representative of at least three consecutive experiments, using at least three parallel samples in each experiment. Release assay of [3H] arachidonic acid and [14C] oleic acids compared to the unstimulated control (in %). Stimulus is 20 μM PAF for 60 min. b Dose–response inhibition by MAFP (pretreatment time 90 min.), c inhibition by indoxam (pretreatment time 90 min), d dose–response inhibition by PACOCF3 (pretreatment time 15 min). Statistical testing and number of experiments as for Fig. 1b

Full size image

In order to determine the importance of the contribution of group IVα PLA2 in the total PAF-response, PAF-mediated [3H]AA and [14C]OA release in cells treated with PAF in the presence or absence of MAFP [11] was determined. We found that MAFP inhibited both [3H]AA and [14C]OA release in a dose-dependent manner by a maximum of 51 and 28.5% respectively (see Fig. 2b). Since MAFP is also known to have some effect on group VI PLA2 [25, 28], we cannot exclude the involvement of PLA2s of this type in the response. However, our previous studies show that it is possible to attenuate AA-release in HaCaT cells completely in the IL-1β pathway using only 10 μM of MAFP [4]. The partial attenuation of PAF-mediated AA-release achieved with MAFP therefore suggests a significant contribution to AA-release by AA-nonspecific enzymes in addition to the contribution by group IVα PLA2. However, our in vitro enzyme assay clearly shows that group IVα PLA2 is activated by PAF, thus our conclusion is that group IVα PLA2 participates in PAF-mediated AA-release together with AA-nonspecific PLA2 subtypes.

To our knowledge, there are no previous reports of PAF-induced sPLA2 activation in the literature. Secretory PLA2 enzymes would be candidate enzymes for the OA release observed; we then examined the role of sPLA2 subtypes in PAF-mediated AA-release. The sPLA2-selective inhibitor indoxam [31] (a generous gift from Shionigi Ltd, Japan), gives a dose-dependent inhibition with a maximum of 33% of PAF-induced AA-release and 15% of OA release at 25 μM (Fig. 2c). Similar results were obtained using SB203347 (a kind gift from Lisa Marshall, SmithKline Beecham, PA, USA), another sPLA2 inhibitor [20, 31] (results not shown). Interestingly, Fig. 2c thus shows that the inhibition found with indoxam predominantly affects AA-release. In several cell types, sPLA2 isoenzymes IIa, IId and V, but not X, have been shown to be more strongly arachidonyl-selective when operating intracellularly, a mechanism which involves sPLA2 isoenzyme selective caveolin-mediated endocytosis [23]. Our data thus suggest that one or several of these three sPLA2s (IIa, IId or V) may participate in PAF-mediated intracellular AA-release in differentiated HaCaT cells.

Lastly, we examined the possible role of group VI PLA2 in the PAF-induced OA response. We found that the PLA2 subtype VI-specific inhibitor palmitoyl trifluoromethyl ketone (PACOCF3) [1] (from Calbiochem) dose-dependently reduced the PAF-induced AA-release by 58% and OA release by 61% at a 25-μM concentration (Fig. 2d). These results were confirmed with application of bromoenol lactone (BEL, from Cayman Chemicals) [1], another group VI inhibitor, which produced comparable levels of maximum inhibition (data not shown). Although PACOCF3 is also known to inhibit group IV PLA2 [11], BEL [1] is not known to do so. As sPLA2 inhibitors were shown to preferentially inhibit AA-release, the incomplete attenuation achieved with the group IVα/VI inhibitor MAFP, and the more successful inhibition with PACOCF3, the conclusion supported is that group VI PLA2 most likely plays a major role in PAF-mediated AA-release. The group VI PLA2 enzyme is probably at least as important as group IVα PLA2, judging by its ability to contribute to the high OA release. The participation as well as the notable significance of group VI PLA2 in the PAF-mediated response is a novel finding.

Taken together, our data suggest the participation of both calcium-dependent and -independent cytosolic PLA2 subtypes IVα and VI, as well as of secretory PLA2 subtypes, in the PAF-induced response in differentiated HaCaT keratinocytes.

The palmitic acid (PA)-derived lipids TTA and TSA [22, 33] are already shown to exhibit anti-inflammatory properties [35]. Most published studies of TTA and TSA show their roles as PPAR ligands [35], however, their potency as anti-inflammatory and anti-apoptotic agents are not fully explained by this mechanism. It would therefore be interesting to test whether their anti-inflammatory properties include inhibition of AA-release. TTA (Fig. 3a) and TSA (Fig. 3b) show a similar overall trend with maximum AA inhibition of 60–70% of the PAF-induced AA-release at 25 μM concentration. Further experiments have therefore been carried out with only one of these two inhibitors. TTA and TSA are both derivatives of PA [22], which was used as a control (Fig. 3c). The data suggests that the inhibitory effect is specific to TTA and TSA, and not shared by their common precursor, PA.

