Abstract

The family of intracellular lipid binding proteins (iLBPs) is comprised of 16 members of structurally related binding proteins that have ubiquitous tissue expression in humans. iLBPs collectively bind diverse essential endogenous lipids and xenobiotics. iLBPs solubilize and traffic lipophilic ligands through the aqueous milieu of the cell. Their expression is correlated with increased rates of ligand uptake into tissues and altered ligand metabolism. The importance of iLBPs in maintaining lipid homeostasis is well established. Fatty acid binding proteins (FABPs) make up the majority of iLBPs and are expressed in major organs relevant to xenobiotic absorption, distribution, and metabolism. FABPs bind a variety of xenobiotics including nonsteroidal anti-inflammatory drugs, psychoactive cannabinoids, benzodiazepines, antinociceptives, and peroxisome proliferators. FABP function is also associated with metabolic disease, making FABPs currently a target for drug development. Yet the potential contribution of FABP binding to distribution of xenobiotics into tissues and the mechanistic impact iLBPs may have on xenobiotic metabolism are largely undefined. This review examines the tissue-specific expression and functions of iLBPs, the ligand binding characteristics of iLBPs, their known endogenous and xenobiotic ligands, methods for measuring ligand binding, and mechanisms of ligand delivery from iLBPs to membranes and enzymes. Current knowledge of the importance of iLBPs in affecting disposition of xenobiotics is collectively described.

SIGNIFICANCE STATEMENT The data reviewed here show that FABPs bind many drugs and suggest that binding of drugs to FABPs in various tissues will affect drug distribution into tissues. The extensive work and findings with endogenous ligands suggest that FABPs may also alter the metabolism and transport of drugs. This review illustrates the potential significance of this understudied area.

Introduction

Intracellular lipid binding proteins (iLBPs) are a family of ubiquitous proteins in animals that solubilize essential cellular lipids (Schaap et al., 2002; Storch and Corsico, 2008; Smathers and Petersen, 2011; Napoli, 2017). Together with avidins and lipocalins, iLBPs belong to the calycin superfamily of structurally related binding proteins. Despite low amino acid sequence homology (<10%), avidins, lipocalins and iLBPs share a common β-barrel structural fold that makes up their ligand binding cavity (Flower et al., 2000; Schaap et al., 2002; Smathers and Petersen, 2011). Avidins and lipocalins are found in both prokaryotic and eukaryotic organisms, but iLBPs are only present in vertebrate and invertebrate animals (Schaap et al., 2002). The ancestral iLBP gene evolved after animals diverged from plants and fungi, and individual isoforms arose through gene duplication and diversification (Schaap et al., 2002; Haunerland and Spener, 2004; Smathers and Petersen, 2011). The primary amino acid sequence identity for the 16 known human iLBPs ranges from 21% to 77% (Fig. 1A). Generally, the amino acid sequence identity for specific iLBPs across different species is greater than the sequence identity of all fatty acid binding proteins (FABPs) within the same species. For example, fatty acid binding protein 1 (FABP1) has >60% amino acid sequence identity across 18 different species (Zhang et al., 2020), but the sequence identity of all FABPs in humans is as low as 21%.

Fig. 1.

Sequence alignment (A) and phylogenetic tree (B) of human iLBPs. The primary amino acid sequences for all human iLBP family members were collected from the National Center for Biotechnology Information protein database (https://www.ncbi.nlm.nih.gov/protein/). The accession numbers for the amino acid sequences used were P09455.2 (CRBP1), P50120.3 (CRBP2), NP_113679.1 (CRBP3), Q96R05.1 (CRBP4), P29762.2 (CRABP1), P29373.2 (CRABP2), P07148.1 9 (FABP1), P12104.2 (FABP2), P05413.4 (FABP3), P15090.3 (FABP4), Q01469.3 (FABP5), P51161.2 (FABP6), O15540.3 (FABP7), P02689.3 (FABP8), Q0Z7S8.1 (FABP9), A6NFH5.2 (FAPB12). The sequences were aligned using Clustal Omega Multiple Sequence Alignment (https://www.ebi.ac.uk/Tools/msa/clustalo/) (Sievers et al., 2011) and visualized using JalView (Waterhouse et al., 2009). The black bars above the sequence alignment show the three motifs (FATTYACIDBP1-3) that make up the highly conserved fingerprint common to all iLBPs. The colored residues indicate conserved residues based on thresholds set by the Clustal X Color Scheme (https://www.jalview.org/help/html/colourSchemes/clustal.html). Red indicates positively charged residues, blue residues are hydrophobic, magenta are negatively charged, green are polar, orange are glycines, yellow are prolines, and cyan are aromatic. Boxed residues indicate locations of a highly conserved G-x-W triplet common to iLBPs and lipocalins and highly conserved residues involved in ionic interactions with hydroxy and carbonyl groups of ligands. The phylogenetic tree shown in (B) was calculated using the UPGMA clustering method in Simple Phylogeny (https://www.ebi.ac.uk/Tools/phylogeny/simple_phylogeny/) using the multiple sequence alignment data for human iLBPs. Evolutionary distances and phylogenetic relationships should not be inferred from this tree. (Figure created with BioRender.com.)

