bioengineering Article Biomimetic 3D Models for Investigating the Role of Monocytes and Macrophages in Atherosclerosis Anna Garcia-Sabaté 1 , Walaa Kamal E. Mohamed 1,2 , Jiranuwat Sapudom 1 , Aseel Alatoom 1 , Layla Al Safadi 1 and Jeremy C. M. Teo 1,3, * 1 Laboratory for Immuno Bioengineering Research and Applications, Division of Engineering, New York University Abu Dhabi, 129188 Abu Dhabi, UAE; anna.sabate@nyu.edu (A.G.-S.); wm1081@nyu.edu (W.K.E.M.); jiranuwat.sapudom@nyu.edu (J.S.); aseel.alatoom@nyu.edu (A.A.); las892@nyu.edu (L.A.S.) 2 Departament de Genètica i Microbiologia, Facultat de Biociències, Universitat Autònoma de Bellaterra, 08193 Barcelona, Spain 3 Department of Mechanical and Biomedical Engineering, Tandon School of Engineering, New York University, New York, NY 11201, USA * Correspondence: jeremy.teo@nyu.edu; Tel.: +971-2-628-6689  Received: 25 August 2020; Accepted: 14 September 2020; Published: 16 September 2020  Abstract: Atherosclerosis, the inflammation of artery walls due to the accumulation of lipids, is the most common underlying cause for cardiovascular diseases. Monocytes and macrophages are major cells that contribute to the initiation and progression of atherosclerotic plaques. During this process, an accumulation of LDL-laden macrophages (foam cells) and an alteration in the extracellular matrix (ECM) organization leads to a local vessel stiffening. Current in vitro models are carried out onto two-dimensional tissue culture plastic and cannot replicate the relevant microenvironments. To bridge the gap between in vitro and in vivo conditions, we utilized three-dimensional (3D) collagen matrices that allowed us to mimic the ECM stiffening during atherosclerosis by increasing collagen density. First, human monocytic THP-1 cells were embedded into 3D collagen matrices reconstituted at low and high density. Cells were subsequently differentiated into uncommitted macrophages (M0) and further activated into pro- (M1) and anti-inflammatory (M2) phenotypes. In order to mimic atherosclerotic conditions, cells were cultured in the presence of oxidized LDL (oxLDL) and analyzed in terms of oxLDL uptake capability and relevant receptors along with their cytokine secretomes. Although oxLDL uptake and larger lipid size could be observed in macrophages in a matrix dependent manner, monocytes showed higher numbers of oxLDL uptake cells. By analyzing major oxLDL uptake receptors, both monocytes and macrophages expressed lectin-like oxidized low-density lipoprotein receptor-1 (LOX1), while enhanced expression of scavenger receptor CD36 could be observed only in M2. Notably, by analyzing the secretome of macrophages exposed to oxLDL, we demonstrated that the cells could, in fact, secrete adipokines and growth factors in distinct patterns. Besides, oxLDL appeared to up-regulate MHCII expression in all cells, while an up-regulation of CD68, a pan-macrophage marker, was found only in monocytes, suggesting a possible differentiation of monocytes into a pro-inflammatory macrophage. Overall, our work demonstrated that collagen density in the plaque could be one of the major factors driving atherosclerotic progression via modulation of monocyte and macrophages behaviors. Keywords: atherosclerosis; monocyte; macrophage; disease model; collagen; 3D cell culture; immunomechanobiology Bioengineering 2020, 7, 113; doi:10.3390/bioengineering7030113 www.mdpi.com/journal/bioengineering Bioengineering 2020, 7, 113 2 of 20 1. Introduction Atherosclerosis is the dominant underlying causation of coronary heart disease and cerebrovascular disease [1] and statistically a major cause of morbidity and mortality worldwide [2,3]. The systematic formation of detrimental atherosclerotic plaques, via monocyte to macrophage differentiation to foam cell formation, which eventually narrows and occludes arteries, is well-reviewed elsewhere [4,5]. In short, monocytes are recruited to lesion sites, where low-density lipoproteins (LDL) and apolipoprotein B-containing lipoproteins have accumulated and activated the endothelium. Monocytes then infiltrate the lesion and differentiate into macrophages, which will uptake lipoproteins, ultimately resulting in foam cells that comprise the central core of atheromas that will progressively occlude the vessel. Atherosclerosis is considered a non-resolving inflammatory condition [2,6], akin to chronic wounds that will not resolve towards healing. On account of this close association with wound healing, and evidence that atherosclerotic plaques are vastly populated by pan-macrophage marker CD68+ cells, macrophages are thought to play other lead roles beyond foam cell formation that progresses plaque buildup [7,8]. In atherosclerosis, macrophages are implicated in progression or regression of plaques, as well as in plaque stabilization or its rupture [9,10]. Pro-inflammatory macrophages, or M1 macrophages, are responsible for progression, leading towards instability and finally resulting in rupture [11]. Conventionally, at the complete opposite end of the macrophage spectrum, sits the anti-inflammatory or M2 macrophages [12], which mediate a more favorable outcome of atherosclerosis [13]. It is now accepted that macrophage phenotypes are not dichotomous but lie along this M1-M2 spectrum, and depending on the atherosclerotic cellular niche, macrophages can dynamically traverse this M1-M2 continuum [12,14], affecting the outcome of this disease. While significant advancements have been made in terms of disease management, research must now shift towards understanding the mechanisms of atherosclerosis regression and the repair of atherosclerotic lesions [15]. It is, therefore, imperative to understand macrophage biology in the interest of this new pivot against atherosclerosis. Animal models have provided abundant information for atherosclerosis research, however, to understand atherosclerosis healing mechanistically, relevant in vitro models are a must. Models of atherosclerosis have progressed from simple 2D culture to 3D multi-cellular cultures [16,17], the latter a better representation considering the dimensionality of the disease and anatomical features involved. To mimic in vivo-like conditions, in vitro models have encapsulated the relevant cells of interest within naturally-derived and polymer-based hydrogels. Several 3D models of atherosclerosis have been published [18–21]. While they mostly tackled the interaction between different cells in atherosclerotic plaques at different stages of plaque progression, they overlooked some factors which we aim to address. The 3D extracellular matrix (ECM) used for scaffolding the models was not assessed in terms of its biomechanical influence on monocytes and macrophages in regard to atherosclerosis’ cellular niche. This is especially important since collagen comprises up to 60% of the plaque protein [22], and the amount of collagen in the plaque dictates its mechanical stability [23], where a deficiency of collagen may lead to plaque rupture and an excess results in narrowing of the artery. Moreover, aged or obese individuals are known to exhibit notably stiffened and thickened arteries [24]. Also, our previous work has specifically shown that macrophage functionality and immune phenotype is regulated by physical parameters of the surrounding ECM [25]. While the majority of biological readouts from 2D and 3D in vitro atherosclerosis models have focused on LDL uptake by the immune cells and secretion of specific pro-inflammatory cytokines, there is a lack of comprehensive secretome studies. Owing to the complexity of the biochemical milieu in atherosclerotic microenvironments, we want to examine adipokines and growth factors, two groups of cytokines that are currently vague, but crucial to understanding the mechanisms of regression and repair of atherosclerotic plaques. Furthermore, most 3D models commonly employ monocytes or resident macrophages (M0) for LDL exposure to simulate atherosclerosis. However, there are pathological situations, for example, dyslipidemia, diabetes, hypertension, obesity, and smoking, Bioengineering 2020, 7, 113 3 of 20 whereby transendothelial permeability to LDLs increases [26,27] and resident macrophages of varying phenotypes within the intima could uptake LDLs, thus also modulate atherosclerotic plaque [14]. In this study, we employ a system of collagen-based hydrogels with different matrix densities that recapitulate the porosity of early (low tissue density) and advanced (high tissue density) atherosclerotic tissue in vivo [28,29]. Within these fibrillar collagen hydrogels, a commonly used and genetically uniform human monocytic cell line, THP-1, is embedded, differentiated into various macrophage phenotypes, and later exposed to oxidized LDLs (oxLDL). Here, we aim to elucidate the effects of the predominant ECM, collagen, on these immune cells by isolating them from other factors such as different cell types or proteins. The assessment was performed to evaluate oxLDL uptake, scavenger receptors and immune phenotype gene expressions, and their secretions of adipokine and growth factors. 2. Materials and Methods 2.1. Cell Culture Human monocytic cell line THP-1 were maintained in RPMI-1640 (Gibco, Thermo Fisher Scientific, Inc., Waltham, MA, USA) with 10% fetal bovine serum (FBS), 1% HEPES, 1% sodium pyruvate, 0.01% beta-mercaptoethanol, and 1% penicillin/streptomycin (all from Invitrogen, Thermo Fisher Scientific, Inc.) at 37 ◦ C, 95% humidity and 5% CO2 (standard cell culture conditions). 