Heterogeneity of foam cell biogenesis across diseases

. Foam cells are dysfunctional, lipid-laden macrophages associated with chronic inflammation of infectious and non-infectious origin. For decades, the paradigm of foam cell biology has been atherogenesis, in which macrophages accumulate cholesteryl esters. Our previous work showed that foam cells in tuberculous lung lesions are surprisingly triglyceride-rich, suggesting multiple modalities of foam cell biogenesis. In the present study, we used matrix-assisted laser desorption/ionization mass spectrometry imaging to assess the spatial distribution of storage lipids relative to foam-cell-rich areas in murine lungs infected with the fungal pathogen Cryptococcus neoformans and in human papillary renal cell carcinoma resection tissues. We also analyzed neutral lipid content and the transcriptional program of lipid-laden macrophages generated under corresponding in vitro conditions. The in vivo data were consistent with in vitro findings showing that C. neoformans -infected macrophages accumulated triglycerides, while macrophages exposed to human renal cell carcinoma-conditioned medium accumulated both triglycerides and cholesteryl esters. Moreover, macrophage transcriptome analyses provided evidence for condition-specific metabolic remodeling. The in vitro data also showed that although both Mycobacterium tuberculosis and C. neoformans infections induced triglyceride accumulation in macrophages, they did so by different molecular mechanisms, as evidenced by different sensitivity of lipid accumulation to the drug rapamycin and the characteristics of macrophage transcriptome remodeling. Collectively, these data demonstrate that the mechanisms of foam cell formation are specific to the disease microenvironment. Since foam cells have been regarded as targets of pharmacological intervention in several diseases, recognizing that their formation is disease-specific opens new research directions of biomedical significance.

Chronic inflammation of infectious and non-infectious origin is often associated with the presence of foam cells, lipid-laden macrophages that exhibit impaired immune function and can contribute to pathogenesis (1). Foam cells form when, due to dysregulated metabolism, lipids accumulate beyond the homeostatic capacity of macrophages. The lipids are stored as droplets that confer a foamy appearance to the macrophages (2).
Our understanding of foam cell biology has been largely based on studies of atherogenesis, a disease in which uptake of normal and proinflammatory lipoproteins by macrophages in the arterial wall leads to imbalanced cholesterol metabolism and formation of cholesterol-laden foam cells (3). The accumulation of foam cells in the arterial intima leads to chronic inflammation, cell death, and tissue necrosis (3). A similar situation is observed in tuberculosis, a chronic inflammatory disease of the lung caused by Mycobacterium tuberculosis. In the tuberculous lung lesions, which are called granulomas, the presence of tissue necrosis is associated with foam cell accumulation (1). Indeed, foam cells are a hallmark of both the atherosclerotic plaque and the necrotizing tuberculous granuloma (3,4). We were surprised to find that the foam cells of necrotic tuberculous lung lesions are enriched in triglycerides (TAG) (5).
Work with cultured human macrophages and mice further established that M. tuberculosis infection induces TAG accumulation in macrophages (5,6). Thus, the atherosclerosis and tuberculosis models of foam cells are fundamentally different, indicating that foam cells may form via different mechanisms in different diseases.