The human iLBPs are divided into four subfamilies (Fig. 1B) based on phylogenetic analysis and amino acid sequences (Schaap et al., 2002; Liu et al., 2008; Smathers and Petersen, 2011; Ragona et al., 2014). Subfamily I is comprised of the cellular retinol binding proteins (CRBPs) and cellular retinoic acid binding proteins (CRABPs). Subfamily II contains liver FABP (FABP1) and ileal FABP (FABP6, also called I-BABP). Intestinal FABP (FABP2) is the lone iLBP to make up subfamily III and heart (FABP3), adipocyte (FABP4), epidermal (FABP5), brain (FABP7), myelin (FABP8), and testis (FABP9) FABPs, and FABP12 make up subfamily IV (Schaap et al., 2002; Smathers and Petersen, 2011). The FABPs were originally named after the organs from which they were cloned but have been later found to have broader expression.

The human iLBP genes are located in several different chromosomes (Table 1) and, like most iLBP genes in animals, have four exons with three introns (Schaap et al., 2002; Babin, 2009; Smathers and Petersen, 2011; Zhang et al., 2020). The second and third exons are conserved in nearly all FABP genes (Zhang et al., 2020). Phylogenetic studies suggest that FABP genes evolved from a common ancestor likely through tandem duplication (Babin, 2009; Zhang et al., 2020). FABP4, 5, 8, 9, and 12 form a gene cluster on the same chromosome in humans and several other mammals. Some of these FABP genes also form clusters in aves, amphibians, and reptiles. This supports the hypothesis that vertebrate FABP genes may have arisen through continuous tandem duplication from a common ancestor (Zhang et al., 2020).

Table 1

Tissue expression patterns, genomic localization, and endogenous ligands of iLBPs

The complete physiologic functions of iLBPs have yet to be defined, but iLBPs appear to facilitate the efficient uptake of endogenous lipids into tissues, acting as carriers to shuttle ligands through the cytosol and modulating rates of ligand metabolism (Kushlan et al., 1981; Luxon and Weisiger, 1993; Martin et al., 2003; Kaczocha et al., 2009; Yu et al., 2014; Gajda and Storch, 2015). Altered iLBP function and expression have been associated with dyslipidemia, metabolic syndrome, obesity, diabetes, atherosclerosis, and inflammation (Furuhashi and Hotamisligil, 2008; Atshaves et al., 2010; Peng et al., 2012; Wang et al., 2016; Furuhashi, 2019; Valizadeh et al., 2021). Several iLBP isoforms also bind xenobiotics (Chuang et al., 2008; Trevaskis et al., 2011; Velkov, 2013; Lee et al., 2015; Huang et al., 2018; Elmes et al., 2019). Based on their high and ubiquitous expression in tissues, iLPBs may be determinants of xenobiotic distribution and uptake into tissues. This review focuses on ligand binding to iLBPs, tissue expression of iLBPs, methods to determine ligand binding, and the biochemical roles of iLBPs as they relate to the potential of iLBPs to be determinants of drug disposition.