2.2. Embedding and Differentiating THP-1 towards Macrophages in 3D Collagen Matrices 3D collagen matrices at concentrations of 1 and 3 mg/mL were prepared as previously published [25,30]. In short, a collagen solution was prepared by mixing of type I rat tail collagen (Advanced BioMatrix, Inc., Carlsbad, CA, USA) with 250 mM of phosphate buffer and 0.1% acetic acid (both from Sigma-Aldrich, Inc., St. Louis, MO, USA). Afterward, 1 × 105 THP-1 cells were suspended in the prepared collagen solution and then transferred onto glutaraldehyde-coated coverslips. Cell-free collagen matrices were analyzed prior to use for cell culture regarding their topological and mechanical properties, using an image-based analysis toolbox [31] and non-destructive rheometer [25,32], respectively. Macrophage differentiation was performed using an established protocol [25]. In brief, THP-1 cells were differentiated into uncommitted macrophages (M0) by culturing in RPMI 1640 media without FBS supplemented with 300 nM phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich, Inc.) for 6 h. To achieve pro-inflammatory (M1) and anti-inflammatory (M2) macrophages, PMA containing media was removed, and M0 were rested in RPMI-1640 cell culture media without FBS for 24 h. Afterward, cells were cultured in activating media for 48 h. For M1 activation, RPMI-1640 cell culture media without FBS was supplemented with 10 pg/mL lipopolysaccharide (LPS, Sigma-Aldrich, Inc.) and 20 ng/mL interferon-gamma IFN-γ (Biolegend, Inc., San Diego, CA, USA), while 20 ng/mL interleukin 4 (IL-4, Biolegend, Inc.) and 20 ng/mL interleukin 13 (IL-13, Biolegend, Inc.) was used for M2 activation. 2.3. Topological and Mechanical Characterization of 3D Collagen Matrices Cell-free 3D collagen matrices were analyzed regarding their topological and mechanical properties based on previously published protocols [32]. In brief, 5- (and-6)- carboxytetramethylrhodaminesuccinimidylester (5(6)-TAMRA-SE) (Sigma-Aldrich, Inc.) stained 3D collagen matrices were imaged using confocal microscopy (SP8; Leica Microsystems GmbH, Wetzlar, Germany) at 63X. Z-stacked (50 µm depth, at 5 µm intervals) image datasets (of pixel size 0.13 × 0.13 µm) were analyzed using a custom-built image analysis toolbox [31] to obtain mean pore size. For each type of 3D collagen matrix, analysis was performed in triplicate with three positions per sample. Measurements of mechanical properties was performed on cell-free label-free 3D collagen matrices non-destructively (ElastoSense™ Bio; Rheolution Inc., Montreal, QC, Canada). Bioengineering 2020, 7, 113 4 of 20 Briefly, fibrillation of 3 mL of collagen solution was initiated at 37 ◦ C and the elastic modulus of collagen was measured during polymerization. Three samples per matrix type were analyzed. 2.4. Treatment with oxLDL Monocyte (THP-1), plus uncommitted and activated macrophages were cultured in the presence of 5 µg/mL oxidized low-density lipoproteins from human plasma (oxLDL; Molecular Probes™, Thermo Fisher Scientific, Inc.) for 5 days at standard cell culture condition. oxLDL conjugated 1,10 -dioctadecyl-3,3,30 ,30 -tetramethylindocarbocyanine perchlorate (DiI) was used for quantification and visualization of LDL uptake. As a control, cells were cultured at similar conditions without oxLDL. 2.5. Quantitative Analysis of oxLDL Uptake After 5 days of LDL treatment, collagen matrices were digested with 6 mg/mL of collagenase (Worthington Biochemical Corp., Lakewood, NJ, USA) prepared in cell culture media for 15 min at standard cell culture conditions to digest the collagen. Afterward, cells were analyzed using Attune NxT Flow Cytometer equipped with autosampler (Thermo Fisher Scientific, Inc.). Percentage of oxLDL uptake cells and mean fluorescent intensity were quantified using FlowJow Software (Build 10.5.3; BD Life Sciences). Experiments were performed in 5 replicates. 2.6. Quantitative Analysis of Lipid Droplet Size For analysis of lipid droplet size, cells were first fixed with 4% paraformaldehyde and stained with 40 ,6-diamidino-2-phenylindole (DAPI; 1:10,000 dilution in PBS) for 2 h at room temperature. Afterward, cells were imaged using a confocal laser scanning microscope (cLSM) (SP8; Leica Microsystems GmbH). The cLSM stacked images were gathered and analyzed regarding the lipid size using DiI-oxLDL fluorescence signal by a home-built MATLAB script (MATLAB 2019a; The MathWorks, Inc., Natick, MA, USA). 2.7. Quantitative Analysis of Adipokines, Chemoattractants, and Growth Factors Secretion of adipokines, chemoattractants, and growth factors were analyzed from cell culture supernatant after 5 days of cell culture in the presence and absence of oxLDL. Custom bead-based multiplex immunoassays were used to quantify secretome (Biolegend, Inc.), following the manufacturer’s instructions. Samples were analyzed using Attune NxT Flow Cytometer equipped with autosampler (Thermo Fisher Scientific, Inc.). Data analysis was done using LEGENDplex™ Data Analysis Software (Biolegend, Inc.). 