Structure, Biochemistry, and Gene Expression Patterns of the Proline Biosynthetic Enzyme Pyrroline-5-Carboxylate Reductase (PYCR), An Emerging Cancer Therapy Target

Abstract

Proline metabolism features prominently in the unique metabolism of cancer cells. Proline biosynthetic genes are consistently upregulated in multiple cancers, while the proline catabolic enzyme proline dehydrogenase has dual, context-dependent pro-cancer and pro-apoptotic functions. Furthermore, the cycling of proline and Δ¹-pyrroline-5-carboxylate through the proline cycle impacts cellular growth and death pathways by maintaining redox homeostasis between the cytosol and mitochondria. Here we focus on the last enzyme of proline biosynthesis, Δ¹-pyrroline-5-carboxylate reductase, known as PYCR in humans. PYCR catalyzes the NAD(P)H-dependent reduction of Δ¹-pyrroline-5-carboxylate to proline and forms the reductive half of the proline metabolic cycle. We review the research on the three-dimensional structure, biochemistry, inhibition, and cancer biology of PYCR. To provide a global view of PYCR gene upregulation in cancer, we mined RNA transcript databases to analyze differential gene expression in 28 cancer types. This analysis revealed strong, widespread upregulation of PYCR genes, especially PYCR1. Altogether, the research over the past 20 years makes a compelling case for PYCR as a cancer therapy target. We conclude with a discussion of some of the major challenges for the field, including developing isoform-specific inhibitors, elucidating the function of the long C-terminus of PYCR1/2, and characterizing the interactome of PYCR.

The enzymes of proline metabolism

Proline metabolism is central to the metabolic shift that occurs in cancer cells, and certain enzymes of proline metabolism have emerged as potential cancer therapy targets (Tanner et al. 2018; Phang 2019; D’Aniello et al. 2020; Burke et al. 2020; Bergers and Fendt 2021). The subject of this review article, the enzyme Δ¹-pyrroline-5-carboxylate reductase, known in humans as PYCR, catalyzes the final step of proline biosynthesis. PYCR isoforms are consistently upregulated in many cancers, and recent studies have shown that lowering PYCR activity by genetic knockdown or chemical inhibition diminishes the growth of cancer cells. The accumulating evidence supports the investigation of the tractability of PYCR as a cancer therapy target, hence the time seems right to summarize our knowledge of the structure and biochemistry of PYCR in the context of cancer research.

Proline metabolism refers to the enzyme-catalyzed interconversion of L-proline and L-glutamate (Fig. 1). The biosynthetic and catabolic arms have distinct enzymes but share the common intermediate Δ¹-pyrroline-5-carboxylate (P5C) and its hydrolysis product, L-glutamate-γ-semialdehyde (GSAL).

Fig. 1.

The enzymes and reactions of proline metabolism. (A) Proline biosynthesis from glutamate and ornithine. (B) Proline catabolism. Abbreviations: α-KG, α-ketoglutarate; GSAL, L-glutamate-γ-semialdehyde; GSALDH, L-glutamate-γ-semialdehyde dehydrogenase; OAT, ornithine-δ-aminotransferase; L-Orn, L-ornithine; PLP, pyridoxal-5’-phosphate; PMP, pyridoxamine-5’-phosphate; PRODH, proline dehydrogenase; P5C, Δ¹-pyrroline-5-carboxylate; P5CS, Δ¹-pyrroline-5-carboxylate synthase; PYCR, Δ¹-pyrroline-5-carboxylate reductase.

Proline biosynthesis starts from either L-glutamate or L-ornithine (Fig. 1A). L-glutamate may be derived (via the enzyme glutaminase) from L-glutamine, a well-known nutrient used by cancer cells to increase proliferation as well as survival under metabolic stress conditions (Choi and Park 2018; Phang et al. 2015; Kuo et al. 2021; Lieu et al. 2020). The glutamate route to proline begins with the formation of GSAL via the intermediate γ-glutamate phosphate. This transformation is done in two distinct enzymatic steps catalyzed by ATP-dependent glutamate-5-kinase and NADPH-dependent γ-glutamate phosphate reductase. In later evolved eukaryotes, these two enzymatic activities are combined in the bifunctional enzyme P5C synthase (P5CS) (Perez-Arellano et al. 2010). γ-glutamate phosphate is highly labile, and it is likely channeled between the kinase and reductase active sites of P5CS, although this has not been studied. GSAL is also a labile intermediate, and at physiological pH, it cyclizes with the loss of water to form P5C (Bearne and Wolfenden 1995). The ornithine route to proline also yields GSAL/P5C via the enzyme ornithine-δ-aminotransferase (OAT), thus linking proline metabolism and arginine metabolism. The glutamate and ornithine pathways converge on PYCR, which catalyzes the reduction of P5C to proline using NAD(P)H as the hydride donor.

Proline catabolism consists of two enzymes (Fig. 1B). Proline dehydrogenase (PRODH) is a flavoenzyme that catalyzes the oxidation of proline to P5C. The enzyme-bound flavin (FAD in human PRODH, FMN in some bacterial PRODHs (Huijbers et al. 2017)) is reduced during this reaction, and the electrons stored in the flavin are transferred to the electron transport chain in the membrane to regenerate the oxidized flavin for another round of catalysis. In eukaryotes, PRODH is an inner mitochondrial membrane protein, whereas bacterial PRODH associates with the membrane in a peripheral manner. The structural basis of PRODH’s interactions with membranes is not well understood. The second enzyme of proline catabolism is GSALDH (a.k.a. P5C dehydrogenase, P5CDH). GSALDH belongs to the aldehyde dehydrogenase (ALDH) structural superfamily and is known as ALDH4; human GSALDH is encoded by the ALDH4A1 gene. GSALDH catalyzes the NAD⁺-dependent oxidation of GSAL to glutamate. In some bacteria, PRODH and GSALDH are combined into the bifunctional enzyme, proline utilization A (Liu et al. 2017).

