2012-01-Thesis Final

The Pennsylvania State University The Graduate School College of Medicine

LIGAND-DIRECTED FUNCTIONAL SELECTIVITY AT THE OPIOID RECEPTOR FAMILY: AN EPIC APPROACH TO UNDERSTANDING OPIOID RECEPTOR SIGNALING

A Dissertation in Neuroscience by Megan Morse

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

May 2012

The Dissertation of Megan Morse was reviewed and approved* by the following:

Robert G. Levenson Professor of Pharmacology Co-director, MD/PhD Program Dissertation Advisor Chair of Committee

Patricia S. Grigson Professor of Neural and Behavioral Sciences Co-Chair graduate program in Neuroscience

Victor Ruiz-Velasco Associate Professor of Anesthesiology

Kevin Alloway Professor of Neural and Behavioral Sciences

*Signatures are on file in the Graduate School

Abstract

Opioid receptors are G-protein coupled receptors (GPCRs) that are activated by opioid ligands. These ligands offer powerful medical benefits due to their analgesic properties, but activation of opioid receptors can also lead to many negative side effects, including tolerance and dependence. Conventional theory ascribes most of the analgesic and addictive effects of opioids to the activation of the mu opioid receptor subtype (MOR). However, it has recently been suggested that the other classic opioid receptors, the delta opioid receptor (DOR) and the kappa opioid receptor (KOR), may also contribute important functionality. We hypothesize that all of the classic opioid receptors play a vital role in understanding the full functionality of the opioid signaling system. Directed drug design is the future of therapeutic development, and in order to be capable of creating targeted opioid analgesics, we must be able to break down the signaling pathway. Using dynamic mass redistribution (DMR) assays and biosensor technology, we characterized all three classic opioid receptors using a library of known opioid ligands. We also studied the roles of cellular context in opioid receptor signaling by characterizing the endogenous population of opioid receptors found in SH-SY5Y neuroblastoma cells. Utilizing 13 different assay formats, this technology allowed us to examine receptor specificity, G-protein coupling, and downstream pathway selectivity. It has been suggested in the literature that ligand-directed functional selectivity provides a sufficient basis for understanding GPCR activity and molding the future of drug design. We hypothesize that the use of biosensor high throughput technology presents a viable opportunity to fully decipher the complexities of the opioid signaling cascade.

Table of Contents

List of Figures………………………………………………………………………...iv List of Tables……………………………………………………………………….....v List of Abbreviations…………………………………………………………………vi Chapter 1: Literature Review……………………………………………………….1 Opioid Drugs…………………………………………………………………..1 Endogenous Opioids………………………….……………………………….3 Opioid Receptor Structure and Function……………………………………...4 MOR…………………………………………………………………………..8 KOR……………………………………………………………….…………..9 DOR…………………………………………………………………………..10 ORL-1…………………………….…………………………………………..11 Cellular and Anatomical Distribution………………………………………...11 G-protein Coupling and Signaling…………………………………………....12 Opioid Receptor Specificity and Functional Selectivity…………………..….14 Receptor Internalization: First Implication for Functional Selectivity…….....15 Opioid Receptor Dimerization………………………………………..………19 Introduction to Biosensors………………………………………………..…..21 Theory behind Resonance Biosensors…………………………………….….22 Dynamic Mass Redistribution…………………………………………….….23 Biosensors: A revolutionary way to study the opioid receptors? ………..…..25 Rationale and Hypothesis………………………………………………..…...26 Chapter 2: Ligand-directed functional selectivity at the mu opioid receptor revealed by label-free integrative pharmacology on-target…………...…28 Introduction……………………………………………………………...…...28 Experimental Procedures………………………………………………….....30 Results…………………………………………………………………...... …39 Discussion…………………………………………………………….…...…56 Chapter 3: Label-free integrative pharmacology on-target of opioid ligands at the opioid receptor family …………………………….…….….59 Introduction………………………………………………………………...... 59 Experimental Procedures………………………………………………...…..60 Results…………………………………………………………………....…..64 Discussion………………………………………………………………..…..85 Chapter 4: Closing Discussion…………………………………………………...... 89 References…………………………………………………………………………...103 Appendix…………………………………………………………………………….116

