Effect of MRP4 G487E Variant on Atpase Activity

ii Abstract

ATP-binding cassette (ABC) transporters are a diverse family of transmembrane proteins

that facilitate the ATP powered translocation of substrates across cellular membranes. Multidrug

resistance-associated protein 4 (MRP4) is a highly polymorphic, human ABC transporter with a

wide array of substrates, including, antiviral drugs and prostaglandins. Previously, altered antiviral drug transport capacities of nonsynonymous single nucleotide polymorphic MRP4 variants, relative to reference MRP4 transport activity, were reported.

This project focused on characterizing the ATP hydrolysis capabilities of the G487E MRP4 variant (rs11568668); G487 is located in an ATP binding cassette, also known as the nucleotide binding domain, of MRP4. Previously published works report a 0.008 minor allele frequency in

Asian American individuals (n = 120) and show decreased transport activity with G487E. Novel data on G487E MRP4 activity from this study corroborate the decreased drug transport findings.

This study characterized prostaglandin E2 – stimulated ATP hydrolysis by human MRP4

reference and G487E variant expressed in insect cells. MRP4 specific activity was determined by

its selective inhibition with beryllium fluoride. Reference MRP4 reached a maximal ATP

hydrolysis velocity of 27.4 ± 1.07 nmoles Pi/mg protein/minute; for comparison, ATP hydrolysis

in Sf21 membranes absent of exogenous transporters reached a maximum velocity of 18.4 ± 0.55

nmoles Pi/mg protein/minute. The G487E MRP4 variant had an intermediate Vmax value of 24.4 ±

0.93 nmoles Pi/mg protein/minute. As measured by an ANOVA, Vmax values for the Sf21, MRP4

reference and G487E MRP4 membranes were significantly different (p < 0.0001).

In contrast, Km values were similar for the three sample types and ranged from 0.34 ± 0.16

µM for reference MRP4 to 0.44 ± 0.18 µM for G487E MRP4. Structural rationales for these data

iii were explored using novel homology models of the first nucleotide binding domain of MRP4. The

G487E mutation was determined to only affect the MRP4 binding affinity of ATP; prostaglandin

E2 binding was not impacted. These results suggest that the reduced transport phenotype of G487E

MRP4 relative to reference MRP4 is due, at least in part, to impaired ATP hydrolysis.

Acknowledgements

This document marks the end of an incredible journey in academic research and the

beginning of an exciting new career in pharmaceutical consulting. I want to thank my friends,

family, and mentors who enhanced this experience and helped make this transition possible.

To my thesis committee, thank you for your support in my decision to leave UCSF a few

years earlier than originally planned. Deanna Kroetz, thank you for being an incredibly supportive

PI and friend. Despite my career interests being far outside of your expertise, you facilitated the

series of informational interviews that led me to health care consulting and gave me the time to

explore and enter this career path. Thank you for your encouragement through this entire project;

I am thrilled that I could leave the lab with detailed instructions for the successful execution of

MRP4 ATPase assays and preliminary MRP4 variant ATPase activity data. Jaime Fraser, thank

you for your continual support through my UCSF journey, from my recruitment in 2013 to my

Masters thesis submission in 2015. Kathy Giacomini, thank you for your insight into career

transitions in non-research science careers and your feedback on my Masters thesis.

To my PSPG cohort, thank you for being involved in every step of this journey – becoming

a California resident, balancing fellowship applications, rotations, project proposals, with journal

clubs and 220 presentations, surviving TAing pharmacy students, and solidifying our next career steps.

iv To my friends and family, thank you for always being a phone call, text, or video call away

– moving 3,000 miles did not feel so far because you were always so close. Grandpa, thank you for inspiring my resilience and determination. Dad, thank you for trying to give advice even when my questions were beyond your expertise and your unwavering confidence in me.

The work described in Chapter 2 was performed with guidance from published work:

Sauna ZE, Nandigama K, Ambudkar S V. Multidrug resistance protein 4 (ABCC4)-mediated ATP hydrolysis: Effect of transport substrates and characterization of the post-hydrolysis transition state. J Biol Chem. 2004;279(47):48855-48864. doi:10.1074/jbc.M408849200. I appreciate the helpful advice of the Ambudkar laboratory at the National Cancer Institute while setting up this assay.

Finally, I would like to thank everyone in the Kroetz and Giacomini labs for their support in and out of the laboratory. Without you, I would still be looking for the MilliQ water (hidden in the fish room).

v Table of Contents

Abstract ...... iii

Acknowledgements ...... iv

List of Tables ...... viii

List of Figures ...... ix

Chapter 1 : Introduction ...... 1

ABC Transporters ...... 1

Human ABC Transporters ...... 1

Functional Characteristics ...... 5

ABC Transporter Associated Diseases...... 6

Multidrug Resistance-Associated Protein 4 (MRP4) ...... 8

Tissue Distribution ...... 8

Interaction with Drugs ...... 8

ATP Binding and Hydrolysis ...... 11

MRP4 Homology Modeling ...... 13

ABCC4 Genetic Variants ...... 15

Goals and Rationale for Project...... 19

References ...... 21

Chapter 2 : Effect of MRP4 G487E Variant on ATPase Activity ...... 26

Materials and Methods ...... 26

Sf21 Cell Culture ...... 26

Sf21 Cell Infection and Harvest ...... 27

Membrane Isolation...... 27

vi Western Blotting ...... 29

ATPase Activity Assay ...... 30

Homology Template Search and Selection...... 32

Model Building and Quality Assessment ...... 32

Results ...... 33

Assay Development...... 33

MRP4 Specific ATPase Activity ...... 35

Substrate-Dependent MRP4 ATPase Activity ...... 37

Proposed Reaction Mechanism ...... 39

Structural Observations ...... 40

Discussion ...... 46

Assay Optimization ...... 46

Impact of G487E Variant ...... 47

Future Directions ...... 50

References ...... 52

vii List of Tables

Table 1.1 Select MRP4 Substrates ...... 9

Table 2.1 Kinetic Parameters of PGE2 stimulated ATP hydrolysis with Crude Membranes ...... 39

Table 2.2 MRP4 Transport Model Parameters ...... 39

viii List of Figures

Figure 1.1 ABC transporter gene dendrogram ...... 2

Figure 1.2 ABC Transporter Superfamily Genotype-Tissue Expression Plot ...... 4

Figure 1.3 Organization of ABC Transporter Domains...... 6

Figure 1.4 Gene Expression of ABCC4…………………………………………………………..7

Figure 1.5 Relative ABCC4 gene expression in major tissue types ...... 8

Figure 1.6 MRP4 export of DHEAS ...... 10

Figure 1.7 Antiviral drug concentrations in Abcc4 (Mrp4) knockout mouse kidneys ...... 10

Figure 1.8 ABC exporter conformational change initiated by ATP and substrate binding ...... 12

Figure 1.9 Effect of substrate concentration on MRP4 ATPase activity ...... 13

Figure 1.10 Homology model of human MRP4 in membrane – Outward-Facing ...... 14

Figure 1.11 Comparison of human MRP4 homology model and P-gp structure ...... 15

Figure 1.12 MRP4 transmembrane prediction and location of SNPs ...... 16

Figure 1.13 MRP4 variants affect intracellular accumulation of antiviral drugs ...... 17

Figure 1.14 Location of the MRP4 variants G487E, K498E, and V1071I ...... 18

Figure 1.15 Effect of MRP4 variants on [3H]- tenofovir transport ...... 19

Figure 2.1 Membrane vesicle preparation protocol ...... 29

Figure 2.2 Linear range of Pi detection...... 33

Figure 2.3 Impact of pH on ABC transporter-dependent ATPase activity ...... 34

Figure 2.4 Time-dependent ABC transporter ATPase activity ...... 34

Figure 2.5 Impact of ATP concentration on Pi levels ...... 35

Figure 2.6 MRP4 Expression in Baculovirus membranes ...... 36

ix Figure 2.7 MRP4 Specific ATPase Activity ...... 37

Figure 2.8 PGE2 stimulated ATP hydrolysis for MRP4 Reference and the G487E variant ...... 38

Figure 2.9 Simple Reaction Scheme Proposed for MRP4 ...... 40

Figure 2.10 MRP4 with nsSNP locations ...... 41

Figure 2.11 Template alignments for G487 and G487E NBD models ...... 42

Figure 2.12 Z-scores for reference and G487E models ...... 43

Figure 2.13 NBD features for G487 and G487E relative to MRP4 homology model...... 44

Figure 2.14 NBD1 ATP binding features relative to residue 487 and neighbors ...... 45

Figure 2.15 Impact of ATP concentration on MRP4 ATPase Activity ...... 47

Figure 2.16 Reference MRP4 NBD1 with G487E substitution ...... 49

Figure 2.17 MRP4 highlighting NBD Variants ...... 50

x Chapter 1 : Introduction

Membrane transporter proteins, which exist in all eukaryotes (and most prokaryotes), can be classified as ion channels, transporters, aquaporins, or ATP-powered pumps. Their function is to move molecules across cellular and organelle membranes that are too large or too hydrophilic to diffuse directly through membranes [1].

ABC Transporters

ATP-powered pumps use the energy generated from ATP hydrolysis to translocate

substrates across membranes against their concentration gradients. The diverse and ubiquitous

superfamily of ATP-binding cassette (ABC) transporters is within this category. ABC transporters

function as importers or exporters; importers move extracellular substrates into the cell, while

exporters push cytosolic substrates outside of the cell [2]. Most eukaryotic ABC transporters are

exporters with wide arrays of substrates including hormones, lipids, drugs, and other xenobiotics.

Many ABC transporters are involved in maintaining cellular homeostasis. Under benign

conditions, they transport lipids and broad-spectrum metabolites, but under toxic conditions, ABC

transporters eradicate toxic exogenous compounds from the cell [3]. In addition to their roles in

homeostasis, ABC transporters are involved in multi-drug resistance (MDR). MDR occurs when

cancers, bacteria, and viruses become resistant to several structurally unique compounds [4].

Human ABC Transporters

Forty-nine genes within the human genome code for ABC transporters; these transporters

are divided into seven subfamilies designated A through G (Figure 1.1). The diversity within this

1 transporter family and their ubiquitous expression in many human cell types and tissues (Figure

1.2) facilitates their involvement in a variety of cellular processes.

ABC

Figure 1.1 ABC transporter gene dendrogram Dendrogram showing the evolutionary relatedness of all members of the human ABC transporter superfamily. Image is taken from Vasiliae et al. [1].

Subfamily A transporters are found in a variety of organs and cell types; they are mostly

involved in lipid trafficking. ‘MDR family of ABC transporters’ is the alias for the mammalian

specific superfamily B transporters. This non-de plume stems from their documented involvement in multidrug resistance in cancer cells [5]. One of the most highly studied ABC transporters, P-

glycoprotein (P-gp), is a member of this subfamily. P-gp mediates multi-drug resistance through

efflux of a variety of structurally diverse compounds [6].

2 Subfamily C (MRP – multidrug resistance-associated proteins) transporters are involved

in toxin excretion, ion transport, and cell surface reception [5]. In particular, MRP1-MRP6 play

important roles in drug and xenobiotic transport [7]. All four human half-transporter genes of

the subfamily D are alternatively spliced to form 49 unique transporters that function in

peroxisomes [1]. Both subfamily E and subfamily F contain ATP-binding domains, but lack transmembrane domains. Without transmembrane domains, they cannot function as transporters; instead, ABCE and ABCF proteins are involved in the regulation of protein synthesis and expression [8]. Members of subfamily G have suggested involvement in cellular lipid homeostasis, multidrug resistance transport, control of intestinal absorption, and promotion of biliary excretion of sterols [9].

