GDF-15 Deficiency Reduces Autophagic Activity in Human Macrophages In Vitro and Decreases p62-Accumulation in Atherosclerotic Lesions in Mice

(1) Background: Growth differentiation factor-15 (GDF-15) is associated with cardiovascular diseases and autophagy in human macrophages (MΦ). Thus, we are interested in investigating autophagic mechanisms with special respect to the role of GDF-15. (2) Methods: Recombinant (r)GDF-15 and siRNA GDF-15 were used to investigate the effects of GDF-15 on autophagic and lysosomal activity, as well as autophagosome formation by transmission electron microscopy (TEM) in MΦ. To ascertain the effects of GDF-15−/− on the progression of atherosclerotic lesions, we used GDF-15−/−/ApoE−/− and ApoE−/− mice under a cholesterol-enriched diet (CED). Body weight, body mass index (BMI), blood lipid levels and lumen stenosis in the brachiocephalic trunk (BT) were analyzed. Identification of different cell types and localization of autophagy-relevant proteins in atherosclerotic plaques were performed by immunofluorescence. (3) Results: siGDF-15 reduced and, conversely, rGDF-15 increased the autophagic activity in MΦ, whereas lysosomal activity was unaffected. Autophagic degradation after starvation and rGDF-15 treatment was observed by TEM. GDF-15−/−/ApoE−/− mice, after CED, showed reduced lumen stenosis in the BT, while body weight, BMI and triglycerides were increased compared with ApoE−/− mice. GDF-15−/− decreased p62-accumulation in atherosclerotic lesions, especially in endothelial cells (ECs). (4) Conclusion: GDF-15 seems to be an important factor in the regulation of autophagy, especially in ECs of atherosclerotic lesions, indicating its crucial pathophysiological function during atherosclerosis development.


Introduction
Atherosclerosis, a chronic-inflammatory disease of the vascular wall, may lead to stroke, myocardial infarction or other cardiovascular events [1]. The progressive development of atherosclerotic plaques implies a deregulated apoptotic reaction of macrophages (MΦs), endothelial (ECs) and smooth muscle cells (SMCs), resulting in a formation of the necrotic lipid core, diminution of the fibrous cap and an inflammatory reaction [2].

SDS-PAGE and Western Blot
PMA-differentiated THP-1 MΦ were washed in ice-cold phosphate-buffered saline (PBS) and lysed using a radioimmunoprecipitation assay (RIPA) buffer of pH 7.5 (Cell Signaling Technology, Frankfurt, Germany), containing a protease/phosphatase inhibitor cocktail (Cell Signaling Technology). The protein concentrations were determined spectrophotometrically using the Pierce BCA (bicinchoninic acid) Protein Assay (Thermo Scientific, Rockford, IL, USA). Proteins were loaded on Novex™ WedgeWell™ 14%, Tris-Glycine (Thermo Scientific). Proteins were transferred onto 0.45 µm nitrocellulose membranes (Millipore, Billerica, MA, USA). Primary antibodies (Table S1) were added and incubated overnight at 4 • C in blocking buffer (5% fat-free milk). Membranes were incubated with enhanced ECL-anti-rabbit IgG-POD antibody. The peroxidase reaction was visualized by AceGlow chemiluminescence substrate (PEQLAB GmbH, Erlangen, Germany) and was documented by the Fusion-SL Advance™ imaging system (PEQLAB GmbH) according to its instruction manual. The intensities of the specific Western blot bands were quantified using the software ImageJ, from the National Institutes of Health (Bethesda, MD, USA) and normalized against α-tubulin.

GDF-15 ELISA
The intracellular level of human GDF-15 in THP-1 MΦ was quantified by the DuoSet ® ELISA Development System (R & D Systems, Inc., Abingdon, UK). The capture antibody was coated to a 96-well MaxiSorp-ELISA Microplate (Nunc, San Diego, CA, USA) and incubated overnight at room temperature. According to manufacturers' instructions, after the blocking step, the samples (2.5 µg protein/well) or standards were added to the well. After incubation with a detection antibody and streptavidin-HRP, we added the substrate solution (SigmaFast™ OPD, Sigma-Aldrich Chemie GmbH) to each well and there incubated for 30 min in the dark. The reaction was stopped with 50 µL 3 M HCl. The GDF-15-protein level (pg/mL) was measured with an ELISA reader (Tecan Deutschland GmbH, Crailsheim, Germany) at OD 490/655nm .

