Assay Development and HTS Flow Cytometry
We developed an HTS flow cytometry assay to select functional modulators of ABCB6. The assay takes advantage of the unique ability of ABCB6 to regulate cellular porphyrin biosynthesis 
. Cells over expressing ABCB6 exhibit increased porphyrin biosynthesis, which directly correlates with the level of ABCB6 expression ( and 
). Furthermore, the increased cellular porphyrin synthesis requires a functional protein, because a non-functional transport mutant of ABCB6 is incapable of promoting porphyrin biosynthesis 
. Accumulation of porphyrin in cells results in increased cell associated fluorescence intensity (). Cellular porphyrin representative of porphyrin levels was confirmed using succinylacetone (SA) as described 
. SA is a potent inhibitor of cellular porphyrin biosynthesis and we have previously demonstrated that SA at a concentration of 200 µM completely blocks cellular porphyrin biosynthesis in less than 24 hours 
Assay development at University of New Mexico Center for Molecular Discovery (UNMCMD).
Feasibility of the HTS assay to identify compounds that functionally alter ABCB6 mediated porphyrin biosynthesis was evaluated using the PCL 2000 library. Although the central idea was to evaluate compounds that decreased the median channel fluorescent (MCF) intensity by FL2 (, Fig. S5a, S5b and S5c
), the assay also identified compounds that increased the MCF intensity (Fig. S5c
) suggesting that the assay is capable of identifying both inhibitors and activators of ABCB6. The controls were K562 cells overexpressing ABCB6 treated with DMSO alone (negative control). Our analysis showed a tight data distribution with Z’ factors averaging 0.61+/−0.08 and a hit rate of 0.4%. Calculations of percentage response were set to 100% at the positive control level and 0% at the negative control level. Cutoffs were set at three times higher than the standard deviation of mean fluorescence values, which corresponded to a percentage response of 58%.
To minimize false positive hits we selected only compounds that satisfied the cutoff criteria because they concurrently decrease fluorescence. By using the criteria used in FL2, 12 compounds from the entire PCL set qualified as hits. However, 8 of these compounds were flagged as potentially fluorescent as the FL2 MCF in the vector was greater than average +2X standard deviation of DMSO control (which corresponded to the DMSO control value measured in the ABCB6 cells). Four compounds met the criteria (0.4% hit rate). The identified compounds were verteporfin, benzethonium chloride, tomatine hydrochloride and piperlongumine ( and Fig. S5a, S5b, S5c and S6
Confirmatory Dose Responses
Compounds that displayed activity in the primary HTS assay were re-tested for dose response. The four hit compounds that decreased MCF intensity by FL2 in the primary screen were evaluated employing ABCB6 over-expressing and vector cells to identify compounds decreasing MCF by FL2 in a dose response manner (dose concentration ranging from 48.8 nM to 100 µM). All the four compounds (verteporfin, benzethonium chloride, tomatine hydrochloride and piperlongumine) showed a dose dependent ability to reduce ABCB6 mediated PPIX accumulation (). In contrast, none of the compounds had any effect on vector cells. Verteporfin exhibited the lowest EC50, while piperlongumine exhibited the highest. Interestingly an increase in tomatine hydrochloride concentration more than ~1.0 µM leads to a rapid decrease of the MCF (). The differences in MFC at each dose of these compounds for three replicates were significant (p<0.05).
Dose response analysis of active compounds identified in the primary HTS.
To confirm the results from our primary HTS assay and to demonstrate that the decrease in PPIX fluorescence mediated by the identified compounds were not associated with their ability to alter ABCB6 expression per se we measured ABCB6 expression in ABCB6-overexpressing cells following treatment with each of the four compounds. As shown in tomatine hydrochloride and benzethonium chloride dose dependently reduced ABCB6-mediated PPIX fluorescence without affecting ABCB6 expression. Similar results were observed for verteporfin (data not shown).
