Interleukin-17A-induced production of acute serum amyloid A by keratinocytes contributes to psoriasis pathogenesis

Background Acute-serum Amyloid A (A-SAA), one of the major acute-phase proteins, is mainly produced in the liver but extra-hepatic synthesis involving the skin has been reported. Its expression is regulated by the transcription factors NF-κB, C/EBPβ, STAT3 activated by proinflammatory cytokines. Objectives We investigated A-SAA synthesis by resting and cytokine-activated Normal Human Epidermal Keratinocytes (NHEK), and their inflammatory response to A-SAA stimulation. A-SAA expression was also studied in mouse skin and liver in a model mimicking psoriasis and in the skin and sera of psoriatic and atopic dermatitis (AD) patients. Methods NHEK were stimulated by A-SAA or the cytokines IL-1α, IL-17A, IL-22, OSM, TNF-α alone or in combination, previously reported to reproduce features of psoriasis. Murine skins were treated by imiquimod cream. Human skins and sera were obtained from patients with psoriasis and AD. A-SAA mRNA was quantified by RT qPCR. A-SAA proteins were dosed by ELISA or immunonephelemetry assay. Results IL-1α, TNF-α and mainly IL-17A induced A-SAA expression by NHEK. A-SAA induced its own production and the synthesis of hBD2 and CCL20, both ligands for CCR6, a chemokine receptor involved in the trafficking of Th17 lymphocytes. A-SAA expression was increased in skins and livers from imiquimod-treated mice and in patient skins with psoriasis, but not significantly in those with AD. Correlations between A-SAA and psoriasis severity and duration were observed. Conclusion Keratinocytes could contribute to psoriasis pathogenesis via A-SAA production, maintaining a cutaneous inflammatory environment, activating innate immunity and Th17 lymphocyte recruitment.

Results IL-1α, TNF-α and mainly IL-17A induced A-SAA expression by NHEK. A-SAA induced its own production and the synthesis of hBD2 and CCL20, both ligands for CCR6, a chemokine receptor involved in the trafficking of Th17 lymphocytes. A-SAA expression was increased in skins and livers from imiquimod-treated mice and in patient skins with psoriasis, but not significantly in those with AD. Correlations between A-SAA and psoriasis severity and duration were observed.

