The role of metabolic reprogramming and de novo amino acid synthesis in collagen protein production by myofibroblasts: Implications for organ fibrosis and cancer

Abstract

Fibrosis is a pathologic condition resulting from aberrant wound healing responses that lead to excessive accumulation of extracellular matrix components, distortion of organ architecture, and loss of organ function. Fibrotic disease can affect every organ system; moreover, fibrosis is an important microenvironmental component of many cancers, including pancreatic, cervical, and hepatocellular cancers. Fibrosis is also an independent risk factor for cancer. Taken together, organ fibrosis contributes to up to 45% of all deaths worldwide. There are no approved therapies that halt or reverse fibrotic disease, highlighting the great need for novel therapeutic targets. At the heart of almost all fibrotic disease is the TGF-β-mediated differentiation of fibroblasts into myofibroblasts, the primary cell type responsible for the production of collagen and other matrix proteins and distortion of tissue architecture. Recent advances, particularly in the field of lung fibrosis have highlighted the role that metabolic reprogramming plays in the pathogenic phenotype of myofibroblasts, particularly the induction of de novo amino acid synthesis pathways that are required to support collagen matrix production by these cells. In this review, we will discuss the metabolic changes associated with myofibroblast differentiation, focusing on the de novo production of glycine and proline, two amino acids which compose over half of the primary structure of collagen protein. We will also discuss the important role that synthesis of these amino acids plays in regulating cellular redox balance and epigenetic state.

Fibrosis

Fibrosis is defined by the excessive deposition of extracellular matrix that replaces normal healthy tissue with scar tissue. While matrix deposition is required for normal wound healing responses, severe or repetitive injury may lead to the continued accumulation of matrix proteins, which ultimately impair tissue function. Fibrosis can occur in any organ system and is associated with a broad range of chronic diseases, including heart disease, interstitial lung disease, scleroderma, hepatitis, inflammatory bowel disease, and diabetes. Organ fibrosis is estimated to be associated with as many as 45% of all deaths (Rockey et al. 2015; Henderson et al. 2020; Distler et al. 2019; Jun and Lau 2018). Although genetic susceptibility loci have been identified for some fibrotic diseases and exposures such as alcohol, cigarette smoke, asbestos, silica, and viral infection are associated with increased risk, the etiology of most fibrotic diseases remains unknown. Furthermore, there are no effective therapies for halting or reversing fibrotic progression in any tissue. Thus, there is tremendous need to identify novel targets for the treatment of organ fibrosis.

Fibrosis and Cancer

Similar to fibrosis, cancer has also been described as “a wound that does not heal” (Dvorak 1986). Solid tumors are closely associated non-transformed vasculature, immune cells, and fibroblasts that constitute the stromal compartment. The aberrant matrix environment of the tumor stroma enhances cellular proliferation, promotes angiogenesis, inhibits anti-tumor immune responses, and promotes metastasis and resistance to therapeutics (Piersma et al. 2020; Cox and Erler 2014; Valkenburg et al. 2018). The importance of a wound-like environment has long been appreciated to promote cancer growth. For instance, Rous sarcoma virus injection into chickens leads to tumor formation only at the injection site (a wound) despite viremia. Wounding at distal sites resulted in tumor formation at these sites (Dolberg et al. 1985). Similarly, it has been shown that fibrotic lung disease is an independent risk factor for lung cancer and that stromal gene signatures indicate prognosis of pancreatic and colorectal cancer (Karampitsakos et al. 2017; Ballester et al. 2019; Moffitt et al. 2015; Calon et al. 2015). The increased understanding of the role of the stroma in promoting cancer progression has led to therapeutics that target the stromal compartment (Valkenburg et al. 2018).