The proline cycle

PYCR is essential to the cycling of proline and P5C to maintain redox homeostasis between the cytosol and mitochondria, known as the proline cycle (Fig. 2) (Phang 1985, 2019). The catabolic half-cycle is the oxidation of proline to P5C catalyzed by PRODH in mitochondria. The synthetic half-cycle is the reduction of P5C to proline catalyzed by PYCR isoforms both inside mitochondria (PYCR1/2) and in the cytosol (PYCR3). It is proposed that the cycle transfers reducing equivalents from cytosolic NAD(P)H into the mitochondrial respiratory chain, although the redox shuttling mechanism remains to be clarified. By impacting cytosolic NADP⁺ levels, the cycle links proline metabolism to the pentose phosphate pathway and nucleotide biosynthesis (Liu et al. 2015; Phang et al. 2012; Phang 2019; Hagedorn and Phang 1986). The proline cycle has been shown to enhance oxidative phosphorylation, maintain cytosolic pyridine nucleotide levels, and generate reactive oxygen species leading to activation of numerous cell-signaling pathways (Elia et al. 2017; Liu et al. 2015; Phang 2019; Hagedorn and Phang 1986; Miller et al. 2009). The central role of the proline cycle in cell survival, proliferation, and metastasis has motivated investigations of PRODH and PYCR as potential cancer therapy targets (Tanner et al. 2018).

Fig. 2.

Overview of the proline cycle.

PYCR isoforms

The human genome encodes three isoforms of human PYCR: PYCR1, PYCR2, and PYCR3 (a.k.a. PYCRL) (Fig. 3). PYCR1 is localized to the mitochondria and has 319 amino acids (UniProt P32322, chromosome 17q25.3, gene symbol PYCR1). PYCR2 (UniProt Q96C36, 1q42.12, gene symbol PYCR2) has 320 amino acids, is 85% identical to PYCR1, and is also mitochondrial. PYCR3 (UniProt Q53H96, 8q24.3, gene symbol PYCR3) is quite different from the other two isoforms. It has only 274 amino acids, shares just 45% identity with PYCR1/2, and is cytosolic. The shorter polypeptide length of PYCR3 is due primarily to a shortening of the C-terminus by ~50 residues (Fig. 3). PYCR3 functions primarily in the ornithine pathway of proline biosynthesis (De Ingeniis et al. 2012).

Fig. 3.

Sequence alignment of PYCR1, PYCR2, and PYCR3 isoforms. The secondary structure elements were obtained from the structure of PYCR1 (PDB ID 5UAV). This figure was made with ESPript 3.0 (Robert and Gouet 2014) from a multiple sequence alignment calculated in Clustal Omega (Sievers et al. 2011). Blue transparent boxes represent residues in the active site of PYCR1. The α-helix-breaking proline, Pro178, is denoted with a star.

The N-terminal Met of all three PYCR isoforms is most likely co-translationally cleaved by the enzyme methionine aminopeptidase. This modification is suggested by the small size of the residue adjacent to the N-Met (Wingfield 2017), which is Ser in PYCR1/2 and Ala in PYCR3 (Fig. 3).

Although PYCR1 and PYCR2 are mitochondrial proteins, the identities of their mitochondrion-targeting signals within the polypeptide chain are uncertain. Analysis of the PYCR1 amino acid sequence with TPpred2.0 (Savojardo et al. 2014), MitoProt II-V1.101 (Claros and Vincens 1996), and TargetP-2.0 (Almagro Armenteros et al. 2019) revealed no mitochondrion-targeting signal. Analysis of PYCR2 suggested N-terminal mitochondrion-targeting peptides with possible cleavage sites at residues 26, 48, or 98. It is unlikely that the suggested targeting peptide is cleaved, because this would remove a large section of the NAD(P)H-binding domain. Although the precise locations of the targeting signals are unknown, presumably PYCR1 and PYCR2 are imported into mitochondria in non-native conformations, and once inside the mitochondrial matrix, they fold into their native conformations and assemble into the appropriate quaternary structures (dimer and decamer, as discussed below).

Certain mutations in the genes encoding proline metabolic enzymes cause inherited metabolic diseases. Mutations in the PYCR1 gene (MIM 179035) have been identified in patients with autosomal recessive cutis laxa disorders (Mohamed et al. 2011; Kariminejad et al. 2017; Reversade et al. 2009). Mutations in the PYCR2 gene (OMIM 616420) are associated with hypomyelinating leukodystrophy 10 (Escande-Beillard et al. 2020; Zaki et al. 2016; Patel et al. 2021b). To our knowledge, there are no reports of inherited metabolic diseases associated with mutations in the PYCR3 gene.

The three-dimensional structure of PYCR

The three-dimensional structure of PYCR1 has been extensively characterized by X-ray crystallography and solution biophysics. Early low resolution crystal structures (3.1 Å) provided the first information about the fold of PYCR1, but the quality of the electron density for ligands was insufficient to locate the active site (Meng et al. 2006). A decade later, high resolution structures (1.85 – 1.90 Å) of PYCR1 complexed with active site ligands confirmed the fold and furthermore provided unequivocal identification of the active site and the basis of cofactor and substrate recognition (Christensen et al. 2017). A key to obtaining high resolution crystals was the use of a truncated PYCR1 construct lacking the C-terminal 19 residues, a region of the protein that is disordered in the earlier PYCR1 structures. Apparently, removing these residues improves the crystallizability of PYCR1.

PYCR1 has a two-domain fold consisting of a Rossmann dinucleotide binding domain (residues 1–162) followed by an α-helical domain (Fig. 4A). The former domain is named for the late Michael Rossmann, an eminent structural biologist who was among the first to appreciate evolutionary structural conservation (Rossmann et al. 1974; Wu and Arnold 2019). The Rossmann domain binds the cofactor NAD(P)H. The α-helical domain mediates oligomerization, both dimerization (Fig. 4C) and the assembly of dimers into a pentamer-of-dimers decamer (Fig. 4D). The α domain also plays a key role in binding P5C. The substrate P5C binds in a loop connecting the K and L α-helices of the α domain, and at first glance, P5C and NADPH appear to be very far apart from each other (Fig. 4A). However, the active site is located in the dimer interface, where the P5C-binding loop of one protomer meets the NAD(P)H-binding site of the other protomer in the dimer (Fig. 4B). Thus, dimerization is essential for catalysis by PYCR1.

Fig. 4.

Fold and oligomeric structure of PYCR1. (A) The structure of a PYCR1 protomer showing the protein fold (PDB ID: 5UAV). The polypeptide chain is colored in a rainbow scheme with dark blue at the N-terminus and red at the C-terminus. A black star represents the α-helix-breaking Pro178 (see Figure 3). NADPH is shown in white sticks. Helices K, L, and M are labeled. (B) Close-up view of the active site of PYCR1 with NADPH and (S)-(−)-tetrahydro-2-furoic acid (THFA) bound. Interactions made by THFA with NADPH, Ser233, and Thr238 are shown in black dotted lines. Note the active site is fully formed by the presence of the other protomer completing the dimer. (C) The oligomeric structure of PYCR1 showing the dimer, with NADPH and THFA shown in spheres. (D) The full oligomeric assembly of PYCR1, a pentamer-of-dimers decamer.

