Pharmacological correction of misfolding of ABC proteins.

The endoplasmic reticulum (ER) quality control system distinguishes between correctly and incorrectly folded proteins to prevent processing of aberrantly folded conformations along the secretory pathway. Non-synonymous mutations can lead to misfolding of ABC proteins and associated disease phenotypes. Specific phenotypes may at least partially be corrected by small molecules, so-called pharmacological chaperones. Screening for folding correctors is expected to open an avenue for treatment of diseases such as cystic fibrosis and intrahepatic cholestasis.

Introduction

About one-third of the eukaryotic genome encodes for proteins that are processed along the secretory pathway. The endoplasmic reticulum quality control system ascertains correct folding as a prerequisite for coat protein complex II (COPII) dependent ER-export. Equating proper folding with correct function allows the ER to check thousands of proteins for proper assembly with a limited set of auxiliary proteins. Nevertheless, function of ABC proteins is often only modestly reduced in misfolded mutants relative to wild type. Even a minor fraction of wild type activity often suffices to alleviate disease related symptoms. The pharmacological chaperone concept proposes folding correction by small molecules to trigger anterograde trafficking of proteins to their final cellular destinations. It also relies on the assumption that the residual function of the proteins suffices to suppress a disease phenotype. Here, development in cell-based high-throughput screening techniques is discussed with respect to correction of ABC protein folding.

Aberrant folding of ABC proteins and human disease

ER-associated degradation (ERAD) of mutant ABC proteins results in several potentially lethal or debilitating human diseases. The available evidence indicates misfolding and premature degradation caused by missense mutations as a significant cause of membrane protein deficiencies [1]. Disease causing mutations in several human ABC proteins, including among others ABCA1 (Tangier disease), ABCB4 (progressive familial intrahepatic cholestasis type 3), ABCB11 (progressive familial intrahepatic cholestasis type 2), ABCC2 (Dubin-Johnson syndrome), ABCC7 (cystic fibrosis), ABCC8 (hyperinsulinemic hypoglycemia of infancy) and ABCG2 (gout), have been linked to aberrant folding, retrotranslocation of proteins into the cytoplasm and subsequent proteasomal degradation [1].

Architecture of ABC-proteins, misfolding and correction

The minimum functional form of ABC-exporters comprises four domains. Two membrane spanning (transmembrane) domains confer solute (substrate) specificity, while two nucleotide binding domains, which bind and hydrolyze ATP in two composite nucleotide binding sites, provide the energy for the movement of solutes across membranes. An analogous architecture can be observed in all presently available ABC exporter crystal structures. The folding process may be considered a stepwise process, in which (sub)domain folding precedes correct positioning of domains or subdomains relative to each other. Impairment of either of these processes can result in the absence of functional protein [2]. A crucial point in adopting the native conformation is a partial domain-swap in the architecture of ABC exporters, which requires engagement of the second intracellular loop (ICL) of each transmembrane domain (TMD) (ICL 2 and 2′ in half transporters and ICLs 2 and 4 in full transporters) into a socket of the contralateral nucleotide binding domain (NBD) via so-called coupling helices. Thereby coupling helices insert between the core- and the α-helical subdomains of each NBD. Charge interactions in conserved positions of the NBDs and ICLs stabilize this conformation. Figure 1a illustrates the architecture in a schematic way. Note that only the front side of the protein is visible, but the process has to proceed in an identical manner on the back side also. In ABCB1, a multispecific drug efflux transporter, this process of contralateral engagement of ICLs 2/4 contributes to the formation of two pseudosymmetric solute binding sites, which are formed by contributions from residues located in the membrane spanning portion of helix-pairs 5/8 and 2/11 respectively [3] (one of these is shown in Fig. 1d). Successful folding has been shown to require both, correct folding of the NBD socket and ICLs, and proper formation of the NBD-TMD interfaces. If either of these processes – domain interface formation (Fig. 1b) or domain folding (Fig. 1c) – fails, the protein misfolds. Active site compounds can correct folding deficient mutants of ABCB1 (unpublished) (Fig. 1e). Despite the presence of a transmission-interface destabilizing missense mutation, the protein adopts a functionally active conformation. For the rescue of misfolded (sub)domains a corrector might be required to bind to the misfolded domain directly (Fig. 1f). Circumstantial evidence for binding to an exosite (non-active site) has been presented in a recent publication on ABCC7, in which direct binding of the corrector compound VX-809 to the misfolded NBD of the delta F508 mutant has been postulated by docking [4].