Intracellular Lipid Binding Protein Structures and Endogenous Ligand Binding

The tertiary structures of iLBPs are virtually superimposable and have two characteristic structural features, a β-barrel domain and helix-turn-helix motif (Fig. 2). Ten anti-parallel β-strands fold into two β-sheets to form the β-“clam-like” cavity of the iLBPs (Fig. 2) (Furuhashi and Hotamisligil, 2008; Storch and McDermott, 2009; Ferrolino et al., 2013; Napoli, 2016). The two alpha-helices along with nearby loops form a portal region for ligand entry and egress into the interior binding cavity (Fig. 2). The iLBPs have a characteristic fingerprint composed of three separate motifs termed FATTYACIDBP1-3 (Fig. 1A). The G-x-W triplet in the first FATTYACIDBP1 motif is highly conserved between iLBP members (Fig. 1A) and homologous with a similar motif in lipocalin family of binding proteins (Smathers and Petersen, 2011). FABP5 is unique in the iLBP family in that it is the only FABP known to form an intramolecular disulfide bond (C120-C127) (Hohoff et al., 1999). The dynamics of iLBP structures and consequences on ligand binding have been extensively studied, and several comprehensive reviews are available on this topic (Storch and McDermott, 2009; Atshaves et al., 2010; Smathers and Petersen, 2011; Ragona et al., 2014).

Fig. 2.

The crystal structure of human holo-CRABP2 (PDB 1CBS) showing the overall structural features of iLBPs. The figure shows the β-barrel cavity composed of 10 β-strands (βA-βJ), the helix-turn-helix cap consisting of the alpha helices (α1 and α2) and the portal to the ligand binding domain with the neighboring loops (loop βC-βD and βE-βF). Two β-sheets, each made up of five β-strands, fold to form the β-clam of the iLBP structure. (Structures generated from PDB using ChimeraX; figure created with BioRender.com.)

iLBP Structures

Crystal structures and NMR solution structures of iLBPs show that ligands are stabilized within the binding cavity by ionic interactions, hydrogen bonding networks with water molecules, and interactions with hydrophobic regions (Kleywegt et al., 1994; Cai et al., 2012; Nossoni et al., 2014; Silvaroli et al., 2016). Hydrophobic interactions between the ligand and amino acid sidechains that line the iLBP binding cavity are important for ligand binding (Thumser et al., 1996). The residues identified as part of the hydrophobic interaction network are shown for representative iLBPs in Fig. 3A. The importance of the hydrophobic interactions is also illustrated in the general observation that binding affinities with FABPs correlate with increasing hydrophobicity (Storch and Corsico, 2008; Smathers and Petersen, 2011). Ionic and hydrogen bonding interactions for ligand binding typically involve charged residues. The arginine residue in the FATTYACIDBP3 (Fig. 1A) is highly conserved in the iLBPs that bind acidic ligands (R122 in FABP1, R126 in FABP3 and FABP4, R129 in FABP5 and R132 in CRABP2) and is located on the βJ strand of these proteins (Fig. 3A). For CRBP1 that binds nonacidic ligands all-trans-retinol and all-trans-retinal, the Q128 appears to be the corresponding residue important for ligand binding (Silvaroli et al., 2016). For CRABP1 and CRABP2, the amino acids R132 and Y134 coordinate with the carboxylic acid of all-trans-retinoic acid (atRA) and R111 appears to coordinate with the carboxylic acid via an ordered water molecule (Fig. 3C) (Kleywegt et al., 1994). Analogous amino acids in some FABPs also coordinate with the carboxylate of fatty acids bound to FABPs (Hanhoff et al., 2002; Smathers and Petersen, 2011). However, these residues are not essential for ligand binding in all iLBPs. Mutations of the conserved arginine in the βJ strand confer different effects on ligand binding depending on the iLBP isoform and the ligand in question. A single R132A or R132Q mutation completely abolishes binding of atRA to CRABP2 (Chen et al., 1995). Similarly, an R126Q mutation on the analogous residue in FABP4 reduces the binding affinity for cis-parinaric acid by >10-fold (Sha et al., 1993). In contrast, an R122Q mutation in FABP1 only moderately decreases fatty acid binding and increases binding of bulkier ligands (Thumser et al., 1996). Charged or polar residues in the βH strand also interact with hydroxy and carbonyl head groups (Fig. 3A), except for subfamily II FABPs (FABP1 and FABP6), and likely contribute to ligand binding. The hydroxy group of all-trans-retinol interacts with Q108 in CRBP1 (Fig. 3A) and Q109 in CRBP2. Ligands appear to interact with the conserved residue R111 in CRABPs (Figs. 1A and 3A). In addition, ligands interact with R106 in FABP2, and this residue is also conserved in subfamily IV FABPs (Figs. 1A and 3A).