Dissection and Tissue Harvesting
At the age of 30 weeks, the mice were weighed, narcotized and analgized with a combination of ketamine (150 mg/kg) and xylazine (20 mg/kg). Local intercostal anesthesia was performed with lidocaine 2%. The thoracic cavity was opened, the apex of the left ventricle was incised and a cannula (8 G, B. Braun Melsungen AG, Melsungen, Germany) was introduced and clamped. After the right atrial incision, the vascular system was perfused with a solution consisting of PBS with 5 UL/mL heparin (Liquemin ® 25,000 UL/5 mL, Roche, Grenzach, Germany), with a delivery volume of 30 mL and a rate of 100 mL/h, using an automated syringe-pump (Secura, B. Braun, Melsungen AG). Afterwards, 250-300 µL 0.9% NaCl sterile physiological solution containing methylene blue (0.25%; Riedel-de Haën, Seelze-Hannover, Germany) was injected into the vascular system. The blue-colored BT was excised under direct observation through a binocular loupe embedded in Tissue-Tek ® (Sakura Finetek, Stauffen, Germany) and snap-frozen in liquid nitrogen-cooled isopentane. Immediately, after opening the right atrium, heparinized (0.25 I.U./mL, Roche) blood samples were taken. Plasma was obtained by centrifugation (10 min, 650× g), stored at −80 • C. Plasma cholesterol, triglyceride (TG) as well as high-density lipoprotein (HDL) and low-density lipoprotein/very low-density lipoprotein (LDL/VLDL) levels were analyzed spectrophotometrically by using commercially available kits (Total Cholesterol/Cholesteryl ester quantitation assay kit, TG quantification assay kit and Cholesterol kit-HDL and LDL/VLDL, Abcam, Cambridge, UK) in a Cytation™ 3 microplate reader (BioTek Instruments GmbH). Body size was determined by measuring nasal-to-anal length, and body mass index (BMI) was calculated as the ratio between body weight and surface area (g/cm 2 ) [19]. The body surface area was derived from the DuBois equation: body surface (m 2 ) = 0.007184 × weight (kg 0.425 ) × body size (cm 0.725 ) [19].

Morphometry and Immunohistology
For morphometrical and immunohistological investigations, cryo cross-sectional series (6 µm) of the BTs were performed. The extent of atherosclerotic plaques in the BT was measured by computer-assisted morphometry. These images were evaluated and quantified with the software Fiji [20]. Standard hematoxylin-eosin (HE) and van-Giesson-Elastica staining were performed. Immunohistochemical staining method was carried out using the antibodies listed in Table 1.  [21]. The p62-accumulation of atherosclerotic plaque was analyzed by counting all p62-accumulation per plaque or cell type in the plaque area [(p62-accumulation)/DAPI) × 100% = p62 accumulation (%)] [9].

Statistical Analyses
Statistical analyses were performed using SigmaPlot 12 (Systat Software Inc., San José, CA, USA). After testing for normality (by Shapiro-Wilk), the unpaired Student's t-test or one-way analysis of variance (ANOVA) was used. Data are reported as mean + standard error of the mean (SEM), and p ≤ 0.05 was considered as statistically significant.

GDF-15 Silencing Reduces Autophagic Activity in Human THP-1 MΦ
Based on previously published data of the transfection efficacy with siGDF-15 RNA in PMA-differentiated human THP-1 MΦ [9], we performed autophagic activity assays in combination with lysosomal activity assays after siRNA transfection. After transient siGDF-15 transfection, we found a 22% (p = 0.04) decrease in autophagic activity in comparison with siControl MΦ (Figure 1a). Lysosomal activity showed no significant changes in siGDF-15 MΦ compared with siControl MΦ (Figure 1b). Serum starvation was used as positive control for an induction of autophagic and lysosomal activities as described by others [13]. After starvation for 4 h human THP-1 MΦ showed a 34% (p = 0.03) increase in autophagic activity ( Figure 1a) and 18% (p = 0.01) increase in the lysosomal activity ( Figure 1b) compared with siControl MΦ. Using TEM after starvation, THP-1 MΦ showed numerous vacuoles in the cytoplasm, depletion of organelles, as well as the presence of large autophagosomes containing membranous whorls and remnants of cytoplasmic material. These findings were less observed in siControl MΦ and siGDF-15 THP-1 MΦ (Figure 1c