Many different research groups catalogued ABC transporter gene expression; initial

catalogs were generated using microarray data, followed by qPCR analyses, and most recently by

RNA-seq. The NIH-sponsored Genotype-Tissue Expression (GTEx) project includes a

comprehensive survey of ABC transporter mRNA levels in multiple tissues [10]. The UCSF

Pharmacogenomics of Membrane Transporters project compiled these data and Figure 1.2

provides a visual representation of relative expression levels of ABC genes in human tissues and

organs, illustrating the broad tissue distribution of these important transporters.

Figure 1.2 ABC Transporter Superfamily Genotype-Tissue Expression Plot

4 Figure 1.2 continued The heat map above was assembled by the UCSF Pharmacogenomics of Membrane Transporters project and summarizes relative expression of ABC transporter genes in major human tissue types. The represented values are the medians of quantile normalized (QN) reads per kilobase transcript per million mapped reads (RPKM). http://pharmacogenetics.ucsf.edu/gtex/index.html [11].

Functional Characteristics

All human ABC transporters contain two nucleotide-binding domains (NBD), known as

ATP-binding cassettes, and two transmembrane domains (TMD). Generally, these domains are ordered N-term-TMD-NBD-TMD-NBD-C-term in ‘full’ human ABC transporters (Figure 1.3)

[5]. ATP hydrolysis powering substrate transport occurs in the NBD. Interwoven alpha helices compose TMDs to form a hydrophilic path transecting lipid membranes for hydrophilic substrates

[12]. Most human ABC transporters contain two TMDs composed of six helices. TMDs enable transport by swapping two helices of one domain to the four remaining helices of the other domain to create two distinct helical clusters [13]. The initial helical conformation is depicted in Figure

1.3.

Structures used to demonstrate inward (open to the cytosol) and outward (open to

extracellular space) conformations are not of human ABC transporters, but of crystalized relatives.

In 2006, the structure of a homodimeric ABC transporter from Staphylococcus aureus, Sav1866,

was solved in the presence of ADP and AMP-PNP, a non-hydrolysable ATP mimetic (PDB:

2HYD) [14]. Sav1866 is a human transporter homolog and is most closely related to ABCB

transporters involved in drug export. The bound non-hydrolysable nucleotides captured the

transporter in the outward conformation. In 2009, an apo state of P-gp from Mus musculus was the

first inward facing human ABC transporter homolog crystal structure (PDB: 3G5U) [15].

5 A B

Figure 1.3 Organization of ABC Transporter Domains Figure duplicated from Martinez and Falson ABC transporter review article [16]. (A) Domain and helix (protein topology) identification in multidrug resistance (MDR) subfamily sequence. (B) Inward facing 3D-structure cartoon of mouse P-gp with transporter domains colored in yellow and blue. Topology labels are included.

ABC Transporter Associated Diseases

NBD-NBD regions and NBD-TMD regions are considered putative ABC transporter

interfaces; point mutations and deletions within these areas would likely lead to altered transporter

function. Multiple point mutations in ABC transporters are associated with a wide array of

diseases. Some of these diseases include: Dubin-Johnson syndrome (DJS), Cystic Fibrosis (CF), and X-linked adrenoleukodystrophy (ALD). Mutations in ABCC2/MRP2 (multidrug-resistance- associated protein 2) can lead to DJS, an autosomal recessive disorder characterized by chronic or intermittent conjugated hyperbilirubmenia with stable liver enzyme levels [17]. CF is most commonly caused by mutations in ABCC7, aptly named Cystic Fibrosis transmembrane regulator

(CFTR). The disease is characterized by inflammation and tissue damage caused by abnormally viscous secretions in lung airways and pancreatic ducts [18]. Defects in ABCD1 leads to ALD, which affects nervous system white matter and adrenal cortex due to accumulation of saturated, very long chain fatty acids [19].

6 seq, part as of GTEx the project. This graph was -

Gene Expression of ABCC4

from the GTex Portal with data from GTEx Analysis Release V4 V4 Accession phs000424.v4.p1). GTEx GTex from (dbGaP the Analysis Release Portal with from data w.gtexportal.org/home/gene/ABCC4#geneExpression [10] w.gtexportal.org/home/gene/ABCC4#geneExpression 4 xpression ofxpression ABCC4 tissues in was various by measured RNA . 1

Figure mRNA e compiled http://ww

7 Multidrug Resistance-Associated Protein 4 (MRP4)

ABCC4 on chromosome 13q32 encodes the multidrug resistance-associated protein 4

(MRP4), a member of the ABC transporter superfamily. MRP4 was previously known as multi- specific organic anion transporter B (MOAT-B). At 1325 amino acids and nearly 150 kDa, MRP4 is the smallest exporter in the ABCC subfamily.

Tissue Distribution

ABCC4 is expressed ubiquitously (Figure 1.4 and Figure 1.5). It is highly expressed in the bladder, prostate, blood cells, and hematopoietic stem cells [20]; it is also expressed in the brain

[21], liver [22], and kidney proximal tubules [23].

Figure 1.5 Relative ABCC4 gene expression in major tissue types mRNA expression of ABCC4 in various tissues was measured by RNA-seq, as part of the GTEx project. This heat map was compiled by the UCSF Pharmacogenomics of Membrane Transporters project [11]. It summarizes relative expression of ABCC4/MRP4 transporter gene in major tissue types. The represented values are the medians of QN of RPKM.

Interaction with Drugs

MRP4 has been implicated in the export of anti-viral agents, anti-cancer drugs, and multiple endogenous compounds, fueling investigations into its role in pharmacokinetics and

8 pharmacodynamics [24]. Anti-viral agents transported by MRP4 include the

nucleoside/nucleotide analogs azidothymidine (AZT), adefovir (9-(2-phosphonylmethoxyethyl)

adenine or PMEA), and tenofovir [25, 26]. MRP4 also transports 6-mercaptopurine, an important

drug in leukemia treatment [27]. Endogenous compounds exported by MRP4 include bile acids,

steroids, cyclic nucleotides, prostaglandins, and folate [28, 23, 29, 27, 30].

Table 1.1 Select MRP4 Substrates

Drugs Endogenous Competitive Inhibitors

AZT DHEAS AEBSF

Adefovir PGE1 Dipyridamole

Tenofovir PGE2 Indomethacin

6-Mercap- cAMP Celecoxib topurine

1 Abbreviations: AZT, azidothymidine; DHEAS, dehydroepiandrosterone sulphate; PGE1, prostaglandin E1; PGE2 prostaglandin E2,; cAMP, cyclic adenosine monophosphate; AEBSF, 4-(2-aminoethyl)benzenesulfonylfluoride. [24]

Additionally, MRP4 has been implicated in the transport of the endogenous steroid

metabolite dehydroepiandrosterone 3-sulphate (DHEAS) [29]. DHEAS, made in the adrenal

gland, is the most abundant circulating steroid in humans. In 2003, Zelcer et al. showed MRP4

dependent transport of DHEAS with [3H]-DHEA uptake assays using inside-out MRP4 vesicles

[29]. Specifically, they showed the impact of DHEA concentration on MRP4 export capabilities

(Figure 1.6).

Figure 1.6 MRP4 export of DHEAS Data published by Zelcer et al in 2003 shows DHEAS transport by MRP4 using substrate uptake assays in Sf9 membrane vesicles. See publication for experimental conditions [29]. The chemical structure of DHEAS (dehydroepiandrosterone 3-sulphate) is also shown.

Figure 1.7 Antiviral drug concentrations in Abcc4 (Mrp4) knockout mouse kidneys The chemical structures of adefovir and tenofovir are shown. Data from Imaoka et al. reported significant accumulation of [3H]-adefovir and [3H]-tenofovir, respectively, in Mrp4 deficient mouse kidneys after intravenous drug infusion [31].

In the kidney, MRP4 is localized to the apical membrane, where it facilitates the transport of substrates into the urine. A role for MRP4 in renal drug elimination is most apparent in Abcc4 knockout mice. Although Abcc4-/- mice do not exhibit visible phenotypic changes, they show significantly altered pharmacokinetics for multiple antiviral drugs. For example, Abcc4-/- mice

10 showed an increased accumulation of adefovir and tenofovir in the kidneys, which indicates

reduced renal clearance (Figure 1.7) [31]. These data suggest high toxicity risk with nephrotoxic

drugs like adefovir and tenofovir if MRP4 becomes altered or inhibited.

ATP Binding and Hydrolysis

As a member of the ABC transporter superfamily, MRP4 contains two nucleotide-binding

domains (NBD) [16]. To bind and hydrolyze ATP, NBDs contain highly conserved motifs known as Walker A and Walker B motifs [32]. The Walker A motif, also known as the phosphate binding

loop, is defined by the residue pattern GXXXXGK(T/S), where X represents any amino acid. The

Walker B motif binds a metal ion, Mg2+, through a conserved acidic residue (glutamate or

aspartate). Magnesium ions enable coordination of ATP phosphates [33].

To facilitate coordination through the transporter (from the NBDs to the TMDs), ABC

transporters contain a signature motif, LSGGQ [13]. The sequence of events in transporter

conformational changes (Figure 1.8) is as follows: substrate and ATP molecules bind

cooperatively to each Walker A motif in the NBDs. This facilitates NBD dimerizization with two

ATP molecules bound at the dimer interface. In each case, the ATP is sandwiched between the

Walker A motif in one NBD and the signature ABC motif of the opposite NBD in a head-to-tail

arrangement. The NBD dimerization triggers the shift from an inward-facing conformation to an

outward-facing conformation. At NBD dimerization, the substrate is trapped within the transporter

transmembrane pocket and released extracellularly when the transporter adopts the outward-facing

conformation [33]. Figure 1.8 depicts initial and final conformations facilitating drug transport.

Figure 1.8 ABC exporter conformational change initiated by ATP and substrate binding Graphic reproduced from Martinez and Falson ABC transporter review article [16]. 3D cartoon structures of ABC exporters convert between the inward conformation (left) and outward conformation (right). Conformational changes powered by ATP hydrolysis facilitate the transport of drug (green starburst). The inward-conformation structure is of mouse apo-P-gp [15, 34] and the outward-conformation structure captured with a non-hydrolysable ATP mimetic is of Sav1866, a bacterial homodimer [14].

MRP4 is one of many of the ‘full’ human ABC transporters with a degenerate signature motif within the second NBD, which prevents ATP hydrolysis, but still allows ATP to bind [32].

Despite the loss of hydrolysis capabilities in NBD2, the transporter is able to function with the energy obtained from hydrolyzing one ATP molecule.

Since a single hydrolysis of ATP initiates MRP4 conformational changes, ATP hydrolysis activity can indicate individual substrate transport events. MRP4 ATPase activity experiments were adapted from assays developed for P-glycoprotein. In 2004, Sauna et al characterized and optimized multiple conditions for MRP4 ATP hydrolysis activity [35]. Amongst these conditions were the quantification of substrate stimulated ATP hydrolysis, specifically, stimulation with prostaglandin E1, prostaglandin E2, cGMP, and GMP (Figure 1.9).