Introduction
Serum amyloid A (SAA) is the circulating precursor of amyloid fibril protein AA [1]. The human SAA protein family contains different isoforms. SAA3, a pseudogene, is not transcribed whereas SAA4 is a constitutive, non-inducible protein (C-SAA). Conversely, the production of SAA1 and SAA2, also known as acute-phase protein-A (A-SAA) is inducible under inflammatory conditions. Because of their extensive homology (95%), neither SAA mRNA nor the protein isoforms 1 and 2 can be distinguished from each other. The mouse SAA family is also composed of four genes: SAA1, 2 and 3 genes encode A-SAA protein and SAA4 gene encodes C-SAA protein [2].
Mainly produced by hepatocytes, A-SAA extra-hepatic production has been reported in humans, specifically in the skin [3]. In vitro, A-SAA synthesis by hepatoma, fibroblast, epithelial or endothelial cell lines is stimulated by proinflammatory cytokines such as IL-1β, IL-6 and TNF-α [4]. Cytokines induce the transcription of SAA1 and 2 genes through the activation of transcription factors including NF-κB, C/EBPβ and STAT3 [5]. In mouse, SAA1 and 2 proteins are mainly synthesized by hepatocytes, whereas SAA3 is mostly detected in extrahepatic tissues [6].
Normal human serum A-SAA concentrations are less than 1 mg/L, but they are dramatically enhanced during the acute phase response up until 1000 mg/L [7]. A-SAA serum levels are also increased in chronic inflammatory diseases such as familial Mediterranean fever [8], rheumatoid arthritis [9], ankylosing spondylitis [10] and psoriasis [11]. Prolonged high serum A-SAA concentrations associated with insufficient degradation can promote A-SAA β-sheet conformation. The deposition of amyloid fibrils causes secondary amyloidosis [12], a serious complication of chronic inflammatory disorders, with forty cases secondary to psoriasis described [13].
We previously shown that a combination of IL-1α, IL-17A, IL-22, oncostatine-M (OSM) and TNF-α target normal human epidermal keratinocytes (NHEK) in monolayer culture or in differentiated reconstituted human epidermis to generate a specific transcriptional profile and histological characteristics reproducing features of psoriasis [19,20]. Interestingly, these cytokines activate the same NF-κB, C/EBPβ and STAT3 signaling pathways as those involved in A-SAA synthesis.
During the acute phase response and through binding to different receptors, A-SAA displays several activities. Binding to HDL affects cholesterol transport. By replacing apolipoprotein A-1 in HDL, it induces a lower affinity for hepatocytes in favor of inflammatory macrophages [21] that express the Formyl Peptide Receptor Like 1 (FPRL1) also named formyl peptide receptor 2 [22]. Through FPRL1, A-SAA also promotes chemotaxis of neutrophils, lymphocytes and monocytes [23], production of IL-1β, IL-6, IL-8 and TNF-α by neutrophils [24], synthesis of matrix metalloproteinases by fibroblasts [25] and angiogenesis [26]. By binding to the TLR2, A-SAA stimulates the synthesis of IL-12, IL-23, TNF-α and IL-18 by mouse macrophages [27] and G-CSF by human monocytes [7]. The engagement of TLR4 by A-SAA activates Nitric Oxide (NO) production by mouse macrophages [28]. A-SAA interaction with scavenger receptors promotes the synthesis of IL-6, IL-8 and TNF-α by CLA-1 (CD36 and LIMPII Analogous-1)-expressing HeLa cells [29], as well as IL-8 by the human monocyte cell line THP1, in a CD36-dependent manner [30]. A-SAA also binds to the Receptor for Advanced Glycation End products (RAGE) resulting in the activation of the NF-κB signaling pathway in rheumatoid fibroblast-like synovial cells [31]. A-SAA can also function as an opsonin by binding to the outer membrane protein A of gram-negative bacteria, facilitating phagocytosis [32]. Interestingly, TLR2, TLR4 and CD36 have been reported to be expressed by keratinocytes [33,34].
We have shown that several proinflammatory cytokines could activate A-SAA production by NHEK and that in turn, A-SAA could induce an inflammatory phenotype in NHEK. These inflammatory properties are underscored in vivo by its increased expression in skin and liver in a mouse model of psoriasiform dermatitis and in skin of patients with psoriasis, but not in atopic dermatitis (AD), another common chronic inflammatory skin disease with a different immune profile.

Material and methods
Cell cultures and supernatants NHEK were obtained from surgical samples of healthy breast or abdominal skins, as described previously [18]. NHEK were cultured to 80% of confluency in Keratinocyte Serum-Free Medium (K-SFM; Invitrogen Life Technologies), supplemented with epidermal growth factor (5 ng/ml) and bovine pituitary extract (50 μg/ml; all purchased from Invitrogen Life Technologies) at 37˚C, 5% CO 2 in a humidified incubator. NHEK were starved for 24 hours in K-SFM without addition of growth factors. NHEK were stimulated for different time-periods with human recombinant cytokines alone or in combination (IL-1α, IL-17A, IL-22, OSM, TNF-α: M5), as described previously (final concentration 10 ng/ml of each cytokine; R&D Systems) [20] or with human rA-SAA (10 μg/ml, purity > 98% and endotoxin level is <0.1 ng/μg of protein, Peprotech,) associated or not with rIL-17A (10 ng/ml, R&D Systems).

Subjects, skin and serum samples
We obtained 37 lesional skin biopsies from psoriatic patients and 28 controls from surgical samples of healthy abdominal or breast skins. We collected 17 sera from psoriatic patients and 11 controls from healthy donors. Patient characteristics are presented in Table 1. Lesional AD skin biopsies were obtained from 12 adults and children. None of the patients received any therapy for at least four weeks. The use of skin samples was approved by the Ethical Committee of the Poitiers Hospital and they were collected after informed consent.