Discussion
GDF-15 has been hypothesized to play a crucial role in cardiovascular diseases [8,23,24], especially in the development of atherosclerosis [18,[25][26][27]. Previous studies shown, an increased level of GDF-15 is associated with the development and progression of atherosclerotic plaques, possibly through the regulation of autophagic and apoptotic processes [8,9]. Our previous data show that GDF-15 plays an important role during "basal or cytoprotective" autophagy, an initial barrier against stress induced-apoptosis [9]. Increased stress conditions, e.g., oxidative stress, lead to the induction of apoptosis, which results in the damage of cytoprotective mechanisms by cleavage of essential autophagyrelated (ATG) proteins [28] and in consequence, inhibition of the autophagic flux [29,30]. Moreover, oxidative stress also induces GDF-15 expression [8]. Therefore, our present study focused on the role of GDF-15 on autophagic processes because both are predominantly involved in the development and progression of atherosclerotic lesions. For detailed investigations, we used an in vitro human atherosclerosis MΦ model, and in vivo

Discussion
GDF-15 has been hypothesized to play a crucial role in cardiovascular diseases [8,23,24], especially in the development of atherosclerosis [18,[25][26][27]. Previous studies shown, an increased level of GDF-15 is associated with the development and progression of atherosclerotic plaques, possibly through the regulation of autophagic and apoptotic processes [8,9]. Our previous data show that GDF-15 plays an important role during "basal or cytoprotective" autophagy, an initial barrier against stress induced-apoptosis [9]. Increased stress conditions, e.g., oxidative stress, lead to the induction of apoptosis, which results in the damage of cytoprotective mechanisms by cleavage of essential autophagy-related (ATG) proteins [28] and in consequence, inhibition of the autophagic flux [29,30]. Moreover, oxidative stress also induces GDF-15 expression [8]. Therefore, our present study focused on the role of GDF-15 on autophagic processes because both are predominantly involved in the de-velopment and progression of atherosclerotic lesions. For detailed investigations, we used an in vitro human atherosclerosis MΦ model, and in vivo GDF-15 −/− /ApoE −/− mice under CED [18], to study the role of GDF-15 on autophagic processes in atherosclerotic lesions. Therefore, we used different analyses (activity assays, TEM) and autophagy-relevant markers (ATG5, p62) to investigate and verify the influence of GDF-15 on the autophagic pathway in atherosclerosis.
In any case, TEM is a useful method to identify the autophagic process, because the formation of autophagic vacuoles is by far the most important morphological feature of autophagic cells. Autophagy starts with the sequestration of portions of the cytoplasm by a phagophore (special double-membrane structure), which matures into the autophagosome. Our present data show an increasing degree of concentric membranous lamellar formations "myelin figures" and represent the autophagic degradation of membranous cellular components, in human THP-1 MΦ, after starvation or after rGDF-15 treatment. After rGDF-15 treatment and especially after starvation terms we detected numerous late/degradative autophagic vacuoles or autolysosomes, with a typical one limiting membrane and electron-dense cytoplasmic material and/or organelles at various stages of degradation [31,32]. However, the membranous lamellar formations are not frequently present in human plaques but can be found in lesions of cholesterol-fed rabbits [33] or after treatment of SMCs in culture with oxidized lipids [34]. Additionally, we showed in nsiGDF-15 and siGDF-15 MΦ fewer autophagosomes, with the typical double-membrane [29], then in rGDF-15 treated MΦ.
These morphological results were confirmed by the autophagic activity by using the auto-fluorescent substance Monodansylcadaverine (MDC), a useful marker for latestage autophagosomes and autophagolysosomes [29,[35][36][37][38][39] and the lysosomal activity by LysoBrite™ Red, another fluorescent substance that selectively accumulates in lysosomes and can be used as evidence of autophagic cargo delivery to lysosomes [29]. Our data show that the autophagic activity in siGDF-15 MΦ was decreased, whereas rGDF-15 had an opposite effect. Interestingly, lysosomal activity was not affected upon GDF-15 treatment. The proper usage and functionality of the autophagic and lysosomal activities were evaluated, using starvation as a positive control.