PYCR1 exists in solution mainly as a pentamer-of-dimers decamer (Fig. 4D) when assayed by sedimentation velocity analytical ultracentrifugation performed at the relatively high enzyme concentration of 6 mg/mL (Christensen et al. 2017). Smaller molecular weight species are observable at lower enzyme concentrations, consistent with a dynamic self-association equilibrium. The crystalline protein is decameric, as expected for the solid state, where the protein concentration is very high. We note that crystal structures of bacterial and plant homologs of PYCR1 are also decameric in crystallo, suggesting the pentamer-of-dimers decamer is a broadly conserved structural feature of P5C reductases (Ruszkowski et al. 2015; Nocek et al. 2005).

For PYCR2, a low resolution crystal structure (3.4 Å) of the apo enzyme has been determined and used to understand a novel missense p.Gly249Val mutation identified in a patient with hypomyelinating leukodystrophy-10 (Escande-Beillard et al. 2020). The folds of PYCR1 and PYCR2 are identical, as expected for two proteins with 85% amino acid sequence identity (Fig. 5A). The amino acid sequence differences between PYCR1 and PYCR2 are mostly located outside of the active site and are largely solvent exposed (Fig. 5B). The pentamer-of-dimers decamer is also observed in the crystal structure of PYCR2.

Fig. 5.

Comparison of the structures of PYCR1 and PYCR2. (A) Cartoon representation of an alignment of the protomers of PYCR1 (PDB ID 5UAV, green) and PYCR2 (PDB ID 6LHM, yellow). (B) A dimer of PYCR1 with pairwise sequence differences from PYCR2 indicated by color. Light green coloring of the left protomer represents positions of non-conserved physicochemical changes and gold are conserved physicochemical sequence changes. The accompanying protomer is included in white to emphasize dimerization is required to form the full active site and to highlight interchain interactions between NADPH and THFA.

The active sites of PYCR1 and PYCR2 are 97% identical in sequence. For this calculation, we defined active site residues as those within 4.0 Å of NADPH or the P5C/proline analog (S)-(−)-tetrahydro-2-furoic acid (THFA) in the PYCR1 ternary complex structure (PDB ID 5UAV). Given this definition, the active site consists of 32 residues, which are indicated by blue shading in the sequence alignment (Fig. 3). The active site sequences of PYCR1 and PYCR2 differ only in one position, which is a conservative substitution: Asp36 of PYCR1 is Glu in PYCR2. This residue is located in the loop that binds the 2’-phosphoryl of NADPH. The high active site sequence identity suggests that the modes of cofactor and substrate binding to PYCR2 likely resemble PYCR1, but this should be confirmed with high resolution crystallography.

Less is known about the structure of PYCR3, as no crystal structures of the enzyme have been reported. With 45% global amino acid sequence identity to PYCR1, there is little doubt that PYCR3 has the same fold as PYCR1/2. The active site of PYCR3 is 53% identical and 72% similar to PYCR1. The PYCR3 active site contains nine residues that are not conserved in PYCR1/2 (Fig. 6). The differences appear mainly in the NADPH binding site. In particular, the loop that binds the 2’-phosphoryl of NADPH has two variations, S34A and D36T. Also note this loop is two residues shorter in PYCR3 (Fig. 3). These structural differences may contribute to the pronounced cofactor-dependence of the K_m for P5C (Table 1). Also, the Q10R and N230D sequence differences are close together in space and could result in an ion pair between Arg and Asp in the active site of PYCR3.

Fig. 6.

Three-dimensional structural representation of nonconservative amino acid variations in the active site of PYCR. The structure of the PYCR1 dimer is shown with the two protomers colored dark gray and white. The light green spheres indicate active site residues that are not conserved in PYCR3. The notation lists the amino acid type and number in PYCR1, followed by the amino acid type in PYCR3. NADPH and the P5C/proline analog THFA are shown in sticks.

Table 1.

Kinetic constants of PYCRs

Enzyme	Variable Substrate	Fixed Substrate	Km (μM)	kcat (s−1)	kcat/Km (M−1s−1)
PYCR1a	T4C	NADP+	1260	55	4 × 104
PYCR1a	NAD+	T4C	151	-	-
PYCR1a	NADP+	T4C	3060	-	-
PYCR1 E221Aa	T4C	NADP+	730	13	2 × 104
PYCR1 E221Aa	NAD+	T4C	235	-	-
PYCR1 E221Aa	NADP+	T4C	480	-	-
PYCR1b	P5C	NADH	1720	70	4 × 104
	P5C	NADPH	2150	29	1 × 104
	NADH	P5C	260	64	2 × 105
	NADPH	P5C	1200	46	4 × 104
PYCR1c	P6C	NADH	146	-	-
PYCR1d	NADH	P5C	70	218	3 × 106
	NADPH	P5C	283	74	3 × 105
	P5C	NADPH	667	31	5 × 104
PYCR1 T238Ad	NADPH	P5C	159	23	1 × 105
	P5C	NADPH	2,887	14	5 × 103
PYCR1e	P5C	NADH	185–374	35–69	2 × 105
PYCR2b	P5C	NADH	1000	149	1.5 × 105
	P5C	NADPH	1700	85	5 × 104
	NADH	P5C	220	219	1 × 106
	NADPH	P5C	240	93	4 × 105
PYCR2f	NAD+	L-proline	1110	52.2	5 × 104
PYCR2 G249Vf	NAD+	L-proline	3280	22.2	7 × 103
PYCR2g	P5C	NADH	1509	61.3	4 × 104
	P5C	NADPH	994	26.0	3 × 104
	NADH	P5C	298	47.9	2 × 105
	NADPH	P5C	216	24.0	1 × 105
PYCR2 R251Cg	P5C	NADH	1315	3.2	2 × 103
	P5C	NADPH	1499	3.1	2 × 103
	NADH	P5C	953	5.9	6 × 103
	NADPH	P5C	119	1.8	1 × 104
PYCR2 R119Cg	P5C	NADH	317	0.08	2 ×102
	P5C	NADPH	334	0.13	4 × 102
	NADH	P5C	34	0.015	4 × 102
	NADPH	P5C	537	1.5	3 × 103
PYCR3b	P5C	NADH	4640	197.0	4 × 104
	P5C	NADPH	380	35.0	9 × 104
	NADH	P5C	420	196.4	5 × 105
	NADPH	P5C	370	24.9	7 × 104

^a(Meng et al. 2006)

^b(De Ingeniis et al. 2012)

^c(Struys et al. 2014)

^d(Christensen et al. 2017)

^e(Christensen et al. 2020)

^f(Escande-Beillard et al. 2020)

^g(Patel et al. 2021a)

Another interesting feature of the PYCR3 sequence is that it lacks the α-helix-breaking proline found near the active sites of PYCR1/2 (replaced with Val). Pro178 of PYCR1 introduces a kink in an otherwise helical section of polypeptide chain, and the kink forms part of the roof of the P5C binding site (star in Fig. 4A) (Christensen et al. 2017). This variation could result in differences in the shape of the P5C pocket.

