A comprehensive machine-readable view of the mammalian cholesterol biosynthesis pathway

Cholesterol biosynthesis serves as a central metabolic hub for numerous biological processes in health and disease. A detailed, integrative single-view description of how the cholesterol pathway is structured and how it interacts with other pathway systems is lacking in the existing literature. Here we provide a systematic review of the existing literature and present a detailed pathway diagram that describes the cholesterol biosynthesis pathway (the mevalonate, the Kandutch-Russell and the Bloch pathway) and shunt pathway that leads to 24(S),25-epoxycholesterol synthesis. The diagram has been produced using the Systems Biology Graphical Notation (SBGN) and is available in the SBGN-ML format, a human readable and machine semantically parsable open community ﬁle format


Introduction
Cholesterol is an intensively studied, multi-functional lipid that is key to many aspects of immunological, neuronal, viral and hepatocyte biology. It is an essential component of cellular membranes and is a precursor to steroids, bile acids and oxysterols whilst its own precursors contribute to prenylation and dolichylation and the formation of vitamin D 3 . One of the oxysterols known to be involved in linking sterol metabolism to innate immunity [1,2] is 25-hydroxycholesterol. However its place in the sterol metabolism has not yet been well established.
Despite the importance of the cholesterol synthesis pathway to cellular function and its value in pharmaceutical therapies, an integrative picture of how the pathway is structured has not been well described in the literature, impeding the development of a more rigorous understanding of the role of the cholesterol metabolism in cellular processes. Publications typically focus on segments of the cholesterol biosynthesis pathway showing variable level of details. Kovacs and co-authors focus on the mevalonate section of the pathway and on the subcellular location of the enzymes involved [3]. Wang and co-authors concentrate on the steps leading to 24(S),25-epoxycholesterol synthesis and their similarity to steps in the cholesterol biosynthesis pathway [4]. Previous work studying the role of the cholesterol biosynthesis pathway has shown a modest level of detail on the sterol arms of the pathway [5][6][7] in innate immunity. The LIPID MAPS consortium offers the most detailed descriptions of the Bloch and Kandutsch-Russell branches of cholesterol biosynthesis, but these lack cell compartment information and lack integration with the 24(S),25-epoxycholesterol shunt arm and other branching pathways [8].
Here we present a comprehensive literature review of the cholesterol synthesis pathway and we implement this as a detailed pathway that captures enzymatic activity and compartmental localization and summarizes all intermediate metabolic forms. Our review also clearly indicates what information is missing and where additional research is required.

Materials and methods
The model of the cholesterol biosynthesis pathway presented in this work has been assembled using a variety of publicly available resources including the research findings of the LipidMaps Biochemical Pharmacology 86 (2013) [56][57][58][59][60][61][62][63][64][65][66] consortium [8] and results obtained from thorough searches of the published literature that have been manually curated and validated by domain experts.
In cases where there were conflicting reports, preference was given to the more recent papers and to the works in which more reliable and advanced methods were used. The suggested order of events is supported by a number of independently obtained research results. The principles of the Evidence Ontology (ECO) [9], the Gene Ontology Evidence Codes [10] and the Evidence Code Decision Tree [11] were considered during the pathway reconstruction.
A brief summary is provided for each enzyme and the corresponding metabolic reactions involved in the pathway. For each enzyme we endeavored to capture the following information where available: corresponding gene name approved by HUGO Gene Nomenclature Committee [12], alternative names, enzymological activities according to the Enzyme Nomenclature Committee of the IUBMB [13], enzyme function description, subunit structure, subcellular location and related disorders.
We have included a list of UniProt IDs for the proteins captured in the model (Table 1) and a list of metabolite names (common and systematic) as used in the LipidMaps database [8] (Table 2). Common names are used on the map where available.
The pathway that we present here is described using the Systems Biology Graphical Notation (SBGN) [14], a community driven consensus graphical schema for capturing the molecular details of pathway systems. In particular, we use the SBGN Process Description language [15]. A machine-readable model is available as part of the supplementary material in SBGN-ML format [16] and we present it graphically in Figs. 1-3, in an enhanced form. The SBGN-ML format files can be read using a variety of software packages.
In particular, the supplementary files provide a description of the pathway that can be edited and modified in accordance with the SBGN standard in order to be of future use to the research community. The SBGN-ML file format encodes the biological meaning associated with each component of the model. This allows the model to be parsed by software (i) to ensure that modification is biologically valid and (ii) to facilitate automatic generation of mathematical descriptions of the pathway biology. It should be possible to open these files in any software designed to comply with the SBGN-ML standard, including but not limited to VANTED and Cytoscape [17,18]. For the purpose of this review, we compiled and tested the files using the VANTED software tool [17]. Here we shall outline how the files can be opened and accessed using the VANTED and CYTOSCAPE [18]  In the search bar, type sbgn and hit return. Folders will appear in the window and under 'Available for install' will appear a Utility folder.