To test the hypothesis that foam cell biogenesis is disease-specific, we began a study of foam cells resulting from another infectious disease, cryptococcosis, and from a form of cancer. Cryptococcosis is a clinically heterogeneous disease caused by the fungal pathogen Cryptococcus neoformans. It affects the lung and other organ systems, including the central nervous system, particularly in immunocompromised individuals (7). Foam cells have been observed in human tissue biopsies from pulmonary and extrapulmonary cryptococcosis (8,9) and in the lungs of infected mice (10). Foam cells have also been associated with several forms of cancer of many organ systems that include liver, lung, colon/rectum, and kidney (11)(12)(13)(14). Indeed, the presence of foam cells is a histological feature of papillary renal cell carcinoma (pRCC) (15, 16). Factors released by cultured pRCC cells induce lipid accumulation in macrophages (13), . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 8, 2023. ; https://doi.org/10.1101/2023.06.08.542766 doi: bioRxiv preprint indicating that the microenvironment of this tumor is lipogenic for macrophages. The nature of storage lipids and the mechanism of foam cell formation are poorly understood in these pathologies. In the present work, we assessed the spatial distribution of foam cells and storage lipids in C. neoformans-infected murine lungs and in human pRCC specimens. We then analyzed lipid content and the transcriptional program of lipid-laden macrophages generated under in vitro conditions that corresponded to these two diseases. The data established that the mechanism underlying foam cell formation varies with disease context. We can no longer base our understanding of foam cell biogenesis only on work focused on atherogenesis. Expanding our view of foam cell biogenesis may provide new targets for therapeutic intervention into diseases --such as atherosclerosis, tuberculosis, multiple sclerosis, and certain cancers --in which foam cell appearance and poor clinical outcome are associated (reviewed in (1)).

Materials and Methods.
The supplementary materials include the description of the materials and methods utilized to generate, culture, infect and/or treat human monocyte-derived macrophages; to perform cell culture processing and measurements of released cytokines and chemokines; neutral lipid content; RNA extraction and bulk RNA sequencing with the associated statistical analyses; to conduct mouse infections with C. neoformans; to obtain and process cryptococcus-infected murine lungs and human cancerous kidney specimens for histopathology and analysis of spatial distribution of cholesteryl esters and triglycerides by matrix-assisted laser desorption/ionization mass spectrometry. relationship between foam cells and neutral lipids (CE and TAG) in infected lungs. At 7 days post intranasal infection, infected mouse lungs exhibited several granulomatous nodular lesions visible at low magnification (Fig. S1). The lesions consisted of large aggregates of fungal cells surrounded by inflammatory infiltrates comprised mostly of polymorphonuclear cells, macrophages and lymphocyte aggregates, and epithelioid cells (Fig. 1A). Foam cells tended to form clusters in peri-or extra-lesional areas of the infected foci in the lungs [hematoxylin and eosin (H&E)-stained lung slices in Fig. 1BC].
When we used matrix-assisted laser desorption/ionization mass spectrometry (MALDI) imaging of sections adjacent to those used for H&E staining, we detected multiple TAG and CE species in the infected lungs (Table S1). All CE species localized in the fungusrich lesions (e.g., compare H&E staining and MALDI imaging for CE 16:0 in Fig. 1C). In contrast, TAG species were distributed throughout the lung tissue, with some species, such as TAG 46:0, more prominently found within the lesions and others, such as TAG 50:1, found extra-lesionally ( Fig. 1C, with corresponding ion counts in Fig. 1D) (see Fig.   S2 for uninfected control tissue). Localization of some TAG species and CE species in the fungus-rich lesions is consistent with the presence of TAG and sterols in fungal cells (17,18). In addition, the spatial distribution of some TAG species, such as TAG 50:1 in

Cryptococcus neoformans infection induces accumulation of TAG-rich lipid droplets in macrophages via an mTORC1-independent pathway.
MALDI imaging provides information about the spatial distribution of analytes in tissues, but it does not have the single-cell resolution needed to precisely assign a particular neutral lipid to a specific cell type. Thus, we utilized an in vitro infection model to study neutral lipid accumulation in macrophages infected with C. neoformans. We infected primary human monocyte-derived macrophages (MDM) with mCherry-expressing C. neoformans H99, quantified lipid droplet content by imaging flow cytometry, and . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 8, 2023. ; observed a significant lipid droplet accumulation in infected macrophages (3.5-fold increase relative to uninfected cells) (Fig. 2AB). Lipid-droplet-enriched macrophages in the infected culture wells included both those containing fungal cells and those that did not ( Fig. 2A and quantitative data in Fig. S3A). These data indicated that lipid droplet formation does not require internalization of fungal cells. Since lipid droplets accumulated in macrophages exposed to cell-free C. neoformans culture filtrate ( Fig.   S3B) but not in macrophages exposed to heat-killed fungi (Fig. S3C), lipid droplet accumulation in macrophages appears to involve a factor(s) released from live C. neoformans cells.