Aside from the amino acid sequence differences in the active site, PYCR3 differs from PYCR1/2 by having a shorter C-terminus (Fig. 3). PYCR3 lacks a ~50-residue section of polypeptide at the C-terminus compared to PYCR1/2. Interestingly, these additional PYCR1/2-specific residues are disordered in the crystal structures, so the conformation of this region of the polypeptide chain is unknown. The residues immediately N-terminal to the disordered region form a ~20-residue α-helix in PYCR1/2 (αM), which is involved in oligomerization (Fig. 4). The role of the flexible C-terminus for the function of PYCR1/2, and the meaning of its absence in PYCR3, are both unknown.

Catalytic properties of PYCRs

PYCR catalyzes the NAD(P)H-dependent reduction of P5C to proline. The structure of the ternary complex of PYCR1 with NADPH and the P5C/proline analog THFA provided insight into the catalytic mechanism (Christensen et al. 2017). The ring of THFA stacks in parallel with the nicotinamide such that the C5 of THFA, which represents the hydride acceptor atom of P5C, is 3.7 Å from the C4 of the nicotinamide (Fig. 4B). This arrangement is consistent with a direct hydride-transfer mechanism, as expected for an oxidoreductase that utilizes an obligate 2-electron cofactor.

Patel et al. recently studied the kinetic mechanism of PYCR2 (Patel et al. 2021a). Steady-state kinetic data suggest a sequential binding mechanism with L-P5C binding the enzyme first, followed by NAD(P)H to form a ternary complex. After the hydride transfer step, NAD(P)⁺ dissociates before proline.

The kinetic constants of recombinant PYCRs have been measured in several studies (Table 1). Caution is recommended when comparing the values from different studies because the experimental details can vary with respect to the concentration of the fixed substrate, the temperature, and whether the full-length enzyme or truncated version was used.

Most of the enzymology of PYCRs has focused on the reductase reaction, using P5C as the substrate and either NADPH or NADH as the cofactor. These include studies of all three PYCR isoforms as well as a few site-directed mutant variants of PYCR1 and PYCR2. One study also investigated whether PYCR1 could play a role in lysine catabolism by catalyzing the NADH-dependent reduction of Δ¹-pyrroline-6-carboxylate (P6C) to pipecolate (Struys et al. 2014). Not only is PYCR1 active with P6C, but the K_m for P6C is within the range found for P5C (Table 1). A challenge with assaying the reductase reaction is that P5C is not commercially available and must be synthesized. This has motivated investigations of the reverse reaction, in which substrates such as L-thiazolidine-4-carboxylate (T4C), L-proline, or 3,4-dehydro-L-proline are oxidized, and NAD(P)⁺ is reduced (Meng et al. 2006; Escande-Beillard et al. 2020; Nocek et al. 2005).

Some general trends are evident in the kinetic data for the reductase reaction (i.e., P5C as the substrate). None of the isoforms stand out as being significantly more active than the others. The average apparent catalytic efficiencies (k_cat/K_m) from studies performed in different labs using NADH as the cofactor are within the same order of magnitude: 9 × 10⁵ M⁻¹s⁻¹ for PYCR1, 4 × 10⁵ M⁻¹s⁻¹ for PYCR2, and 3 × 10⁵ M⁻¹s⁻¹ for PYCR3. PYCRs exhibit robust catalytic activity with either NADH or NADPH as the cofactor. PYCR1 and PYCR2 tend to have higher catalytic efficiency when using NADH as the cofactor. For PYCR1, this result reflects both a lower K_m for NADH versus NADPH, and a lower K_m for P5C when NADH is used. For PYCR2 and PYCR3 the higher efficiency with NADH is due more to k_cat.

Product inhibition of PYCR has been investigated (Table 2). L-proline inhibits PYCR1 with an inhibition constant (K_i) in the range of 1–2 mM. For reference, the K_m for P5C when NADH is the cofactor is in the range of 0.1–0.4 mM. L-proline is a stronger inhibitor of PYCR2 (~0.1 mM) and a much weaker inhibitor of PYCR3 (8 mM). NAD⁺ inhibits PYCR2 with K_i of 0.8 mM, compared to a K_m for NADH of 0.3 mM.

Table 2.

Inhibition constants for PYCR inhibitors

Inhibitor	Enzyme	Ki (μM)
L-proline	PYCR1	600–1700a,b
(S)-(−)-tetrahydro-2-furoic acid	PYCR1	2200b
cyclopentanecarboxylate	PYCR1	1200b
L-thiazolidine-4-carboxylate	PYCR1	600b
L-thiazolidine-2-carboxylate	PYCR1	450b
N-formyl-L-proline	PYCR1	100b
L-proline	PYCR2	96 – 145a,c
NAD+	PYCR2	800c
L-proline	PYCR3	8500a

^aP5C was the variable substrate and NADH was fixed (De Ingeniis et al. 2012)

^bP5C was the variable substrate, NADH was fixed, and truncated PYCR1 was used (Christensen et al. 2020)

^cP5C was the variable substrate and NADH was fixed (Patel et al. 2021a)

^dNADH was the variable substrate and P5C was fixed (Patel et al. 2021a)

Inhibitor development targeting PYCRs

Inhibitor discovery targeting PYCR is in an early stage. Milne et al. recently reported the results of a small-scale screening study that identified pargyline and as an inhibitor of PYCR1 (Milne et al. 2019). Subsequent optimization yielded pargyline derivatives with IC₅₀ values against the purified enzyme of 9–400 μM, such as 1 in Scheme 1. The kinetic mechanism by which 1 inhibits PYCR1 was not determined. Cell-based studies showed that 1 lowered intracellular proline levels in SUM159PT human breast cancer cells and inhibited proliferation of MDA-MB-321 and SUM159PT breast cancer cell lines.