Fig. 3.

Binding characteristics of endogenous ligands with iLBPs. (A) The distribution of residues shown to interact with endogenous ligands for iLBPs are depicted along their structural features. Residues labeled in red font are involved in coordinating with the hydroxy or carbonyl groups of endogenous ligands via ionic interactions. (B) A top-down perspective into the atRA binding site of hCRABP2 (PDB 1CBS) with side chains that interact with atRA labeled. The position of R111, R132, and Y134 residues that coordinate with the carboxyl group of atRA are shown along with the position of R59 which interacts with the β-ionone ring. (C) Side view and positions of residues R111, R132, and Y134 are shown relative to the carboxylate of atRA along with hydrogen bonding interactions. (D) The amino acid sidechains that interact with atRA and form the atRA binding pocket in CRABP2 are shown. (Structures generated from PDB using ChimeraX; figure created with BioRender.com.)

The helix-turn-helix motif, in conjunction with nearby βC-βD and βE-βF loops, form the portal region of the iLBP that permits ligand entry and egress from the interior binding cavity (Fig. 2) (Vaezeslami et al., 2006; Storch and Corsico, 2008; Silvaroli et al., 2016). Different hypotheses describe the extent of the dynamics and flexibility of the portal region. Early observations from NMR solution structures of FABP2 showed disorder and flexibility in the portal region leading to the “dynamic portal hypothesis” (Hodsdon and Cistola, 1997a,b). This hypothesis suggests that the disordered portal region in the apo-protein could undergo large structural fluctuations to permit ligand entry but shifts to an ordered closed state upon ligand binding. Processes that destabilize the helical cap, such as interactions with cationic membranes, would then shift the protein toward the disordered state and, hence, facilitate ligand release. Later studies with FABP1 supported this hypothesis and showed the apo- and holo-protein structures to have an open and closed “helix cap”, respectively (He et al., 2007). However, the dynamic portal hypothesis is not sufficient to reconcile observations that some FABPs have similar structures between ligand bound and unbound forms (Vaezeslami et al., 2006; Gillilan et al., 2007; Cai et al., 2012).

Internal protein dynamics may also have a major role in influencing ligand accessibility as major fluctuations in the portal region are not observed in structural studies with retinoid binding proteins and some FABPs (Vaezeslami et al., 2006; Cai et al., 2012; Ragona et al., 2014). In CRABP2, the portal appears large enough to allow entry of all-trans-retinoic acid (atRA) in both the apo- and holo-structures with little change in the overall protein backbone (Vaezeslami et al., 2006). However, the sidechain of R59 (Fig. 3), which is located at the entry of the portal region (βC-βD loop) in the apo-protein, appears to rotate in the holo-protein to form stabilizing interactions with atRA (Fig. 3). An analogous residue (F57) in FABP4 appears to have a similar function and supports the importance of sidechain dynamics in the internal binding cavity. Structural studies with FABP4 suggest that locking the internalized ligand in the holo-protein is controlled by the F57 sidechain on the βC-βD loop that rotates into the binding cavity in the holo-conformation (Gillilan et al., 2007) despite little conformational change between apo- and holo-proteins. Indeed, ligand binding kinetics with the fluorescent ligand 8-anilinonaphthalene-1-sulfonic acid (ANS) are faster in an FABP4 portal mutant (V32G, F57G, K58G), which has an enlarged portal region, than with wildtype (WT) FABP4 (Jenkins et al., 2002).

Solution NMR studies show that the backbone dynamics in the portal region in apo-proteins vary between iLBP isoforms. FABP6 has a relatively rigid portal region while FABP1, 3, and 4 portal regions are more flexible, and the FABP2 portal is virtually disordered (Ragona et al., 2014). However, in general the changes in the backbone dynamics of FABPs upon ligand binding are consistent with disordered to ordered stabilizing interactions.

