FATP4 inactivation in cultured macrophages attenuates M1- and ER stress-induced cytokine release via a metabolic shift towards triacylglycerides.

Fatty acid transport protein 4 (FATP4) belongs to a family of acyl-CoA synthetases which activate long-chain fatty acids into acyl-CoAs subsequently used in specific metabolic pathways. Patients with FATP4 mutations and Fatp4-null mice show thick desquamating skin and other complications, however, FATP4 role on macrophage functions has not been studied. We here determined whether the levels of macrophage glycerophospholipids, sphingolipids including ceramides, triacylglycerides, and cytokine release could be altered by FATP4 inactivation. Two in vitro experimental systems were studied: FATP4-knockdown in THP-1-derived macrophages undergoing M1 (LPS+IFNγ) or M2 (IL-4) activation and bone marrow-derived macrophages (BMDMs) from macrophage-specific Fatp4-knockout (Fatp4M-/-) mice undergoing tunicamycin (TM)-induced ER stress. FATP4-deficient macrophages showed a metabolic shift towards triacylglycerides and were protected from M1- or TM-induced release of pro-inflammatory cytokines and cellular injury. Fatp4M-/- BMDMs showed specificity in attenuating TM-induced activation of inositol-requiring enzyme1α, but not other unfolded protein response pathways. Under basal conditions, FATP4/Fatp4 deficiency decreased the levels of ceramides and induced an upregulation of mannose receptor CD206 expression. The deficiency led to an attenuation of IL-8 release in THP-1 cells as well as TNF-α and IL-12 release in BMDMs. Thus, FATP4 functions as an acyl-CoA synthetase in macrophages and its inactivation suppresses the release of pro-inflammatory cytokines by shifting fatty acids towards the synthesis of specific lipids.