Figure 1.9 Effect of substrate concentration on MRP4 ATPase activity (A-C) Data displayed above come from Sauna et al.’s characterization of the effect of putative substrates on MRP4 ATP-hydrolysis. Beryllium-fluoride sensitive ATP hydrolysis activity was measured under defined concentrations of PGE1, PGE2, cGMP, and GMP (respective chemical structures shown below the graphs) in control (open circles) and MRP4 (black circles) containing crude membranes. [35]

MRP4 Homology Modeling

Most ABC transporters are not structurally characterized; to date, only two whole

transporter crystal structures have been solved. In 2008, Azza et al. created a hybrid template of

the Sav1866 x-ray crystal structure (PDB: 2HYD), a bacterial homolog with 26% sequence

identity, with a truncated and solubilized MRP1 nucleotide-binding domain x-ray crystal structure

(PDB: 2CBZ) with the YASARA for homology modeling [36]. Sav1866’s outward conformation

was frozen by the addition of a non-hydrolysable ATP mimetic [14]. The 2HYD-2CBZ hybrid

template enabled the creation of an ATP bound MRP4 model [36]. Figure 1.10 depicts MRP4’s two sets of six helices open to extracellular space; the helical clustering stems from the merged

ATP bound, cytosolic NBDs. This conformation facilitates substrate export into extracellular space.

Figure 1.10 Homology model of human MRP4 in membrane – Outward-Facing The MRP4 outward model is colored to emphasize the dimerization of monomers reported with Sav1866 [14]; each monomer provided a template for one half of the full MRP4. The first half is colored by chain highlighting transmembrane helices 1 through 6 and ATP-bound NBD1. The second half is colored in grey; it indicates helices 7-12 and ATP-bound NBD2. A 180° rotation around the vertical axis would yield a nearly identical image of helices 7 through 12 and NBD2 [35]. Graphics created with www.YASARA.org.

After the P-gp crystal structure was released in 2009 [34], Jan Koendrink from the Radboud

University Nijmegen Medical Center created an inward-facing homology model for MRP4. This

model is not published, but was provided to the Kroetz lab by request. Koendrink generated the

model using a hybrid template of P-gp structure Mus musulus (PDB: 3G5U) with a structure of

MRP1’s nucleotide-binding domain (PDB: 2CBZ) [37]. Figure 1.11 provides a superimposed

image of the rainbow colored MRP4 homology model over the pink P-gp crystal structure.

MRP4’s inward-facing conformation is characterized by NBD separation and two distinct

groupings of six nearly parallel transmembrane helices. This separation provides an opening for

substrate binding, but the converging of the two helical groupings prevents substrate transport.

90º

Figure 1.11 Comparison of human MRP4 homology model and P-gp structure Inward-facing MRP4 model was kindly provided by the Koendrink laboratory. The MRP4 homology model (rainbow chains) is superimposed on the inward-facing P-gp structure (pink). Unlike the conventional orientation, the structure on the left is imaged from “behind”; NBD2 (red, left domain) and NBD1 (green, right domain) are separated. The structures on the right are rotated 90° and emphasize the parallel orientation of helices 1 through 6 relative to NBD1. Graphics generated in MacPyMOL.

ABCC4 Genetic Variants

A 2-D transmembrane predicted representation of MRP4 is depicted in Figure 1.12. This

depiction highlights multiple MRP4 single nucleotide polymorphisms (SNPs) that exist in the

population. Several evolutionarily conserved, non-synonymous SNPs (nsSNP) in MRP4 have

either ≥ 5% minor allele frequencies in at least one population or high Grantham scores. These

nsSNPs include P78A, G187W, K293E, K304N, P403L, G487E, K498E, E757K, C956S, and

V1071I [38].

Figure 1.12 MRP4 transmembrane prediction and location of SNPs Predicted transmembrane folding for MRP4 highlighting NBDs and localized sSNPs (red circles) and nsSNPs (green circles) from the UCSF Pharmacogenomics of Membrane Transporters project database [37]. Abla et al. added additional notations to the graphic: NBDs: nucleotide binding domains, EC: evolutionarily conserved, EU: evolutionarily unconserved, black circle: inactive MRP4 mutation, G538D [38].

Previously, the Kroetz lab examined the effects of MRP4 coding variants on transporter

function. The SOPHIE (Studies of Pharmacogenetics in Ethnically Diverse Populations) project

sequenced 270 healthy, ethnically diverse individuals. Ten MRP4 nsSNPs were selected for further

investigation based on allele frequency, evolutionary conservation, and potential impact of residue

substitution (quantified by Grantham scores) [38].

Initial variation of transporter function was quantified using an intracellular accumulation

assay and published in 2008 (Figure 1.13). HEK293 cells transfected with reference or variant

ABCC4 cDNA were exposed to either AZT or PMEA (anti-viral drugs known to be MRP4

substrates, Figure 1.9). Variants with higher intracellular accumulation than MRP4 reference had

16 decreased drug export capabilities. All of the variants retained some export capabilities. Reduced

function variants were P78A, G187W, P403L, and G487E. Western blots revealed that G187W

was the only variant with lower expression, which confirms the functional impact of these non-

synonymous single nucleotide polymorphisms on MRP4 transport activity [38]. The functional

and clinical implications of additional MRP4 variants have been explored in other laboratories.

Some findings include clinical association of 6-mercaptopurine sensitivity with MRP4 E757K,

MRP4 variants associated with platelet function, and ABCC4 variants affecting risk of Kawasaki

disease [40, 41, 42]. These findings are not central to the studies in this thesis and will not be discussed further.

Grey = AZT White = PMEA

Figure 1.13 MRP4 variants affect intracellular accumulation of antiviral drugs Intracellular accumulation of AZT (grey) and PMEA (white) for MRP4 variants (P78A, G187W, K293E, K304N, P403L, G487E, K498E, M744V, E757K, C956S, and V1071I) relative to reference MRP4. (*, p < 0.005; **, p < 0.001.) See publication for experimental conditions [38]. Reproduced from Abla et al. J Pharm Exp Ther 2008.

Subsequent functional studies of MRP4 variants in the Kroetz lab expanded upon the

potential disease possibilities and efflux capabilities of G487E, K498E, and V1071I. G487E and

K498E are located in NBD1, while V1071I is located in NBD2, adjacent to the degenerate ATP

17 hydrolysis site (Figure 1.14). Of these variants, G487E was the only variation predicted to lead to disease. G487E was predicted to be deleterious with a glutamic acid residue substituting for a glycine residue. K498E was predicted to be neutral, where lysine, a basic residue, is substituted with glutamic acid. V1071I, where valine is replaced with a slightly larger aliphatic residue, isoleucine, was also predicted to be neutral [43].

Figure 1.14 Location of the MRP4 variants G487E, K498E, and V1071I Cartoon representations of regional MRP4 homology models focusing on MRP4 variants: K498E, G487E, and V1081I. Reproduced from Kelly et al. Protein Sci 19:2110-21, 2010. Human MRP1 (PDB: 2CBZ) [37] was the template for (A) K498E and (B) G487E. Plasmodium yoelii MRP2 (PDB: 2GHI) was the template for (C) V1071I [44]. Structural alignment of the MRP4 models with bacterial ABC (PDB: 1L2T) [45] provided ATP coordinates. Residues within 3.5 Å of the cognate NBD (light blue) and ATP (yellow) are highlighted [43].

In the same publication, two of the three nsSNP disease and structural impact predictions were corroborated with [3H]-tenofovir efflux activity. The G487E variant showed reduced function, while K498E did not affect MRP4 transport (Figure 1.15). Unlike the original prediction,

V1091I showed significantly reduced MRP4 transport capabilities [43].

Figure 1.15 Effect of MRP4 variants on [3H]- tenofovir transport MRP4 G487E, K498E, and V1071I tenofovir efflux activity is expressed relative to reference MRP4; data reported as mean with S.D. (n = 9 – 22). Reproduced from Kelly et al. Protein Sci 19:2110-21, 2010. Significant differences were assessed by one-way ANOVA and Bonferroni’s multiple comparison test (*p < 0.05, **p < 0.001) [43].

Goals and Rationale for Project

From published literature and active exploration in the Kroetz lab by Michael Martin,

MRP4 is an interesting transporter for pharmacogenetic and pharmacokinetic study. As described above, some studies have characterized nonsynonymous MRP4 variants and the effects on export function. These studies only explore the phenotypic impact of nsSNPs related to drug translocation; they have not explored the root of these functional changes. This project sought to explain the molecular basis for these functional shifts. Since MRP4 transport is powered by ATP hydrolysis, testing ATPase activity could reveal a molecular explanation of altered export function.

This project focused on the G487E MRP4 variant (rs11568668) found in the SOPHIE project at a 0.008 allele frequency (n = 120) amongst Asian American individuals [38]. The G487E variant is of interest because of its location and the severity of the residue substitution. G487 is located in the first nucleotide-binding domain of MRP4 and falls within a loop adjacent to the transmembrane domain [43]. A Grantham score of 98 is due to the substitution of an insignificant

19 -H side chain of glycine with a large, negatively charged ( ) side chain of glutamate in this putative region [38]; this replacement is likely to affect transporter function.

Previously reported transport phenotypes of MRP4 variants, including G487E, are shown in Figures 1.13 and 1.15. MRP4 G487E expressing cells show 20-30% higher intracellular accumulation of AZT and adefovir compared to reference MRP4 expressing cells (Figure 1.13)

[38]. In 2010, Kelly et al. further characterized impaired export capacity for cells containing

MRP4 variants with tenofovir; the G487E variant showed 40% lower tenofovir efflux relative to reference MRP4 (Figure 1.15) [43]. The goal of the studies performed for this thesis was to quantify the effect of the G487E variant on MRP4 ATPase activity. Noted differences in ATPase activity would provide further support for the functional and clinical significance of this variant and extend our understanding of the molecular basis for differences in transport associated with the MRP4 G487E variant.

20 References

1. Vasiliou V, Vasiliou K, Nebert DW. Human ATP-binding cassette (ABC) transporter family. Hum Genomics. 2009;3(3):281-290. doi:10.1186/1479-7364-3-3-281.

2. Verrier PJ, Bird D, Burla B, Dassa E, Forestier C, Geisler M, Klein M, Kolukisaoglu Ü, Lee Y, Martinoia E, Murphy A, Rea PA, Samuels L, Schulz B, Spalding EJ, Yazaki K, Theodoulou, Frederica L. Plant ABC proteins - a unified nomenclature and updated inventory. Trends Plant Sci. 2008;13(4):151-159. doi:10.1016/j.tplants.2008.02.001.

3. Dietrich CG, Geier A, Oude Elferink RPJ. ABC of oral bioavailability: transporters as gatekeepers in the gut. Gut. 2003;52(12):1788-1795. doi:10.1136/gut.52.12.1788.

4. Alekshun MN, Levy SB. Molecular Mechanisms of Antibacterial Multidrug Resistance. Cell. 2007;128(6):1037-1050. doi:10.1016/j.cell.2007.03.004.

5. Dean M, Hamon Y, Chimini G. The Human ATP-Binding Cassette transporter superfamily. J Lipid Res. 2001;42:1007-1017. doi:10.1101/gr.184901.

6. Smyth MJ, Krasovskis E, Sutton VR, Johnstone RW. The drug efflux protein, P-glycoprotein, additionally protects drug-resistant tumor cells from multiple forms of caspase-dependent apoptosis. Proc Natl Acad Sci U S A. 1998;95(12):7024-7029. doi:10.1073/pnas.95.12.7024.

7. Dietrich CG, Geier A, Oude Elferink RPJ. ABC of oral bioavailability: transporters as gatekeepers in the gut. Gut. 2003;52(12):1788-1795. doi:10.1136/gut.52.12.1788.

8. Sauvage V, Millot JM, Aubert D, Visneux V, Marle-Plistat M, Pinon JM, Villena I. Identification and expression analysis of ABC protein-encoding genes in Toxoplasma gondii. Toxoplasma gondii ATP-binding cassette superfamily. Mol Biochem Parasitol. 2006;147(2):177-192. doi:10.1016/j.molbiopara.2006.02.008.