iii

List of Figures

1.1: Structure Homology of classic opioid receptors…………………………...…………..…6 1.2: Schematic overview of genetic knockdown……………………………...…………….....8 1.3: Anatomical distribution of opioid receptors in the rodent brain………...... ……………..12 1.4: 25 years of understanding in GPCR advancements…………..……………………...…..15 1.5: GPCR signaling: G-protein vs β-arrestin pathways………………………………..….....16 1.6: Principles of two types of label-free biosensors…………………………………..…...... 22 1.7: Principles of DMR……………………………………………………….…………...... 24 2.1: Dose dependent DMR responses of ligands in HEK-MOR cells……..………………....40 2.2: Blockage of MOR agonist-induced DMR by naloxone…………………..…………..…42 2.3: Numerical descriptor of opioid ligand pharmacology………………………………..….43 2.4: False colored heat map of 42 opioid ligands on HEK-293 and HEK-MOR………...... 45 2.5: The DMR characteristics of BNTX and 0.1% DMSO………………………………..…47 2.6: DMR response of levallorphan, β-funaltrexamine, and naltrindole………….……….....47 2.7: ICI 199,441 induced DMR responses…………………………………….………….….50 2.8: Comparison on DMR and cAMP responses induced by ligands………………….….…51 2.9: Sensitivity of opioid ligand-induced DMR to PTx…………………….………….…….52 2.10: Sensitivity of opioid ligand-induced DMR to CTx and Fsk………………………..….53 2.11: Comparison of immediate responses of buffer and inhibitor pretreated HEK-MOR cells……………………………………………………….……….……54 2.12: Comparison of 30 min poststimulation responses of buffer and inhibitor pretreated HEK-MOR cells……………………………….………….……55 3.1: Extracting DMR parameters for effective similarity analysis…………………….…..…65 3.2: A false colored heat map of DMRs of opioid ligands in five cell lines……………...….68 3.3: Dose-dependent responses of agonists in distinct cell lines…………………...... 70 3.4: A false colored heat map based on the selectivity of opioid ligands to the opioid receptor family…………………………………………………...…....71 3.5: The inhibition pattern by opioid ligands………………………………………………...74 3.6: A false colored heat map of functional selectivity of opioid ligands at the DOR…….....75 3.7: A false colored heat map of functional selectivity of opioid ligands at the KOR…….....79 3.8: A false colored heat map of functional selectivity of opioid ligands at the endogenous receptors in SH-SY5Y………………………………………….…...….81 3.9: Dose responses of a panel of opioid ligands in HEK-DOR cells…………….....…….…82 3.10: Dose responses of a panel of opioid ligands in HEK-KOR cells…………….….....…..83 3.11: Dose responses of a panel of opioid ligands in SH-SY5Y cells………………...... …..85 Appendix A: DMR responses of ligands with activity in parental HEK293 cells…………..116

List of Tables

2.1: Assay layout for MOR cells……………………………………………………..………38 3.1: Assay design for all cells………………………………………………………...... …….63 Appendix B…………………………………………………………………………………..117 Appendix C…………………………………………………………………………………..119 Appendix D…………………………………………………………………………………..121

List of Common Abbreviations

GPCR G-protein coupled receptor

DOR delta-opioid receptor

KOR kappa opioid receptor

MOR mu opioid receptor

ORL1 opioid receptor-like 1

DMR dynamic mass redistribution

P-DMR positive DMR

N-DMR negative DMR iPOT label-free integrative pharmacology on-target

HEK human embryonic kidney

CTx cholera toxin

PTx pertussis toxin

TCT tissue culture treated

DMSO dimethyl sulfoxide

DMEM Dulbecco’s modified Eagle’s medium

HBSS Hank’s Balanced Salt Solution

HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

KO Knockout

CNS Central Nervous System cAMP cyclic adenosine monophosphate

Chapter 1 Literature Review

Opioid Drugs Opioids are powerful compounds that have numerous, unparalleled medicinal benefits. These compounds can be endogenous or exogenous, natural or derived. In clinical settings, opioid-based drugs are currently the best choice for the treatment of severe, chronic, and unremitting pain. Unfortunately, alongside the medicinal benefits of opioids are a number of side effects caused by their use and misuse. These side effects, which can by physical or behavioral, are often as mild as constipation and nausea, but have the potential to be as severe as tolerance, withdrawal, and ultimately addiction (Neve, 2009; Kieffer & Evans, 2002; Palos et al., 2004; Stein et al., 2003). With this in mind, the current direction of opioid research is focused on separating the beneficial functions of opioids from their detrimental counterparts. (Kieffer & Evans, 2002; Corbett et al., 2006). This ‘holy grail’ of addiction research has thus far been elusive, but a growing body of knowledge regarding opioids and their activity is beginning to open the door on a potential breakthrough in drug discovery. The medicinal properties of opioids have been known and applied for over 5000 years. The early opioids were derived from the poppy plant, Papaver somniferum, to relieve pain and produce euphoria (Waldhoer et al., 2004). Currently, morphine is the most common medicinally used opioid derivative. First isolated in 1806 from opium alkaloids, the drug was named for Morpheus, the Greek god of dreams. In the intervening centuries, morphine has had a profound impact, gaining prevalence in both clinical settings, where it is used for pain treatment, and illicit settings, where it is used as a drug of abuse (Contet et al, 2004). Morphine has been well established as the prototypical example of an exogenous non-peptide opioid drug (Kieffer, 1995). Yet, while morphine remains the gold standard of analgesics, there is still much to learn about opioid receptor activation and the mechanisms of analgesia. Opioid drugs are powerful pain relievers. This is because opioid receptors are part of a descending inhibitory system which modulates spinal cord pain transmission 1