2.8. Gene Expression Analysis Gene expression analysis was performed using an established protocol, as published [25]. Briefly, total RNA was extracted using TRIzol (Invitrogen, Thermo Fisher Scientific, Inc.) and converted into complementary DNA (cDNA) using a high-capacity cDNA reverse transcription kit (Applied Biosystems, Thermo Fisher Scientific, Inc.). The cDNA concentration and purity (the ratio of absorbance at 260 nm and 280 nm) were quantified using nanodrops (Thermo Fisher Scientific, Inc.) prior to performing gene expression analysis. The primers used in this study were synthesized from Bioneer Inc. The primer sequences are listed in Supplementary Table S1. qPCR was performed using the SYBR Green PCR Master Mix (Applied Biosystems, Thermo Fisher Scientific, Inc.). The qPCR procedure was set as follows: denaturation for 5 min at 95 ◦ C; 45 cycles of denaturation (95 ◦ C, 15 s), annealing under primer-specific conditions (30 s), and target gene-specific extension (30 s at 72 ◦ C). Fluorescence signals were measured for 20 s at 72 ◦ C. To confirm the specificity of the PCR products, a melting curve analysis was performed at the end of each run. The Beta-actin gene was used as a reference gene. Experiments were performed in four replicates. Bioengineering 2020, 7, 113 5 of 20 2.9. Data and Statistical Analysis Unless otherwise stated, all experiments were performed in at least four repeats, and data are represented as mean ± standard deviation (SD). Statistical significance has been determined by two-way ANOVA followed by Tukey’s test using Prism 8 (GraphPad Software), and the level of significance was set to p < 0.05, unless otherwise stated. 3. Results With atherosclerosis being the main cause of cardiovascular diseases [1], there is an increasing need for platforms that can mimic in vivo conditions. Furthermore, these platforms must be thoroughly characterized, and a profound understanding of individual contributions of the cells involved in atherosclerosis development must be reached before venturing to co-culture models. Most atherosclerosis studies are carried on 2D tissue culture plates [16]. While meaningful contributions have been made towards 3D modeling atherosclerosis in vitro, most studies have focused on co-culture of endothelial cells (ECs), smooth muscle cells (SMCs), and monocytes, in either physiological [18] or synthetic [20] scaffolds, while detailed individual contributions of macrophages still remain unclear. Here, we embedded monocytic THP-1 cells into 3D type I collagen matrices of low and high concentrations, surrogates for the plaque of low and high densities. As mentioned earlier, collagen I is known to make up for approximately two-thirds of the total collagen in atherosclerotic plaques [22], and the progression of atherosclerosis leads to an increase in collagen concentration within the different layers of diseased vessels. By reconstituting collagen matrices at low and high densities, we were able to mimic the changes of the ECM at early and advanced stages of atherosclerosis. The pore size of the collagen matrices was 11.54 ± 0.88 µm and 3.67 ± 0.95 µm for 1 mg/mL and 3 mg/mL collagen concentration, respectively, and as expected, matrix elasticity was enhanced with an increase of collagen concentration and was 52.10 ± 10.38 Pa and 211.43 ± 15.62 Pa for 1 mg/mL and 3 mg/mL collagen concentration, respectively. The resulting pore sizes span the pore sizes found in vivo [28,30]. Those of atherosclerotic plaque have not been experimentally measured as far as we know, likely due to the complexity of such tissues. However, it has been reported that regressive plaques (i.e., more advanced) have an increased content of collagen [33]. Embedded cells were differentiated into macrophages and further activated into distinct pro- and anti-inflammatory phenotypes. Cells were cultured in the presence of oxLDL simulating atherosclerosis conditions in vitro for 5 days. The duration of the treatment was experimentally determined by measuring oxLDL uptake at different timepoints, we observed that engulfment of LDL by all cells plateaued from 5 days onwards (data not shown). To understand the differences in response by each cell type, we evaluate the uptake of lipids, further assess their expression of relevant oxLDL receptors, quantified their secretomes, and finally examined their expression of selected surface markers to discern if there are any phenotypic changes due to oxLDL exposure. 3.1. %oxLDL + Cells Are Greatest in Monocytes and M1 Accumulates the Most oxLDL Monocytes and macrophages were cultured in the presence and absence of DiI fluorescent-tagged oxLDL (DiI-oxLDL) for 5 days. As shown in Figure 1A, it can be observed that macrophages uptake a higher amount of oxLDL than monocytes, and particularly M1 appears to engulf more oxLDL than other macrophage phenotypes. No differences in uptake could be visually observed between low and high matrix density for the same cell phenotypes. To confirm the visual impressions, cells were harvested from 3D collagen matrices by digestion with collagenase and oxLDL uptake was analyzed by flow cytometry. The percentage of oxLDL positive cells (oxLDL+ cells; cells that uptake oxLDL) and the uptake capacity were assessed using mean fluorescence intensity (MFI). As shown in Figure 1B, we observed a significantly high percentage of monocytes that uptake oxLDL when compared to macrophages (see Supplementary Figure S1A). However, monocytes engulf a lower amount of oxLDL, as demonstrated through MFI. M1 showed significantly higher uptake compared to M0 and M2 Bioengineering 2020, 7, 113 6 of 20 macrophages (Supplementary Figure S1B) but without significant difference across matrix densities. We further quantified the size of lipid droplets using a custom-built image analysis toolbox written Bioengineering in MATLAB. 