9. Schmitz G, Langmann T, Heimerl S. Role of ABCG1 and other ABCG family members in lipid metabolism. J Lipid Res. 2001;42(10):1513-1520.

10. Pierson E, Koller D, Battle A, Mostafavi S. Sharing and Specificity of Co-expression Networks across 35 Human Tissues. PLOS Comput Biol. 2015;11(5):e1004220. doi:10.1371/journal.pcbi.1004220.

11. The UCSF Pharmacogenomics of Membrane Transporters project database is supported by NIH/NIGMS (U01 GM61390); http://pharmacogenetics.ucsf.edu/gtex/index.html.

12. Ludewig U, Frommer WB. Genes and Proteins for Solute Transport and Sensing. Arab B. 2002;3(1):1. doi:10.1199/tab.0092.

13. Biemans-Oldehinkel E, Doeven MK, Poolman B. ABC transporter architecture and regulatory roles of accessory domains. FEBS Lett. 2006;580(4):1023-1035. doi:10.1016/j.febslet.2005.11.079.

21 14. Dawson RJP, Locher KP. Structure of a bacterial multidrug ABC transporter. Nature. 2006;443(7108):180-185. doi:10.1038/nature05155.

15. Aller SG, Yu J, Ward A, Weng Y, Chittaboina S, Zhuo R, Harrell PM, Trinh YT, Zhang Q, Urbatsch IL, Chang G. Structure of P-glycoprotein reveals a molecular basis for poly-specific drug binding. Science. 2009;323(5922):1718-1722. doi:10.1126/science.1168750.

16. Martinez L, Falson P. Multidrug resistance ATP-binding cassette membrane transporters as targets for improving oropharyngeal candidiasis treatment. Adv Cell Mol Otolaryngol. 2014;2(23955):1-8. doi:10.3402/acmo.v2.23955.

17. Lee JH, Chen HL, Chen HL, Ni YH, Hsu HY, Chang MH. Neonatal Dubin-Johnson syndrome: Long-term follow-up and MRP2 mutations study. Pediatr Res. 2006;59(4 PART 1):584-589. doi:10.1203/01.pdr.0000203093.10908.bb.

18. Cutting GR. Cystic fibrosis genetics: from molecular understanding to clinical application. Nat Rev Genet. 2014;16(1):45-56. doi:10.1038/nrg3849.

19. Steinberg SJ, Moser AB RG. X- linked adrenoleukodystrophy. Metab Mol bases Inherit Dis. 1999:3257-3301.

20. Su AI, Wiltshire T, Batalov S, Lapp H, Ching KA, Block D, Zhang J, Soden R, Hayakawa M, Kreiman G, Cooke MP, Walker JR, Hogenesch JB. A gene atlas of the mouse and human protein-encoding transcriptomes. Proc Natl Acad Sci U S A. 2004;101(16):6062-6067. doi:10.1073/pnas.0400782101.

21. Leggas M, Adachi M, Scheffer GL, Sun D, Wielinga P, Du G, Mercer KE, Zhuang Y, Panetta JC, Johnston B, Scheper RJ, Stewart, CF, Schuetz JD. Mrp4 confers resistance to topotecan and protects the brain from chemotherapy. Mol Cell Biol. 2004;24(17):7612-7621. doi:10.1128/MCB.24.17.7612-7621.2004.

22. Rius M, Nies AT, Hummel-Eisenbeiss J, Jedlitschky G, Keppler D. Cotransport of reduced glutathione with bile salts by MRP4 (ABCC4) localized to the basolateral hepatocyte membrane. Hepatology. 2003;38(2):374-384. doi:10.1053/jhep.2003.50331.

23. van Aubel RAMH, Smeets PHE, Peters JGP, Bindels RJM, Russel FGM. The MRP4/ABCC4 gene encodes a novel apical organic anion transporter in human kidney proximal tubules: putative efflux pump for urinary cAMP and cGMP. J Am Soc Nephrol. 2002;13(3):595-603.

24. Russel FGM, Koenderink JB, Masereeuw R. Multidrug resistance protein 4 (MRP4/ABCC4): a versatile efflux transporter for drugs and signalling molecules. Trends Pharmacol Sci. 2008;29(4):200-207. doi:10.1016/j.tips.2008.01.006.

25. Schuetz JD, Connelly MC, Sun D, Paibir SG, Flynn PM, Srinivas RV, Kumar A, Fridland A. MRP4: A previously unidentified factor in resistance to nucleoside-based antiviral drugs. Nat Med. 1999;5(9):1048-1051. doi:10.1038/12487.

22 26. Adachi M, Sampath J, Lan L Bin, Sun D, Hargrove P, Flatley R, Tatum A, Edwards MZ, Wezeman M, Matherly L, Drake R, Schuetz J. Expression of MRP4 confers resistance to ganciclovir and compromises bystander cell killing. J Biol Chem. 2002;277(41):38998-39004. doi:10.1074/jbc.M203262200.

27. Chen ZS, Lee K, Walther S, Raftogianis RB, Kuwano M, Zeng H, Kruh GD. Analysis of methotrexate and folate transport by multidrug resistance protein 4 (ABCC4): MRP4 is a component of the methotrexate efflux system. Cancer Res. 2002;62(11):3144-3150.

28. Reid G, Wielinga P, Zelcer N, van der Heijden I, Kuil A, de Haas M, Wijnholds J, Borst P. The human multidrug resistance protein MRP4 functions as a prostaglandin efflux transporter and is inhibited by nonsteroidal anti-inflammatory drugs. Proc Natl Acad Sci U S A. 2003;100(16):9244-9249. doi:10.1073/pnas.1033060100.

29. Zelcer N, Reid G, Wielinga P, Kuil A, van der Heijden I, Schuetz JD, Borst, P. Steroid and bile acid conjugates are substrates of human multidrug-resistance protein (MRP) 4 (ATP- binding cassette C4). Biochem J. 2003;371(Pt 2):361-367. doi:10.1042/BJ20021886.

30. Denk GU, Soroka CJ, Takeyama Y, Chen WS, Schuetz JD, Boyer JL. Multidrug resistance- associated protein 4 is up-regulated in liver but down-regulated in kidney in obstructive cholestasis in the rat. J Hepatol. 2004;40(4):585-591. doi:10.1016/j.jhep.2003.12.001.

31. Imaoka T, Kusuhara H, Adachi M, Schuetz JD, Takeuchi K, Sugiyama Y. Functional involvement of multidrug resistance-associated protein 4 (MRP4/ABCC4) in the renal elimination of the antiviral drugs adefovir and tenofovir. Mol Pharmacol. 2007;71(2):619-627. doi:10.1124/mol.106.028233.

32. Procko E, O’Mara ML, Bennett WFD, Tieleman DP, Gaudet R. The mechanism of ABC transporters: general lessons from structural and functional studies of an antigenic peptide transporter. FASEB J. 2009;23:1287-1302. doi:10.1096/fj.08-121855.

33. Locher KP. Structure and mechanism of ABC transporters. Curr Opin Struct Biol. 2004;14(4):426-431. doi:10.1016/j.sbi.2004.06.005.

34. Li J, Jaimes KF, Aller SG. Refined structures of mouse P-glycoprotein. Protein Sci. 2014;23(1):34-46. doi:10.1002/pro.2387.

35. Sauna ZE, Nandigama K, Ambudkar S V. Multidrug resistance protein 4 (ABCC4)-mediated ATP hydrolysis: Effect of transport substrates and characterization of the post-hydrolysis transition state. J Biol Chem. 2004;279(47):48855-48864. doi:10.1074/jbc.M408849200.

36. El-sheikh A a K, Heuvel JJMW Van Den, Krieger E, Russel FGM, Koenderink JB. Functional Role of Arginine 375 in Transmembrane Helix 6 of Multidrug Resistance Protein 4 ( MRP4 / ABCC4 ). 2008;74(4):964-971. doi:10.1124/mol.107.043661.protein.

37. Ramaen O, Leulliot N, Sizun C, Ulryck N, Pamlard O, Lallemand JY, Tilbeurgh HV, Jacquet E. Structure of the Human Multidrug Resistance Protein 1 Nucleotide Binding Domain 1

23 bound to Mg2+/ATP Reveals a Non-productive Catalytic Site. J Mol Biol. 2006;359(4):940- 949. doi:10.1016/j.jmb.2006.04.005.

38. Abla N, Chinn L, Nakamura T, Liu L. The human multidrug resistance protein 4 (MRP4, ABCC4): functional analysis of a highly polymorphic gene. J Pharmacol Exp Ther. 2008;325(3):859-868. doi:10.1124/jpet.108.136523.such.

39. The UCSF Pharmacogenomics of Membrane Transporters project database is pharmacogenetics.ucsf.edu; ABCC4 transporter

40. Krishnamurthy P, Schwab M, Takenaka K, Nachagari D, Morgan J, Leslie M, Du W, Boyd K, Cheok M, Nakauchi H, Marzolini C, Kim RB, Poonkuzhali B, Schuetz E, Evans W, Relling M, Schuetz JD.Transporter-Mediated Protection Against Thiopurine-Induced Hematopoietic Toxicity. Cancer Res. 2008;68(13):4983-4989. doi:10.1016/j.jsbmb.2011.07.002.Identification.

41. Gieger C, Kühnel B, Radhakrishnan A, Cvejic A, Serbanovic-Canic J, Meacham S, Voss K, Stephens J, Attwood A, Jolley J, Lloyd-Jones H, Sambrook J, Rendon A, Ouwehand WH, Radhakrishnan A, Cvejis A, Serbanovic-Canic J, Meacham S, Ramirez-Solis R, Shin S-Y, Kanoni S, Attwood A, Memari Y, Yang T-P, Deloukas P, Stemple D, Ouwehand WH, Soranzo N, Tang W, Folsom AR, Porcu E, Cucca F, Maschio A, Uda M, Sanna S, Pistis G, Sala C, Toniolo D, Elling U, Penninger J, Goodall AH, Samani NJ, Goodall AH, Samani NJ, Labrune Y, Bonnefond A, Froguel P, Lopez LM, Porteous DJ, Starr JM, Visscher PM, Deary IJ, Lopez LM, Davies G, Deary IJ, Mägi R, Okada Y, Takahashi A, Kamatani N, Okada Y, Pirastu N, Dadamo P, Ulivi S, Gasparini P, Sorice R, Ruggiero D, Ciullo M, Teumer A, Völker U, Zhang W, Chambers JC, Elliott P, Bis JC, Wiggins KL, Ellinghaus D, Schreiber S, Franke A, Gögele M, Pramstaller PP, Hicks AA, Hottenga J-J, De Geus EJC, Willemsen G, Boomsma DI, Langenberg C, Loos RJF, Hua Zhao J, Wareham NJ, Kovacs P, O'Reilly PF, Jarvelin M-R, Shin S-Y, Hammond C, Mangino M, Memari Y, Spector TD, Mägi R, Esko T, Pullat J, Metspalu A, Esko T, Metspalu A, Esko T, Metspalu A, Hartiala J, Allayee H, Hartiala J, Allayee H, Kanoni S, Dedoussis GV, Murgia F, Biino G, Pirastu M, Parsa A, Gibson QD, Mitchell BD, Shuldiner AR, Van Der Harst P, De Boer RA, Mateo Leach I, Van Veldhuisen DJ, van Der Schoot CE, Balkau B, Bastardot F, Basu S, Feng W, Baumeister SE, Biino G, Pirastu M, Biino G, Toniolo D, Bomba L, Cambien F, Döring A, Wichmann H-E, Döring A, Meisinger C, Elliott P, Erdmann J, Schunkers H, Evans DM, Ring SM, Falchi M, Froguel P, Frazer IH, Glazer NL, Hartikainen A-L, Pouta A, Heckbert SR, Smith NL, Psaty BM, Hengstenberg C, Hersch M, Illig T, Tee Khaw K, Kyrtsonis M-C, Lagou V, Prokopenko I, Lumley T, Mcknight B, Montgomery GW, Visscher PM, Martin NG, Ferreira MA, Nakamura Y, Nauck M, Navis G, Nöthlings U, Nolte IM, Snieder H, Porteous DJ, Pouta A, Rotter JI, Taylor K, Ruokonen A, Sambrook J, Bradley JR, Ouwehand WH, Schlessinger D, Scott J, Kooner JS, Smith NL, Starr JM, Stumvoll M, Tönjes A, Hazen SL, Wilson Tang WH, Tenesa A, Lay Thein S, Wichmann H-E, Zitting P, Shuldiner AR, Joost Van Pelt L, Zwaginga J-J, Ganesh SK, Jarvelin M-R, Jarvelin M-R, Psaty BM, Akkerman J-WN, Greinacher A, Jupe S, Khadake J, Wernisch L, Rendon A, Erdmann J, Schunkert H, Pirastu N, Dadamo P, Gasparini P. New gene functions in megakaryopoiesis and platelet formation. Nature. 2011;480(7376):201-208. doi:10.1038/nature10659.