Introduction
Macrophages are generated in the bone marrows and mononuclear phagocytes in circulation under specific programming by cytokines, transcription factors, and epigenetics [1]. Macrophages are important in the physiopathology of microbial infections and regulation of innate and adaptive immune response [2]. The stimulation of resting macrophages (M0) with Toll-like receptor (TLR) ligands (i.e., LPS) and Th1 cytokines (i.e., IFN-) skews M0 towards M1-like macrophages which mediate host defense, antitumor immunity, and inflammatory responses [3]. The alternative polarization towards M2-like macrophages is induced by Th2 cytokines IL-4 and IL-13. M2-like macrophages regulate anti-inflammatory response to control infections and to mediate tissue remodeling and fibrosis [3]. In contrast to anti-inflammatory functions, M2-like macrophages are reported to mediate proangiogenic processes that favor tumor progression [4] and the development of allergic inflammation such as asthma [5]. Therefore, the plasticity of macrophages and a shift between M1-like and M2-like activation states underlie the pathogenesis of diverse inflammatory diseases.
Emerging evidence indicates that intracellular metabolism plays an integral control of immune response, for an example, aerobic glycolysis provides metabolites needed for cell proliferation and the syntheses of cytokines [6]. Nutrients such as glucose, glutamine, and fatty acids have been shown to regulate the activities of the regulators involved in immune metabolism and functions [6,7]. Fatty acids are utilized not only for the generation of ATP but also the activation of M1-like inflammatory response [8].
M1-like macrophage activation is also regulated by acylated lipids including phosphatidylcholine (PC) [9,10], phosphatidic acid [11], triacylglycerides (TAGs) [12,13], and ceramides (Cer) [14]. Moreover, IL-4induced polarization to M2-like macrophages is associated with alterations of lysophosphatidic acid [15] and sphingosine 1-phosphate [16]. Hence, the activation of fatty acids to various metabolic pathways for synthesis of glycerophospholipids (GPLs), spingolipids (SPLs), and TAGs is critical for M1-like and M2like macrophage polarization. Activation of fatty acids by thioesterification to form acyl-CoenzymeA (acyl-CoA) is catalyzed by members of long chain acyl-CoA synthetases (ACSL) and fatty acid transport protein (FATP/SLC27) families [17]. These enzymes are involved in macrophage functions as a pan-ACSL inhibitor Triacsin inhibits macrophage inflammatory response [18]. The most studied FATPs are FATP1 (SLC27A1) and FATP4 (SLC27A4) which are ubiquitously expressed in many tissues [17]. These two FATPs are thought to have similar functions in skin physiology [19]. The skin of global [20] and keratinocyte-specific [21] Fatp4-deficient mice show features of restrictive demopathy. Fatp4 also regulates skin permeability barrier by activating the synthesis of acylceramides [22]. Although FATP4 expression is reported in human monocytic leukemia THP-1 cells [23] and mouse bone marrow-derived macrophages (BMDMs) [24], its functions in immune cells have not been studied. Here, we investigated whether FATP4 inactivation could elicit any effects on macrophage polarization, cytokine release, and cellular injury. Hence, fatty-acid activation by FATP4 may regulate changes in GPLs and TAG levels depending on cellular demands such as during stress. Specifically, FATP4 inactivation induces a metabolic shift from GPLs and Cer towards TAGs as reported in skin fibroblasts [27] and skin [21,22] of Fatp4-null mice as well as in subcutaneous fat of adipose-specific Fatp4-deficient mice fed with high-fat diet [30]. As GPLs, SPLs, and TAGs are critical for macrophage functions [9][10][11][12][13][14], we hypothesized that FATP4 inactivation may affect the cytokine release by altering these lipids. Because of FATP4 localization in the ER [25,26], we chose a stress model which involves unfolded protein response (UPR) activation in the ER [31] in two in vitro systems using small interfering RNA (siRNA) against FATP4 in THP-1-derived macrophages and BMDMs of myeloid-specific Fatp4-deficient (Fatp4 M-/-) mice. Here, we showed that FATP4-deficient macrophages were protected from M1 (LPS+IFN)-or ER-stress-induced cytokine release and cellular injury associated with a metabolic shift towards TAGs. Moreover, our data revealed that FATP4 catalyzed the synthesis of VLFA Cer in macrophages. AL, USA). Internal standard peak areas were monitored for quality control and used for quantification of analytes of samples and standards. Data acquisition and processing was performed with Masslynx version 4.1 software. The data were exported to Excel sheets and analyte / internal standard ratios were used to determine concentrations. Lipid contents were normalized to cellular mg protein.

Western blotting
Cell lysates were subjected to Western blot analyses. FATP4 antibody was from Abcam (ab200353).

Gene expression
For quantitative reverse transcription-polymerase chain reaction (qRT-PCR) analyses, total RNA and cDNA were prepared. mRNA expression was analyzed in quadruplets on an Applied Biosystems 7500 using TaqMan® assay-on-demand primers. The expression level was calculated using Δ−Ct transformation

Statistical analyses
Data were mean ± SEM. Significant difference was considered at p < 0.05 by using Mann-Whitney U test of GraphPad Prism 5.

FATP4 KD in THP-1 cells enhances M2 phenotypes and attenuates M1 cytokines
Human monocytic THP-1 cells were differentiated by PMA and subjected to FATP4 knockdown (KD) by Regarding macrophage functions, FATP4 KD at M0 and M2-like states did not markedly alter mRNA expression of pro-inflammatory markers TNF- ( Figure 1D), CCL5, TLR4, TLR9, and IL-6 (data not shown). FATP4 KD caused a decrease trend of IL-8 mRNA expression ( Figure 1E) but a significant attenuation of IL-8 release ( Figure 1E). M2-like activation by IL-4+IL-13 treatment attenuated the release of TNF- similarly by siCon and siFATP4 cells ( Figure 1D) indicating the lack of effects by FATP4 KD.
FATP4 KD also did not alter the release of IL-6, IL-10, and IL-12 (data not shown). Interestingly, we observed that siCon and siFATP4 cells secreted high levels of IL-8 compared with TNF- and other The effects of FATP4 KD on M1-like macrophages were further studied by LPS+IFN- treatment of siCon and siFATP4 cells for 24 h. FATP4 KD during M1-like activation was also able to suppress FATP4 expression and ACS activity (Figure 2A). We found that LPS+IFN- treatment of siCon cells upregulated ACS activity and FATP4 expression, and similar FATP4 mRNA upregulation has also been previously  Figure   S1). From the calculated coefficients and p-values (p<0.001), a significant negative correlation was obtained indicating that FATP4 KD attenuated pro-inflammatory cytokine release under M1 by an upregulation of M2 phenotypes.