2020, 7, x FOR PEER we Interestingly, REVIEW found larger lipid droplets in low matrix density in all macrophage 6 of 20 phenotypes, while no differences could be observed in monocytes. In sum, our In sum, ourresults results suggest suggest higher higher oxLDL oxLDL uptakeuptake in all macrophages, in all macrophages, especiallyespecially in M1 in M1 phenotypes, phenotypes, than in monocytes. This finding is in line with the high phagocytic than in monocytes. This finding is in line with the high phagocytic activity of M1 macrophages activity of [34], M1 macrophages [34], while it contradicts reports that showed M2 macrophages have while it contradicts reports that showed M2 macrophages have a higher capacity to accumulate LDL a higher capacity to accumulate than M1 in theLDL than M1 2D culture in the[35,36]. model 2D culture modelreports Literature [35,36].that Literature reports that M1 macrophages areM1themacrophages predominant are the predominant phenotype present in human and mouse plaque lesions phenotype present in human and mouse plaque lesions and directly contribute to atheroscleroticand directly contribute plaque to atherosclerotic plaque formation in vivo [11,37–40], which supports our findings formation in vivo [11,37–40], which supports our findings in the 3D cell culture model. Both %oxLDL+ in the 3D cell culture cells, asmodel. well asBoth lipid%oxLDL+ size, was cells, as well reduced as lipid in higher size, was density reduced in higher density matrices. matrices. Figure 1. oxLDL uptake by monocytes and macrophages in 3D collagen matrices. (A) Representative Figure 1. oxLDL uptake by monocytes and macrophages in 3D collagen matrices. (A) Representative confocal images of monocytes and macrophages with DiI-oxLDL uptake. (B) In this infographic, confocal images of monocytes and macrophages with DiI-oxLDL uptake. (B) In this infographic, the the number of oxLDL+ cells (y-axis) was greatest in THP-1 (grey) followed by M2 (green) macrophages, number of oxLDL+ cells (y-axis) was greatest in THP-1 (grey) followed by M2 (green) macrophages, then M1 (red), and finally M0 (blue). The amount of oxLDL in the cells, as quantified through mean then M1 (red), and finally M0 (blue). The amount of oxLDL in the cells, as quantified through mean fluorescence intensity (MFI, solid circle) (circle size, see legend), was greatest in M1 macrophages, fluorescence intensity (MFI, solid circle) (circle size, see legend), was greatest in M1 macrophages, followed by M0, then M2, and finally THP-1. The dotted circle is the corresponding standard deviation followed by M0, then M2, and finally THP-1. The dotted circle is the corresponding standard of the MFI. A sample size of 5 was performed for each condition. (C) Lipid droplet size as measured deviation of the MFI. A sample size of 5 was performed for each condition. (C) Lipid droplet size as from images obtained with confocal microscopy using a custom-built image analysis toolbox. Data were measured from images obtained with confocal microscopy using a custom-built image analysis shown as a boxplot (whiskers represent minimum and maximum, + represents the mean, line inside toolbox. Data were shown as a boxplot (whiskers represent minimum and maximum, + represents each box plot represents the median). Significance is represented as * p < 0.05 between samples in the mean, line inside each box plot represents the median). Significance is represented as * p < 0.05 1 mg/mL collagen concentration. between samples in 1 mg/mL collagen concentration. 3.2. oxLDL Induced LOX1 Expression in Monocytes and Macrophages, while It Enhanced CD36 Expression in M2oxLDL 3.2. Phenotypes Induced LOX1 Expression in Monocytes and Macrophages, while It Enhanced CD36 Expression in M2ToPhenotypes better understand the oxLDL uptake and lipid size formation, we assessed if the presence of oxLDL affectsunderstand To better its uptake receptors. the oxLDLWe analyzed uptake the gene and lipid expressionwe size formation, profile of two assessed main if the receptors presence of oxLDL affects its uptake receptors. We analyzed the gene expression profile of two main receptors responsible for oxLDL uptake, namely lectin-like oxidized low-density lipoprotein receptor-1 (LOX1) and scavenger receptor class B member 3 (CD36) through qPCR [41,42]. LOX1 is a transmembrane glycoprotein that binds to and internalizes oxLDL resulting in distinct cell-type-specific downstream signaling, which leads to differential cellular behaviors [43]. As shown in Figure 2A, in the absence of Bioengineering 2020, 7, 