Benzethonium chloride is a broad range antiseptic with a variety of antimicrobial properties 
and a tendency to be potentially toxic 
. The plant steroidal alkaloid α-tomatine has a variety of reported activities spanning from antifungal to immune adjuvant 
. Verteporfin (trade name Visudyne), is a benzoporphyrin derivative used as a photosensitizer for photodynamic therapy to eliminate the abnormal blood vessels in the eye associated with conditions such as the wet form of macular degeneration 
. Piperlongumine has anti-cancer properties and selectively targets and kills cancer cells but leaving normal cells unharmed 
. It is intriguing to note that the assay detected verteporfin, which is similar in structure to carboxylated porphyrin derivatives such as hemin and pheophorbide A, which have been shown to be ABCB6 substrates 
A follow-up cytotoxicity assay was performed employing ABCB6 overexpressing and vector cell lines (). This experiment aimed to identify in principal those compounds that are not toxic at the concentrations exhibiting their functional inhibitory role by decreasing the intracellular accumulation of porphyrin mediated by ABCB6. Interestingly, the normalized viability of vector cells was reduced more than 80% when the concentration of tomatine and benzethonium chloride was >390 nM. However, viability was not reduced more than 25% even at the highest concentration (100 µM) of verteporfin and piperlongumine. The differences in percent-normalized viability at each dose of these compounds for three replicates were significant (p<0.05).
Toxicity profile for ABCB6 overexpressing and vector cells treated with the four compounds identified in the primary HTS screen.
In case of ABCB6 overexpressing cells, benzethonium, verteporfin, tomatine and piperlongumine exhibited EC50 values of 1.9, 0.17, 0.08 and 2.23 uM, respectively. It is true that both verteporfin and tomatine are more toxic to ABCB6 cells than the other two compounds (). However, the selection of a less toxic compound in between benzethonium and piperlongumine is also complex. The normalized viability of ABCB6 expressing cells was not reduced significantly up to a dose of 780 nM dose of benzethonium, which was reduced by 40% at this concentration of piperlongumine.
Topology and Homology Modeling of ABCB6
ABC transporters have a characteristic architecture that consists minimally of four domains: two transmembrane domains (TMDs), and two cytoplasmic ATP Binding domains 
. At the sequence level, the superfamily of ABC transporters is identified by a characteristic set of highly conserved motifs that are present in the ABCs.
Although the detailed folds of the TMDs can vary, they all interact with the helical domains of the ABC domain through coupling helices that are located in the loops between TM helices 
. In addition to the sequence comparisons of the ABC domains, the structures of the binding proteins also suggest that the hydrophobic amino acid (HAA; such as Leu, Ile, Val system) and monosaccharide (MOS; such as ribose (RBS)) branch could exhibit a distinct TMD fold. The structures of the binding proteins for ABC transporters vary in their topological threading between domains 
and can be classified into three distinct categories (designated I, II and III, which confusingly overlaps with the nomenclature used for the TMD folds).
The homology modeling and docking results show that the best scoring poses of the docked ligands (), are positioned in the upper part of the transmembrane region, close to the corresponding binding site of the QZ59 inhibitor in the ABCB1a crystal structure 
. This is the site where verapamil is known to bind to the ABCB1 transporter 
. While we do not intend to make quantitative predictions based on this docking model, we did notice a higher overlap between the endogenous substrate, coproporphyrinogen III, with tomatine and verteporfin, compared to the overlap with benzethonium and piperlongumine.
Topology and homology model of ABCB6 dimer with the docked ligands.
Selectivity and Validation of Compounds Identified in the Primary HTS assay – Hemin Agarose Affinity Chromatography Using Mitochondrial Fractions
The compounds identified in the primary HTS were evaluated for their ability to act as functional inhibitors of ABCB6 using hemin-agarose affinity chromatography and mitochondrial transport assay 
. We have previously demonstrated that these two assays when used in combination have the ability to identify substrates and potential inhibitors of ABCB6 
. We found that following hemin-agarose affinity chromatography, verteporfin and tomatine hydrochloride showed a dose dependent displacement of ABCB6 from hemin agarose [>50% displacement of ABCB6 at 5 µM ()]. In contrast, neither succinylacetone () a known inhibitor of porphyrin biosynthesis nor benzethonium chloride (data not shown) one of the compounds identified in the HTS screen were capable of displacing ABCB6 even at a concentration of 20 µM. These results suggest that verteporfin and tomatine hydrochloride could be potential substrates of ABCB6 and might be capable of interacting with the same substrate-binding site as hemin.