Myofibroblasts

While fibrosis is a complex disease involving multiple cell types, myofibroblasts represent the cellular site of pathologic matrix production that leads to distortion of tissue architecture and loss of function (Darby et al. 2016). Myofibroblasts are also present in the tumor stroma and are responsible for the aberrant matrix environment of tumors (Piersma et al. 2020; Liu et al. 2019). Myofibroblasts are contractile cells with features of both fibroblasts and smooth muscle cells that promote wound closure during normal wound healing. These cells are eliminated by apoptosis during normal wound closure; however, they persist during hypertrophic scar development and in fibrotic disease (Hinz and Lagares 2020). Myofibroblasts are not present in healthy tissues and their cell(s) of origin remain a topic of debate; however, tissue resident fibroblasts have been shown to give rise to myofibroblasts in all injured organs (Pakshir et al. 2020). This transformation is largely governed by signaling downstream of Transforming Growth Factor-β (TGF-β), which is highly expressed in fibrotic tissues and is sufficient to induce myofibroblastic differentiation and collagen matrix production in vitro, and to induce fibrosis in vivo (Sporn et al. 1983; Ignotz and Massague 1986; Roberts et al. 1986; Sime et al. 1997).

Recent single cell RNA sequencing and lineage tracing experiments have highlighted heterogeneity and plasticity in fibroblast populations in multiple fibrotic tissues as well as in cancer stroma (Shaw and Rognoni 2020; Biffi and Tuveson 2021). This complexity may complicate the design of myofibroblast-specific therapies; thus, targeting of specific pathologic functions such as matrix production may hold promise for developing new therapies for fibrotic disease. Aberrant production of matrix proteins by myofibroblasts results in an altered mechanical environment that is sufficient to promote further myofibroblast activation in a feed forward cycle (Tschumperlin et al. 2018). Thus, the myofibroblast and its production of matrix proteins is an important target of fibrosis therapy.

Structure of Collagen Matrix Proteins

Collagen is the most abundant component of extracellular matrix and comprises one third of all protein in humans (Shoulders and Raines 2009; Ricard-Blum 2011). The structural stability of collagen is derived from its quaternary structure consisting of three parallel polypeptide strands forming triple helices known as tropocollagens. This helical structure of collagen depends on triple-helical domains consisting of a repeating G-X-Y primary structure where G represents glycine at every third amino acid residue and X and Y can represent any amino acid but are most commonly proline at the X position and hydroxyproline at the Y position. Together, glycine, proline, and hydroxyproline (which is produced through posttranslational hydroxylation of collagen proline residues), constitute over half of all amino acids in the primary structure of collagen protein (Bradley et al. 1974) (Figure 1). The repeating glycine residues allow for interstrand hydrogen bonds while post-translational proline hydroxylation at the Y position provides increased thermostability through stereoelectronic effects. All collagens contain at least one triple-helical domain. Fibril-forming collagens such as collagen I, III, and V which are the primary collagens produced by myofibroblasts, contain one major triple-helical domain. Other collagens are characterized by the presence of several smaller triple-helical domains interrupted by non-helical domains which are associated with molecular recognition and give these collagens plasticity and flexibility. Thus, collagens are able to impart both structural and signaling cues that regulate both normal and pathogenic cellular behavior (Karsdal et al. 2017; Xu et al. 2019).

Figure 1. Percentages of amino acids in human collagen 1.

Amino acid composition of (a) collagen proteins and (2) non-collagen proteins in human lung (Bradley et al. 1974). While glycine and proline/hydroxyproline make up 33.37% and 20.51% of total amino acid residues in collagens, their percentage is much lower in other proteins.

Metabolic Reprogramming during Myofibroblast Differentiation

Glycolysis and de novo glycine synthesis

Myofibroblastic differentiation is associated with transcriptomic and morphologic changes as cells develop an extensive stress fiber network, expand their rough endoplasmic reticulum, and produce extracellular matrix proteins. These cells also alter the way in which they use metabolites in order to support this contraction and matrix production. Upon treatment with TGF-β, normal human lung fibroblasts exhibit increased rates of glycolysis and mitochondrial oxygen consumption (Nigdelioglu et al. 2016; Xie et al. 2015; Bernard et al. 2015; Hamanaka et al. 2019). This increase in metabolic activity supports not only the energetic needs of myofibroblasts but also plays a crucial role in producing the molecular building blocks required by myofibroblasts for the production of matrix proteins.