Endogenous Ligand Binding in iLBPs

The divergence of ligand specificity in iLBPs arises from nuanced differences in the β-barrel cavity and portal regions. Figure 4 shows representative structures of CRBP1, FABP1, FABP2, and FABP4 bound with their endogenous ligands illustrating general features of ligand binding with iLBPs. Historically protein fractionation, gel filtration, ion-exchange chromatography, and electrophoresis techniques were used to isolate and identify iLBPs in tissues and tissue homogenates, and bound ligands were identified from this isolated protein (Bashor et al., 1973; Ockner and Manning, 1974; Maatman et al., 1991; Veerkamp and Maatman, 1995). Subsequent characterization of ligand binding has been largely done with fluorescent probes or radiolabeled ligands in tissue homogenates or with purified recombinant iLBPs (MacDonald and Ong, 1987; Giguère et al., 1990; Nemecz et al., 1991; Sanquer and Gilchrest, 1994; Folli et al., 2001) and with x-ray crystallography and NMR (Kleywegt et al., 1994; LaLonde et al., 1994; Thompson et al., 1997; Hohoff et al., 1999).

Fig. 4.

Binding orientations of endogenous ligands in the binding cavity of intracellular lipid binding proteins for (A) hCRBP1 with all-trans-retinol (PDB 5H8T), (B) hFABP1 with two oleate molecules (PDB 2LKK), (C) rFABP2 with myristate (PDB 1ICM), and (D) mFABP4 with arachidonic acid (PDB 1ADL). (Structures generated from PDB using ChimeraX; figure created with BioRender.com.)

The known endogenous ligands of iLBPs are summarized in Table 1. The retinoid binding proteins appear to be selective toward vitamin A and its metabolites while all FABPs bind long chain fatty acids (LCFAs) (Table 1). Some FABPs also bind a variety of other endogenous ligands (Table 1). It is important to note that the list of known ligands is limited to those ligands that have been explicitly tested for their binding and may not be comprehensive for all endogenous ligands. Several studies have explored synthetic derivatives of the endogenous ligands of FABPs (Wang et al., 2016; Floresta et al., 2017), but these synthetic derivatives are not discussed in this review. Additionally, the summary here includes binding data from species other than human proteins, as many ligand binding studies with iLBPs were done with rat, mouse, and bovine recombinant protein.

Endogenous Ligands of Subfamily I

Vitamin A (retinol) or its biologically important metabolites retinaldehyde and the pharmacologically active atRA bind the proteins in subfamily I with high affinity. Notably, all proteins in this subfamily are intracellular, in contrast to the lipocalin retinol binding protein 4 (RBP4), which is the circulating carrier for retinol. all-trans-retinol, all-trans-retinaldehyde, and their 13-cis and 9-cis isomers bind to CRBP1 (Fig. 4A) and CRBP2 with nanomolar affinity. Yet, neither atRA nor its 13-cis and 9-cis-isomers bind to CRBPs (Kane et al., 2011; Napoli, 2016, 2017; Menozzi et al., 2017). all-trans-retinol binds to CRBP3 (Folli et al., 2001) and all-trans-retinol along with 13-cis and 9-cis retinol bind to CRBP4 (Vogel et al., 2001; Folli et al., 2002). Although CRBPs appear to be specific for retinol and retinal ligands, monoacylglycerols were also recently shown to bind to CRBPs (Lee et al., 2020). This suggests CRBPs and CRABPs may have broader ligand specificity than previously assumed. atRA and its isomers and metabolites bind specifically to CRABP1 and CRABP2 with atRA having higher binding affinity toward CRABP1 and CRABP2 (Kd = 0.4–39 nM) (Fiorella and Napoli, 1991; Fogh et al., 1993; Norris et al., 1994; Wang et al., 1997) than 9-cis-RA (Kd = 51–69 nM) (Norris et al., 1994) or 13-cis-RA (Kd = 156–238 nM) (Fiorella et al., 1993). Generally, atRA appears to bind slightly tighter to CRABP1 than CRABP2 (Fiorella et al., 1993; Dong et al., 1999; Yabut and Isoherranen, 2022). Retinol or retinal isomers do not bind to CRABP1 or CRABP2 (Fiorella and Napoli, 1991; Fiorella et al., 1993; Napoli, 2017).