(Inturrisi, 2002). The antinociceptive actions of opioids occur at numerous sites in the brain including the periaquaeductal gray, the rostral ventral medulla, the substantia nigra, and within the spinal dorsal horn (Eriksen & Sjogren, 1997). At the cellular level, opioids act to inhibit pain by decreasing neurotransmission release and decreasing synaptic transmission (Inturrisi, 2002). Opioids cause inhibition of noxious stimulation at the brainstem level, influence the ascending forebrain systems and direct cortical or thalamic inhibition (Erikson & Sjogren, 1997). While opioids are clinically unsurpassed as analgesics, they are also illegally used and abused on the street. Along with their pain-reliving characteristics, opioids can also induce a sense of euphoria and well-being (Bailey & Connor, 2005). This state of euphoria and relaxation after abusing opioids is thought to be the motivating factor which drives their misuse. It has also been suggested that the use of opioids can be correlated with the abuse of other drugs, increased risky drug-related behavior, and functional problems. This correlation was documented when assessing over 50,000 clients at substance abuse centers (Bulter et al., 2010). Due to their prevalence in clinical settings, the availability of opioids has dramatically increased in the past decade. In 2007, it was estimated that out of all new drug users, 2.14 million abused pain relievers (especially Oxycontin and Vicodin), while only 2.09 million used marijuana. This staggering statistic from the 2008 National Survery on Drug Use and Health demonstrates that opioids have recently become the top gateway drug in the United States. Also, in 2007 overdoses of medicinal opioids led to 11,499 known deaths, more than heroin and cocaine combined (Okie, 2010). This evidence suggests that the availability of medicinal opioids has allowed them to become just as dangerous, if not more dangerous, than illicit substances like heroin and cocaine. In addition, an increase in medicinal opioid prescription, sales and usage has led to the widespread availability of the drugs. It has brought abusable drugs into rural areas where they were previously unavailable. This increase in the recreational use of opioids has become a severe societal risk, reiterating the need for the “holy grail” of opioid drugs. Scientific attempts to develop clinically functional morphine analogues that are safer and less addictive have been unsuccessful at best, and at times perilous, as was the case with heroin. Heroin, a derived compound, was initially hailed as a safer and less

addictive alternative to morphine but has since become one of the most dangerous and most profitable illicit drugs worldwide (Corbett et al., 2006; Jordan & Devi, 1998; Keiffer & Gaveriaux-Ruff, 2002). The claims for reduced respiratory depression and reduced dependence liability for heroin were clearly ill-founded, yet this impactful mistake did not deter researchers from continuing the search for the ultimate designer opioid drug. This desire to design more appropriate opioid drugs has led to a surge in interest to understand the endogenous opioid system. It is apparent that we must extensively study the endogenous ligands and receptors to fill in our gaps in knowledge, and use that information to derive novel ligands for the system. The current overriding hypothesis of the opioid system states that opioid ligands behave very complexly by exhibiting functional selectivity and acting on multiple opioid receptors (Gharagozlou et al., 2003). Yet, there is a disparity in the field in determining the best way to derive these potentially miraculous drugs. Some believe that we need to isolate the effects of opioid ligands to singular receptors and singular pathways while other believe the answers lie in manipulating the system so that a ligand will bind with significant affinity to all of the opioid receptors. This will lead to agonists that activate various combinations of agonistic and antagonistic effects in multiple opioid receptors (Filizola et al., 2001).