A-SAA protein measurement
A-SAA protein was quantified by ELISA (Human SAA CytoSet, Invitrogen Life Technologies) in NHEK supernatants and by immunonephelemetry assay in patient sera (Siemens).

Statistical analysis
Statistical analysis was performed using GraphPad Prism 5 Software (Inc, San Diego, CA, USA). The p values 0.05 were considered as significant and all data are represented as mean ± SEM.
The kinetic study showed that M5 stimulation induced A-SAA mRNA expression reaching a maximum after 24 hours (88-fold increase over unstimulated NHEK) ( Fig 1A) followed with a maximum A-SAA protein concentration in culture supernatants at 48 hours (12-fold increase) ( Fig 1B). Tested independently, IL-1α, IL-17A or TNF-α induced A-SAA mRNA and protein levels whereas IL-22 or OSM were ineffective. Together, the M5 mix had an additive effect on A-SAA production (Fig 1C and 1D). By sequentially subtracting each of the cytokines of M5, only IL-17A was found to significantly decrease A-SAA production (Fig 1E and 1F). This increased A-SAA synthesis by M5-stimulated NHEK, as compared to resting control was confirmed by intracellular staining (Fig 1G and 1H).

Skin and liver A-SAA expression in a mouse model of psoriasiform dermatitis
We reported increased A-SAA mRNA expression in inflammatory mouse skin. A similar increased expression of the SAA1/2 isoform was observed both in the skin and in the liver (respectively a 55 and a 45-fold increase; Fig 3A), whereas the increase in SAA3 transcripts were about 300 times more expressed in the skin than in the liver (respectively a 3265 and a 10-fold increase; Fig 3B) as compared to untreated controls. These results are in accordance with an higher extra-hepatic tissues SAA3 expression previously reported [6].

A-SAA expression in skin and serum of psoriatic patients
A-SAA mRNA expression was 9-fold increased in lesional skins of psoriatic patients, compared to healthy skins (Fig 4A), and A-SAA protein serum levels were 26-fold increased in psoriatic patients, as compared to healthy donors (Fig 4B). We observed a positive correlation between circulating levels of A-SAA and C Reactive Protein (CRP), a commonly used serum marker for inflammation (Fig 4C). Of note, in psoriatic patients, A-SAA mRNA skin expression was found to be positively correlated with circulating A-SAA levels ( Fig 4D). Further analysis showed that A-SAA expression of both skin transcript and serum protein were higher in severe psoriatic patients (Psoriasis Area Severity Index (PASI) > 10) as compared to those with mild psoriasis (PASI<10) (Fig 4E, serum data not shown). A-SAA mRNA skin expression was higher with long disease duration > 10 years (Fig 4F), with cigarette smoking (Fig 4G) or with the presence of metabolic syndrome (Fig 4H), as diagnosed according to the American Heart Association [37]. We did not find significance with respect to the latter parameters in sera (data not shown). Finally, as compared to healthy skin, A-SAA mRNA expression in lesional skin samples from patients with AD was not significantly increased, contrary to those from psoriatic patients (Fig 5A), in which A-SAA levels paralleled the expression of IL-17A transcripts (Fig 5B).