Significant changes of GPLs, SPLs, and TAGs in M1 and M2 states by FATP4 inactivation in THP-1
Since accumulating evidence has shown that FATP4 inactivation shifts the composition from GPLs and  Table I.
It is long known that phorbol esters stimulate PC synthesis in cultured human promyelocytic leukemia cells [39]. Since PMA was used to differentiate THP-1 cells in our studies, we surmise that these cells would maintain on-going synthesis of PC and GPLs, and that M1-like induction by LPS+IFN enhances and M2 but not M1 state (Figures 3A,B). This suggests that FATP4 deficiency may provide fatty acids directed for the on-going PMA-activated PC synthesis [39] resulting in increased levels of these GPLs under basal M0 and anti-inflammatory M2-like conditions.

FATP4 deficiency in mouse BMDMs alleviates ER stress-induced inflammatory cytokine release
It is shown that LPS ligation to TLRs 2 and 4 specifically activates the ER-stress sensor kinase inositolrequiring enzyme1 (IRE1) and its downstream transcription factor X-box binding protein1 (XBP1) [31].
We therefore further studied whether FATP4 deficiency in macrophages could alter IRE1 activation during ER stress, in a similar manner as siFATP4 cells undergoing M1 inflammation (Figure 2). Instead of Because ACSL activity of Fatp4 M-/-BMDMs may alter fatty acids pools [8] and downstream lipids [9][10][11][12][13][14], we therefore measured genes related to cellular synthesis and uptake of fatty acids. Results showed that TM treatment markedly suppressed mRNA expression of uptake genes Cd36 and of Fas mRNA expression. This elevation was attenuated in Fatp4 M-/-BMDMs which is consistent with attenuated cytokine release (Figures 4E-H).

FATP4 deficiency in mouse BMDMs alleviates ER stress-induced apoptosis by inhibiting p-IRE1
We further analyzed ER stress signaling via IRE1α UPR activation [31] which is primarily a repair mechanism in response to unfolded proteins and ER membrane aberrancy [45]. We showed that TMinduced phosphorylation of IRE1α in WT BMDMs was attenuated by Fatp4 deficiency; whereas CHOP, BIP, and p-elF2 activation was not significantly affected ( Figure 5A) WT BMDMs was also reversed by the deficiency (Figure 5D). Thus, Fatp4 deficiency in BMDMs rendered protection against TM-induced cellular apoptosis and the release of pro-inflammatory cytokines by suppressing IRE1α activation.