Endogenous Ligands of Subfamily II

FABP1 and FABP6 make up subfamily II. Generally, bulky ligands in addition to LCFA bind to FABP1 and FABP6 (Smathers and Petersen, 2011). The binding pockets of FABP1 and FABP6 are larger (≥639 and 460 Å3, respectively) (Lücke et al., 2000) than other FABPs that have solvent accessible binding pockets of approximately 230 to 330 Å3 (Smathers and Petersen, 2011). FABP1 and FABP6 can accommodate larger ligands found in the liver such as bile acids, cholesterol, bilirubin, and heme (Bernlohr et al., 1997; Smathers and Petersen, 2011). Other ligands of FABP1 include branched fatty acids, endocannabinoids, acyl-CoA, prostaglandins, and vitamin K (Khan and Sorof, 1990; Thumser and Wilton, 1996; Martin et al., 2003; Storch and Corsico, 2008; Atshaves et al., 2010; Smathers and Petersen, 2011). While fatty acid ligands appear to bind to all other FABPs in a 1:1 ratio, two fatty acids can bind to FABP1 simultaneously (Fig. 4B) (Bernlohr et al., 1997; Thompson et al., 1997; Cai et al., 2012). FABP1 has a high-affinity fatty acid binding site (Kd = 4–60 nM) located deep within its interior cavity and a low affinity site (0.3–12 μM) closer in proximity to the alpha-helical domain and opening of the portal region (Fig. 4B) (Atshaves et al., 2010; Smathers and Petersen, 2011; Cai et al., 2012). With larger ligands such as bile acids, this stoichiometry appears to be reduced (1:1) along with reduced binding affinities (Kd 4–50 μM) (Richieri et al., 1995). FABP6 is structurally similar to FABP1, but due to differences in interior amino acid side chains between the two proteins, preferential ligands of FABP6 include bile acids over LCFAs (Lücke et al., 2000). Due to the size of bile acids, only a single ligand is typically observed in the FABP6 binding cavity.

Endogenous Ligands of Subfamily III

FABP2 is the sole member of subfamily III. In contrast to the iLBPs in subfamily II, FABP2 has a small solvent accessible binding pocket (234 Å3) (Smathers and Petersen, 2011), and its preferential ligands include saturated LCFAs (Fig. 4C) (Lowe et al., 1987; Richieri et al., 1994; Velkov et al., 2005; Smathers and Petersen, 2011). Measured fatty acid binding affinities with FABP2 range between 0.02 and 1.5 µM based on fluorescence displacement assays (Nemecz et al., 1991; Velkov et al., 2005, 2007).

Endogenous Ligands of Subfamily IV

The seven members of subfamily IV collectively bind diverse lipids. The size of the solvent accessible binding pockets of subfamily IV FABPs appear to be larger than subfamily III (FABP2) but smaller than subfamily II (FABP1 and FABP6). FABP3, 4 and 8 have 323, 310, and 330 Å3 binding pockets, respectively (Smathers and Petersen, 2011). Saturated and unsaturated fatty acids bind to FABP3 with nanomolar affinity, and oxygenated fatty acids (epoxyeicosatrienoic acid, hydroxyeicosatetraenoic acid, dihydroxyeicosatrienoic acid) bind to FABP3 with Kd values from 0.4 to 14 µM (Widstrom et al., 2001; Smathers and Petersen, 2011). FABP4 appears to be more ligand selective, and only LCFAs bind to FABP4 with nanomolar affinity (Kd = 22–196 nM) (Fig. 4D) (Richieri et al., 1994; Gillilan et al., 2007; Storch and Corsico, 2008; Smathers and Petersen, 2011). However, other ligands such as atRA also bind to FABP4 but with a considerably lower binding affinity (Kd = 50 µM) (Matarese and Bernlohr, 1988; Veerkamp et al., 1999).

With FABP5, stearic acid and docosahexaenoic acid have nanomolar affinity to FABP5 (Kd = 0.17–0.29 and 0.16 µM, respectively) and oleic acid, lauric acid, and arachidonic acid binding affinity range from nanomolar to micromolar (Kd = 0.15–1.6, 2.5 and 0.12–1.7 µM, respectively) (Hohoff et al., 1999; Smathers and Petersen, 2011; Kaczocha et al., 2012; Pan et al., 2015; Lee et al., 2018). atRA has also been reported to bind to FABP5 in fluorescence displacement assays with ANS (Kd = 35 nM) (Schug et al., 2007). However, FABP5 did not to sequester atRA from metabolism by cytochrome P450 (CYP) enzymes, suggesting binding may not be as tight as suggested by the displacement assay (Yabut and Isoherranen, 2022). FABP7 prefers polyunsaturated fatty acids with longer chains (docosahexaenoic acid, eicosapentaenoic acid, arachidonic acid), and these fatty acids bind to FABP7 with affinities ranging from 27 to 250 nM (Smathers and Petersen, 2011).