113 7 of 20 responsible for oxLDL uptake, namely lectin-like oxidized low-density lipoprotein receptor-1 (LOX1) and scavenger receptor class B member 3 (CD36) through qPCR [41,42]. LOX1 is a transmembrane Bioengineering 2020, 7, x FOR PEER REVIEW 7 of 20 glycoprotein that binds to and internalizes oxLDL resulting in distinct cell-type-specific downstream signaling, which leads to differential cellular behaviors [43]. As shown in Figure 2A, in the absence oxLDL, no significant change in LOX1 could be observed between monocytes and macrophages as well of oxLDL, no significant change in LOX1 could be observed between monocytes and macrophages as between matrix density. After treatment with oxLDL, all cells exhibited up-regulation of LOX1 as well as between matrix density. After treatment with oxLDL, all cells exhibited up-regulation of expression with the highest expression in M1, tying in with our MFI data, followed by M2, M0 LOX1 expression with the highest expression in M1, tying in with our MFI data, followed by M2, macrophages, and then monocytes. Our finding is corroborated by other reports suggesting an enhanced M0 macrophages, and then monocytes. Our finding is corroborated by other reports suggesting LOX1 expression via activated nuclear factor kappa-light-chain-enhancer of activated B cells (NF-kB) [44], an enhanced LOX1 expression via activated nuclear factor kappa-light-chain-enhancer of activated pro-inflammatory cytokines mediated LOX1 expression [45], as well as oxidative stress [46]. The B cells (NF-kB) [44], pro-inflammatory cytokines mediated LOX1 expression [45], as well as oxidative mentioned mechanisms might also explain the higher expression of LOX1 in M1 phenotypes. stress [46]. The mentioned mechanisms might also explain the higher expression of LOX1 in M1 Contradicting our MFI data, we found slightly higher expression in the dense matrix in monocytes and phenotypes. Contradicting our MFI data, we found slightly higher expression in the dense matrix in macrophages, except M2 phenotypes. We postulate that this observation might be caused by an enhanced monocytes and macrophages, except M2 phenotypes. We postulate that this observation might be modulation of pro-inflammatory response by macrophages in a dense matrix, as previously reported [25]. caused by an enhanced modulation of pro-inflammatory response by macrophages in a dense matrix, Further evidence that collagen concentration can modulate the phenotype of macrophages is found in as previously reported [25]. Further evidence that collagen concentration can modulate the phenotype atherosclerotic plaque progression and regression studies [47–49], where progressive plaques (lower of macrophages is found in atherosclerotic plaque progression and regression studies [47–49], where collagen content) contain more pro-inflammatory macrophages, while more anti-inflammatory progressive plaques (lower collagen content) contain more pro-inflammatory macrophages, while macrophages are found in regressive plaques (higher collagen content). more anti-inflammatory macrophages are found in regressive plaques (higher collagen content). Figure 2. Expression of oxLDL receptors of monocytes and macrophages cultured in 3D collagen Figure 2.(A) matrices. Expression There is an of increase oxLDL receptors of monocytes in gene expression and macrophages of lectin-like cultured in lipoprotein oxidized low-density 3D collagen matrices. (LOX1) receptor-1 (A) There is anmonocytes in both increase in(THP-1) gene expression of lectin-like and significantly oxidized low-density in all macrophage lipoprotein subtypes when cells receptor-1 (LOX1) in both monocytes (THP-1) and significantly in all macrophage are cultured in 3D collagen hydrogels and exposed to oxLDL. Significantly, M1+LDL exhibited the subtypes when cells highest are cultured in 3D collagen hydrogels and exposed to oxLDL. Significantly, fold induction and THP-1+LDL, the lowest, with M0+LDL and M2+LDL comparable. LOX1 expression M1 +LDL exhibited the highest only foldsignificantly changed induction and in M1THP-1 +LDL, the lowest, with M0+LDL and M2+LDL comparable. LOX1 with collagen concentration (x-axis, 1 mg/mL to 3 mg/mL). (B) Genetic expression only changed significantly expression of CD36, another oxLDL receptor, was onlyin M1 with measurable collagen concentration in M2+LDL, and(x-axis, 1 mg/mL significantly to 3 higher mg/mL). (B) Genetic expression of CD36, another oxLDL receptor, was only measurable expression was found in lower density collagen hydrogels. All significances represented as p < 0.05, in M2 +LDL, and significantly higher expression was found in lower density collagen hydrogels. * compared to untreated sample at the same matrix condition, & between oxLDL samples of different All significances represented cell types at theassame p < 0.05, matrix * condition, compared and to untreated sample # significant changeatwith the same matrix collagen condition,The concentration. & between dotted oxLDL