FATP4 deficiency in macrophages causes a metabolic shift towards TAGs during ER stress
It has been known that cellular GPL profiles are different between PMA-versus M-CSF-differentiated macrophages and between human versus mouse macrophages [46]. Therefore, FATP4/Fatp4 inactivation may affect GPL profiles differently between PMA-differentiated THP-1 cells (Figure 3) versus M-CSFdifferentiated BMDMs. Moreover, M-CSF is shown to severely induce the breakdown of GPLs including PC via the activation of phospholipase C [47] and phospholipase D [48]. Opposite from siFATP4 cells (Figures 3A,B), naïve BMDMs from Fatp4 M-/mice showed a decrease in PC, PI, SM, and total Cer levels ( Figures 6A,B). This suggests that Fatp4 inactivation in BMDMs may enhance the on-going hydrolysis of During ER stress, TM treatment of WT BMDMs led to a significant decrease in PC, PE, PI, SM, Cer d18:1/24:1, and total Cer levels (Figures 6A,B). TM-treated Fatp4 M-/-BMDMs showed a further decrease in PC, SM, Cer d18:1/24:1, and total Cer levels (Figures 6A,B), and a further increase in TM-induced TAGs ( Figure 6C). Fatp4 deficiency did not alter TM-induced elevation of cellular cholesterol and nonesterified fatty acids (Supplementary Figure S4). From the obtained results (Figures 6A-C), we determined the ratios between TAGs with each of GPLs and SPLs to indicate a metabolic shift in relation to TAGs during TM treatment (Supplementary Figure S5). From averaged lipid ratios, the effects of FATP4 KO in response to TM were subsequently calculated based on the average lipid ratio of WT-Con as 100% (see Figure 6E legend, Supplementary Figure S5). By this calculation, a response was obtained without an error bar (Figure 6E, Left). FATP4 KO effects in response to TM showed a strong positive response of TAGs/Cer and much lesser extent TAGs/SM, and TAGs/GPLs suggesting a shift from Cer, SM, and GPLs towards TAGs (Figure 6E, Right). This shift led to attenuated release of pro-inflammatory cytokines, inactivation of IRE-1, and apoptosis.
As two different experiments were carried out in our studies, we further confirmed the effects of