Scheme 1.

Two recently-discovered inhibitors of PYCR1.

Our group recently used a focused target-specific in crystallo screening approach to discover inhibitors of PYCR1 (Christensen et al. 2020). Focused screening (a.k.a. knowledge-based) involves selecting subsets of molecules from chemical libraries that are likely to have activity based on prior knowledge of the target protein and chemical classes that have activity at the target (Deng et al. 2006; Orry et al. 2006; Harris et al. 2011; Hughes et al. 2011). This approach leveraged the prior knowledge of ligand recognition learned from high resolution crystal structures of PYCR1 (Christensen et al. 2017). The initial screen was performed in crystals rather than using activity assays. Crystals of apo PYCR1 were soaked with 27 commercially-available proline analogs, and the electron density in the known P5C site was used as a proxy for inhibition. Compounds that bound the enzyme in crystallo were studied further using enzyme assays to determine the mechanism of inhibition. One advantage of the in crystallo approach over the traditional inhibitor screening pipeline is that problems validating hit compounds are minimized, since the initial assay selects for compounds that bind specifically to the target.

The focused target-specific in crystallo screening campaign uncovered N-formyl-L-proline (2 in Scheme 1) as a promising lead inhibitor of PYCR1 (Christensen et al. 2020). Compound 2 is a competitive inhibitor (competitive with P5C) with an inhibition constant of 100 μM. The mechanism of inhibition includes an unexpected cascade of protein conformational changes emanating from the P5C site to an oligomer interface, which are needed to expand the active site to accommodate the formyl group of 2. Furthermore, 2 was shown to phenocopy the PYCR1 knockdown in MCF10A H-RASV¹² breast cancer cells. To our knowledge, 2 is the only inhibitor of PYCR1 that has been thoroughly validated by demonstrating the kinetic mechanism of action against the purified enzyme, the mode of binding to the enzyme by X-ray crystallography, and activity in cancer cells.

Differential expression of PYCR genes in cancer

PYCR isoforms are consistently upregulated in many cancers. To provide a global view of this phenomenon, we utilized Gene Expression Profiling Interactive Analysis (GEPIA) 2 (http://gepia2.cancer-pku.cn/#index) to analyze differential gene expression across a variety of cancers (Tang et al. 2019). The server provides RNA sequencing expression data of 9736 tumors and 8587 normal samples from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) projects. We compiled a list of PYCR gene expression in 28 cancer types using a log₂(Fold Change) (log2FC) cutoff of 0.1 (equivalent to FC of 1.07), a q-value cutoff of 0.01, and the ANOVA differential method.

The clear trend in the data is that PYCR gene expression tends to increase in cancer cells, especially for PYCR1 (Fig. 7A). PYCR1 expression increased by log2FC > 1.0 in 22 of the 28 cancers (79%, Fig. 7B). Among these, sixteen cancers are notable in showing large increases in PYCR1 expression of log2FC > 2.0, i.e., greater than 4-fold higher. The largest increase in PYCR1 expression occurs in uterine carcinosarcoma (UCS) (log2FC = 4.0). PYCR1 expression decreases only in acute myeloid leukemia (LAML), which has one of the largest changes in expression.

Fig. 7.

Differential gene expression across 28 cancer types for PYCR isoforms. (A) The log2FC is plotted for each cancer type. (B) The data from panel A are combined to show the number of cancer types with increased expression of PYCR genes. (C) Pairwise scatter plot representations of the data from panel A indicating possible associations between the expression patterns of PYCR genes in cancer cells. Abbreviations used in panel A: ACC, adrenocortical carcinoma; BLCA, bladder urothelial carcinoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; COAD, colon adenocarcinoma; DLBC, lymphoid neoplasm diffuse large B-cell lymphoma; ESCA, esophageal carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; KICH, kidney chromophobe; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LAML, acute myeloid leukemia; LGG, brain lower grade glioma; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; OV, ovarian serous cystadenocarcinoma; PAAD, pancreatic adenocarcinoma; PRAD, prostate adenocarcinoma; READ, rectum adenocarcinoma; SKCM, skin cutaneous melanoma; STAD, stomach adenocarcinoma; TGCT, testicular germ cell tumors; THCA, thyroid carcinoma; THYM, thymoma; UCEC, uterine corpus endometrial carcinoma; UCS, uterine carcinosarcoma.

PYCR2 is also upregulated in cancer cells, but to a lesser extent than PYCR1 (Fig. 7). PYCR2 is upregulated by log2FC > 1.0 in six cancers (21%, Fig. 7B). The largest increases occur in thymoma (THYM) and lymphoid neoplasm diffuse large B-cell lymphoma (DLBC).

The differential expression of PYCR3 is between those of PYCR1 and PYCR2 (Fig. 7). PYCR3 expression increases by log2FC > 1.0 in thirteen cancers (46%, Fig. 7B). The largest increase is in DLBC. Two cancers (7%) show a slight downregulation of the PYCR3 gene.

We also examined correlations among the expression patterns of the three PYCR genes. For example, in DLBC and THYM, all three genes are substantially increased by log2FC of 2.0 – 3.0. Conversely, there are no cancers that show downregulation of all three isoforms, although both PYCR1 and PYCR2 are substantially decreased in LAML. Considering pairwise relationships, scatter plots of the expression data suggest positive associations between all three pairs of PYCR genes (Fig. 7C).

Consistent with our analysis of the transcript data, numerous studies have reported increased PYCR1 expression in cancer (summarized in SI Table S1). To our knowledge, the first such report was a 2002 study using prostate cancer tissue samples (Ernst et al. 2002). Upregulation of PYCR1 in prostate cancer was confirmed in subsequent studies (Jariwala et al. 2007; Zeng et al. 2017), and knockdown of the PYCR1 gene was shown to inhibit prostate cancer cell growth (Zeng et al. 2017). An open mechanistic question is whether PYCR1 mediates the action of androgen receptor signaling in prostate cancer (Jariwala et al. 2007; Zeng et al. 2017).