24 42. Khor CC, Davila S, Shimizu C, Sheng S, Matsubara T, Suzuki Y, Newburger JW, Baker A, Burgner D, Breunis W, Kuijpers T, Wright VJ, Levin M, Hibberd ML, Burns JC. Genome- wide linkage and association mapping identify susceptibility alleles in ABCC4 for Kawasaki disease. J Med Genet. 2011;48(7):467-472. doi:10.1136/jmg.2010.086611.

43. Kelly L, Fukushima H, Karchin R, Gow JM, Chinn LW, Pieper U, Segal MR, Kroetz DL, Sali A. Functional hot spots in human ATP-binding cassette transporter nucleotide binding domains. Protein Sci. 2010;19:2110-2121. doi:10.1002/pro.491.

44. Vedadi M, Lew J, Artz J Amani M, Zhao Y, Dong A, Wasney GA, Gao M, Hills T, Brokx S, Qiu W, Sharma S, Diassiti A, Alam Z, Melone M, Mulichak A, Wernimont A, Bray J, Loppnau P, Plotnikova O, Newberry K, Sundararajan E, Houston S, Walker J, Tempel W, Bochkarev A, Kozieradzki I, Edwards A, Arrowsmith C, Roos D, Kain K, Hui R. Genome-scale protein expression and structural biology of Plasmodium falciparum and related Apicomplexan organisms. Mol Biochem Parasitol. 2007;151(1):100-110. doi:10.1016/j.molbiopara.2006.10.011.

45. Smith PC, Karpowich N, Millen L, Moody JE, Rosen J, Thomas PJ, Hunt JF. ATP binding to the motor domain from an ABC transporter drives formation of a nucleotide sandwich dimer. Mol Cell. 2002;10(1):139-149. doi:10.1016/S1097-2765(02)00576-2.

46. Karpowich N, Martsinkevich O, Millen L, Yuan YR, Dai PL, MacVey K, Thomas PJ, Hunt JF. Crystal structures of the MJ1267 ATP binding cassette reveal an induced-fit effect at the ATPase active site of an ABC transporter. Structure. 2001;9(7):571-586. doi:10.1016/S0969- 2126(01)00617-7.

25 Chapter 2 : Effect of MRP4 G487E Variant on ATPase Activity

The G487E MRP4 variant has reduced antiviral drug transport activity, relative to reference

MRP4 [1, 2]. The goal of this project was to form a mechanistic explanation for this variant’s

transport phenotype. To achieve this, ATPase activity was characterized for G487E MRP4. To

facilitate a true mechanistic explanation, the previously reported transport activity and newly

quantified Michaelis-Menten kinetic characterization of ATPase activity will be supported by

novel nucleotide binding domain 1 (NBD1) homology models for both reference and G487E

MRP4.

Materials and Methods

Sf21 Cell Culture

Sf21 insect cells were kindly provided by the Fujimori lab at University of California, San

Francisco (UCSF). Cell growth was maintained in vent-cap sealed 1 L baffled flasks at 28 °C in an incubator with orbital shaker at 125 rpm. Baffled flasks were neither cleansed with soap, nor sterilized with bleach. Flasks were rinsed out and scrubbed repeatedly with water, then cleansed with 1 L of 70% ethanol overnight. The following day, the flasks were scrubbed again and 1 L of

70% ethanol was added overnight. To sterilize, the clean empty flasks and caps were autoclaved twice at 250 °C using a 30 minute sterilization time and 30 minute dry time.

Sf21 cells were grown in Sf II 900 SFM media (Life Technologies, Carlsbad, CA)

supplemented with 1X penicillin-streptomycin. For healthy cell growth, cell density was

maintained between 0.5 and 4 x 106 cells/mL by splitting every 2-3 days. Cell density was

monitored with Trypan blue staining.

26 Sf21 Cell Infection and Harvest

Baculovirus containing genes encoding MRP4 variants were previously made by Michael

Martin in the Kroetz lab using standard cloning, selection and propagation strategies.

Recombinant viruses encoding the following MRP4 transporters were used in the studies described

below: reference MRP4, catalytically dead G538D MRP4, and G487E MRP4. A recombinant virus expressing the empty pFastBac1 vector was used as a background control. Passage 3 viral stocks stored at 4 °C were used to infect 500 mL of Sf21 cells (~2 x 106 cells/mL > 90% alive) grown in

vent-cap sealed 1 L baffled flasks as described above.

Over the following 48-72 hours, the cells were monitored (via Trypan blue staining) for

arrested cell growth and cell death to suggest successful viral infection and protein expression.

When cell density was < 70% alive, the cells were harvested. Cell suspensions were divided into

50 mL Falcon tubes and spun at 500g for 5 minutes in an Eppendorf Centrifuge 5430 (Hamburg,

Germany) to pellet the cells. Cell pellets were resuspended in 30 mL of the remaining supernatant

in a 50 mL Falcon tube. The single tube containing the cell suspension was spun for 5 minutes at

500g. For highest protein yield, the isolated cell pellet was immediately lysed for membrane

isolation. Alternatively, the pellets could be frozen in a dry ice-ethanol bath and stored at -80 °C

until membrane isolation.

Membrane Isolation

The following protocol was adapted from methods reported earlier for expression and

isolation of ABC transporters in baculovirus (Figure 2.1) [3]. Freshly harvested cells (or quickly

thawed pellets) were resuspended in 50 mL of hypotonic lysis buffer (0.5 mM Tris, 0.5 mM

HEPES, 0.1 mM EDTA, pH 7.4) supplemented with protease inhibitors (2 mM PMSF from freshly

prepared 200 mM ethanol stock, 5 µg/mL Aprotinin from 10 mg/mL water stock at 4 °C, 1 µg/mL

27 Pepstatin A from 5 µg/mL DMSO stock at -20 °C, and 5 µg/mL Leupeptin from 25 mg/mL

water stock at -20 °C). Resuspended cells were placed on an orbital shaker at 4 °C for 30 minutes.

A Dounce tissue grinder with glass pestle “A” was used to homogenize the 50 mL of cell

lysate on ice; to achieve complete homogenization the pestle was depressed >20 times. To pellet

the membrane, the homogenate was transferred into large ultracentrifuge tubes and spun at

approximately 100,000g using the Ti-60 rotor in an Optima LE-80k Ultracentrifuge (Beckman

Coulter, Inc, Palo Alto, CA) for 30 minutes at 4 °C. After the supernatant was removed, the membrane pellets were resuspended in 20 mL of “isolation” buffer (250 mM sucrose, 10 mM Tris,

10 mM HEPES, pH 7.4) and homogenized further on ice with glass pestle “B”.

A volume of 10 mL of separation buffer (40% W/V sucrose, 10 mM Tris, 10 mM HEPES, pH 7.4) was added to three clean ultracentrifuge tubes. Approximately 8 mL of the post-“B”

homogenate was gently layered on the “separation” buffer. The layered solutions were spun at

approximately 100,000g using the Ti-60 rotor in an Optima LE-80k Ultracentrifuge for 30

minutes at 4 °C. After gently removing the tubes from the rotor, a cloudy middle layer containing

the membrane was visible under a strong light source. This middle layer was carefully removed in

500 µL volumes (ensuring that only the middle and top layers were removed) and added to a 15

mL Falcon tube. The membranes were diluted to 15 mL with the “isolation” buffer and divided

equally into fresh ultracentrifuge tubes, then spun again for 30 minutes at 100,000g to pellet the

isolated membranes. Supernatant was carefully removed and the pellets were resuspended in a

small amount of the remaining “isolation” buffer.

BCA assays were performed to determine the protein concentrations within the purified

membrane samples (diluted 2:25). From the calculated concentrations, the membranes were

28 diluted to 2 µg/µL in 160 µL aliquots and frozen on dry ice. The small aliquots were stored at -

80 °C until needed for experiments.

Figure 2.1 Membrane vesicle preparation protocol Crude membrane preparation was adapted from Ishikawa’s protocol shown in this pictorial representation [3].

Western Blotting

Samples were loaded onto 4-12% BisTris-acetate gels (Novex by Life Technologies) and run at 150 V using a Criterion electrophoresis cell and NuPAGE MOPS SDS running buffer

(Novex by Life Technologies). Gels were run until the 50 kDa band on the ladder standard

(Precision Plus ProteinTM Standards Kaleidoscope, Bio-Rad) was just above the gel’s edge. For supernatants of crude membrane preps, 15 µg of protein was diluted in 2X Laemmli Sample Buffer

(Bio-Rad) to a total volume of 30 µL.

Samples were transferred overnight at 4 °C onto nitrocellulose membranes (BioTraceTM

NT, PALL Life Sciences) using conditions of 200 mA and 10 V in Tris/glycine buffer with 10% methanol. Membranes were washed in 10 mL of 1X PBS-1% Tween for 5 minutes and then blocked for 30 minutes at room temperature in 10 mL of 5% milk, 1X PBS-1% Tween. MRP4

29 primary antibodies (clone M4I-10, Enzo Life Sciences; Farmingdale, NY) were added at a 1:500

dilution and incubated for 1 hour on a blot rocker at room temperature. Blots were quickly rinsed

and then washed three times with 10 mL 1X PBS-1% Tween 20 for 5 minutes. After the washes,

the nitrocellulose membrane was incubated with secondary antibodies in 1X PBS-1% Tween for

30 minutes on a blot rocker at room temperature in a light tight tray. Goat anti-rat 800CW (Licor;

Lincoln, NE) was used to detect the MRP4 primary antibodies. Blots were imaged on a Licor

Odyssey and densitometry was performed using Image Studio Lite version 5.0. Statistical analysis was done using GraphPad Prism v5.0. Dunnett’s test was used to test for significant differences in normalized blot density between MRP4 variants and reference.