Discussion
It is recognized that FATP4 plays a major role in skin physiology as reported in skin of Fatp4-null mice [20][21][22] and patients with FATP4 mutations manifested as a rare autosomal recessive disorder, ichthyosis prematurity syndrome (IPS) [51]. The skin abnormalities were associated with a marked decrease in keratinocyte GPLs and Cer but an increase in TAG levels. Here, a similar metabolic shift towards TAGs was demonstrated in FATP4/Fatp4-deficient macrophages undergoing M1-like or TM-induced stress. This metabolic shift led to an attenuation of stress-induced pro-inflammatory cytokine release and cellular injury (Figures 3E,6E) For a possible mechanism for the decreased Cer levels in siFATP4 cells observed at M0 and M2-like states, we propose that FATP4 inactivation may divert fatty acids towards the synthesis of GPLs likely catalyzed by other ACSLs and FATPs (Figure 3E, Top). While GPL accumulation can lead to upregulation of MR CD206 [52], it is possible that FATP4 KD may increase certain lipids, such as, PPAR agonists [53] and prostaglandin E [54] that are capable of upregulating CD206 expression. At M0 and M2like states, FATP4 KD was able to suppress the release of IL-8, a chemotactic factor for leukocyte emigration to endothelium [55]. As a pattern recognition receptor, MR CD206 is reported to regulate IL-8 and immune response by interacting with TLR2 [56] or PPAR [57]. It is however not clear how CD206 upregulation by FATP4 KD could suppress IL-8 release. FATP4 KD's ability to increase GPLs and SM and decrease Cer levels may lead to an alteration of membranes in the ER-Golgi thereby resulting in a defect in the secretion of IL-8 [37]. Consistently, it has been shown that the secretory pathways for IL-8 and TNF- release can be altered by targeting an ER protein, such as, V-ATPase [58]. Moreover, it has been shown that IL-8 release is linked to lysophosphatidic acid [59] which is a precursor of diacylglycerol and subsequently TAGs. Hence, the observed suppression of IL-8 release was consistent with a directed metabolic shift away from TAGs by FATP4 KD at M2-like state (Figures 3D and 3E, Top).
During pathological M1-like activation, FATP4 KD in THP-1 cells attenuated the elevated proinflammatory cytokines in an M2 CD206-and PPAR-dependent manner. This is consistent with the reported negative role of CD206 [60] and PPAR [36] on LPS response in vitro. Such attenuation by FATP4 KD was associated with a metabolic shift towards TAGs and Cer (d18:1, 24:1) likely by the diversion of fatty acids for syntheses of these lipids (Figure 3E, Bottom). It is known that LPS treatment in macrophages increases TAG levels [12,13] catalyzed by glycerol phosphate acyltransferase [61]. However, the deletion of this gene further exacerbates the release of pro-inflammatory cytokines [12]. Thus, TAG species play a protective role in LPS response in vitro, although LPS-induced TAG storage persists long after pro-inflammatory response have subsided [13]. TAG storage in macrophages has also been shown to be protective in vivo during high-fat diet induced inflammation [62] and seipinopathy-related ER stress [63]. Similar to TAGs, the elevation of Cer d18:1/24:1 levels by FATP4 KD is consistent with reported beneficial role of very long chain SPLs [64]. Conversely, the depletion of Cer d18:1/24:1 in macrophages increases pro-inflammatory cytokines [65]. Thus, our results also demonstrated FATP4 specificity for VLFA Cer metabolism during M1-like activation.
Due to the use of M-CSF rendering M2-like phenotypes [43], Fatp4-deficient BMDMs showed a different profile of GPLs and SPLs compared to siFATP4 THP-1 cells, possibly involving the on-going GPL hydrolysis during M-CSF differentiation [47,48] as well as downregulation of lipid synthesis genes Agpat2 [49] and Lxr [50] (Supplementary Figure S8). Consequently, naïve BMDMs from Fatp4 M-/mice showed a decrease in PC, PI, SM, and Cer levels which were associated with attenuated TNF- and IL-12 release ( Figure 6D). Such attenuation could be linked to an upregulation of Cd206 which is shown to trigger anti-inflammatory program during functional maturation of monocyte-derived dendritic cells [66] and TLR activation [67]. Moreover, suppressed PC synthesis is shown to attenuate LPS-induced cytokines [9,10], it is possible that the decrease in PC, PI, SM, and Cer levels in mutant BMDMs may affect ER-Golgi secretory pathways thereby attenuating the TNF- and IL-12 release.
During TM-induced ER stress, WT BMDMs showed a decrease in most GPLs, SM and Cer d18:1/24:1 and an increase in TAG levels. The changes of these lipids are consistent with reports in hepatocytes undergoing apoptosis and ER stress [68,69]. Interestingly, the KD of CerS2 even further increases ER stress [69] thus supporting the protective role of its product Cer d18:1/24:1 under ER stress. However, Fatp4 M-/-BMDMs showed a further increase in TAG levels leading to attenuation of pro-inflammatory cytokines ( Figure 6E). This suggests that TAG storage synthesis is coordinated with membrane biogenesis depending on cellular demands during stress [28,29], and we showed that these events were promoted by Fatp4 deficiency. As TM amplifies M1-like responses to LPS [70], the attenuation of pro-inflammatory cytokines by FATP4/Fatp4 inactivation was consistently observed in both LPS (Figures 1-3) and ER stress (Figures 4-6) models (Supplementary Figures S6,S7). Consistent with a previous report [71], the observed protection in Fatp4 M-/-BMDMs is associated attenuated activation of ER transmembrane protein p-IRE1.
Further investigations may be warranted to measure lipidomics and p-IRE1 expression in the ER fractions of WT and Fatp4 M-/-BMDMs (with and without LPS or TM treatment) to clarify the effects of Fatp4 deficiency at the ER membrane level.
Under in vivo conditions, M2 macrophages have been linked to allergies [5], and that M2 cytokines such as IL-4 and IL-5 can regulate eosinophilia and IgE [75]. Our in vitro results showing upregulated CD206 expression and attenuated cytokine release in FATP4-deficient macrophages may have some implications regarding their interactions with other innate and adaptive immune cells under in vivo conditions. These interactions could lead to immune-related abnormalities seen in adult IPS patients [72][73][74]. Accordingly, genetic variants of CD206 have been shown to be associated with susceptibility for asthma [76]. Although WT, Wild-Type; XBP1, X-box binding protein 1

Conflict of interests
The authors have no conflicts of interest associated with the manuscript.    Data are mean ± SEM, N= 6; * p < 0.05, ** p < 0.01.