samples horizontal of different line indicates a levelcell types of no at the same matrix condition, and # significant change with change. collagen concentration. The dotted horizontal line indicates a level of no change. CD36 is a plasma membrane glycoprotein that binds a diverse array of ligands, including oxLDLCD36 is aAs [50,51]. plasma shownmembrane in Figure glycoprotein 2B, we foundthat binds higher CD36a diverse arraysolely expression of ligands, in M2including oxLDL macrophage in [50,51]. the As and absence shown in Figure presence 2B, wewhereby of oxLDL, found higher CD36 expression a significant up-regulation solely in M2expression of CD36 macrophage in be could the absence after observed and presence treatmentof oxLDL, with oxLDL.whereby a significant The expression up-regulation of CD36 of CD36 in THP-1 derived M2expression could be has been reported; observed after treatment with oxLDL. The expression of CD36 in THP-1 derived these cells were polarized based on a similar protocol [52]. Although M2 showed an increase in both M2 has been reported; LOX1 these expression, and CD36 cells were polarized the amount based on a uptake of oxLDL similar isprotocol [52].the lower than Although M2 showed M1 counterpart, which an increase in both LOX1 and CD36 expression, the amount of oxLDL uptake is lower than the M1 counterpart, which exhibited up-regulation of LOX1. The reason could be on account of M2 being anti-atherosclerotic [13] or because of LDL degradation after CD36-mediated uptake [51]. Although the contribution of LOX1 in the oxLDL uptake is reported to be minimal compared to other receptors [42], an increase of soluble LOX1 can contribute up to 40% in oxLDL uptake [43]. Similar to LOX1, we observed a slight increase in CD36 expression of M2 in the denser matrix after treatment with Bioengineering 2020, 7, 113 8 of 20 exhibited up-regulation of LOX1. The reason could be on account of M2 being anti-atherosclerotic [13] or because of LDL degradation after CD36-mediated uptake [51]. Although the contribution of LOX1 in the oxLDL uptake is reported to be minimal compared to other receptors [42], an increase of soluble LOX1 can contribute up to 40% in oxLDL uptake [43]. Similar to LOX1, we observed a slight increase in CD36 expression of M2 in the denser matrix after treatment with oxLDL. A report suggested a correlation between vessel stiffening and CD36 expression in endothelial cells [53]. To further understand the contributing role of monocytes and macrophages within the atherosclerotic lesion, we performed a quantitative analysis of their secretome using custom bead-based multiplex ELISA. 3.3. RBP4 Is Up-Regulated by All Cell Types with the Addition of oxLDL and Enhanced by High-Density Matrices for THP-1+LDL and M1+LDL Adipokines retinol-binding protein 4 (RBP4), resistin, and leptin are reported to support the progression of atherosclerosis, while adiponectin has protective actions against plaque formation [54]. These factors are also known to be released along with other pro-inflammatory factors during the development of obesity [55–57], and within adipose tissue, they have been shown to induce inflammation by activating macrophages [58]. While no significant statistical differences were found in adipokines, some distinct qualitative trends were observed, which are discussed below. Clinically, RBP4 is a marker for atherosclerotic-associated cardiovascular disease [59]. In animal models, RBP4 has been found to positively contribute towards atherosclerotic progression and mediate LDL uptake [60]. In our model, we found an increase in RPB4 secretion by all cell types in response to oxLDL (Figure 3A); generally, higher secretions were measured in the denser matrix for THP-1+LDL and M1+LDL . It has been demonstrated that RBP4 expression was enhanced in the areas rich in macrophage foam cells in atherosclerotic lesions of aortic specimens from both humans and apolipoprotein E-deficient mice [60], suggesting foam cell-like formation in our biomimetic atherosclerosis models. Leptin reportedly exacerbates inflammation [61] in low matrix density; our results showed that THP-1+LDL and M0+LDL reduced leptin secretion (Figure 3B) as with resistin (Figure 3C). It is reported that human macrophages secrete resistin on their own, and this contributes to atherosclerosis development directly by causing endothelial and smooth muscle cell (SMC) dysfunction [62,63], potentially increasing the permeability of LDL into adjacent tissue. Only THP-1+LDL showed elevated leptin secretion in denser matrices. Evidence suggests that leptin and resistin may initiate the recruitment of monocytes, macrophages, endothelial cells, and smooth muscle cells towards the atherosclerotic site [64,65]. At site, these cell types all individually progress atherosclerosis, although beyond the scope of this study, co-culture experiments with the above-mentioned cell types with our biomimetic model have to be performed to enhance the understanding of