ATPase Activity Assay

The following assay was adapted from previously published methods [5, 4]. ATP hydrolysis in isolated plasma membranes was determined by measuring the release of inorganic phosphate (Pi) from Mg-ATP in the presence of F- and P- type ATPase inhibitors (sodium azide,

EGTA, and oubain) with PiColorLock Gold Phosphate Detection System (Innova Biosciences,

Cambridge, UK). MRP4-mediated ATPase activity is inhibited by incubation with 200 µM beryllium fluoride [5]. The difference in Pi measured in the presence and absence of beryllium fluoride represents the MRP4-specific ATPase activity.

ATPase assays were all performed at 37 °C in a 96-well plate Flat Bottom Cell Culture

Plate with Low Evaporation Lid (Denville Scientific Inc., South Plainfield, NJ) while shaking.

Membrane protein (6 µg) was used in final reaction volume of 60 µL. Prior to setting up the experiment, 10 mL of MilliQ water and 10 mL of the 2X ATPase Assay Buffer (100 mM MES, 100 mM Tris HCl, 100 mM KCl, 4 mM EGTA, 20 mM MgCl2 pH 6.8) were pre-warmed to 37 °C and

30 the frozen membrane stocks were thawed slowly on ice over ~2 hours. Thawed membrane stocks

were kept on ice and only added to the reaction mixture just before incubation.

The following concentrations of non-ABC transporter ATPase inhibitors were used in the

reactions: 1 mM oubain (from 30 mM water stock), 2 mM DTT (from 60 mM water stock), 5 mM

NaN3 (from 100 mM water stock). These inhibitors were required since the crude membranes

contain multiple ATPases. Ethylene glycol tetraacetic acid (EGTA) was used to inhibit Ca2+-

ATPase activity, oubain and dithioreitol (DTT) to inhibit Na+/K+-ATPases, and sodium azide

(NaN3) to inhibit mitochondrial ATPases. Additional components for the reaction include 30 µL

of 2X ATPase buffer, 3 µL of 2.0 µg/µL crude membrane, varying concentrations of PGE2, and

MilliQ water up to a final volume of 60 µL. This reaction mixture was incubated for 5-10 minutes

in a 37 °C room while shaking.

To determine MRP4-specific ATPase activity the reaction conditions (varying PGE2

concentrations or time) were tested in the presence and absence of 200 µM beryllium fluoride (an

MRP4 specific ATPase inhibitor). Reactions were initiated by the addition of 2 µL of freshly

prepared 60 mM ATP (Adenosine 5' triphosphate disodium salt hydrate, BioXtra ≥ 99%, Sigma,

St Louis, MO). For each reaction condition, background amounts of inorganic phosphate from

ATP were determined by substituting 6 µL membrane with 6 µL of water.

Reactions were terminated by the addition of very acidic PiColorLock GoldTM; the exact

time for each reaction was noted and incorporated in the data processing. Stabilizer from the kit

was added 5 minutes after termination. Following incubation with shaking at 37 °C for 30 minutes

the yellow-green color intensity was measured at 620 nm. Background contamination was

measured in the reaction mixture without membrane and was subtracted from the samples. The

color claims to be stable for multiple hours following addition of the PiColorLock GoldTM reagent,

31 but ATP breaks down in acidic conditions leading to higher Pi levels [5]. To minimize contributions of non-catalytic ATP hydrolysis all measurements were completed within 10 minutes of stopping the reaction.

Homology Template Search and Selection.

Residues within the first nucleotide binding domain (NBD1) were input into SWISS-

MODEL’s template search; specifically, residues L428 through D634 were used. Templates for this primary amino acid sequence were searched for with Blast [6] against the SWISS-MODEL template library (SMTL, last update: 2015-07-30, last included PDB release: 2015-07-24 – when models were generated). A total of 174 templates were found. From these templates an initial

HHBlits profile was built [7] and then iterated against NR20 [8]. The new profile was compared to all SMTL profiles and yielded 2009 templates. Three templates with the highest sequence identity and the most refined crystal structures were manually selected to build models.

Model Building and Quality Assessment

SWISS-MODEL was set to auto-model mode. Auto-model mode built models based on the selected target-template alignments using Promod-II [9]. Coordinates conserved between the

target and template were copied directly from the template to the model. Meanwhile, insertions

and deletions were remodeled using a fragment library; the side chains were then rebuilt. Finally,

regularization of the model’s geometry was performed with a force field. In situations where loop

modeling with ProMod-II did not give satisfactory results, alternative models were built using

MODELLER [10]. The QMEAN scoring function was used to assess the overall and the per-

residue model quality [11]. Additional, weights of the individual QMEAN terms have been trained

specifically for SWISS-MODEL. Finally, models were downloaded as PDBs, and then

stylistically modified in PyMOL Molecular Graphics System, Version 1.7.4 Schrödinger, LLC.

32 Results

Assay Development

The ATPase assay used in these experiments was adapted from published protocols [4],

then validated and optimized for use in the Kroetz laboratory. The linear range for measurement

of Pi concentrations was up to 55 µM (Figure 2.2 A). Above this concentration, the signal was

saturated. Measurements of Pi produced in preliminary MRP4 ATPase assays were above the linear range (Figure 2.2 B), requiring the optimization of assay conditions.

Figure 2.2 Linear range of Pi detection. Pi standards (black circles) ranging from 0 µM to 100 µM were used to determine the linear range of detection. (A) The black trendline represents the linear relationship for values between 0 and 50 2 µM Pi (r = 0.997). (B) The red squares represent calculated Pi levels from preliminary MRP4 ATPase assays based on the standard curve between 0 and 50 µM Pi. Pi levels were measured using the PiColor Lock Gold Phosphate Detection System.

ATPase activity in crude membranes was sensitive to pH. ABC transporter specific activity

measured in the presence of EGTA, DTT, oubain, and sodium azide was slightly higher at pH 6.8

than at pH 7.2 (Figure 2.3) for all of the membranes tested.

Appropriate incubation times for measurement of Pi were also established. ABC transporter specific ATP hydrolysis during 20 minute reactions with 5 mM ATP generated Pi concentrations

33 above the linear measurable range for the assay (Figure 2.4A). Decreasing the reaction time to 16

minutes reduced Pi production to levels within the linear range (albeit on the high end; Figure

2.4B)

Figure 2.3 Impact of pH on ABC transporter-dependent ATPase activity ATPase activity was measured for 20 minutes at 37 °C using 10 µg of crude membrane (Empty Vector (EV), G538D MRP4, and Reference (Ref) MRP4) in the presence of 1 mM oubain, 2 mM DTT, and 5 mM NaN3. Reactions were initiated with the addition of 25 µM PGE2 and 5 mM ATP and were performed at pH 7.2 and pH 6.8. Pi from ABC transport activity was measured using the PiColor Lock Gold Phosphate Detection System and are shown for a single assay.

Figure 2.4 Time-dependent ABC transporter ATPase activity ATPase assays were performed at pH 6.8 and 37°C with 6 µg of crude reference MRP4 membrane and background ATPase inhibitors. Reactions were initiated with 25 µM PGE2 and 5 mM ATP and were run for either (A) 20 minutes (orange squares) or (B) 16 minutes (blue squares). Pi generated from ABC transporter activity was calculated based on a linear standard curve between 0 and 50 µM (black circles). Pi levels were measured using the PiColor Lock Gold Phosphate Detection System.

34 ATP concentrations also required optimization to reduce background Pi. Reactions with 5

mM ATP (Figure 2.5A) had high Pi contamination and Pi produced by ABC transporter ATP

hydrolysis fell above the linear range for measurement. A reduction of ATP to 2 mM (Figure

2.5B) significantly decreased background Pi contamination and ABC transporter specific Pi

formation was detected within the linear range of the assay.

Figure 2.5 Impact of ATP concentration on Pi levels Absorption measurements and estimated Pi concentrations for background Pi (red squares) and ABC transporter specific Pi (green triangles) production are shown for different ATP concentrations relative to defined Pi standard curve values. Pi levels were measured using the PiColor Lock Gold Phosphate Detection System in 16 min reactions at pH 6.8 using 6 µg of crude reference MRP4 membrane, background ATPase inhibitors, 25 µM PGE2 and (A) 5 mM or (B) 2 mM ATP.

Based on the initial assays described above, final conditions for measuring ABC

transporter-specific ATPase activity were selected as follows: 6 µg crude membrane protein, 1

mM oubain, 2 mM DTT, 5 mM NaN3, and 2 mM ATP run for 15 minutes at pH 6.8. Samples with

Pi concentrations out of the standard range of 0-50 µM were diluted and remeasured.

MRP4 Specific ATPase Activity

After assay development was complete, MRP4 specific ATPase activity was determined.

MRP4 expression in the crude membranes was confirmed for MRP4 reference and MRP4 G487E

via Western blotting with MRP4 antibodies (Figure 2.6). Absence of detectable human MRP4 in

35 Sf21 membranes was also confirmed. Immunoreactive MRP4 was also detected in membranes

containing inactive G538D MRP4 (data not shown).

Optimized ATPase assays with crude membranes were performed under three conditions:

no inhibitors, non-ABC transporter inhibitors (oubain, DTT, and sodium azide), and MRP4

inhibitor (beryllium fluoride) plus non-ABC transporter inhibitors. Under these three conditions,

ATPase activity was highest without inhibitors (Figure 2.7). The addition of oubain, DTT and sodium azide significantly reduced ATPase activity by 35% on average and this was reduced 70%

further with the addition of beryllium fluoride. Baculovirus membranes from cells that expressed

active (reference and G487E) and inactive (G538D) MRP4 behaved in a similar fashion (Figure

2.7). The difference in ATPase activity between oubain/DTT/azide and BeFx/oubain/DTT/azide

represents MRP4-specific activity [4].

Figure 2.6 MRP4 Expression in Baculovirus membranes MRP4 expression was detected by Western blotting using the MRP4 primary antibodies (clone M4I-10) with Goat anti-rat 800CW for the secondary antibody. Baculovirus membranes expressing the empty pFastBac1 vector were used as a negative control.

Figure 2.7 MRP4 Specific ATPase Activity ATPase activity was measured in four different crude Sf21 membranes (EV membranes without MRP4 (blue circles), ATPase inactive G538D MRP4 (red squares), MRP4 Reference (green triangles), and the MRP4 G487E variant (purple inverted-triangles) in the absence of any ATPase inhibitors, with background ATPase inhibitors (1 mM oubain, 2 mM DTT, and 5 mM NaN3), and with background ATPase inhibitors plus a MRP4 specific inhibitor (200 µM beryllium fluoride). A total of 1 - 5 measurements were made for each condition and are indicated by the individual symbols. Median values for each condition (for n >2) are indicated with horizontal lines. Pi levels were measured using the PiColor Lock Gold Phosphate Detection System in 15 min reactions at pH 6.8, 6 µg of crude membrane, 25 µM PGE2 and 2 mM ATP.

Substrate-Dependent MRP4 ATPase Activity

Substrate-stimulated MRP4 ATPase activity was measured in Sf21 membranes expressing

MRP4 reference and the G487E variant using varying concentrations of PGE2; similar assays were performed in Sf21 membranes expressing no exogenous proteins. ATPase activity followed single-substrate Michaelis-Menten kinetics (Figure 2.8); Vmax and Km estimates were obtained by

fitting the data to this equation (Table 2.1). The Vmax was 1.5-fold higher in membranes expressing

MRP4 reference transporter compared to Sf21 membranes from cells expressing no exogenous transporters (27.4 ± 1.07 nmoles Pi/mg protein/minute and 18.4 ± 0.55 nmoles Pi/mg protein/minute, respectively). The G487E MRP4 variant had an intermediate Vmax value which

was reduced 11% relative to the reference MRP4 (24.4 ± 0.93 nmoles Pi/mg protein/minute). Vmax

values for the Sf21, MRP4 reference and G487E MRP4 membranes were significantly different as

37 measured by ANOVA (p < 0.0001). The Vmax values for the MRP4 reference and G487E variant

were significantly different (p < 0.05, extra sum of squares F-test), indicating reduced transport capacity of this variant transporter. In contrast, Km values were similar for the three membrane

preps and ranged from 0.34 ± 0.16 µM for reference MRP4 to 0.44 ± 0.18 µM for G487E MRP4

(Table 2.1). It should be noted that the relatively large error rates for the Km estimates reflect the

lack of tested substrate concentrations below the Km.