Inhibition of glycolysis with 2-deoxyglucose prevents TGF-β-induced collagen protein production by lung fibroblasts independent of any effect on collagen mRNA expression, which remains induced by TGF-β (Nigdelioglu et al. 2016). While glycolytic metabolism likely serves many roles in myofibroblasts, including production of ATP, one crucial role for glucose in these cells is as a substrate for de novo glycine synthesis. TGF-β induces the expression of all of the enzymes required for the conversion of the glycolytic intermediate 3-phosphoglycerate to the amino acid glycine using an amino group donated from glutamate. These enzymes include phosphoglycerate dehydrogenase (PHGDH), phosphoserine aminotransferase 1 (PSAT1), phosphoserine phosphatase (PSPH), and serine hydroxymethyltransferase 2 (SHMT2) (Nigdelioglu et al. 2016) (Figure 2). TGF-β-induced collagen synthesis occurs independent of extracellular glycine and knockdown of these enzymes is sufficient to inhibit collagen production even in the presence of extracellular glycine, suggesting that de novo-synthesized glycine is preferentially used by myofibroblasts for collagen protein production (Nigdelioglu et al. 2016; Hamanaka et al. 2019). Supporting this notion, metabolic tracing experiments using ¹³C-glucose in the presence of extracellular ¹²C-glycine, demonstrated that TGF-β stimulates the production of glycine from glucose with glucose-derived glycine comprising up to 80% of cellular glycine pools (Schworer et al. 2020). Furthermore, carbon derived from glucose is incorporated into collagen protein (Nigdelioglu et al. 2016; Schworer et al. 2020; Selvarajah et al. 2019; Tsurufuji and Mori 1965).

Figure 2. Contributions of glycolytic intermediates and glutamine to collagen protein synthesis.

Both glucose and glutamine contribute to the synthesis of two most abundant amino acids (glycine and proline) in collagen protein. Glycine can be synthesized from the glycolytic intermediate, 3-phosphoglycerate (3P glycerate) by the enzymes of serine glycine synthesis pathway including phosphoglycerate dehydrogenase (PHGDH), phosphoserine aminotransferase 1 (PSAT1), phosphoserine phosphatase (PSPH), and serine hydroxymethyltransferase 2 (SHMT2). Glutamine is converted to glutamate by glutaminase (GLS) in glutaminolysis. Glutamate is not only a precursor for proline but also contributes to the generation of glycine by donating nitrogen through the PSAT1 transamination reaction.

Glutamine Metabolism

Glutamine is the most abundant amino acid found in human plasma and a key substrate used by cells to replenish the TCA cycle to support mitochondrial metabolism (Altman et al. 2016). Glutamine is metabolized in a pathway termed glutaminolysis in which glutamine is first converted to glutamate by the enzyme glutaminase (GLS). Glutamate is then metabolized to the TCA cycle intermediate α-ketoglutarate by either glutamate dehydrogenase (GLUD) or by aminotransferase enzymes, including PSAT1 (Figure 2). The glutaminolytic pathway has been studied extensively in the context of cancer and GLS inhibitors are in clinical trials for colorectal and non-small cell lung cancers.

Multiple reports have demonstrated that GLS expression is induced in lung fibroblasts after TGF-β treatment (Bernard et al. 2018; Ge et al. 2018; Hamanaka et al. 2019; Choudhury et al. 2020). Inhibition of glutaminolysis by either culturing cells in the absence of extracellular glutamine or by inhibition of GLS prevents TGF-β-induced myofibroblast differentiation (Bernard et al. 2018; Ge et al. 2018; Hamanaka et al. 2019; Choudhury et al. 2020). Interestingly, while glutaminolysis is an important metabolic pathway that feeds the TCA cycle, it appears that this anaplerotic role of glutamine is dispensable for myofibroblast differentiation. TGF-β increases the entry of carbons from extracellular glutamine into the TCA cycle (Schworer et al. 2020); however, knockdown of α-ketoglutarate dehydrogenase, the TCA cycle enzyme that converts α-ketoglutarate to succinyl-CoA, had no effect on TGF-β-induced expression of collagen or α-SMA protein (Hamanaka et al. 2019). Furthermore, while TGF-β increases mitochondrial oxygen consumption in lung fibroblasts, this increase was shown to be independent of the presence of glutamine in the media or on the activity of GLS (Hamanaka et al. 2019). Thus, mitochondrial glutamine catabolism, although induced by TGF-β, is not required for myofibroblast differentiation.