Another method to monitor autophagy machinery is western blot with the specific marker LC3. LC3 is initially synthesized in an unprocessed form (proLC3). Due its lack of amino acids on the C terminus, proLC3 is converted into LC3-I, a proteolytically processed form. Finally, LC3-I was modified into the PE-conjugated form, LC3-II. The LC3 conversion from cytosolic soluble form (LC3-I) to membrane-bound form (LC3-II) is the key issue during autophagosome formation [40,41]. Autophagic flux is the final stage, in which the autophagosome is digested by lysosomes such that the LC3-II protein will be partially decreased. During this stage, if the autophagic flux was disturbed, especially by lysosomal protease inhibitors such as E64d and pepstatin A, LC3-II increased. Mouse embryonic fibroblasts (MEFs) show, during starvation, a decreased protein level of LC3-I, in which LC3-II protein level increased [42]. In the present study we demonstrated that rGDF-15 induce LC3 conversion and resulted in an increased LC3-II protein level. The amount of LC3-II is closely correlated with the number of autophagosomes and corroborates the process of autophagosome formation [40]. Our current results are consistent with a previous study, which showed that GDF-15 silencing in MΦ (siGDF-15 MΦ) led to an impaired ATG5, ATG12/ATG5-complex and p62-protein level, as well as lesser p62 accumulations compared with nsiGDF-15 MΦ, whereas rGDF-15 promoted p62 accumulation [9]. Based on this previous data, the current study verifies the influence of GDF-15 on autophagic activity with consequences on p62 turnover in human THP-1 MΦ. One important step of autophagy is the autophagosome-lysosome fusion, which leads to lysosomal degradation of the sequestered materials by various lysosomal hydrolytic enzymes [43]. This process of "autophagic-flux" can be documented by investigating the autophagy receptor p62, which is a marker of autophagic status [43][44][45][46]. Defect autophagy increases the quantity of p62, a protein that, if overexpressed, can stimulate the production of reactive oxygen species and cell death [44]. In relation to any GDF-15 effects on the lysosomal activity, we imply that GDF-15 promotes apoptotic processes in human MΦ as a consequence of increasing autophagic activity when the autophagic cargo delivery to lysosomes (lysosomal activity) is constant, which will reach an impaired autophagy flux with increased p62accumulation [9]. Therefore, in this study, we further proved the influence of GDF-15 on autophagy by using relevant markers, like p62, in the context of plaque progression by using a GDF-15 −/− /ApoE −/− mouse model.
Adult GDF-15 −/− /ApoE −/− mice showed increased body weight and BMI compared with the ApoE −/− genotype. The effect of GDF-15 deficiency on body weight corresponds with previous data [18] and with the observation that transgenic mice, which overexpressed GDF-15, showed hypophagia as well as reduced body weight [26,47]. In accordance with previous observations, the triglyceridemia was elevated in mice lacking GDF-15, whereas the cholesterol level was unchanged after 20 weeks CED, compared with ApoE −/− mice [18]. Additionally, the finding that GDF-15 deficiency considerably inhibits lumen stenosis, was previously described in the literature [18,27]. Therefore, we conclude that the inhibition of lesion progression in the absence of GDF-15 is obviously not due to a modification of plasma lipid levels. Consequently, other mechanisms or factors must be involved. One of these mechanisms could be autophagy. In this context, it was of high interest for this study to address the question of whether GDF-15 influenced cell death processes such as autophagy.
At first, we investigated the cellular composition and distribution in atherosclerotic plaque, because GDF-15 has been supposed to be involved in orchestrating atherosclerotic lesion progression [18]. Immunohistomorphometric analyses of atherosclerotic lesions in the BT show that GDF-15 deficiency did not affect the percentage of the sm-α-actin + area, CD31 + area (EC), CD68 + area (MΦ), as well as ATG5 + area (autophagy) in atherosclerotic plaques after 20 weeks CED. Other studies showed, by immunohistological investigation, a significant reduction in the number of APG5L/ATG + cells in GDF-15-deficient mice after 20 weeks of CED [18]. We suppose that the discrepancies may be due to the use of different staining methods.