When we measured storage lipid content in C. neoformans-infected cells by an enzymatic assay, we found that infection increased the content of intracellular TAG but not cholesterol derivatives (Fig. 2C). Moreover, lipid droplet accumulation in C.
neoformans-infected cells was essentially abrogated by treatment with an inhibitor of diglyceride acetyl transferase (A9222500), the enzyme that catalyzes the conversion of di-to tri-glycerides (Fig. 2D). This finding supports the conclusion that C. neoformansinduced lipid droplets are TAG enriched, as previously seen with M. tuberculosis infection ((5) and Fig. 2D).
Our previous work showed that the accumulation of TAG-rich lipid droplets in macrophages infected with M. tuberculosis requires mTORC1 signaling, as it is inhibited by rapamycin treatment (5). Unlike the M. tuberculosis case, however, rapamycin had no effect on lipid droplet accumulation in C. neoformans-infected macrophages (Fig.   2D). Thus, even though M. tuberculosis and C. neoformans both induce accumulation of TAG-rich lipid droplets, the two pathogens do so by utilizing different signaling mechanisms.

M. tuberculosis-and C. neoformans-infected macrophages.
We investigated the pathways underlying TAG accumulation in M. tuberculosis-and C.
neoformans-infected macrophages by conducting transcriptomics analyses of . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 8, 2023. ; https://doi.org/10.1101/2023.06.08.542766 doi: bioRxiv preprint macrophages obtained from the same donors and infected in vitro with either pathogen. When we analyzed the Gene Ontology (GO) annotations related to metabolic processes, we found that the most informative signatures of macrophage metabolic reprogramming associated with TAG accumulation derived from the downregulated pathways in M. tuberculosis-infected macrophages and from the upregulated pathways in C. neoformans-infected macrophages. The former included lipid catabolism, fatty acid oxidation, oxidative phosphorylation, and electron transport chain (Fig. 2E), while the latter were enriched for glycolysis (Fig. 2F).
Metabolism-related analyses at the gene level (Table S2) showed that, in M.
tuberculosis-infected macrophages, the top three downregulated metabolic genes encoded: (i) acyl-CoA synthase (ACSM5), (ii) carnitine octanoyl transferase (CROT), which converts acyl-CoA to acyl-carnitine, a step required for transport across the mitochondrial membrane, and iii) aldehyde hydrogenase (ALDH3A2), which oxidizes fatty aldehydes to fatty acids. Downregulation of these genes likely leads to defective fatty acid oxidation. In C. neoformans infection, the top five upregulated metabolic genes all encoded glycolytic enzymes (Table S3). We also found indicators of reduced mitochondrial functions in C. neoformans-infected macrophages, including downregulation of polyribonucleotide nucleotidyl transferase 1 (PNPT1) and a glutaminyl-tRNA amidotransferase subunit 1 (QRSL1) ( Table S3). The PNTP1 product regulates mitochondrial homeostasis and the abundance of electron transport chain components (19). Missense mutations in the human QRSL1 locus have been associated with defects in oxidative phosphorylation (20). In addition, several aldehyde dehydrogenases were downregulated in C. neoformans-infected macrophages, an indicator of reduced fatty acid oxidation (Table S3).
We identified additional gene expression markers of TAG accumulation in the two infections. For example, in M. tuberculosis infection, we observed downregulation of lipolytic genes, upregulation of sirtuins and sirtuin-stabilizing functions, and expression changes in genes signifying increased production of ceramide and altered cellular redox. These can all lead to TAG accumulation (see Table S2). In C. neoformans infection, additional indicators of metabolic remodeling toward TAG biosynthesis . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 8, 2023. ; https://doi.org/10.1101/2023.06.08.542766 doi: bioRxiv preprint included (i) upregulation of genes for the production of dihydroxyacetone phosphate, which can be routed toward TAG biosynthesis, (ii) upregulation of glycolytic enzymes, such as hexokinase (HK2) and lactate dehydrogenase (LDHA), that indirectly inhibit lipolysis, and (iii) downregulation of AMP-activated protein kinase (AMPK), which inhibits de novo biosynthesis of fatty acids and stimulates fatty acid oxidation (21) (see Table S3).