Phang’s group has examined the involvement of PYCRs in tumor growth, stressing the importance of the proline cycle (Liu et al. 2015). They showed that the oncoprotein MYC induces proline biosynthesis from glutamine in P493 human B lymphoma cells by increasing the expression of all three PYCRs and P5CS. The knockdown of any proline biosynthetic gene markedly decreased growth of several cancer cell lines, apparently by decreasing ATP production. Paradoxically, the growth inhibitory effects of proline biosynthesis knockdown persisted even when proline was added to the cultures, suggesting that proline itself was not of primary importance. Instead, the authors concluded that proline biosynthesis contributes to tumor metabolic reprogramming through the cycling of proline and P5C (Fig. 2).

Proline metabolism is central to the metabolic alterations that occur in breast cancer. A 2011 study using negative selection RNAi-screening identified PYCR1 among metabolic genes associated with aggressive breast cancer and stemness (Possemato et al. 2011). Recent work confirmed the upregulation of PYCR1 gene expression and PYCR1 protein in breast cancer cells and showed that the knockdown of PYCR1 reduced the invasion and migration capabilities of breast cancer cell lines and increased drug sensitivity of orthotopically injected ER-positive tumors in vivo (Ding et al. 2017; Craze et al. 2018; Shenoy et al. 2020). Also, higher PYCR1 expression was found to be correlated with tumor size, higher risk of tumor relapse, and poorer patient survival (Ding et al. 2017; Shenoy et al. 2020).

The landmark 2017 study from Fendt’s group established the central role of the proline cycle in breast cancer metastasis (Elia et al. 2017). They found that the knockdown of either PRODH or PYCR1 impaired spheroidal growth of MCF10A H-RAS^V12 breast cancer cells. Further, the inhibition of PRODH enzyme by the proline analog THFA impaired spheroidal cell growth and inhibited metastases in a mouse model of breast cancer. A subsequent study showed that targeting PYCR1 with a small molecule (2 in Scheme 1) induced proline accumulation and impaired proline biosynthesis as well as spheroidal growth in MCF10A H-RAS^V12 breast cancer cells (Christensen et al. 2020). Together, these studies showed the importance of the proline cycle in breast cancer and provided clear support for investigating PRODH and PYCR1 as breast cancer therapy targets.

Several studies have also found increased expression of PYCR1 in lung cancer (Cai et al. 2018; Guo et al. 2019; Sang et al. 2019; Wang et al. 2019; Gao et al. 2020; Lu et al. 2021). Some of these studies also found that increased PYCR1 expression correlated with poor patient outcomes (Cai et al. 2018; Wang et al. 2019; Gao et al. 2020). In one study, the microRNA miR-328–3p was shown to target PYCR1, and the level of miR-328–3p was decreased in lung adenocarcinoma cells, leading to increased PYCR1 expression (Lu et al. 2021). Knockdown of PYCR1 inhibited lung cancer cell proliferation and tumorigenesis, and increased cell apoptosis (Cai et al. 2018; Guo et al. 2019; Sang et al. 2019; Wang et al. 2019; Gao et al. 2020). Two recent studies have focused on the interaction of PYCR1 with Kindlin-2 in lung adenocarcinoma. They found that increased stiffening of the extracellular matrix, which is associated with the pathological progression of cancer (Lampi and Reinhart-King 2018), promotes translocation of Kindlin-2 into mitochondria where it interacts with PYCR1 (Guo et al. 2019). Apparently, this protein-protein interaction did not affect the catalytic activity of PYCR1, as measured in vitro by monitoring the reverse reaction at a single substrate (3,4-dehydro-L-proline) concentration using a glutathione S-transferase fusion of Kindlin-2. Instead, it was suggested that the interaction with Kindlin-2 enhances proline biosynthesis by preventing degradation of PYCR1. A subsequent study by the same group has linked the enhancement of proline biosynthesis by the PYCR1-Kindlin-2 interaction with PINCH-1-mediated mitochondrial dynamics (Guo et al. 2020).

PYCR1 involvement in colorectal cancer has been investigated. One study used peptide mass fingerprinting and found increased PYCR1 protein in an in vitro model of the colorectal adenoma-to-carcinoma progression (Roth et al. 2010). More recently, upregulation of PYCR1 was confirmed in colorectal cancer tissues and cells, and knockdown of PYCR1 inhibited the proliferation, drug resistance, and epithelial-mesenchymal transition of colorectal cancer (Yan et al. 2019). This study also implicated the interaction of PYCR1 enzyme with STAT3 as part of the signaling mechanism in colorectal cancer cells.

Ribosome profiling has been used to identify metabolic vulnerabilities in tumors that could be leveraged in cancer therapy; this approach applied to kidney cancer discovered a vulnerability for proline linked to high levels of PYCR1 (Loayza-Puch et al. 2016). A subsequent study of renal cell carcinoma also found increased expression of PYCR1 and connected PYCR1 levels to increased metastasis and poorer overall patient survival (Weijin et al. 2019).

The involvement of PYCR in liver cancers has received attention recently (see the review in this issue, (Ding et al. 2021)). PYCR2 was identified as a prognostic biomarker in hepatitis B virus-related hepatocellular carcinoma (Gao et al. 2019). PYCR1 expression is also increased in hepatocellular carcinoma, and the knockdown of PYCR1 reduced cell proliferation of multiple hepatocellular carcinoma cell lines in vitro and tumor growth in vivo (Zhuang et al. 2019; Ding et al. 2020). Also, the combination of high PRODH expression, and low expression of both PYCR1 and P5CS (a.k.a. ALDH18A1) was associated with better patient outcomes (Ding et al. 2020).

Numerous other studies have found upregulation of PYCRs in various cancers, which lends support to the idea from global analysis of transcript data (Fig. 7) that proline biosynthesis is broadly important in the unique metabolism of cancer cells. PYCR3 is upregulated in taxol-resistant nasopharyngeal carcinoma cell lines (Li et al. 2015). PYCR1 expression is increased in human malignant melanoma cell lines and is an indicator of poor prognosis (Ye et al. 2018). Knockdown of PYCR1 or PYCR2 decreased proliferation of melanoma cells via increased apoptosis (Ye et al. 2018) or AMPK/mTOR-induced autophagy (Ou et al. 2016). Recently, the correlated upregulation of the Rho-family GTPase Rac3 and PYCR1 has been found in bladder cancer (Cheng et al. 2020), and a genomics study uncovered PYCR1 as a prognostic factor in bladder cancer patients (Liu et al. 2021). Enhanced PYCR1 activity has been found in glioma cells carrying mutations in isocitrate dehydrogenase 1 (Hollinshead et al. 2018). In situ metabolomics has been used to identify proline as an enhanced metabolite in tissues from esophageal cancer patients, and then immunohistochemistry staining revealed increased PYCR2 protein in regions of high proline (Sun et al. 2019). Finally, in samples from gastric cancer patients, PYCR1 expression was found to be upregulated and correlated with advanced tumor stage and aggressive histological type (Xiao et al. 2020). The same study found that the knockdown of PYCR1 inhibited proliferation of gastric cancer cells by promoting apoptosis (Xiao et al. 2020).