Figure 1

Schematic representation of ABC protein architecture, misfolding and correction with pharmacochaperones. (a) Architecture of folded protein. The two halves of a full transporter are shown in light and dark blue. TMDs and ICLs are shown as rectangles, NDBs as circles. C and α refer to the core (F1 type) and the α-helical subdomains of NBD, respectively. Parts of the TMD-NBD interface are formed by engagement of ICL2 (ICL4) into the socket between C and α subdomain of the respective contralateral NBD. (b) Misfolded protein – incorrect formation of TMD-NBD interface. (c) Misfolded protein – misfolding of one NBD. (d) Position of one substrate binding site in the TMD of correctly folded protein (orange circle). (e) Correction of misfolded domain interface by pharmacochaperone (yellow circle) bound to substrate binding site (orange). (f) Correction of misfolded protein with one disrupted NBD. Rescue might require direct binding of corrector molecule (green circle) to exosite of NBD (blue circle). Alternatively, an active site compound might also correct folding, by providing the scaffold of a correctly positioned ICL 2/4 to the respective NBD. NBD, nucleotide binding domain; TMD, transmembrane domain; ICL, intracellular loop.

Circumvention of potential functional inhibition by corrector molecules

Figure 2a shows a misfolded ABC protein destined for proteasomal degradation. The presence of a corrector allows its trafficking to the plasma membrane. At the same time, correctors can potentially inhibit function of the protein, for example by occupying the active sites. The half-life of an ABC transporter is determined by ER-export rates on one hand and internalization/recycling/lysosomal degradation on the other hand. An intermittent treatment regimen with correctors would, however, allow a potential inhibitory action of compounds to be released and the protein to become active, if compound elimination were shorter than the resident time of the transporter at its correct cellular membrane localization (Fig. 2b). Activity of the protein would be expected for non-overlapping portions of the blue and red curves. These gray-shaded areas refer to time intervals, during which drug levels decrease, but protein is retained at its site of action (Fig. 2b).

Figure 2

(a) Pathways for folded and misfolded ABC proteins. (b) Schematic representation of the dynamics (change in time) in protein amount (at correct cellular localization, red curve) and pharmacochaperone plasma level (blue). Activity can be expected for non-overlapping areas (shaded in gray).

Is there a proof of principle for folding correction by pharmacological chaperones?

A proof of principle for use of small molecule correctors has been established in vivo by recent approval of a corrector for the treatment of phenylketonuria (sapropterin (tetrahydrobiopterin) is the cofactor for the enzyme phenylalanine-hydroxylase) [5]. A small molecule trafficking rescue approach is also actively pursued for cystic fibrosis, the most frequent lethal inherited disease, in which an airway epithelial chloride channel, CFTR (cystic fibrosis transmembrane conductance regulator; ABCC7) is impaired. Screening programs have been initiated in academic/industrial joint ventures [6,7]. Two corrector molecules have entered clinical studies. For one of the compounds, VX-809 (Fig. 3a) a phase II clinical study has been concluded in 2012. A phase III study will begin in 2013, in which this corrector is combined with the recently approved functional potentiator compound VX-770. A second compound, VX-661 (Fig. 3b) is presently evaluated in a clinical phase II study (CF foundation: http://www.cff.org/research/ClinicalResearch/). Successful mutation-specific chaperone therapy with 4-phenylbutyrate has recently been reported as proof of principle in a child with progressive familial intrahepatic cholestasis type 2 [8].

Figure 3

Chemical structures of two corrector molecules which entered clinical studies: (a) VX-809 and (b) VX-661.

Which ABC protein related diseases and which mutations should be used for screening of pharmacochaperones?

A search for correctors of ABC transporter misfolding is primarily indicated for diseases devoid of treatment alternatives. These include the most frequent lethal inherited disease cystic fibrosis, pseudoxanthoma elasticum, gout, intrahepatic cholestasis and liver disease. The most recent addition to successful folding-correction is congenital hyperinsulinism. In vitro data suggest that for a subset of sulfonylurea receptor 1 (SUR1) mutations potassium channel activity can be corrected by carbamazepine [9].

Screens are usually conducted with mutations that are frequent in populations. The ΔF508 mutation in ABCC7 is present in at least one allele in 90% of cystic fibrosis patients [10] while the E297G and D482G mutations in ABCB11 are found (individually or in combination) in close to 60% of European families affected by progressive familial intrahepatic cholestasis type II [11]. These frequent mutations represent logical starting points for screening efforts.