The Opioid System Endogenous Opioids There are three families of genes which encode endogenous opioid peptides: pro- enkephalins, prodynophins, and pro-opiomelanocortin (Kieffer, 1999) The genes encode peptides (such as enkephalins, dynorphins, and endorphins) which bind to the delta opioid receptor, kappa opioid receptor and mu opioid receptor, respectively. These peptides all contain a similar pentapeptide sequence, Tyr-Gly-Gly-Phe-Met/Leu (YGGFM/L) (Waldhoer et al., 2004). This conserved sequence is found at the amino- termini (Snyder & Childers, 1979; Waldhoer et al., 2004). Endogenous opioid peptides are implicated in numerous biological events such as modulation of pain, control of mood, drive and reinforcement, and control of attentional mechanisms (Akil et al., 1984). The most well known function of the endogenous opioid system is the control of pain,

which became apparent after studies of stimulation-produced analgesia (SPA) (Akil et al., 1984; Kieffer, 1995; Mayer & Liebeskind, 1974). It was seen that stimulation of specific locations in the brain led to a reduction in responsiveness to pain, and this analgesic effect was inhibited by pretreatment with naloxone (a known opioid antagonist). The endogenous opioids are also involved in stress-induced analgesia (Akil et al., 1984). Opioid antagonists can also reverse stress-induced analgesia, which supports the hypothesis for the involvement of endogenous opioid systems (Akil et al., 1984). Interestingly, endogenous peptides will interact with multiple opioid receptors, even though they are most selective for one receptor (Lord et al., 1977). This suggests that even in the endogenous opioid system the receptors do not function individually, and we need to remember the complexity when translating the endogenous properties to designer drugs.

Opioid Receptor Structure and Function Though the medicinal properties of opioids have been utilized for centuries, little characterization of the opioids themselves had been done prior to the late 20th century. The opioid receptors were discovered before the characterization of opioid ligands (Mansour et al., 1988). The opioid receptors were first identified in the 1970’s, when researchers discovered receptors in the brain and peripheral tissues that were sensitive to morphine (Pert & Snyder, 1973). This sparked an interest in finding the endogenous ligands of these receptors (as morphine is a derived exogenous ligand). Recent advances have quickly forwarded opioid research, and we now understand the fundamental components of the endogenous opioid system. However, much remains enigmatic when looking more closely at the fine details and mechanisms of opioid activity. Opioids bind to a group of endogenous G-protein coupled receptors (GPCRs) known as the opioid receptor family. GPCRs are the largest family of cellular surface molecules involved in signaling transmission (Marinissen & Gutkind, 2001). All GPCRs share similar primary amino acid sequences, a seven-transmembrane spanning structure, and the ability to mediate intracellular signaling through the activation of heterotrimeric GTP-binding proteins (Gainetdinov et al., 2004). This mediation leads to amplification of signaling due to a cascade of downstream events that are triggered by GPCR activation

(Law & Loh, 1999). This class of signaling molecules is important because of its vast range of therapeutic benefits. At least one quarter of all drug-targeted therapeutics are designed around this large class of receptors (Hopkins & Groom, 2002; Neubig & Siderovski, 2002). The classic opioid receptor family consists of three members: Mu (MOR), Kappa (KOR), and Delta (DOR) (Keiffer, 1995). There is also a fourth opioid receptor, ORL-1, which is similar in structure but not necessarily in function to the classic opioid receptors. This has led researchers to believe that it is atypical (Mollereau et al., 1994). Sequence analysis of the opioid receptors has proven that MOR, DOR, KOR, and ORL1 all belong to the class A subfamily (rhodopsin-like) of GPCRs (Law et al., 2000). The seven transmembrane spanning regions of each of the opioid receptors are connected with three intracellular and three extracellular loops. Their sequences all begin with an extracellular N-terminus and end with an intracellular C-terminal tail. The classic opioid receptors are 372-398 amino acids in length (Keiffer, 1995). The genes that encode the opioid receptors were isolated and mapped in the early 1990’s. Results from this research showed that the three classic opioid receptors (MOR, KOR, and DOR) were highly homologous at the structural level. On average, the receptors are about 60% identical (Fig 1.1) (Kieffer, 1995; Law et al., 2000; Reisine & Bell, 1993). This high degree of conservation of structure suggests similar conservation of the functional properties of each of the receptors.

Figure 1.1 Schematic view of the overlay of MOR, DOR, and KOR, in order to visualize structural homology. Black circles represent regions which are conserved in all three types of classic opioid receptors; red circles are conserved in 2 of the 3 receptors, and white circles are not homologous. Source: Kieffer, 1995 (modified).