The transcriptomics data also shed light on the requirement in M. tuberculosis infection for signaling by mechanistic target of rapamycin complex 1 (mTORC1) (Fig. 2D), which is lipogenic in multiple ways (22). M. tuberculosis-infected macrophages showed downregulated TP53 gene and upregulated TP53-specific E3 ligases that target this factor for proteasomal degradation (see Table S2). Decreased activity of TP53 correlates well with increased mTORC1 signaling, since TP53 induces expression of Deptor (Table S2) and leads to activation of AMPK, two factors that inhibit mTORC1 infection, we found no quantitative differences for 43 cytokines and chemokines in the culture media of non-infected vs. infected macrophages ( Fig. 2G and not shown). In contrast, in response to M. tuberculosis infection, macrophages produced many cytokines and chemokines that were mostly proinflammatory (Fig. 2G). Thus, at least in vitro, the cytokine/chemokine milieu of macrophages infected with C. neoformans and M. tuberculosis differed greatly. That difference may contribute to the different mechanisms of lipid accumulation seen in the two infections.  (Fig. S6, Table S1). In particular, the two monounsaturated species (CE 16:1 and CE 18:1), which were the most abundant in the tissues, associated with highly localized, intense signals (CE 16:1 tissue localization is shown in Fig. 3A). In contrast, TAG species yielded only localized signals, which were similar for all detected TAGs (Fig. 3B shows the distribution of TAG 52:2, which is representative of all TAG species; see Fig. S7 for other TAG species). H&E staining revealed that the intense localized CE signals corresponded to large foam cell aggregates (Fig. 3CEG show one such area at increasing magnification), while the TAG signals corresponded to tissue regions containing large numbers of foam cells interspersed among cancer cells ( Fig.   3DFH show a representative area at increasing magnification). Given that CE species were detected throughout the tissue (Fig. S6), the TAG-rich regions also contained CE, albeit at lower levels than the large foam cell aggregates shown in the left panels of Fig.   3. In summary, MALDI imaging showed associations between foam-rich areas with TAG . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  foam-cell-inducing factor (13), we investigated the relationship among pRCC, foam cell formation, and IL-8 in our experimental system. We found that IL-8 is lipogenic for macrophages, but at a concentration ~10-fold higher than that measured in the ACHNconditioned medium (~0.5ng/ml) ( Fig. S8BC and Table S4). Moreover, blocking IL-8 activity with neutralizing antibodies did not prevent lipid droplet formation in macrophages exposed to ACHN medium, and IL-8 depletion from the ACHN medium had no effect on lipid droplet formation (Fig. S8D-F). Thus, the lipogenic effect of the ACHN medium cannot be ascribed to IL-8. We cannot exclude, however, that IL-8 is lipogenic via paracrine-autocrine mechanisms in the tumor microenvironment in vivo since, for example, ACHN-medium-treated macrophages produce lipogenic concentrations of this cytokine (~5ng/ml) ( Fig. S8A and Table S4).