2.2.4
Open the utility folder and select the latest version of CySBGN before hitting the install button. The CySBGN plugin will then be downloaded and installed. Once it is installed, close the 'Manage Plugins' window. 2.2.5 From the File menu select import followed by 'Network (Multiple File Types)'. In the window that opens, make sure that the 'Local' option is chosen and high the 'Select' button to bring up a file selector. Choose the downloaded file with the 'sbgn' file extension in the usual way. Fig. 1 shows the mevalonate arm of the cholesterol biosynthesis pathway and includes enzymatic activity in the mitochondria, peroxisome, cytoplasm and endoplasmic reticulum. The arm starts with the consumption of acetyl-CoA, which occurs in parallel in three cell compartments (the mitochondria, cytoplasm and peroxisome) and terminates with the production of squalene in   Fig. 2 shows the sterol arms of the cholesterol biosynthesis pathway and this includes the Bloch pathway, the Kandutsch-Russell pathway and the shunt pathway. This arm starts with Squalene and terminates with cholesterol production on the Bloch and Kandutsch-Russell pathways and with 24(S),25-epoxycholesterol on the shunt pathway. Fig. 3 provides a legend for the SBGN schema, explaining the various nodes and edges.

Mevalonate arm of the cholesterol biosynthesis pathway
3.2.1. Acetyl-CoA acetyltransferase (ACAT1; ACAT2; Acetoacetyl-CoA thiolase; EC 2.3.1.9) is an enzyme that catalyzes the reversible condensation of two molecules of acetyl-CoA and forms acetoacetyl-CoA. This reaction is an important step in ketone body formation. Both mitochondrial ACAT1 and cytosolic ACAT2 enzymes are homotetramers [19,20]. Kovacs et al. suggest a possibility of distribution of ACAT1 between peroxisomes and mitochondria as experimental evidence supports the formation of acetoacetyl-CoA in peroxisomes [3]. The proposed step in peroxisomes is shown in Fig. 1 by a reaction glyph with a question mark. Mutations of the ACAT1 gene cause alpha-methylacetoacetic aciduria, an autosomal recessive disorder [21].
3.2.2. Hydroxymethylglutaryl-CoA synthase (HMGCS1; HMGCS2; EC 2.3.3.10) forms HMG-CoA from acetyl-CoA and acetoacetyl-CoA. The two proteins with this enzymological activity are HMGCS1 and HMGCS2 (Table 1). HMGCS1 is a cytoplasmic enzyme and HMGCS2 is localized to mitochondria and peroxisome [3]. Ortiz and co-authors provide evidence for the involvement of HMGCS2 in producing cholesterol-convertible HMG-CoA [22]. Peroxisomal localization of this enzyme was subsequently confirmed and the significance of the peroxisomal pathway in cholesterol production was demonstrated [39,40]. The schema proposed by Kovacs and co-authors implies that the mitochondrial component of HMG-CoA is being converted into acetyl-CoA and acetoacetate by HMGCL (see 3.2.3) and is not likely to be involved in further steps contributing to cholesterol formation [3]. The possibility of HMG-CoA transport from the mitochondria to the endoplasmic reticulum or peroxisome requires further study.
3.2.3. Hydroxymethylglutaryl-CoA lyase, mitochondrial (HMGCL; EC 4.1.3.4) is a key enzyme in the ketone body formation pathway that provides fuel to extrahepatic tissues [23]. It transforms HMG-CoA into acetyl-CoA and acetoacetate. HMGCL is a mitochondrial enzyme and Kovacs et al. suggest peroxisomal localization in addition to mitochondrial [3]. Since the peroxisomal localization is not confirmed yet, we show this step with a question mark on the diagram (Fig. 1). The enzyme deficiency (HMGCLD) or hydroxymethylglutaric aciduria may be due to a variety of mutations and can be fatal [24].