The central findings of the previous observation suggest downregulation of apoptotic and autophagic processes in the absence of GDF-15 [18] because both apoptosis and autophagy are involved in cell death processes. Blocking autophagy renders MΦ in apoptosis, worsens the recognition and clearance of the dead cells by efferocytosis, and promotes plaque necrosis in a mouse model of advanced atherosclerosis [48]. For that reason, we investigated the accumulation of p62 in atherosclerotic lesions, because p62 is a critical checkpoint of the extrinsic apoptosis pathway, to control cell death and/or survival [49]. p62-protein regulates cell survival by the packing and delivery of polyubiquitinated, misfolded, aggregated proteins and dysfunctional organelles [50][51][52]. Increased levels of p62 in atherosclerotic plaques likely reflect dysfunctional autophagy because this reflects defects during the fusion stage with the lysosomes [53]. Here we show that the accumulation of p62 was significantly decreased in atherosclerotic plaques of GDF-15 −/− /ApoE −/− mice after 20 weeks CED, as well as after the silencing of GDF-15, which leads to a reduction of p62 accumulation in THP-1 MΦ [9]. Moreover, Bonaterra et al. observed significantly fewer apoptotic cells in the atherosclerotic plaques of GDF-15 −/− /ApoE −/− mice after 20 weeks of CED compared with ApoE −/− mice [18]. However, the induction of apoptosis results in the damage of cytoprotective mechanisms by the cleavage of essential ATG proteins [28,54] and consequently an inhibition of the autophagic flux, followed by increased p62 accumulation. Interestingly, only in CD31 + cells p62 accumulation significantly decreased in atherosclerotic plaques of GDF-15 −/− /ApoE −/− mice after 20 weeks CED. Grootaert et al. hypothesized that defective endothelial autophagy plays a direct role in aging-related arterial dysfunction, because defective autophagy in ECs leads to apoptosis and senescence [53], whereas and vice versa, functional endothelial autophagy limits plaque formation under high shear stress conditions [22].
Ionizing radiation (IR) induces mitochondrial ROS production in ECs and therefore causes damage and cellular senescence [55]. GDF-15, released in senescent ECs, contributes to the pathogenesis of atherosclerosis via its pro-senescent activity, implicating endothelial loss of function [55,56]. Also, senescent ECs expressed an increased level of GDF-15, whereas the paracrine effect of GDF-15 was associated with EC proliferation, migration and nitric oxide by non-senescent ECs [57]. Another study shows that GDF-15 causes endothelial dysfunction by impairing vascular contraction and relaxation [58]. Our study shows that GDF-15 −/− /ApoE −/− mice have increased survivin expression in atherosclerotic plaques, especially increased percentage of survivin positive ECs compared with ApoE −/− mice. Survivin, also known as Birc5, is a member of an inhibitor of the apoptosis protein family [59]. The general function of survivin is to inhibit cell apoptosis and promote proliferation [60,61]. It was previously suggested that survivin is not expressed in the normal adult vascular wall of mice and rabbits [62]. Several publications indicate that survivin is a negative regulator of autophagy that interacts with different proteins of the autophagic machinery, such as LC3, and interferes in the formation of autophagosomes, preventing LC3-I's cleavage into LC3-II. The survivin inhibitor YM155 increases the conversion of LC3-II and promotes autophagy-mediated ROS production, DNA damage and cell death in breast cancer cells [63,64]. Also, survivin inhibits the conjugation and complexation between ATG12, ATG5, and ATG16L1 which are crucial for the elongation of autophagophores during canonical autophagy [65]. It seems that survivin can interfere with the elongation of autophagosomes in ECs and prevent excessive autophagy, apoptosis and/or senescence following endothelial dysfunction in GDF-15 −/− /ApoE −/− mice after 20 weeks CED. On the other hand, we analyzed p53 in ECs of GDF-15 −/− /ApoE −/− and ApoE −/− mice after 20 weeks CED. p53 protein induces apoptosis by regulating the expression of several apoptotic genes. In particular, p53 binds specific elements of the survivin promoter and represses survivin expression [66,67]. In our study, the expression of p53 in atherosclerotic plaque was not detectable in ECs of both mice genotypes. These findings may imply a linkage between survivin and GDF-15 in relation to autophagy and apoptosis in arteriosclerotic plaques.
Our study suggests that GDF-15 is involved in establishing atherosclerotic lesions by the regulation of autophagic processes, which may have important pathophysiological consequences for atherosclerotic plaque progression and, thus, may be useful in developing novel strategies for therapeutic intervention.

Institutional Review Board Statement:
The study and procedures were approved by the Gießen Regional Commission (V54-19c2015h01MR20/26Nr.G40/2017; 9.10.2017) and were carried out in accordance with the regulations for the care of animals and their use for experiments at the Philipps University of Marburg.

Informed Consent Statement: Not applicable.
Data Availability Statement: All relevant data are presented in this paper.