Fig. 3
figure 3

Tetradecylthioacetic acid and tetradecylselenoacetic acid inhibit arachidonic acid release. The response is measured as % release of [3H] arachidonic acid and [14C] oleic acids compared to the unstimulated control. Pretreatment time for all inhibitors is 90 min. Effects on the PAF-induced response (left) (20 μM PAF, 60 min). Dose–response inhibition by a TTA and b TSA. c Dose–response result for Palmitic acid. PA acts as a control, since TTA and TSA are PA derivatives. Effects on the calcium ionophore (A23187)-induced response (right) (1 μM A23187, 60 min). d Dose–response inhibition by MAFP, e dose–response inhibition by TSA. Statistical testing and number of experiments as for Fig. 1b

Full size image

We also tested these inhibitors in the calcium ionophore (A23187) (Sigma) response, to compare with the PAF response. Calcium ionophore acts by releasing calcium from both the mitochondrion and the extracellular matrix [28], and can induce activation of calcium-dependent enzymes such as group IVα PLA2 [17]. Twenty-five micromolar of the group IVα PLA2 inhibitor MAFP [11] was found to strongly inhibit calcium ionophore induced AA-release (73.5%; Fig. 3d, right). In cells pre-treated with 35 μM of TSA, a near-complete attenuation of A23187-induced AA-release, 98%, was found (Fig. 3e).

Thus, TTA and TSA give a strong arachidonyl-selective inhibition in both the PAF and calcium ionophore responses, suggesting that they affect the group IVα PLA2-regulated pathway. The data suggest a prominent role for AA-specific PLA2 subtypes in the response to calcium ionophore, consistent with our previous studies [32].

We then wished to determine whether TSA actually acts as inhibitor of the group IVα PLA2 enzyme. The effect of TSA on a recombinant group IVα PLA2 substrate was tested using the in vitro enzyme activity assay [18], and compared to the effects of AACOCF3 and PACOCF3. AACOCF3 (Cayman Chemicals) is a potent, reversible group IVα PLA2 inhibitor, with an effect similar to that of MAFP [11]. Forty nanomolar AACOCF3 gives a 70% reduction in group IVα PLA2 activity (Fig. 4a). A similar concentration of the group VI-selective inhibitor PACOCF3 gives a 55% reduction in group IVα PLA2 activity, which is consistent with previous studies showing that PACOCF3 is a poorer inhibitor of group IVα PLA2 than AACOCF3 is [11]. TSA shows only a 20% reduction in group IVα PLA2 activity, suggesting that it has a poor direct effect on group IVα PLA2. Thus, the observed strong reduction of cellular AA-release by TSA must mainly be due to a signalling mechanism leading to the activation of group IVα PLA2, rather than on the enzyme itself.

Fig. 4
figure 4

Tetradecylthioacetic acid and tetradecylselenoacetic acid point of action in AA-release is upstream of group IVα PLA2. a In vitro activity assay applying inhibitors directly to lysate from group IVα PLA2 overexpressing insect cells (95% significance compared to control indicated by asterisk as determined by Student’s t test). b Simultaneous application of TTA and PACOCF3 is additive. The response is measured as % release of [3H] arachidonic acid compared to the unstimulated control. Stimulus is 20 μM PAF for 60 min. Pretreatment time for TTA is 90 min, PACOCF3 was added 15 min, before PAF stimulation. Results shown are representative of at least three consecutive experiments, using at least three parallel samples in each experiment

Full size image

Since PACOCF3 and TTA are both PA derivates, we lastly wanted to examine whether these inhibitors act through similar or different mechanisms. TTA and PACOCF3 were applied to cells both separately and together, and AA-release was induced with PAF. Both TTA and PACOCF3 inhibited AA-release by approx 30%, respectively, while both together inhibited AA-release by almost 60% (Fig. 4b). The effect of TTA and PACOCF3 combined is clearly additive. This suggests that these inhibitors work on parallel, separate mechanisms, where TTA acts on signalling involving group IVα PLA2, while PACOCF3 works on the group VI PLA2 enzyme.

Measuring cellular AA and OA release in the presence of different PLA2 inhibitors is an established method within the PLA2 research community [18, 30]. We have used two structurally and mechanistically different inhibitors against each class of PLA2 in our experiments. Both of the cPLA2 (MAFP, AACOCF3) [11] and iPLA2 inhibitors (PACOCF3, BEL) [1, 5] used here are very well established. The two latter inhibitors display similar inhibitory properties, and knowing that BEL has high selectivity for iPLA2 over cPLA2, our conclusions are strongly supported. The selectivity of the sPLA2 inhibitor indoxam has been thoroughly characterized [31, 37], and as a result has become a preferred inhibitor for the secretory enzyme. The in vitro assay utilizes knowledge on substrate specificity, enzyme structure and calcium requirements to positively identify cPLA2: The calcium requirement for sPLA2 is 100-fold lower than for cPLA2, the assay buffer contains dithiothreitol which reduces the disulfide bridges stabilizing the sPLA2 structure, while iPLA2 activity requires ATP [18]. The total evidence obtained with cell culture as well as in vitro assays compared with known information on isoenzyme substrate specificity and inhibitor selectivity against the same isoenzyme thus adds up to a meaningful overall interpretation. In summary, PAF induces arachidonic and oleic acid release in differentiated HaCaT keratinocytes. The main participating PLA2 isoenzyme is group VI, with contribution from group IVα PLA2 and to some extent the sPLA2 subtypes. The important role of group VI PLA2 in PAF-mediated AA-release is a novel finding, and may therefore represent a novel intervention point in inflammatory dermatoses.