Discussion
In the present study, we report that keratinocytes produce A-SAA, thereby confirming and extending a previous study reporting A-SAA synthesis in the skin using in situ hybridization [3]. In vitro, A-SAA production is stimulated by IL-1α, TNF-α and especially IL-17A with an additive effect in M5, previously described as mimicking psoriasis in vitro [19,20]. These five cytokines activate the same signaling pathways as those reported to be involved in the transcription of A-SAA genes in the liver [5,20]: NF-κB which is activated by IL-1α [38], IL-17A [39] and TNF-α [40], STAT3 which is activated by IL-22 [18] and OSM [14] and C/EBPβ which is activated by IL-17A [39]. In agreement with two recent studies [41,42], we find IL-17A to be the most potent inducer of A-SAA in keratinocytes and its central role in this respect is highlighted by a pronounced reduction in A-SAA synthesis when it is removed from M5. In addition, A-SAA upregulates its own expression leading us to suggest that A-SAA could have an autocrine effect contributing to the maintenance of chronic inflammation. We have shown that A-SAA mainly promotes AMPs expression by keratinocytes, thereby highlighting its role in innate immunity, like other acute-phase proteins such as CRP. A-SAA also has adaptive immunological functions through the induction of various cytokines and chemokines. Contrary to a previous report in which keratinocytes derived from foreskins were used [42], A-SAA has not been found to induce the synthesis of IL-1β. Herein, we have shown increased expression of transcripts for TNF-α, of which the involvement in psoriasis is well-established, as well as CCL20 and hBD2, which are known to activate Th17 chemotaxis through CCR6 [43]. In addition, A-SAA promotes its own synthesis and has a strong synergistic effect with IL-17A. Taken together, we suggest that A-SAA production and biological functions are selfmaintained by a positive feedback.
In parallel, we showed an increased expression of A-SAA transcripts in IMQ-treated mouse skin. In contrast to SAA1/2, SAA3 expression, known to be mainly extrahepatic [6], is much higher in the skin than in the liver. The immunomodulatory effects of IMQ are related to the stimulation of plasmacytoid DCs through TLR7 and TLR8, resulting in upregulation of the type I interferon pathway [44] and the induction of IL-23, IL-17A and IL-1α expression [35,36], thereby explaining the increased production of A-SAA. Very recently and in agreement with our datas, Yu et al, [45], reported that SAA was overexpressed in IMQ treated skin mice, and they further showed that neutralizing anti-SAA antibodies attenuated skin hyperplasia and inflammation this model, demonstrating that SAA contribute to the physiopathology of this psoriasiform-induced dermatitis. In agreement with our in vitro and in vivo results, we show that human skin produce A-SAA in inflammatory conditions, such as those observed in psoriasis confirming previous studies [41,42]. We have shown a positive correlation between cutaneous A-SAA mRNA expression and serum A-SAA levels. It remains an open question whether A-SAA skin production in the skin could contribute to the increased blood levels or if the latter arise exclusively from hepatic synthesis. In addition, A-SAA expression in psoriatic skin is exacerbated with disease severity and duration. Circulating A-SAA levels are correlated with those of CRP, which has been reported to be associated with psoriasis severity [46]. We have also shown an association with smoking and metabolic syndrome. Smoking is significantly associated with psoriasis [47]. In the skin, nicotine binds to nicotinic acetylcholine receptors on DCs, macrophages, endothelial cells and keratinocytes, enhancing the synthesis of proinflammatory cytokines such as IL-12, IL-1β, TNF-α [48] that may lead to local skin synthesis of A-SAA. Psoriasis is known to be associated with metabolic syndrome [49]. A-SAA is synthesized by inflamed adipocytes and promotes lipolysis, while decreasing insulin sensitivity in adipocytes [50].
We questioned whether A-SAA expression was increased in other inflammatory skin conditions such as AD. In AD, the barrier defect leading to antigen penetration activates an adaptive immune response with Th1, Th2, Th22 lymphocyte polarization and induction of IgE synthesis by B lymphocytes [51]. In contrast, the T lymphocytes involved in psoriasis differentiate into Th1, Th17 and Th22 [17]. In our hands, A-SAA expression is comparable in AD and healthy skins. This observation is in line with the absence of described cases of AA amyloidosis secondary to AD, whereas cases have been reported in psoriasis. Furthermore, as demonstrated in vitro, A-SAA synthesis is stimulated mainly by IL-17A which is overexpressed in psoriasis but not AD skins, that could be accountable for the higher SAA expression levels in psoriasis compared to AD.
In conclusion, we report that IL-1α, TNF-α and chiefly IL-17A induce A-SAA expression by NHEK. This production was also increased in the skin and liver in a mouse model of psoriasiform dermatitis and in the skin and serum of psoriatic patients, but not in the skin of AD patients. In turn, A-SAA induced its own production by NHEK and the synthesis of hBD2 and CCL20 involved in the trafficking of Th17 lymphocytes. These results indicate that in psoriatic skin, keratinocytes contribute to the pathogenesis via the production of A-SAA and that its autocrine response maintains a cutaneous Th17-polarized inflammation.