Figure 2.8 PGE2 stimulated ATP hydrolysis for MRP4 Reference and the G487E variant Impact of PGE2 concentration on beryllium fluoride sensitive ATPase activity was measured in crude Sf21 membranes without MRP4 (circles), expressing MRP4 reference (squares), and expressing the MRP4 G487E variant (diamonds). Each value represents the mean ± SE from 3-5 independent assays. The data were fit to a single substrate Michaelis-Menten equation and the fitted line is shown. Michaelis-Menten parameter estimates are provided in Table 2.1. MRP4 specific activity was determined by comparing ATPase activity with background ATPase inhibitors (1 mM oubain, 2 mM DTT, and 5 mM NaN3) and ATPase activity with background ATPase inhibitors plus 200 µM beryllium fluoride (MRP4 specific inhibitor). Pi levels were measured using the PiColor Lock Gold Phosphate Detection System in 15 min reactions at pH 6.8 with 6 µg of crude membrane, 2 mM ATP, and varying concentrations of PGE2.

38 Table 2.1 Kinetic Parameters of PGE2 stimulated ATP hydrolysis with Crude Membranes 1 Vmax (nmoles Pi/mg protein/min) Km (µM) Best-fit Values Std. Error Best-fit Values Std. Error Sf21 18.42 0.55 0.39 0.13 MRP4 Reference 27.35 1.07 0.34 0.16 MRP4 G487E 24.45 0.93 0.44 0.18 1 Vmax values are significantly different for the three membrane preps (p < 0.0001) and between the MRP4 reference and G487E variant (p < 0.05).

Proposed Reaction Mechanism

To explore the causes of phenotypic variation in ATPase activity, a mechanistic model of

MRP4 activity was developed (Figure 2.9). The model design draws from the scheme proposed by Sauna et al. for the related ABC transporter P-gp [4]. Figure 2.9 illustrates the reversible

complexing of MRP4 and ATP (k1) and subsequent ATP hydrolysis (k2). This reaction is not favorable but always takes place, even in the background of MRP4 transport activity (powered by

ATP hydrolysis). Inorganic phosphate molecules (Pi) are produced from this ATPase activity,

independent of transport activity. When intracellular substrate is introduced to the reaction, MRP4

and ATP complex with substrate readily (k3). Under these conditions, ATP hydrolysis facilitates

substrate transport to the extracellular space (k4). Table 2.2 defines the reactions corresponding to

the rate constants used in Figure 2.9. Figure 2.9B depicts the impact of beryllium fluoride (BeFx)

inhibition of MRP4. BeFx prevents MRP4 from hydrolyzing ATP when in complex with substrate,

in turn, preventing substrate transport. Background ATP hydrolysis (k2) still occurs when BeFx

is present; therefore, some Pi molecules are generated.

Table 2.2 MRP4 Transport Model Parameters Rate Constant1 Reaction

k1 ATP binding to MRP4

k2 ATP hydrolysis by MRP4

k3 Substrate and ATP binding to MRP4

k4 ATP hydrolysis and substrate transport by MRP4 1 The rate constants represent those in Figure 2.9.

39 A k 1 k2  MRP4 + ATP MRP4 ATP MRP4ADP + Pi

k Substrate(inside) 3

k4 MRP4ATPSubstrate MRP4ADP+ Substrate + P (exported) i B k1 k  2 MRP4 + ATP MRP4 ATP MRP4ADP + Pi

BeFx k3 Substrate(inside)

k4 MRP4ATPSubstrate MRP4ADP+ Substrate + P (exported) i Figure 2.9 Simple Reaction Scheme Proposed for MRP4 Figure 2.9 continued Panels A and B illustrate reaction schemes representing MRP4 under the conditions of the current experiments. In the presence of ATP, MRP4 reversibly complexes with ATP in the NBD (k1). (For simplicity, only the active NBD, NBD1, is considered). The MRP4•ATP complex disassociates with or without ATP hydrolysis. When ATP is hydrolyzed (k2), one ADP molecule and one Pi molecule are formed. In the presence of substrate, MRP4 and ATP complex with substrate (k3) in a highly favorable reaction. When MRP4 hydrolyzes ATP, one ADP molecule and one Pi molecule are produced and substrate is transported across the membrane. In panel B the inhibition of MRP4 by beryllium fluoride (BeFx) is indicated. BeFx inhibits k4.

Structural Observations

Protein structure dictates protein function. The locations of ten known MRP4 variants are highlighted in pink and noted with white arrows on Koendrink’s inward-facing MRP4 homology model generated from the P-gp crystal structure (Figure 2.10). These variants include: P78A,

G187W, K293E, K304N, P403L, G487E, K498E, M744V, E757K, and V1071I. Most of the variants showed altered transport capacities [1,2]. Based on the locations of the nsSNPs there are a few structural explanations for the phenotypic changes: altered transport cavity, impeded NBD interactions, and decreased ATP binding in NBDs (both functional and degenerate domains).

40 The current project focused on the G487E MRP4 variant; residue 487 is located in the soluble NBD1. To explain the altered transport [1, 2] and PGE2 dependent ATPase activity (Figure

2.8), homology models of Reference MRP4 NBD1 and G487E MRP4 NBD1 were generated in

SWISS-MODEL. Models were based on PDB: 2CBZ [12] (Figure 2.11). MRP4 reference and variant NBD1s have approximately 50% sequence identity to 2CBZ. 2CBZ is a crystal structure of the nucleotide-binding domain 1 in human multidrug resistance protein 1 (MRP1) with Mg2+ and ATP. Since 2CBZ was crystallized with ATP and Mg2+, the domain is assumed to represent the NBD1 in the transporter’s inward-facing conformation.

Figure 2.10 MRP4 with nsSNP locations Cartoon representation of inward facing MRP4 highlighting the location of several nsSNPs. The angle presented is from “behind”. nsSNPs are highlighted in pink and indicated with white arrows. The variants include P78A, G187W, K293E, K304N, P403L, G487E, K498E, M744V, E757K, and V1071I. These variants are found throughout the entire transporter. The MRP4 structure was generated by Associate Professor Koendrink at the Radboud University Nijmegen Medical Center and is colored in rainbow order by chain.

41 Reference MRP4 NBD1

G487E MRP4 NBD1

Figure 2.11 Template alignments for G487 and G487E NBD models Model sequences are aligned to a known nucleotide-binding domain 1, PDB: 2CBZ. 2CBZ is the crystal structure of the nucleotide-binding domain 1 in human multidrug resistance protein 1 (MRP1) with Mg2+ and ATP. “Model_02” is the sequence for Reference MRP4 NBD1 and “Model_01” is the sequence for G487E MRP4 NBD1. Homology models were generated with SWISS-MODEL. Confidence of the individual residues in the models ranges from high (blue) to low (orange) relative to 2CBZ. Reference MRP4 NBD1 shares 51.2% sequence identity with 2CBZ and G487E MRP4 NBD1 shares 50.5% sequence identity with 2CBZ. Important regions are highlighted in both alignments: residue 487 (cyan box), P-loop (green box), and ABC signature motif (pink box). Known secondary structures of 2CBZ are noted with symbols overlapping the residues; arrows indicate beta-sheets and elongated circles indicate alpha helices.

42 Figure 2.11 aligns the MRP4 NBD1 sequences to the 2CBZ template; the confidence of

the homology model is high in the regions of residue 487, P-loop, and ABC signature motif. Figure

2.12 provides Z-scores for the homology NBD1 models. Z-scores were determined for QMEAN4,

Cβ, All Atom, Solvation, and Torsion. Average Z-scores for x-ray structures are zero. Negative values indicate poor scores and positive values indicate good scores.

Reference MRP4 NBD1 G487E MRP4 NBD1

Figure 2.12 Z-scores for reference and G487E models Z-scores for NBD1 homology models generated for reference MRP4 and G487E MRP4 with SWISS-MODEL with 2CBZ as a template. Z-scores were calculated for QMEAN4, Cβ, All Atom, Solvation, and Torsion.

According to Benkert et al, “QMEAN shows a statistically significant improvement over

nearly all quality measures describing the ability of the scoring function to identify the native

structure and to discriminate good from bad models” [13]. Cβ is the second carbon atom in amino

acids; the Z-score values the Cβ distance dependent interactions. Z-scores for all atoms value the

distance-dependence interaction potential for all atoms. Solvation energy identified residues’

accessibility to water accessibility. The torsion score evaluates local geometrics; specifically,

torsion angle potential between three consecutive residues. Most of the Z-scores are negative

values. Poor scores indicate that the model requires additional optimization and should not be used

for docking experiments. Despite the low confidence in the models, they provide initial structural

insights into NBD1.

43 Reference MRP4 G487E MRP4

Figure 2.13 NBD features for G487 and G487E relative to MRP4 homology model Inward-facing homology models of Reference MRP4 and G487E MRP4 represented in cartoon form highlighting important regions of NBD1. For the best view of these NBD regions the structure is flipped 180°, showing the “backside” of the transporter. Inward facing full MRP4 homology structure (light grey) based on mouse P-gp was generated by Associate Professor Koendrink at the Radboud University Nijmegen Medical Center. The full model is hybridized with the new SWISS-MODEL generated NBD models – reference NBD1 (white) and G487E NBD1 (purple). Important regions are highlighted: residue 487 (cyan), neighboring residues (orange), P-loop/Walker A motif (green), and ABC transporter signature motif (pink). Neighboring residues are all within 3.5 Å from residue 487. The P-loop is about 20 Å from residue 487. Graphics generate in Mac PyMOL.

44 Newly generated NBD1 models replaced the NBD1s in Koendrink’s inward facing MRP4

homology model (light grey) in Figure 2.13. Figure 2.14 shows a cartoon representation of the

full transporter from “behind” and zooms into the respective NBD1s. Figure 2.14 focuses on the

NBD1 domains independently with the same color schemes as Figure 2.13. Although 20 Å

separate residue 487 from the P-loop, there is a series of beta sheets that lay between these regions.

Glycine at 487 just facilitates a turn/loop to form; when substituted with glutamate, many sterics

of polar neighbors are impacted. This model shows glutamate 487 in close proximity to polar

neighbors with like charges. If energy minimizations were performed, the series of beta sheets

would shift to accommodate the increased sterics and added polarity. The shifting of the beta sheets

would likely shift the P-loop in the G487E transporter variant.

Reference MRP4 NBD1 G487E MRP4 NBD1

Figure 2.14 NBD1 ATP binding features relative to residue 487 and neighbors Cartoon representation of homology models of Reference MRP4 (white) and G487E MRP4 (purple) NBD1s generated in SWISS-MODEL. Important residues are represented by sticks and highlighted: residue 487 (cyan), neighboring residues (orange), P-loop/Walker A motif (green), and ABC transporter signature motif (pink). Neighboring residues are all within 3.5 Å from residue 487. The P-loop is about 20 Å from residue 487. Graphics generated in MacPyMOL.