Storage lipid analysis by enzymatic assays showed that lipid droplet accumulation in
ACHN-medium-treated macrophages correlated with increased levels of TAG and free cholesterol (Fig. 4C), suggesting yet another context-specific mechanism of foam cell . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 8, 2023. ; https://doi.org/10.1101/2023.06.08.542766 doi: bioRxiv preprint formation. Indeed, metabolism-related gene expression analysis identified mechanisms for both TAG and cholesterol accumulation. Genes encoding enzymes associated with glycolysis were upregulated ( Fig. 4D and Table S5). These included triosephosphate isomerase 1 (TPI1), which catalyzes the conversion of glyceraldehyde-3P to the TAG precursor dihydroxy-acetone phosphate, and enolase 2 (ENO2), which converts phosphoenolpyruvate to pyruvate (Fig. 4D). We also found gene markers of reduced TCA cycle, including upregulation of adenylate kinase 4 (AK4), a key metabolic regulator increasing glycolysis and inhibiting TCA cycle and oxidative phosphorylation (30) and downregulation of PPARGC1A. The latter gene encodes PGC-1α, a master regulator of energy metabolism that promotes fatty acid oxidation and TCA cycle and decreases TAG storage (31) (Fig. 4D). Together, increased glycolysis and reduced TCA cycle would result in routing of pyruvate towards de novo lipogenesis. Additional gene expression markers of TAG accumulation included the upregulation of perilipin 5 (PLIN5) (Fig. 4D), a member of a protein family protecting TAG from enzymatic degradation (32), and downregulation of the phospholipase PLD6 (Fig. 4D), which inhibits YAP/TAZ mediated lipogenesis (33). Moreover, reduced fatty acid oxidation might result from the downregulation of the acyl-CoA synthase ACSM4 (Fig. 4D) and genes expressing associated functions, such as acyl-CoA conversion to acyl-carnitine (carnitine palmitoyltransferase I, CPT1) and oxidation of fatty aldehydes to fatty acids (aldehyde hydrogenases, ALDH genes) ( Fig. 4D and Table S5). The downregulation of N6AMT1 (Fig. 4D), which encodes a methyltransferase mostly implicated in DNA and protein modification, is also consistent with cellular lipid accumulation, since it has been associated with lipid catabolism (34) by yet unknown mechanisms.
The ACHN-medium-treated macrophages also exhibited gene expression changes associated with dysregulation of cholesterol metabolism. For example, the scavenger receptor CD36, which is a key regulator of cholesterol homeostasis, was downregulated (Table S5), presumably as a consequence of PPARGC1A downregulation (Fig. 4D) (31). Additional heterogeneity may result from the combinatorial effects of multiple foam-cellinducing signals in some microenvironments. For example, foam cells may be induced in pRCC by yet unidentified cancer cell products, as indicated by our results with ACHNconditioned medium, together with lipogenic proinflammatory cytokines, such as IL-8, produced by tumor-associated-macrophages. Moreover, additional mediators might be produced by cell types that are not represented in our in vitro system, such as the lipidfilled tumor cells that we observed in the pRCC bioptic tissue. In tuberculosis, we . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 8, 2023. ; https://doi.org/10.1101/2023.06.08.542766 doi: bioRxiv preprint showed that, in addition to bacterial components that presumably trigger TLR2 signaling (38, 39), macrophage lipid accumulation requires another lipogenic proinflammatory cytokine, TNFα, produced by infected macrophages (5). Additional work is needed to identify the exogenous (i.e., generated by microbes, cancer cells, or other cell types) trigger signals and to determine how commonly foam cells are induced by combinations of exogenous and autocrine/paracrine signals.
It is reasonable to assume that, despite the different mechanisms of foam cell biogenesis, their presence represents a maladaptive immune response in the pathological contexts they form. Generally, lipid-laden macrophages tend to lose protective immune functions, including phagocytosis, efferocytosis, and autophagy.
They can also induce tissue damage, contribute to necrosis, exhibit impaired antimicrobial activity, and even sustain survival of intracellular pathogens (reviewed in   was used to measure cytokine/chemokine/growth factor concentration in the supernatants of uninfected (control) and infected cells. Each sample was measured in duplicate. Average Z-score of 5 donors was calculated for each analyte (gradient from green to red represents increasing concentrations). White asterisks indicate significant differences between uninfected and infected cells. *, p < 0.05; **, p < 0.01; ***, p < 0.001 (paired t-test).
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted June 8, 2023. ;