Glutamine-Dependent Glycine and Proline Biosynthesis

In addition to its role as a mitochondrial substrate, glutamine-derived glutamate can be used in anabolic pathways, supporting glutathione synthesis, cellular import of cystine, and as a precursor for amino acid biosynthesis, including the de novo synthesis of glycine and proline. As mentioned above, glutamate is required for the activity of PSAT1 which transfers the amino group of glutamate to 3-phosphohydroxypyruvate (the product of PHGDH), producing 3-phosphoserine, a precursor for glycine (Figure 2). Metabolic labeling experiments using glutamine labeled on its alpha nitrogen demonstrated that TGF-β increases the incorporation of glutamine-derived nitrogen into glycine (Schworer et al. 2020). Twenty-five percent of cellular glycine was labeled after culture in ¹⁵N-glutamine while no label was found in glycine after culture in ¹³C-glutamine. Thus, glycine synthesis in lung fibroblasts utilizes carbon derived from glucose and nitrogen derived from glutamate. The role of glutamate sources other than extracellular glutamine has not been investigated; however, knockdown of PSAT1 in the presence of extracellular glycine, inhibited TGF-β-induced collagen protein production in lung fibroblasts, consistent with the importance of this pathway for collagen synthesis (Hamanaka et al. 2019).

Glutamine-derived glutamate is also an important precursor for de novo synthesis of proline. TGF-β induces the expression of all enzymes required to convert glutamate to proline, including pyrroline-5-carboxylate synthase (P5CS), which converts glutamate to pyrroline-5-carboxylate (P5C), as well as P5C reductases (PYCR1, PYCR2, PYCRL/PYCR3), which convert P5C to proline (Hamanaka et al. 2019; Schworer et al. 2020) (Figure 3). Total cellular levels of P5C and proline are increased in cells after TGF-β and carbon labeling shows that TGF-β increases the percentage of cellular proline derived from extracellular glutamine (Hamanaka et al. 2019; Schworer et al. 2020). Glutamine-derived proline is incorporated into collagen protein, and knockdown of P5CS prevents TGF-β-induced collagen protein production (Hamanaka et al. 2019; Schworer et al. 2020). Interestingly, flux analysis showed that TGF-β increases the incorporation of glutamine-derived carbon into proline by 7-fold and into α-ketoglutarate by only 2-fold, indicating that proline is a major cellular destination for glutamine metabolism in myofibroblasts (Schworer et al. 2020).

Figure 3. Metabolic pathways for the biosynthesis of proline.

Proline is synthesized from pyrroline-5-carboxylate (P5C) by P5C reductases (PYCR1, PYCR2, PYCRL/PYCR3). 5PC is generated from either glutamine or arginine. Glutamine is converted to glutamate and then 5PC by glutaminase (GLS) and pyrroline-5-carboxylate synthase (P5CS), respectively. Arginine is converted to ornithine and then 5PC by arginase 1 and 2 (ARG1/2) and ornithine aminotransferase (OAT), respectively.

While metabolic labeling experiments in fibroblasts suggest TGF-β treatment leads to 70% of cellular proline being derived from extracellular glutamine, these experiments were done in the absence of extracellular proline (Schworer et al. 2020). It is unknown how TGF-β affects the use of extracellular proline; however, the finding that endogenous proline synthesis supports collagen protein production is consistent with older reports using radioactive tracers to study the effect of amino acids on cellular proline pools and incorporation of extracellular proline into collagen. When rat granulomas were cultured in the presence of ³H-proline and ¹⁴C-glutamate, incorporation of both ³H and ¹⁴C was noted in collagen hydroxyproline (Aalto et al. 1973). ¹⁴C-proline is taken up by cultured chicken embryo tibiae and incorporated into proteins. Addition of glutamine to the medium reduced the radioactivity of proline and hydroxyproline hydrolyzed from these tibiae (Blumenkrantz and Asboe-Hansen 1973). Similarly, when human fibroblasts were cultured in ¹⁴C-proline, increasing the medium glutamine concentration from 2 mM to 8 mM reduced the radioactivity of intracellular free proline as well as that of collagens without affecting the total concentration of collagen (Bellon et al. 1985). It has also been noted that the specific activity of intracellular free proline never equilibrates with extracellular labeled proline and that the specific activity of prolyl-tRNAs and collagen hydroxyproline is lower than that of intracellular free proline (Hildebran et al. 1981; Bellon et al. 1987). This lack of correlation with extracellular proline was abolished in CHO-K1 cells which are proline auxotrophs (Hildebran et al. 1981). Glutamine was shown to compete with extracellular proline for incorporation into prolyl-tRNA and collagen independent of free intracellular proline concentration (Bellon et al. 1987). Similarly, tissue slices from cirrhotic rat livers contained 4 times more ¹⁴C-hydroxyproline in collagen protein when cultured in ¹⁴C-glutamate than when cultured in ¹⁴C-proline (Rojkind and Diaz de Leon 1970). Thus, glutamine and glutamate are important precursors for the proline used for collagen production, even in the presence of extracellular proline.