Screening techniques used in the identification of folding correctors

To ascertain correct cellular localization, screening for pharmacological chaperones of membrane proteins requires cell-based assays. Trafficking correction is only beneficial in cases, in which protein function is at least partially retained. Two different types of assays have been described: (i) those that yield information on protein localization, or (ii) those that probe protein function as a reflection of both, correct cellular localization and residual functional activity. The latter represents the preferable option for mutations in which stronger functional impairment is expected. Also, combinations of correctors and potentiators (compounds that activate ABC-proteins) can be assessed.

Most of the available literature on ABC proteins is on the cystic fibrosis transmembrane conductance regulator (ABCC7), which was and is the most extensively studied ABC protein both in academic, industrial and joint academic/industrial ventures. Several reviews on available assay technologies have appeared in print [6,12]. ABCG2 has recently been identified as a major uric acid efflux transporter in renal tubular epithelial cells [13]. Hyperuricemia and gout are linked to a frequent folding deficient single nucleotide polymorphism (Q141K), which leads to a folding defect in the encoded protein and is found at particularly high frequency in the Japanese population. Functional assays of ABCG2 have been reviewed [14]. Techniques described there include vesicular ATPase and transport assays, which can be used for functional assays of ABC proteins, but not all of them have the potential to be adapted to a high-throughput setting for screening of folding correction. Detection of the fully glycosylated mature form of ABC proteins in polyacrylamide gels by Western blotting [9] unfortunately is not easily adapted to the setting of high-throughput screening, although it represents a gold standard for further characterization of hits.

High- or ultra-high-throughput screening technologies afford the identification of hits from compound libraries within the scope of days. In this review we intend to summarize screening approaches for correctors of folding deficient ABC proteins with a focus on high-throughput technologies. Other functional assays that require a more extensive pre-screening effort, such as purification of vesicles or protein, are not subject of the current review.

Assays for detection of protein localization

Misfolding affects protein processing in the endoplasmic reticulum and directs proteins to ERAD. Thus retention of a plasma membrane protein in the ER is indicative of a protein folding defect. The following three approaches, which are depicted in a schematic way in Fig. 4a are commonly used.

Figure 4

Synopsis of ABC protein assays used for screening. (a) Assays for detection of protein localization. From left to right: fluorescent tags: GFP and mCherry in combination with Flag-tag; fluorescent antibodies: antibodies against a genuine extracellular epitope of an ABC protein, or an extracellular HA-tag; fluorogen-activating protein (FAP) (introduced as an extracellular tag of an ABC protein). (b) Functional assays. From left to right: detection of chloride transport of ABCC7 based on the substrate-sensitive fluorescence of coexpressed YFP protein and a voltage-sensitive fluorescent dye; direct detection of fluorescent substrate of the ABC protein (ABCA1 and ABCB1 are given as examples); and simultaneous observation of substrate and GFP-labeled ABC protein. The detection techniques (confocal microscopy, microplate reader, and flow cytometry) are indicated for each approach according to information in the literature as cited in the text. Colors refer to the respective cellular localization of the fluorescent dye (intracellular, or cell membrane).

Fluorescent tags

Fluorescence microscopy with confocal optics is broadly used for localization studies of membrane proteins. Almost 20 years ago GFP-tags have been introduced to trace cellular localization and routing of proteins of interest to different cellular compartments. GFP-tagging has proven useful in confocal microscopy studies of several ABC proteins including ABCG2 [15]. Obviously, modest variations in the nature of the fluorescent proteins do not have any impact, for example, a YFP tag was used for ABCA3, a transporter involved in surfactant production in alveolar type II cells [16]. Early on, the potential for scale-up has been realized and high throughput approaches have been developed that use fully automated fluorescent microscopy imaging [17]. So-called high content microscopy-based assays, which assess a cell phenotype by image analysis, were applied to study trafficking of the CFTR protein [7]. In these studies authors used mCherry and a FLAG-epitope tag on the same protein. This allowed quantification of cell surface expression with an anti FLAG antibody and comparison with the total amount of CFRT in cells as estimated by the mCherry tag (Fig. 4a).

Though a GFP-tag proved useful for a range of ABC proteins, it failed in case of ABCA4, because it caused mislocalization of the protein. Therefore, visualization had to rely on an anti ABCA4 antibody and a fluorescent secondary antibody [18].

Antibodies

The on-cell western (life cell western) assay represents a robust approach, which uses fluorescent antibodies that recognize epitopes exposed to the cell exterior, in combination with multi-well screening. The fluorescence of proteins at the cell surface can be verified without a requirement for microscopic imaging. However, the technique relies on the availability of antibodies directed against extracellular loops of ABC proteins.