In addition to the fatty acid ligands of FABPs, FABP3, FABP5, and FABP7 have also been shown to bind the endocannabinoids 2-archidonylglycerol and anandamide (AEA), and FABPs have been proposed to have a role in modulating endocannabinoid metabolism and signaling. 2-archidonylglycerol and AEA bind to FABP7 with higher affinity (Kd = 0.2 and 0.8 µM, respectively) than to FABP3 (Kd = 1.63 and 3.07 µM, respectively) and to FABP5 (Kd = 1.45 and 1.26 µM, respectively) (Kaczocha et al., 2012; Elmes et al., 2015). FABP8, 9, and 12 have not been extensively studied, and the binding of their endogenous ligands is not well characterized (Storch and Corsico, 2008; Smathers and Petersen, 2011).

Tissue Distribution and Expression of iLBPs

The tissue distribution of iLBPs is broad and expression patterns have been studied in several mammalian species including rat, mice, pig, and human (Paulussen et al., 1988, 1990; Gong et al., 1994; Sanquer and Gilchrest, 1994). However, species differences in tissue expression have not been comprehensively compared for all iLBPs. The following is a summary of the tissue expression of iLBPs in adult mammals determined using a combination of techniques including western and northern blot analysis, immunohistochemistry, enzyme-linked immunosorbent assay, reverse transcriptase-quantitative polymerase chain reaction, and binding assays with radiolabeled ligands. Some iLBPs are expressed in multiple tissues while others are expressed in specific tissues and cell types that may be indicative of specialized biologic functions (Storch and Corsico, 2008). The expression pattern of the FABPs is sometimes evident from the original name of the FABP, as FABPs were named after the tissues from where they were first identified. However, multiple FABPs are often expressed in the same tissues, and the expression patterns are typically broader than what is implied from the original names of the FABPs. Hence, early studies identifying FABPs in tissues often required confirming the specificity of antisera against multiple FABPs (Paulussen et al., 1990; Maatman et al., 1991; Gong et al., 1994).

Although the iLBPs are generally considered to be intracellular, FABP1, 2, 3, 4, and 5 have also been measured in plasma in humans (0.3–13 ng/mL) (Pelsers et al., 2003; Ishimura et al., 2013). Yet their concentrations are much lower than other circulating proteins such as albumin that bind fatty acids in plasma, and the importance of the circulating FABPs is unknown. FABP4 is the only isoform shown to be secreted from tissues (adipose) into circulation (Hotamisligil and Bernlohr, 2015; Shrestha et al., 2018; Villeneuve et al., 2018). For this review only CRABPs and those FABPs that xenobiotics have been shown to bind to are discussed, but the tissue expression for all iLBPs is summarized in Table 1.

CRABP1 protein is found in various tissues including liver, kidney, stomach, lymph, eye, and brain, but it is most abundant in skin and reproductive tissues (seminal vesicles, vas deferens, and testis) (Kato et al., 1985). CRABP2 protein expression appears to be limited to skin (Giguère et al., 1990).

FABP1, or liver FABP, is the major FABP in the liver and the intestine but is also found in the kidney, lung, pancreas, and stomach (Besnard et al., 2002; Pelsers et al., 2003; Gajda and Storch, 2015; Wang et al., 2015). FABP1 is most abundant in the liver and comprises 2% to 11% of all cytosolic protein in the liver (Wang et al., 2015; Schroeder et al., 2016). Expression of FABP1 in the liver is zonal, possibly indicating a unique role in specific areas of the liver (Bass et al., 1989). Peroxisome proliferators, female sex steroids, retinoids, and a diet high in fat increase the expression of FABP1 messenger RNA (mRNA) and protein in the liver (Poirier et al., 1997; Hung et al., 2003; Trevaskis et al., 2011; Velkov, 2013). Interestingly, FABP1 mRNA and protein expression are decreased after dexamethasone treatment, likely due to altered lipid metabolism and concentrations (Foucaud et al., 1998). In the gut, FABP1 mRNA is expressed throughout the length of the small intestines but is highest in the duodenum and jejunum (Agellon et al., 2002; Gajda and Storch, 2015). The expression pattern of FABP1 in the liver and intestines suggests FABP1 may also impact drug metabolism in the liver and drug absorption in the intestines. Additionally, FABP1 expression and function may have a role in metabolic disease progression as FABP1 polymorphisms in humans are associated with dyslipidemia, nonalcoholic fatty liver disease, and hepatocellular carcinoma (Peng et al., 2012; Schroeder et al., 2016; McKillop et al., 2019; Valizadeh et al., 2021). For example, the T94A mutation (allele frequency 26%–38%) in FABP1 alters FABP1 expression, ligand binding characteristics, protein structure and stability, and protein function (Schroeder et al., 2016). The T94A single nucleotide polymorphism is associated with elevated triglycerides, low-density lipoprotein cholesterol, and altered response to fenofibrate (Schroeder et al., 2016).