Ornithine-Dependent Proline Biosynthesis

The non-proteinogenic amino acid ornithine is also an important precursor for proline biosynthesis. Ornithine is a product of the breakdown of arginine by the urea cycle enzyme arginase (Figure 3). To continue the urea cycle, ornithine transcarbamylase (OTC) catalyzes the reaction of carbamoyl phosphate with ornithine, producing citrulline. Alternatively, ornithine can be decarboxylated by ornithine decarboxylase (ODC) producing putrescine, the committed step in polyamine synthesis. Lastly, ornithine aminotransferase (OAT) can transfer the side chain amino group of ornithine to α-ketoglutarate, producing glutamate and P5C (Morris 2002) (Figure 3).

While relatively less is known about the contribution of ornithine-derived proline to collagen synthesis than that from glutamate-derived proline, this pathway may also contribute significantly to de novo synthesis of proline and collagen. When rat xiphoid cartilage was cultured in ³H-proline and ¹⁴C-ornithine, 20% of protein-incorporated proline was found to be derived from ornithine (Smith and Phang 1978). Furthermore, the rate of ¹⁴C-labeled proline incorporation was not inhibited by elevation of extracellular proline concentration. Labeling of lung fibroblasts with ¹⁴C-arginine or ¹⁴C-glutamate showed that although glutamate-derived proline accounted for 2–5 times as much cellular proline as that from arginine, ¹⁴C-arginine led to significant labeling of cellular free proline, protein-incorporated proline, and hydroxyproline (Shen and Strecker 1975). Moreover, treatment of either NIH-3T3 cells or rat vascular smooth muscle cells with TGF-β resulted in the upregulation of arginase expression, the inhibition of which reduced collagen protein synthesis at the posttranscriptional level (Durante et al. 2001; Kitowska et al. 2008).

In Vivo Evidence of a role for De Novo Amino Acid Biosynthesis for Collagen Production

The work outlined above using cultured cells and tissues has suggested that de novo synthesis of glycine and proline plays a major role in the production of the amino acids incorporated into collagen, independent of extracellular levels of glycine and proline, and suggesting a compartmentalization of these endogenously-synthesized amino acids for protein production. Further work has suggested that collagen-producing cells depend on these same pathways in vivo.

Enzymes of the de novo glycine synthesis pathway, including PHGDH, PSAT1, and SHMT2, have been shown to be highly expressed in fibrotic lung tissue from humans and mice (Nigdelioglu et al. 2016; Hamanaka et al. 2018; Hamanaka et al. 2019). Lung tissue from IPF patients was shown to have significantly elevated levels of glycolytic intermediates and glycine than control donor tissues (Kang et al. 2016). Further, inhibition of de novo synthesis of glycine from glucose with the PHGDH inhibitor NCT503 reduced bleomycin-induced lung fibrosis in mice (Hamanaka et al. 2018).

Tissue levels of proline have also been found to be elevated in IPF lung tissue (Kang et al. 2016). Similar findings were found in liver tissue from cirrhosis patients (Kershenobich et al. 1970). This was independent of any effect of cirrhosis on plasma proline levels, suggesting local production of proline increases the liver free proline levels. Infusion of radiolabeled leucine, tyrosine, or proline into rabbits led to accumulation of these free amino acids in the skin. While leucine and tyrosine were readily incorporated into collagen proteins, tissue free proline proved to be a poor precursor for collagen. Calculations suggested that less than 5% of the proline used for collagen production in the skin was at equilibrium with the free proline in the blood (Robins 1979). Thus, de novo proline synthesis contributes to collagen protein production in vivo as it does in culture.