The use of anti ABCC7 antibodies has been reviewed [12,19]. Similar studies have been conducted with ABCA4 [18]. The use of specific cell surface antibodies for ABCB1 and ABCG2 has also been reviewed [14].

In many instances antibodies against extracellular epitopes are not available. This limitation can be overcome by insertion of the small hemaglutinin (HA) epitope into tolerant regions of the protein sequence. This strategy has proven feasible in the analysis of CFTR with on-cell western assays to determine CFTR turnover with a fluorescent secondary antibody [20]. Alternatively, the HA-tag in CFTR can be visualized with an HRP-conjugated primary antibody [12]. The HA-tag in combination with a secondary fluorescent antibody has also successfully been applied for ABC proteins from other ABC subfamilies including ABCA3 [16]. In the latter study, the cellular localization of the transporter was visualized by confocal microscopy.

Fluorogen-activating proteins

An interesting approach is the use of fluorogen-activating proteins (FAPs), which has recently been applied to study the efficiency of compounds in rescuing trafficking of CFTR mutant ΔF508 [21]. FAPs are tags that are fused to proteins so that they are exposed to the cell exterior. Upon binding, these tags increase fluorescence of a fluorogen that is present in solution. The authors used confocal microscopy to visualize the presence of protein at the plasma membrane and suggest that the approach can be adapted to multi-well plate screening.

Approaches using tags and fluorescence detection can in principle be applied to all ABC proteins, provided that the tags themselves do not interfere with trafficking of the targeted proteins.

Functional assays

Even if folding deficiencies are rescued by pharmacological agents, and protein localization is corrected, the transport activity of the protein may still be compromised to variable extent. Functional assays offer the advantage that they verify both the ability of a small molecule corrector to promote ER-export of the protein and functional activity at the correct cellular destination.

Functional assays are based on direct detection of fluorescent substrates, or fluorescent sensors reporting translocation of substrates of ABC proteins (Fig. 4b).

Halide-sensitive fluorescent probe

This approach utilizes a halide sensitive YFP variant to estimate chloride conductance of CFTR. As this approach is limited to the detection of halides, it can only be used to study CFTR. In the presence of chloride ions the probe fluoresces. Upon changing chloride to iodide containing medium, iodide replaces chloride leading to quenching of YFP [22-24]. High throughput screening of small molecules for CFTR ΔF508 trafficking correction has been performed by coexpression of CFTR and halide-sensitive YFP.

Detection of a membrane potential change

Similar to the above described method, this technique is also only applicable to CFTR. Cl− efflux leads to a change in membrane potential, which can be detected by fluorescent voltage sensitive probes [25].

Fluorescent transport substrates

If available, ABC transporters can be studied by using fluorescent substrates. One possible way of detecting fluorescent substrates is by means of flow cytometry. This technique is widely used for the evaluation of the activity of surface expressed ABC transporters and detects transport rates as a surrogate parameter for protein localized at the cell membrane. Table 1 lists examples of substrates of ABC proteins that can be used for functional tracking of folding deficient mutations. The substrates for multidrug transporting proteins have been reviewed elsewhere [26,27].

Table 1

Examples of fluorescent substrates and techniques used to investigate transport activity of ABC proteins

ABC protein	Disease	Substrate (natural)	Substrate in experiment	Technology
ABCA1 (CERP)	TGD	Cholesterol [28]	Pennsylvania Green/N-alkyl-3β-cholesterylamine-derived probe (F-Ch) [29]	Microplate reader, HTS [29]
ABCB1 (Pgp1)	MDR	Large hydrophobic molecules, anticancer drugs, HIV protease inhibitor [26]; reviewed in [27]	Rhodamine 123 [30]	Confocal microscopy [30]
			Calcein acetoxymethyl ester (CaAM) [31], JC1 [31,32]	Flow cytometry, HTS [31,32]
			eFluxx-ID Green, eFluxx-ID Gold [33,34]	Flow cytometry [33,34]
			Reviewed in [26]
ABCB11 (BSEP)	PFIC-2	Bile salts	3α- and 3β-NBD-UDCA [35]	Microplate reader [35]
ABCC2 (MRP2)	DJS;MDR	Organic ions [28];Reviewed in [36]	Lopinavir (LPV), calcein, carboxyfluorescein diacetate [37]
			Glutathione-Methylfluorescein [38], Glutathione-Monochlorobimane [38,39]	Microplate reader [38,39]
ABCC6 (MRP6)	PXE; MDR	Drugs, organic ions;Reviewed in [40]	Glutathione conjugate of N-ethylmaleimide (NEM-GS) [41], leukotriene C4 (LTC4) [41,42]; reviewed in [40]	Vesicular assays, radioactivity [41,42]
ABCG2 (BCRP)	Gout; MDR	Anionic compounds and hydrophobic drugs [26]; reviewed in [27]	Mitoxantrone (MX) [43]	Confocal microscopy [43]
			JC1 [31]	Flow cytometry, HTS [31]
			eFluxx-ID Green, eFluxx-ID Gold [34]	Flow cytometry [34]
			Reviewed in [26]