Finally, in addition to the aforementioned cancer type-specific research, PYCR has been identified as a consistently upregulated gene in pan-cancer analyses, which look for genes that are differentially expressed in multiple cancers. These include an analysis of messenger RNA profiles of 1454 metabolic enzymes in 19 cancer types and a study of transcriptomic changes in samples representing ten major cancer types (Nilsson et al. 2014; Haider et al. 2016). Another study found an association between PYCR1 expression and poor prognosis (Darzi et al. 2021).

Splice variants

Alternative splicing generates multiple protein isoforms from a single gene, and several splice variants of PYCR genes are known. Excluding the splice variants retained as introns, we found twelve variants of PYCR1 that are expressed as proteins, four for PYCR2, and three for PYCR3 (Table 3). Most of the PYCR proteins generated by alternative splicing are unlikely to be active enzymes. For example, the eight PYCR1 proteins predicted to have fewer than 290 amino acids have severe truncations that eliminate parts of the active site and/or dimer interface (see amino acid sequence alignment in SI Fig. S1). Similarly, the three shorter variants of PYCR2 (< 320 amino acids) and the 254-residue form of PYCR3 are probably not active.

Table 3.

Splice variants of PYCR genes used for differential expression analysis

GEPIA ID	Ensembl Transcript Namea	Ensembl Transcript ID	UniProt ID	Protein Residues (aa)
PYCR1-001	PYCR1-201	ENST00000329875.13	P32322b,c	319
PYCR1-002	PYCR1-202	ENST00000337943.9	P32322-2	316
PYCR1-003	PYCR1-203	ENST00000402252.6	P32322-3	346
PYCR1-005	PYCR1-204	ENST00000403172.8	E2QRB3	288
PYCR1-010	PYCR1-214	ENST00000585215.5	J3QL24	225
PYCR1-012	PYCR1-209	ENST00000579698.5	J3QLK9	165
PYCR1-013	PYCR1-215	ENST00000585244.1	J3KTA8	84
PYCR1-014	PYCR1-212	ENST00000583564.5	J3QKT3	94
PYCR1-015	PYCR1-207	ENST00000577756.5	J3QL32	217
PYCR1-016	PYCR1-213	ENST00000584848.5	J3QL23	173
PYCR1-201	PYCR1-216	ENST00000619204.4	P32322b,c	319
PYCR1-202	PYCR1-217	ENST00000629768.2	J3QL32	217
PYCR2-001	PYCR2-201	ENST00000343818.11	Q96C36c	320
PYCR2-006	PYCR2-207	ENST00000489681.5	A0A087WZF0	157
PYCR2-007	PYCR2-205	ENST00000472798.2	A0A087WX69	87
PYCR2-201	PYCR2-208	ENST00000612039.4	A0A087WTV6	246
PYCR3-001	PYCR3-201	ENST00000220966.10	D3DWK4	286
PYCR3-004	PYCR3-203	ENST00000433751.5	Q53H96-2	254
PYCR3-008	PYCR3-207	ENST00000495276.6	Q53H96-1b,c	274

^a(Aken et al. 2017)

^bThe UniProt canonical protein sequence for this isoform.

^cExperimentally shown to be an active enzyme.

We analyzed the differential expression of PYCR splice variants in cancer with GEPIA2 (cutoffs of q-value < 0.01 and log2FC > 0.1). The heatmap in Fig. 8 shows the differential expression data for the PYCR splice variants. Note that some data are lacking or did not meet the threshold for statistically significance (denoted by a zero in the heatmap). The expression trends for the splice variants (Fig. 8) track those of the three PYCR genes (Fig. 7), which is expected since there is overlap between the data sets used in the two analyses. Generally, the data suggest that PYCR1 is the most consistently upregulated in cancer, followed by PYCR3, and then PYCR2.

Fig 8.

Differential gene expression of PYCR splice variants in various cancer types. Numbers in the cells are log2FC values; a value of 0 indicates either that the change was not statistically significant or data were not available. The transcript labels on the horizontal axis refer to the GEPIA IDs listed in Table 3. The corresponding Ensembl transcript IDs and UniProt IDs for the encoded proteins are also listed in Table 3. Asterisks denote transcript variants encoding experimentally-verified catalytically active enzymes. Abbreviations: ACC, adrenocortical carcinoma; BLCA, bladder urothelial carcinoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; COAD, colon adenocarcinoma; DLBC, lymphoid neoplasm diffuse large B-cell lymphoma; ESCA, esophageal carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; KICH, kidney chromophobe; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LAML, acute myeloid leukemia; LGG, brain lower grade glioma; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; OV, ovarian serous cystadenocarcinoma; PAAD, pancreatic adenocarcinoma; PRAD, prostate adenocarcinoma; READ, rectum adenocarcinoma; SKCM, skin cutaneous melanoma; STAD, stomach adenocarcinoma; TGCT, testicular germ cell tumors; THCA, thyroid carcinoma; THYM, thymoma; UCEC, uterine corpus endometrial carcinoma; UCS, uterine carcinosarcoma.

The phenomenon of increased PYCR expression is broadly observed across multiple cancer types and splice variants (Fig. 8). Twenty-two of the 28 cancer types (79%) have at least one PYCR variant with expression increased by log2FC ≥ 1.2, and 16/19 splice variants (84%) are increased by log2FC ≥ 1.2 in at least one cancer type. An exception to the trend of increased PYCR expression in cancer is LAML, where several PYCR splice variants are significantly downregulated (Fig. 8, bottom row).

As noted above, only a subset of the splice variants are predicted to encode active enzymes. It is of interest to examine the expression of these variants separately from the others, because they may provide insight into the relevance of PYCR catalytic activity to cancer metabolism. Both of the splice variants encoding the canonical PYCR1 enzyme show log2FC > 1.2 in more than half of the 28 cancer types (GEPIA IDs PYCR1-001, PYCR1-201). Similarly, for PYCR2 and PYCR3, the variants encoding known active enzymes experience the most consistent upregulation compared to the other variants, although we note the expression data are sparse for many of the presumed inactive variants (Fig. 8). The observation that transcripts encoding active enzymes are upregulated in cancer cells compared to normal cells is consistent with the catalytic activity of PYCRs playing a role in the metabolic transition to cancer.