this complex disease. The atheroprotective adiponectin was below the assay’s theoretical limit of detection using the bead-based ELISA kit (minimum detectable concentration 41.1 pg/mL) in all cells with and without treatment with oxLDL, potentially due to its short half-life [66]. Overall, with the addition of oxLDL there is an increase of secreted adipokines in the dense matrix, particularly by THP-1+LDL and M1+LDL , similarly to what would be expected in an advanced atherosclerotic plaque. Bioengineering 2020,7,7,113 Bioengineering2020, x FOR PEER REVIEW 9 9ofof2020 Figure 3. Adipokine secretome from monocytes and macrophages cultured in 3D collagen hydrogels. Figure Box plots3. of Adipokine absolute secretome from monocytes secretion levels of (A) RBP4,and(B)macrophages cultured leptin, and (C) in to resistin 3Dobserve collagentrends hydrogels. and Box plots of absolute secretion levels of (A) RBP4, (B) leptin, and (C) resistin to observe indicate comparisons that are statistically significant. The dotted line in the box plots marks the lowest trends and indicate comparisons that are statistically significant. The dotted line in the box detectable concentration of assay. Statistical significance has been determined by two-way ANOVAplots marks the lowest detectable followed byconcentration of assay. Tukey’s test, and Statistical the level significance of significance has been determined is represented as * p < 0.05,by two-way ANOVA significantly higher followed than by Tukey’s the untreated test, at sample andthethe levelmatrix same of significance is represented density condition, as * p < 0.05, no significance significantly higher otherwise. than the untreated sample at the same matrix density condition, no significance otherwise. 3.4. In High-Density Matrices THP-1+LDL and M1+LDL Secretion of Immune Cell Chemoattractants Are Elevated and Suppressed 3.4. In High-Density in M2+LDL Matrices THP-1+LDL and M1+LDL Secretion of Immune Cell Chemoattractants Are Elevated and Suppressed We measured IP-10,inMCP-1, M2+LDL and IL-8 levels from our cell culture system as these are known chemoattractants We measured that are involved IP-10, MCP-1, and in the IL-8recruitment levels fromofour inflammatory cell culture cells systemto as thethese atherosclerotic are known site [67–69]. Both IP-10 (Figure 4A) and IL-8 (Figure 4B) concentrations are significantly chemoattractants that are involved in the recruitment of inflammatory cells to the atherosclerotic site elevated in the dense matrix with [67–69]. Both IP-10 (Figure M1 +LDL compared to the other cell types, whereas the chemoattractants 4A) and IL-8 (Figure 4B) concentrations are significantly elevated in the are suppressed with M2 +LDL . IL-8 dense matrix with M1+LDL compared recruits cells to that the participate other cell in acutewhereas types, inflammation [70], while IP-10 isare the chemoattractants a chemoattractant for immune cells from the adaptive immune system to the atherosclerotic suppressed with M2+LDL. IL-8 recruits cells that participate in acute inflammation [70], while IP-10 is site [71]. Itaischemoattractant reviewed in thefor literature immune that thefrom cells adaptive immuneimmune the adaptive system during systematherosclerosis changessite to the atherosclerotic from a [71]. protective to a pathological response, typically when the ratio of effector T-cells to It is reviewed in the literature that the adaptive immune system during atherosclerosis changes from regulatory T-cells increases [72,73]. a protective to aTherefore, pathologicalthe increased response,secretion typicallyofwhen both IL-8 and IP-10 the ratio by M1+LDL of effector T-cellsin denser collagen to regulatory T- matrices, and hence cells increases theTherefore, [72,73]. respective the immune cell types, increased is inofresponse secretion both IL-8to and the stiffer IP-10 bytissue in advanced M1+LDL in denser atherosclerotic collagen matrices,plaques. and hence the respective immune cell types, is in response to the stiffer tissue in advanced atherosclerotic plaques. Bioengineering 2020,7,7,113 Bioengineering2020, x FOR PEER REVIEW 1010ofof2020 Figure 4. Chemokine secretome from monocytes and macrophages cultured in 3D collagen hydrogels. Figure Box 4. of plots Chemokine secretomelevels absolute secretion from monocytes of (A) IP-10,and macrophages (B) IL-8, and (C) cultured MCP-1into3D collagen observe hydrogels. trends and Box plots indicate of absolutethat comparisons secretion levels of significant. are statistically (A) IP-10, (B) TheIL-8, andline dotted (C)inMCP-1 to observe box plots marks thetrends and lowest indicate comparisons that are statistically significant. The dotted line in box plots detectable concentration of assay. Statistical significance has been determined by two-way ANOVA marks the lowest detectable followed byconcentration of assay. Tukey’s test, and Statistical the level significance of significance has been determined is represented by two-way as * p < 0.05 and # p