GLS, P5CS, and OAT expression have all been shown to be elevated in the lung tissue of IPF patients (Ge et al. 2018; Hamanaka et al. 2019; Schworer et al. 2020; Kang et al. 2016; Nojima et al. 2020). The expression of cytoplasmic arginase 1 and mitochondrial arginase 2 is induced in the lungs of mice after bleomycin instillation; however, the evidence of arginase expression in IPF patients is less robust (Endo et al. 2003; Kitowska et al. 2008; Nojima et al. 2020). Expression levels of ALDH18A1 (the gene encoding P5CS) mRNA in IPF lung tissue correlates negatively with pulmonary function measurements including forced vital capacity (FVC) and diffusing capacity for carbon monoxide (DLCO) (Schworer et al. 2020). OAT expression also correlates negatively with FVC (Kang et al. 2016). While inhibition of GLS either pharmacologically, or by conditionally deleting its expression in fibroblasts, reduces bleomycin-induced lung fibrosis in mice (Cui et al. 2019; Choudhury et al. 2020), it remains to be determined whether this is due to the effect of GLS inhibition on proline biosynthesis.

Why do Myofibroblasts Rely on De Novo Amino Acid Biosynthesis?

Redox Balance

The question remains as to why collagen-producing cells such as myofibroblasts engage in de novo amino acid biosynthesis for production of collagen protein even in the presence of extracellular glycine and proline. While no one answer is likely to completely explain this phenomenon, metabolite flux through these metabolic pathways may provide benefits to the cells other than production of the amino acids themselves. One major benefit that glycine and proline synthesis is maintenance of cellular redox balance. Myofibroblasts are characterized by elevated levels of cellular reactive oxygen species (ROS) that are produced both by mitochondria and NADPH oxidases (Jain et al. 2013; Hecker et al. 2009). These ROS act as signaling molecules which are required for the transcriptional events downstream of TGF-β; however, cellular ROS production must be carefully balanced by cellular antioxidants and other compensatory mechanisms so that excessive ROS accumulation does not occur (Hamanaka and Chandel 2010). The de novo synthesis of both glycine and proline is intimately linked with cellular redox status and this link may be beneficial for cells, particularly under stress conditions.

De novo glycine synthesis promotes cellular antioxidant function through reduction of NADP⁺ to NADPH in the downstream one-carbon pathway. NADPH provides reducing equivalents for cellular glutathione and thioredoxin antioxidant systems. Cells lacking SHMT2 are characterized by increased levels of oxidized glutathione, oxidative stress, and increased sensitivity to oxidative stressors (Ye et al. 2014). De novo production of proline is also associated with protection from oxidative stress, a process extensively studied in plants (Szabados and Savoure 2010). While it has been suggested that proline can act as a direct ROS scavenger (Krishnan et al. 2008; Kaul et al. 2008), another mechanism by which proline synthesis can protect against oxidative stress is by acting as a mitochondrial vent to prevent ROS production in the first place. As mentioned above, myofibroblast differentiation is associated with increased mitochondrial oxygen consumption and elevations of TCA cycle intermediates including citrate, α-ketoglutarate, succinate, fumarate, and malate (Xie et al. 2015; Bernard et al. 2015; Schworer et al. 2020; Hamanaka et al. 2019; Selvarajah et al. 2019; O’Leary et al. 2020). Increased oxidation of substrates in mitochondria can lead to an NADH/NAD⁺ ratio that exceeds ATP demand and the capacity of complex IV to transfer the electrons to molecular oxygen. This can lead to premature leaking of single electrons to oxygen, producing superoxide. De novo proline production removes both carbon and reducing equivalents from mitochondria. Instead of oxidizing glutamine-derived glutamate (via α-ketoglutarate), which transfers electrons to NAD⁺ in the TCA cycle, conversion of glutamate to P5C prevents glutamine-derived carbon from entering the TCA cycle. Further, conversion of P5C to proline by mitochondria-localized PYCR1 and PYCR2 oxidizes NADH, reducing the number of electrons transferred to complex I of the electron transport chain. Indeed, when P5CS is knocked out in fibroblasts using CRISPR/Cas9, cells exhibit increased levels of TCA cycle intermediates and cellular ROS content after stimulation with TGF-β (Schworer et al. 2020).