45 Discussion

MRP4 is a key protein involved in the export of endogenous and exogenous compounds in

many tissues including the kidneys, colon, bladder, and prostate. This study focused on the variant,

G487E, found at the allele frequency of 0.008 in Asian Americans (n = 120) from the SOPHIE

project [1]. G487E MRP4 showed impaired anti-viral drug export activity relative to reference

MRP4 [1, 2]. This project quantifies Michaelis-Menten kinetics of G487E MRP4 ATP hydrolysis activity and corroborates the previously published findings [1, 2]. Current data suggest

compromised ATP hydrolysis as a mechanism that facilitates transport reduction by G487E MRP4

Assay Optimization

Protocols reported by Sauna et al. [4] for MRP4 ATPase activity assays were optimized for the aforementioned experiments; the most notable sources of variation are from batch of MRP4 crude membranes and the use of PiColorLock Gold Phosphate Detection System. The optimal range for detection of Pi was determined to be below 55 µM. To ensure final Pi concentrations in

a measurable range, the reaction length was decreased and the initial ATP concentration was

decreased. Sauna et al. found that 5 mM ATP initiating transport yielded optimal activity [4]. In

contrast, a concentration of 2 mM ATP gave optimal results in our assay. Closer inspection of the

published data shows minimal difference in ATPase activity between 2 mM and 5 mM ATP

(Figure 2.15). Therefore, it is unlikely that initiating the transport reaction with only 2 mM ATP

would have significantly lower ATPase activity. These differences in ATP concentration likely

reflect small differences in sensitivity with the reagents used for detection of inorganic phosphate.

Figure 2.15 Impact of ATP concentration on MRP4 ATPase Activity Reproduced from Sauna et al J Biol Chem. 2004;279(47):48855-48864. ATP concentration dependent MRP4 specific ATPase activity data. Km (ATP) was determined to be 0.62±0.07 mM.

Impact of G487E Variant

MRP4 ATPase optimization facilitated the exploration of substrate dependent activity.

Specifically, defining the impact of prostaglandin E2 on ATP hydrolysis in reference MRP4,

G487E MRP4, and Sf21 (sans exogenous transporters) crude membranes. Previously, the Kroetz lab published MRP4 antiviral drug transport data; these data demonstrated reduced transport capabilities of G487E relative to reference MRP4 using AZT, adefovir and tenofovir as substrates

[1, 2]. The data reported in this thesis corroborates these findings and provides a mechanistic explanation of the decreased transport phenotype.

As measured by an ANOVA and confirmed with multiple comparison tests, Vmax values from Sf21, MRP4 reference, and MRP4 G487E PGE2 dependent ATPase activity are significantly different (p < 0.0001). Calculated Vmax values are 18.4 ± 0.55 nmoles Pi/mg protein/minute, 27.4

± 1.07 nmoles Pi/mg protein/minute, and 24.4 ± 0.93 nmoles Pi/mg protein/minute, respectively

(Table 2.1). Vmax represents the maximum rate at which the transporters could hydrolyze ATP while stimulated by PGE2. Identification of altered G487E MRP4 function lead to an interest in potential structure based explanations for these findings.

47 Residue 487 exists in the first nucleotide binding domain (NBD1) of MRP4 [2]. Therefore,

homology models of NBD1 for reference and G487E models were generated using SWISS-

MODEL; MRP1 NBD (PDB: 2CBZ) was used as a template [12]. Note the homology models

generated only copied structurally solved templates in highly conserved regions and then

approximated the differences through fragment modeling or simply by swapping template residues

for the desired sequence where needed. Energy minimizations were not performed.

Homology models provided preliminary insight into the severity of the glycine to

glutamate substitution at position 487 in MRP4. Generally, glycine residues are important in loops

and turns; the lack of a true side chain provides structural flexibility. Indeed, based on these

models glycine-487 facilitates a loop in NBD1. As illustrated in Figure 2.14, glycine 487 in

reference MRP4 has a minimal impact on neighboring residues and motifs.

Figure 2.16 emphasizes this with a direct substitution of glycine-487 with a glutamate

residue in the reference MRP4 NBD1 model. A glutamate substitution introduces large sterics and

polar interactions. When directly substituted in the reference NBD1 model (Figure 2.16), G487E

overlaps a neighboring glutamate; these glutamates would repel each other. These opposing

neighbors abut a series of beta sheets sandwiched between residue 487 and the P-loop, a 20 Å distance (Figure 2.14). I hypothesize that G487E alters this set of parallel beta sheets; the altered beta sheets shift the P-loop, which in turn decreases NBD1 affinity for ATP. Lower affinity for

ATP would prevent maximum ATP hydrolysis.

A decreased ATP affinity hypothesis holds true with the calculated Km values. Km is an

inverse measure of a substrate’s affinity for the enzyme. Although, measured activity was initiated

by substrates, ATP and PGE2, Km refers to the titrated substrate, PGE2. Since Km does not change

48 significantly between Sf21 (sans exogenous transporters), reference MRP4, or G487E MRP4 crude

membranes, PGE2 affinity is not affected by the glycine to glutamate substitution.

Figure 2.16 Reference MRP4 NBD1 with G487E substitution Cartoon representation of Reference MRP4 NBD1 homology model (white) generated in SWISS- MODEL. Glutamate (cyan) was directly substituted for G487 and is emphasized with stick and dot representations. Important regions are represented by sticks and highlighted: neighboring residues (orange), P-loop/Walker A motif (green), and ABC transporter signature motif (pink). Neighboring residues are all within 3.5 Å from residue 487. The P-loop is about 20 Å from residue 487. Graphics generated in MacPyMOL

If maximum hydrolysis capabilities and previously shown transport activity are reduced, while retaining substrate affinity, then we can conclude that G487E affects the affinity of MRP4 for ATP. Therefore the most likely mechanism for impaired transport function via G487E MRP4, relative to reference MRP4, is a lowered affinity for ATP in NBD1.

49 Future Directions

This project provides initial mechanistic insights into the G487E MRP4 variant. Additional pharmacokinetic characterization of this variant needs to be performed. An extension of these reported data would explore if initial ATPase velocity rates are significantly different for reference

MRP4 and G487E MRP4; this will be determined through time dependent ATPase activity and with 25 µM PGE2, the lowest substrate concentration to achieve Vmax.

To corroborate the previously published G487E transport data, substrate dependent

ATPase assays will be performed with tenofovir and adefovir. Optimal substrate concentrations calculated in these experiments will determine the substrate concentrations for time dependent

ATPase assays. I predict that tenofovir and adefovir dependent ATPase activity will show significant decrease in Vmax of G487E MRP4 relative to reference MRP4.

Figure 2.17 MRP4 highlighting NBD Variants Cartoon representation of inward facing MRP4, as shown in Figure 2.10, highlighting the NBDs with nsSNPs; the angles presented are from “behind”. nsSNPs G487E and G187W in NBD1 (green) and V1071I in NBD2 (red) are labeled and identified with white arrows. The MRP4 structure was generated by Associate Professor Koendrink at the Radboud University Nijmegen Medical Center and is colored in rainbow order by chain.

50 ATPase activity assays will then be performed with other MRP4 variants: V1071I

(rs11568653) and G187W (rs11568658). Similar to the G487E variant, the V1071I variant has low allele frequency - 0.006 in African Americans (n = 160) [1]. Phenotypically, this variant is interesting because of its 60% reduction in tenofovir export relative to reference MRP4 [2].

G187W has the highest potential for clinical significance, but was not explored in this study for logistical reasons. In the SOPHIE project, allele frequencies were 0.025 in Caucasian Americans

(n = 160), 0.108 in Asian Americans (n = 120), and 0.130 in Mexican Americans (n = 100) [1].

Phenotypically, G187W has severely impaired export capabilities; Figure 1.13 shows significant anti-viral drug accumulation intracellularly.

I propose that V1071I, in NBD2, and G187W, between NBD1 and the TMD (Figure 2.17), would impact the maximum ATP hydrolysis velocity. Conversely, substrate dependent ATPase activity with transmembrane variants may indicate altered substrate affinities through shifted Km

values. Transmembrane nsSNPs were highlighted in Figure 1.12 and Figure 2.10.

MRP4 is ubiquitously expressed in human tissue (Figure 1.5) and exports a wide array of

endogenous and xenobiotic compounds (Table 1.1). The prevalence and impaired function of

MRP4 variants and the diverse substrate profile for MRP4 reference makes a compelling case for

further study. A long term goal is to apply knowledge of MRP4 genetic variation and function for

precision medicine.

51 References

1. Abla N, Chinn L, Nakamura T, Liu L. The human multidrug resistance protein 4 (MRP4, ABCC4): functional analysis of a highly polymorphic gene. J Pharmacol Exp Ther. 2008;325(3):859-868. doi:10.1124/jpet.108.136523.such.

2. Kelly L, Fukushima H, Karchin R, Gow JM, Chinn LW, Pieper U, Segal MR, Kroetz DL, Sali A. Functional hot spots in human ATP-binding cassette transporter nucleotide binding domains. Protein Sci. 2010;19:2110-2121. doi:10.1002/pro.491.

3. Ishikawa T, Sakurai A, Kanamori Y, Nagakura M, Hirano H, Takarada Y, Yamada K, Fukushima K, Kitajima M. High-speed screening of human ATP-binding cassette transporter function and genetic polymorphisms: New strategies in pharmacogenomics. Methods Enzymol. 2005;400(05):485-510. doi:10.1016/S0076-6879(05)00027-3.

4. Sauna ZE, Nandigama K, Ambudkar S V. Multidrug resistance protein 4 (ABCC4)-mediated ATP hydrolysis: Effect of transport substrates and characterization of the post-hydrolysis transition state. J Biol Chem. 2004;279(47):48855-48864. doi:10.1074/jbc.M408849200.

5. Ambudkar S V. ATPase Activity in Crude Membranes of Human MDR1-Transfected Mammalian Cells. Methods Enzymol. 1998;292(1996):504-514.

6. Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ.Gapped BLAST and PSI-BLAST: A new generation of protein database search programs. Nucleic Acids Res. 1997;25(17):3389-3402. doi:10.1093/nar/25.17.3389.

7. Remmert M, Biegert A, Hauser A, Söding J. HHblits: lightning-fast iterative protein sequence searching by HMM-HMM alignment. Nat Methods. 2011;9(2):173-175. doi:10.1038/nmeth.1818.

8. Wheeler, D. L., T. Barrett, et al. (2005). "Database resources of the National Center for Biotechnology Information." Nucleic Acids Res 33 Database Issue: D39-45.

9. Guex N., Peitsch M.C. (1997) SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 1997; 18:2714-2723

10. Sali A, Blundell TL. Comparative Protein Modelling by Satisfaction of Spatial Restraints. J Mol Biol. 1993;234(3):779-815.

11. Benkert P, Biasini M, Schwede T. Toward the estimation of the absolute quality of individual protein structure models. Bioinformatics. 2011;27(3):343-350. doi:10.1093/bioinformatics/btq662.

12. Ramaen O, Leulliot N, Sizun C, Ulryck N, Pamlard O, Lallemand JY, Tilbeurgh HV, Jacquet E. Structure of the Human Multidrug Resistance Protein 1 Nucleotide Binding Domain 1 bound to Mg2+/ATP Reveals a Non-productive Catalytic Site. J Mol Biol. 2006;359(4):940- 949. doi:10.1016/j.jmb.2006.04.005.

52 13. Benkert P, Tosatto SCE, Schomburg D. QMEAN: A comprehensive scoring function for model quality assessment. Proteins. 2008;71(1):261-277. doi:10.1002/prot.2171