Selectivity and validation of HTS identified compounds by hemin agarose affinity chromatography.
Selectivity and Validation of Compounds Identified in the Primary HTS Assay – Hemin Agarose Affinity Chromatography Using Purified Protein
To confirm hemin interaction with ABCB6 specifically and to extend the studies described above we performed hemin-agarose affinity chromatography using purified ABCB6 protein. The purification and characterization of ABCB6 is part of another manuscript that is currently in preparation and will not be discussed here in detail, except to show that the purified protein (confirmed by mass spectrometry) exists as two isforms of ~ 50 kDa and ~80 kDa proteins (), both of which were recognized by the flag-antibody. Because the flag-antibody is fused to the c-terminal tail of ABCB6, it suggests that the two isoforms have different N-terminal sequences. We found that following hemin-agarose affinity chromatography, only the ~80 kDa isoform was detected in immunoblots ( and Fig. S7
), suggesting isoform specific interaction with hemin. As with the isolated mitochondria, both verteporfin and tomatine hydrochloride showed a dose dependent displacement of purified ABCB6 from hemin-agarose, while no such competition was seen with succinylacetone. Taken together results presented in , and Fig.S7 and S8, demonstrate for the first time that hemin directly interacts with ABCB6. Despite these results it is still quite possible that hemin interaction with ABCB6 might require additional proteins that promote hemin interaction with ABCB6. However, from the results presented here it appears that this interaction might be transient and is potentially not required once the hemin-ABCB6 complex is formed, because no interacting proteins were seen following hemin-agarose-ABCB6 affinity chromatography (Fig. S7
). In addition, these results suggest that verteporfin and tomatine hydrochloride could be potential substrates of ABCB6 and might be capable of interacting with the same substrate-binding site as hemin.
SDS-PAGE analysis of purified ABCB6 and selectivity and validation of HTS identified compounds by hemin-agarose affinity chromatography using purified ABCB6.
The results from hemin-agarose affinity chromatography is further strengthened by the modeling studies described above which demonstrate that both tomatine and verteporfin show higher overlap with the endogenous substrate, coproporphyrinogen III (). Further, given that verteporfin is a carboxylated porphyrin, similar to porphyrins that have previously been demonstrated to be transported by ABCB6, these data suggest that verteporfin is a substrate for ABCB6 that competes for the same site on ABCB6 as does hemin. Thus, the identification of verteporfin in this HTS assay highlights the capability of the assay to identify not only inhibitors of ABCB6 but also potential substrates.
Selectivity and Validation of Compounds Identified in the Primary HTS Assay – Mitochondrial Transport Assay and ATPase Activity Assay
To test further the specificity of the identified compounds to inhibit ABCB6 function we evaluated the ability of verteporfin, tomatine hydrochloride and benzethonium chloride to inhibit ABCB6 mediated transport of hemin into mitochondria. We have previously demonstrated that ABCB6 is capable of transporting radioactive labeled hemin into isolated mitochondria and that this assay is capable of distinguishing ABCB6 substrates from non-substrates 
. As shown in , hemin uptake by isolated mitochondria was inhibited by all three compounds with verteporfin showing the highest inhibition followed by tomatine hydrochloride and benzethonium chloride. The fact that benzethonium chloride acts as a weak inhibitor of ABCB6 mediated transport of hemin into isolated mitochondria, despite the fact that it does not interfere with hemin binding to ABCB6 suggests that benzethonium chloride may function as a potential inhibitor of ABCB6 by interacting with a site that is different from that bound by hemin. These results presented above highlight the recurring theme in the transporter field where the transporters appear to have multiple substrate binding sites that promote co-operativity either in promoting or inhibiting transport of a specific substrate. In contrast to benzethonium chloride, both tomatine hydrochloride and verteporfin were capable of inhibiting transport as well as hemin interaction with ABCB6 suggesting that these compounds could be potential substrates of ABCB6.
Three of the four compounds identified in the primary HTS screen compete for ABCB6 mediated hemin transport.