Epigenetic Regulation

While the contribution of amino acid biosynthesis to cellular redox homeostasis is likely a major reason for this dependence, other mechanisms, including epigenetic regulation are also likely to be involved. It is increasingly understood that cellular epigenetic state is intimately linked with metabolic state. This is due to the fact that metabolite cofactors including acetyl-CoA and S-adenosylmethionine (SAM) are required for the activity of chromatin-modifying enzymes such as acetylases and methyltransferases, respectively (Su et al. 2016; Ye et al. 2014). Removal of acetyl and methyl marks is also regulated by metabolites as the sirtuin family of deacetylases are sensitive to NAD⁺ levels and TET DNA demethylases and Jumanji C (JMJC) histone demethylases require α-ketoglutarate, oxygen, and vitamin C as cofactors (Houtkooper et al. 2012; Islam et al. 2018).

Very little is known about epigenetic regulation during myofibroblastic differentiation or during the pathogenesis of fibrotic disease; however, TGF-β-induced myofibroblastic differentiation as well as fibrotic lung tissue from mice and IPF patients are associated with reductions in histone acetylation and changes in DNA methylation patterns (Jones et al. 2019; Li et al. 2017; Rabinovich et al. 2012; Sanders et al. 2012). De novo amino acid production may be an important regulator of these epigenetic changes. One-carbon metabolism downstream of de novo glycine synthesis supports SAM production and is associated with changes in DNA methylation (Maddocks et al. 2016; Kottakis et al. 2016; Reina-Campos et al. 2019). Furthermore, α-ketoglutarate produced by PSAT1 has been shown to be essential for maintenance of DNA methylation patterns and stemness in mouse embryonic stem cells (mESCs) (Hwang et al. 2016)

Proline metabolism and collagen synthesis may also regulate epigenetic state in myofibroblasts. Treatment of mESCs with proline leads to the development of a mesenchymal phenotype associated with genome-wide increases in histone and DNA methylation (Comes et al. 2013; D’Aniello et al. 2017). This transition is reversed by vitamin C and interestingly does not occur in mESCs with the genes for collagen prolyl hydroxylases knocked out (D’Aniello et al. 2019). Collagen prolyl hydroxylases are α-ketoglutarate-dependent dioxygenases, the same family of enzymes that includes TET DNA demethylases and JMJC histone demethylases (Islam et al. 2018). Thus, collagen synthesis and proline hydroxylation appear to be in competition with chromatin-modifying enzymes for the vitamin C and possibly α-ketoglutarate that are required for their activity. It remains to be determined whether similar regulation occurs through de novo proline production in myofibroblasts; however proline abundance is regulated by P5CS induction as knockout of ALDH18A1 reduces cellular proline levels while overexpression increases them (Schworer et al. 2020).

Caveats and Future Directions

The totality of the evidence suggests that de novo synthesis of glycine and proline is crucial for TGF-β-induced collagen protein production and for organ fibrosis. Genetic approaches have identified metabolic pathways that are regulated by TGF-β in fibroblasts. Metabolic tracing experiments demonstrate incorporation of de novo synthesized glycine and proline into collagen protein, and inhibition of the relevant enzymes prevents TGF-β-induced collagen protein production in vitro and attenuate fibrosis in vivo. It is important to note that the in vitro experiments conducted have been performed in defined environments with defined metabolic conditions that do not accurately reflect in vivo conditions in humans during either health or disease. While much can be learned from monocultures, the complexities of cell-cell interactions as well as the microenvironmental nutrient availability are not representative of the in vivo environment. Culture conditions have been shown to play a major role in regulating the way cells use metabolites. Metabolic pathways that are important in cultured cells are not necessarily important in vivo (Muir et al. 2018; Muir and Vander Heiden 2018). Thus, further study of the nutrient microenvironment found in fibrotic tissues in vivo will be crucial to fully understand the metabolism of myofibroblasts.