Abbreviations: TGD, Tangier disease; MDR, multi drug resistance; PFIC-2, progressive familial intrahepatic cholestasis type 2; DJS, Dubin-Johnson syndrome; PXE, pseudo-xanthoma elasticum; HTS, high throughput screen; JC1, J-aggregate-forming lipophilic cation 5,5,6,6-tetrachloro-1,1,3,3-tetraethylbenzimidazolcarbocyanine iodide; NBD, 4-nitrobenzo-2-oxa-1,3-diazol; UDCA, ursodeoxycholic acid.

Commonly used alternative designations of ABC proteins are given in parenthesis next to the systematic classification identifier.

Flow cytometry has previously been employed for HTS assays not only of multidrug resistance proteins – ABCB1 [31,32], ABCG2, and ABCC1 [31], but also of ABCB6, which recognizes protoporphyrin as a transport substrate [44].

Alternatively, microplate-based HTS platforms have been used to detect the transport of fluorescent substrates by multidrug transporters [45], and also for monitoring flux of a fluorescently labeled cholesterol analog by ABCA1 [29].

Yet another method used to estimate the transport of fluorescent substrates is confocal microscopy. This approach allows for simultaneous recording of cellular Hoechst 33342 or mitoxantrone accumulation and cellular localization of GFP-ABCG2 [15,43]. It was also employed to document transport of nitrobenzoxadiazole-labeled lipids and the localization of ABCA3-DsRed [16].

Conclusions

A vast number of causative mutations have been described for the at least 25 different disease entities associated with ABC-transporter malfunction. However, only recently the importance of protein misfolding due to missense mutation has been appreciated as a significant mechanism in the etiology of ABC transporter-linked diseases beyond the well established example of the inherited disease of cystic fibrosis. The mutations, which cause folding deficiency of different ABC transporters, cluster at hot spots indicative of structural conservation (unpublished data). Small molecule correctors are able to counteract the destabilizing effect of mutations by achieving sufficient thermodynamic stabilization of otherwise misfolded proteins. This allows them to pass the quality control checkpoint of the ER and commence their journey to their final cellular destinations. The pharmacochaperone concept posits residual functional activity to be associated with proteins, even when having lower thermodynamic stability.

Extensive efforts have been directed toward identifying corrector molecules for the treatment of the most frequent lethal inherited disease, cystic fibrosis. Nonetheless misfolding has increasingly been appreciated as a more common cause of ABC-transporter associated diseases including, among others, severe liver diseases.

With respect to screening platforms, whole cell based assays are required for the assessment of corrector action. Available techniques can coarsely be classified into either localization assays or functional assays. The latter have the potential advantage that they not only assess correct protein localization, but also confirm that a functional protein has been delivered to the correct cellular destinations. Hybrid assays that allow simultaneous detection of function and localization have also been described. Microscopy, flow cytometry and microplate assays represent alternatives, in which each has its individual merits. Speed, robustness and amount of information provided by the respective assay system have to be weighed against each other. It may be expected that the refined and rapid assay technologies for CFTR corrector action in high and ultra high throughput screening platforms will be matched by the scale up of technologies for other ABC proteins. Some of these presently rely on more tedious systems, including indirect methods.

It should be kept in mind that mutations and correctors probably form dedicated pairs and that only a subset of known mutations will be amenable to corrector action. It will also be necessary to clarify if an activity ceiling, as described for cystic fibrosis correctors, also exists for other ABC-transporter mutants.

Further development and refinement of assay systems will lay a foundation for the identification and clinical development of correctors, primarily for those severe diseases that are caused by frequent natural variants of ABC-transporters. Expectations are high so that this development will make those diseases, for which treatment options presently do not exist, amenable to therapeutic intervention in the near future.

Conflict of interests

The authors declare that they have no conflict of interest.