Interestingly, a few splice variants predicted to encode inactive proteins are also substantially upregulated. For example, PYCR1-005, which encodes a truncated 288-residue version of the canonical enzyme, is increased by log2FC > 1.2 in 14/28 cancers. This protein lacks a 32-residue section of the α domain in the dimer interface. Because dimerization is essential for catalytic activity, this protein is probably not active. The significance of increased expression of inactive PYCR proteins in cancer remains to be determined. One possibility is that such proteins, although lacking catalytic activity, could still participate in protein-protein interactions. Another possibility is that they regulate catalytic activity by forming mixed oligomers with active PYCRs.

Outlook and challenges

The cancer biology literature from the past 20 years makes a compelling case for PYCR, especially PYCR1, as a novel cancer drug target. PYCR1 gene expression is consistently higher in cancerous tissue compared to normal tissue, the knockdown of PYCR1 impairs cancer cell proliferation both in vitro and in animal models, and PYCR1 gene expression is predictive of undesirable tumor characteristics and poor patient outcomes. These key results appear in dozens of independent studies and have been observed with many different cancer types. Our analysis of PYCR transcript data is not only consistent with this body of research but also suggests that upregulation of PYCR is a highly conserved aspect of cancer cell metabolism (Figs. 7 and 8).

Potent, specific inhibitors of PYCRs are needed as chemical probes to investigate the roles of PYCRs in cancer and as lead compounds for drug discovery. Currently, only one validated chemical probe of PYCR1 is known (2 in Scheme 1), but its potency of 100 μM against the purified enzyme is modest, and its isoform specificity has not been determined. The discovery of 2 from a very small library of 27 compounds showed proof-of-concept for screening proline analogs as inhibitors of PYCR1. It will be interesting to see the extent of inhibition that can be achieved from a more extensive screening of proline analogs versus compounds that access the NADPH site. High-throughput screening of large, diverse chemical libraries should also be given a high priority.

Developing isoform-specific PYCR inhibitors is a challenge for the field. As described here, the active sites of PYCR1 and PYCR2 are virtually identical in structure, which suggests that compounds directed at the active site are unlikely to discriminate between these two isoforms. A larger question is whether specificity between PYCR1 and PYCR2 is necessary for cancer therapeutic applications. The prospects for identifying PYCR3-specific inhibitors are somewhat brighter. Our analysis predicts that PYCR3 may have structural features that distinguish it from PYCR1/2, including ten non-conserved active site residues, a shorter 2’-phosphoryl-binding loop, and a substantially shorter C-terminus (by ~50 residues). A crystal structure of PYCR3 is needed to determine whether these sequence differences result in structural differences that can be leveraged for inhibitor design. It will be important to test new inhibitors against all three PYCR isoforms.

The function of the long C-termini of PYCR1 and PYCR2 is a fundamental unanswered question. The C-terminus of PYCR1/2 contains ~50 extra residues not found in PYCR3 or P5C reductases from plants and bacteria (Fig. 3). Unfortunately, this region is disordered in crystal structures of PYCR1/2. The C-terminus is not essential for catalytic activity, since recombinant PYCR1 lacking the C-terminal 19 residues is catalytically active (Christensen et al. 2017). Nevertheless, a detailed examination of the role of the C-terminus on the catalytic properties of PYCR1 would be informative. The structures of PYCR1/2 indicate that the flexible C-terminus could access the active site, suggesting the possibility that it is a regulatory element. For example, the last resolved residue of helix αM in the PYCR1 structure is only ~20 Å from the active site, and the missing stretch of ~45 residues of polypeptide chain is certainly long enough to reach into the active site. Also, helix αM is in the oligomer interface, so it is possible that the flexible C-terminus mediates inter-subunit communication that could be important for regulating catalytic activity. A third possibility is that the C-terminus is involved in protein-protein interactions. Clearly, more biochemical, cellular, and structural work is needed to understand the function of the flexible C-terminus of PYCR1/2 and the impact of its absence in PYCR3.

Protein-protein interactions of PYCRs is a promising area of research where there is much to learn. Recent work has uncovered possible interacting partners of PYCR1/2, including ORAOV1 in the context of esophageal cancer (Togashi et al. 2014), ribonucleotide reductase small subunit B (Kuo et al. 2016), Kindlin-2 in lung adenocarcinoma (Guo et al. 2019; Guo et al. 2020), the SIRT3 mitochondrial deacetylase (Chen et al. 2019), STAT3 in colorectal cancer (Yan et al. 2019), Lon chaperone (Kuo et al. 2020), Kaposi’s sarcoma-associated herpesvirus K1 oncoprotein (Choi et al. 2020), and several enzymes involved in mitochondrial glycine, glutamate, and fatty acid metabolism (Escande-Beillard et al. 2020). These potential interactors of PYCR1/2 were discovered using cell-based methods, co-immunoprecipitation, and pull-downs. It will be important to validate these protein-protein interactions using biophysical experiments on the purified proteins as part of a broader investigation of the role of PYCR protein-protein interactions in cancer (Mackay et al. 2007).

In summary, PYCR has emerged over the past 20 years as a novel cancer therapy target. The major challenge for the next decade is to determine the tractability of PYCR as a drug target and translate basic science advances on proline metabolism into therapeutic reality.

Abbreviations:

ALDH

Aldehyde dehydrogenase

DLBC

Lymphoid neoplasm diffuse large B-cell lymphoma

GSAL

L-glutamate-γ-semialdehyde

GSALDH

L-glutamate-γ-semialdehyde dehydrogenase

LAML

Acute myeloid leukemia

Log2FC

Log₂(Fold Change)

OAT

Ornithine-δ-aminotransferase

P5C

Δ¹-pyrroline-5-carboxylate

P6C

Δ¹-pyrroline-6-carboxylate

P5CS

Δ¹-pyrroline-5-carboxylate synthase

PRODH

Proline dehydrogenase

PYCR

Human Δ¹-pyrroline-5-carboxylate reductase

T4C

L-thiazolidine-4-carboxylate

THFA

(S)-(−)-tetrahydro-2-furoic acid

THYM

Thymoma

UCS

Uterine carcinosarcoma