To confirm our assertion above and to demonstrate that verteporfin and tomatine hydrochloride are potential substrates of ABCB6 we performed transport assays using these compounds in functional mitochondria isolated from cells expressing either a transport competent ABCB6 protein (referred to as ABCB6 in the figure legends and ) or a transport incompetent ABCB6 protein (referred to as ABCB6-MT) as described above in methods and 
. As shown in both verteporfin () and tomatine hydrochloride () were taken up in an ATP dependent manner into functional mitochondria isolated from ABCB6 expressing cells which was significantly higher than the uptake observed using functional mitochondria isolated from ABCB6-MT cells.
Verteporfin and tomatine hydrochloride are ABCB6 transport substrates.
We have previously demonstrated that ABCB6 mediated transport of substrates is an energy-dependent process, and that ABCB6 is a mitochondrial membrane ATPase that functions as an active drug transporter 
. Based on these observations we hypothesized that ABCB6 might function as an ATPase during the transport of verteporfin and tomatine hydrochloride. To test this hypothesis we measured the ATPase activity of ABCB6 in the presence and absence of ATP, and the drug, using mitochondria isolated from ABCB6 expressing or ABCB6-MT expressing cells. As shown in both verteporfin () and tomatine hydrochloride () stimulated vanadate-sensitive mitochondrial ATPase activity in ABCB6 expressing cells but not in ABCB6-MT expressing or vector control cells. Together results presented in and demonstrate that verteporfin and tomatine hydrochloride could be potential substrates of ABCB6 and highlights the robustness of the HTS assay to identify not only potential inhibitors but also potential substrates of ABCB6.
There is increasing evidence that ABCB6 localizes to multiple cellular compartments including the plasma membrane 
. Although the precise identity and the substrate specificity of ABCB6 localized to the plasma membrane are not defined, there is a potential possibility that the differentially localized ABCB6 might have overlapping substrate specificity 
. This could lead to a possible erroneous generalization of the identified compounds as either inhibitors or substrates of ABCB6 based on the primary HTS assay. Although this is a concern that has to be considered, the secondary assays of specificity described here can eliminate these discrepancies in most cases as these secondary assays utilize isolated mitochondria and purified ABCB6 protein as opposed to whole cells used in the primary HTS assay. Further, it is interesting to note that the purified ABCB6 protein exists as two isoforms (), very similar to what has been described by Paterson et al 
as the mitochondrial and plasma membrane forms of ABCB6. In addition our results suggest that the two isoforms could have different substrate specificities as only the ~80 kDa form of ABCB6 appears to interact with hemin but not the ~50 kDa form. Although it is not clear at this time which of the two isoforms localize to the mitochondria, based on our earlier studies where we have shown that the ~80 kDa form of ABCB6 interacts with heme and localizes to the mitochondria we hypothesize that the ~80 kDa form of the ABCB6 isoform localizes to the mitochondria.
As optimized, the HTS assay is amenable to high-throughput screening applications for identification of ABCB6 modulators and in combination with the secondary assays can identify potential mitochondrial ABCB6 substrates and inhibitors. The assay is also appropriate to establish the relative potencies of active compounds. Potential pitfalls include possibility of false positives due to cytotoxic compounds or pore-forming proteins, which could allow ABCB6 function-independent leakage of PPIX, leading to a decrease in the overall cellular PPIX levels. Finally, although the assay is configured to only detect compounds that block ABCB6, the mechanism of action is not limited nor is specificity ensured. Because ABCB6 activity is dependent on ATP hydrolysis, compounds that affect ATP synthesis and accumulation could also appear as false positives. However, as with the differentially localized ABCB6, these issues can be easily resolved by secondary validation assays and specificity studies described above.
Further analysis of the compounds identified here as well as those resulting from future screening of the small Molecular Libraries Small Molecule Repository (MLPCN) is expected to yield significant inhibitors and activators of ABCB6 and potential new ABCB6 substrates. Availability of these inhibitors, activators and substrates in multiple classes will also be useful in understanding the biochemical, physiological, and clinical implications of ABCB6 in development, tumorigenesis and therapy related drug resistance.