3.2.4. 3-Hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR; EC 1.1.1.34) catalyzes the conversion of 3-hydroxy-3methylglutaryl-CoA into mevalonic acid. The enzyme is highly regulated by relevant signaling pathways that include the SREBP pathway [25]. Kovacs et al. confirm endoplasmic reticulum localization of HMGCR and provide evidence that suggests peroxisomal localization [3]. In our model both locations for this enzyme are included. This enzyme is conventionally regarded as being rate limiting in the pathway and its interactions are targeted by the statin class of drug.
3.2.6. Phosphomevalonate kinase (PMVK; EC 2.7.4.2) catalyzes formation of mevalonate 5-diphosphate from mevalonate 5phosphate, an essential step in the mevalonate pathway. It is a reversible reaction and kinetic constants have been determined for human enzymes, both for forward and reverse reactions [36,37]. Expression of this enzyme is regulated in response to dietary sterol levels and this regulation is coordinated with HMGCR [38]. Peroxisomal localization of the enzyme has been confirmed [3,30,[38][39][40].
3.2.7. Diphosphomevalonate decarboxylase (MVD; mevalonate (diphospho) decarboxylase; EC 4.1.1.33 is an enzyme that decarboxylates mevalonate 5-diphosphate forming isopentenyl diphosphate while hydrolyzing ATP. This enzyme is considered to be a useful target for lowering serum cholesterol levels [41] and is active as a homodimer [41]. Information on peroxisomal localization of diphosphomevalonate decarboxylase is provided in the section on mevalonate kinase (3.2.5).
3.4. Oxysterol 24(S),25-epoxycholesterol synthesis from squalene 24(S),25-Epoxycholesterol is produced in a shunt pathway that is parallel to the two branches of the cholesterol synthesis pathway [4,[106][107][108]. The same set of enzymes is involved in the formation of cholesterol from 24,25-dihydrolanosterol in the Kandutsch-Russell pathway, desmosterol from lanosterol in the Bloch pathway and 24(S),25-epoxycholesterol from 24(S),25-epoxylanosterol in a shunt pathway [4]. Due to its importance in regulatory processes 24(S),25-epoxycholesterol is in the focus of several recent publications [107,[109][110][111][112]. However, further research is necessary to confirm each step of the shunt pathway and the corresponding intermediate metabolites. In our representation, the known intermediates of the shunt pathway are shown and likely missing information is noted with the appropriate SBGN glyph.

Transport of the intermediate metabolites between different cellular compartments
Little is known about transporting of the intermediate metabolites of the cholesterol biosynthesis pathway between different cellular compartments. It is not known whether metabolites in the mitochondria participate in cholesterol biosynthesis and, in a number of cases, it is unclear whether metabolites move between compartments through diffusion or transportation. This is something that should be addressed in the future. It has been suggested that limitations in subcellular fractionation methods are a significant factor in our poor understanding of the subcellular localization of these enzymes [3].

Concluding remarks
The diagrams presented here show a comprehensive view of the mevalonate, Kandutsch-Russell, Bloch and shunt pathways. The diagrams are described using the SBGN schema, an open and community developed graphical language for unambiguously capturing pathway structure. Missing/uncertain information is clearly marked on the pathway diagrams and shows the areas that need to be further explored. Models of these diagrams are available as supplementary material in the SBGN-ML and SBML/ CellDesigner [113] file formats for future development and refinement.
We hope that by elucidating and integrating the detailed structure of this pathway, we will contribute to a finer level of understanding of cholesterol metabolism and its function and that this will serve as a useful resource for future studies of the cholesterol biosynthesis pathway.