Mutations in ALDH18A1 (the gene encoding P5CS) and PYCR1 are associated with the connective tissue disorder cutis laxa, consistent with a role of proline synthesis in collagen and elastin production. PHGDH and PSAT1 mutations are also associated with skin defects, among other clinical manifestations. Whether these defects are the result of an aberrant matrix environment is unknown (Rumping et al. 2020; El-Hattab 2016). While pharmacologic inhibition of either PHGDH or GLS prevents lung fibrosis after bleomycin instillation in vivo, fibroblast-specific knockout of PHGDH or P5CS will be needed to conclusively demonstrate the role of these de novo amino acid synthesis in fibroblasts for the fibrotic response in vivo. Studies of in vivo metabolism of extracellular metabolites after infusion of stable isotope-labeled metabolites will also strengthen the evidence for de novo amino acid synthesis in fibrosis. This technique has been used in tumor-bearing mice as well as in cancer patients to determine how tumors use metabolites in situ (Davidson et al. 2016; Maher et al. 2012; Faubert et al. 2017; Hensley et al. 2016).

Diet has been shown to play a role in microenvironmental metabolite concentrations in cancer models and dietary provision of glycine, proline, and hydroxyproline (which can be converted to glycine in the kidney) has been shown to be important for collagen production and growth of livestock (Sullivan et al. 2019; Li and Wu 2018; Wu 2020; Wu et al. 2019). Elimination of serine and glycine from diets of mice lowers plasma serine and glycine levels to an extent that inhibits cancer cell growth; however, serine and glycine deprivation had no effect on the ability of bleomycin instillation to induce fibrosis in mouse lungs (Maddocks et al. 2013; Maddocks et al. 2017; Hamanaka et al. 2018). Thus, the effect of dietary nutrients on plasma amino acid levels, tissue microenvironment, and fibrotic responses in vivo warrants further investigation. It will also be of interest to examine the relative contributions of dietary versus de novo amino acid synthesis for both homeostatic collagen synthesis and fibrotic collagen synthesis.

Finally, while studies have focused on anabolic pathways, which convert glucose and glutamine into amino acids required for collagen synthesis, the role of catabolic pathways should also be investigated. The branched chain amino acids leucine, isoleucine, and valine can be deaminated by branched chain aminotransferases, producing glutamate which may be used to produce glycine or proline as discussed above (Selwan and Edinger 2017). Another source of amino acids which may be important for collagen production may be breakdown products of extracellular and intracellular proteins. Macropinocytosis has been shown to contribute significantly to the amino acid pools of Kras-driven pancreatic cancer cells (Kamphorst et al. 2015; Davidson et al. 2017). It has also been suggested that recycling of proline from collagen breakdown products by the enzyme prolidase may play a role in supporting collagen synthesis during fibrosis (Karna et al. 2020). Whether this effect is myofibroblast-intrinsic remains to be demonstrated.

Conclusions

Current evidence strongly suggests that de novo amino acid biosynthesis is a major regulator of matrix production by myofibroblasts. These pathways represent novel targets for the treatment of fibrotic diseases and, as inhibitors of these pathways are currently in development for cancer therapy, these drugs may be repurposed for fibrosis therapy. It will be of interest to determine how targeting these pathways affects not only cancer cells, but also the interplay between these cells and the stroma. Cancer-associated fibroblasts participate in a metabolic interplay with tumor cells, producing metabolites such as lactate, alanine, and asparagine that are taken up and metabolized by tumor cells (Li and Simon 2020). Catabolism of collagen peptides produced by stromal cells may also be an important source of nutrients for certain cancer cells (Olivares et al. 2017). Thus, inhibition of collagen synthesis by cancer-associated fibroblasts may deprive cancer cells of the mechanical and microenvironmental niche required for tumorigenesis as well as essential fuel required for cell growth.

While the physical building blocks for collagen protein are a major output of these amino acid synthesis pathways, the requirement for their de novo production even in the presence of extracellular amino acids suggests the non-proteinogenic functions of these pathways also plays a major role in myofibroblasts. Prokaryotes possess the ability to produce all 20 amino acids from carbon and nitrogen sources. Mammals have lost the ability to synthesize 9 of these molecules, making these dietary necessities. The retention of the pathways for the synthesis of the remaining 11 nonessential amino acids suggests that the process of producing these amino acids provides a cellular benefit that is of equal importance to the cell as the actual end product. It has become clear that these amino acid biosynthesis pathways are important regulators of cellular redox homeostasis and epigenetic state. These meta-roles of amino acid synthesis pathways may play important roles in gene expression and apoptosis resistance in myofibroblasts. Continued study of these pathways will thus provide a greater understanding of the biology of fibrosis and has potential to lead to novel drugs to target fibrotic disease.