FABP2, also called intestinal FABP, is solely expressed in the gut, and its expression appears to be similar to FABP1 in rodent intestine but lower than FABP1 in human intestine. FABP2 mRNA is expressed throughout the length of the small intestine, and its expression is highest in the jejunum (Sacchettini et al., 1990; Gajda and Storch, 2015). Along with FABP1, FABP2 expression is highest in the villi of enterocytes, and it is not expressed in the crypt. FABP2 expression in enterocytes may be regulated by the gut peptide tyrosine tyrosine (Halldén and Aponte, 1997). FABP2 expression appears to be diffused throughout enterocytes but localized to the apical side in a fasted state (Alpers et al., 2000). Similar to FABP1, an A54T polymorphism in FABP2 appears to be associated with dyslipidemia, insulin resistance, obesity, and cardiovascular disease and may increase the risk of colorectal cancer (McKillop et al., 2019; Huang et al., 2022). FABP2 has been proposed as a potential biomarker for disruption of intestinal epithelial integrity as FABP2 is released to circulation when intestinal epithelium is compromised (Huang et al., 2022).

FABP3, or heart FABP, protein has been found in the heart, skeletal muscle, brain, kidney, liver, lung, spleen, and placenta (Paulussen et al., 1990). FABP3 is most abundant in the heart where its expression is nearly twofold greater than in skeletal muscle. Protein abundance in the kidney and brain is about half of that in the muscle and even less in the liver and placenta. FABP3 is also found to circulate at elevated levels in plasma in response to myocardial injury, presumably due to release from the heart. As such, it may be a potential biomarker for cardiovascular disease (Pelsers et al., 2005). In the kidney, FABP3 is found to be expressed in the distal and proximal convoluted tubules (Maatman et al., 1991), suggesting FABP3 could play a role in renal handling of drugs and xenobiotics.

FABP4, known as adipocyte FABP, is abundantly expressed in adipose tissue and is also the major FABP found in macrophages (Pelton et al., 1999; Furuhashi and Hotamisligil, 2008). FABP4 is the most abundant FABP in circulation (Ishimura et al., 2013) and is secreted from adipocytes via a membrane-bound pathway independent of the canonical endoplasmic reticulum-Golgi-plasma membrane secretion pathway (Villeneuve et al., 2018). Secreted FABP4 may serve as an adipokine, and lipolysis increases secretion of FABP4 from adipocytes (Furuhashi et al., 2015). Exogenous FABP4 influences hepatocyte glucose production, insulin secretion by pancreatic β cells, and cellular functions of cardiomyocytes and smooth muscle cells (Furuhashi et al., 2015). Indeed, circulating FABP4 levels are associated with the development of insulin resistance, diabetes, atherosclerosis, cardiac dysfunction, and inflammation (Furuhashi and Hotamisligil, 2008; Ishimura et al., 2013; Furuhashi et al., 2015; Saito et al., 2021). Reduced FABP4 appears to reduce the risk of metabolic and cardiovascular disease (Hotamisligil et al., 1996; Furuhashi and Hotamisligil, 2008; Furuhashi et al., 2015), and, hence, FABP4 has been explored as a potential therapeutic target (Floresta et al., 2017). Due to its small size, FABP4 found in circulation is subject to glomerular filtration, but it accumulates in the kidney via megalin-mediated reabsorption from the tubular lumen (Shrestha et al., 2018). Notably, circulating FABP4 levels also showed a sex difference with females having higher concentrations than males (Ishimura et al., 2013).

FABP5, epidermal FABP, is the major FABP found in the epidermis, but FABP5 tissue expression is broad and not restricted to the sk