The greatest structural homology can be found in the transmembrane domains and intracellular loops, while the terminal ends and extracellular loops are the most divergent (Law et al., 2000). All three receptors have DRY binding sequences in the second and third intracellular loops that are conserved among many G-protein coupled receptors, which allows for binding similarities between GPCRs (Reisine, 1995). It has been proposed that the seven transmembrane domains are arranged sequentially and circularly in a counterclockwise fashion. This area provides a dynamic binding interface for opioid receptor ligands (Waldhoer et al., 2004). The similarity in sequence between helical cores of the classic opioid receptors suggest that there will be cross reactivity between ligands, since most ligands bind into the pocket made when the transmembrane domains fold and create a hydrophobic binding region (Kieffer & Evans, 2009). Also notable are the similar structures of the third intracellular loop and the amino acid stretch of the C-terminal tail, which are almost identical in MOR, DOR, and KOR. This suggests that the classic opioid receptors can have similar interacting proteins 6

(Kieffer, 1995). This theory plays a large role in the subsequent breakdown of opioid receptor signaling, as the GPCR interacting proteins are crucial for determining downstream signaling and functionality (Brady & Limbird, 2001; Jin et al., 2010). This theory has been supported by the ability for all opioid receptors to activate similar pathways and downstream signaling. The extracellular domains of MOR, KOR, and DOR, which include three extracellular loops as well as the N-terminal domain, differ strongly in sequence between subtypes. It has been suggested that these variations in extracellular domains lead to receptor selectivity (Waldhoer et al., 2004). While some ligands do act indiscriminately, there are many opioid ligands which are receptor specific. 3D computer modeling of the receptors shows that these extracellular domains likely form a protein gate, regulating which proteins are capable of entering the binding pocket of the helical core. Therefore, while the binding cavity is comprised of hydrophobic regions that exhibit highly conserved structure, the diverse, hydrophilic extracellular loops act as gates that allow specific, specialized ligands to bind to each of the opioid receptors (Waldhoer et al., 2004; Kieffer & Evans, 2009). This has been demonstrated through the generation of chimeric receptors (Kong et al., 1994; Wang et al., 1995). Reisine’s group showed the structural importance of the N-terminus for selectively binding agonists and antagonists by using chimeric receptors with exchanged extracellular NH2 termini (Kong et al., 1994). Also, chimeric and site-mutagenized DORs and MORs determined multiple extracellular domains, mostly in the first and third extracellular loops, determined receptor selectivity (Wang et al., 1995). After the identification of the opioid receptors, the next goal was to understand the functionality of each receptor subtype. Activation of the family of opioid receptors leads to many behaviors, such as responses to noxious information and stress, reward, and motivation. They also control autonomic functions including respiration, thermoregulation, gastrointestinal motility and immune responses (Waldhoer et al., 2004; Pasternak, 1993). It is important to emphasize the fact that agonists of the MOR, KOR and DOR have all been shown to produce analgesia (Ossipov et al., 1997; Pasternak, 1993).

In the early 1990’s, genetic knockout (KO) studies were first utilized to try and tease outR differencesEVIEW in the functional characteristics of each of the classic opioid receptors (Kieffer, 1999; Matthews et al, 1996). See Figure 1.2. long been believed that morphine activates multiple morphine receptors in vivo. The investigation of morphine responses in mice lacking ␮-receptors has now elucidated its molecular mode of action unequivocally (Fig. 1; Table 1; Refs 25–34). Mice that lack ␮-receptors (MOR-deficient mice) have been generated by several laboratories, by disrupting MOR gene exon 1 (Refs 25–27), exon 2 (Ref. 28) or exons 2 and 3 (Ref. 29). Analgesia,Figure. the 1.2 main Schematic therapeutic actionoverview of morphine, of was investigated extensively after acute subcutaneous (s.c.) administrationgenetic knockdown of the drug. blocking Morphine opioid analgesia was abolishedreceptor at doses functionality. that produce potentSource: analgesia in wild-typeKieffe mice r,(up 1999. to 50 mg kg–1, s.c.) in tail immersion28, tail-flick25–27 and hotplate25,28 tests. Similar results were obtained following intrathecal or intracerebroventricu- lar (i.c.v.) administration27. The injection of very high doses of morphine in mutant mice indicated a 110-fold

• analgesia increase in the ED50 value when morphine was injected • reward by the s.c. route, and no analgesia following i.c.v. admin- • withdrawal istration (25 ␮g, that is, 15 times the ED50 value in wild- • respiratory depression type mice)29. Altogether, these data show the absence of • immunosuppression • constipation morphine antinociception in MOR-deficient mice at doses that classically induce strong analgesia in wild- Fig. 1. The molecular mechanism of action of morphine. Disruption of type mice. This demonstrates that the MOR-encoded Bythe ␮ !knockingopioid receptor (MOR) out gene the leads opioid to a complete receptors loss of the main singly and in combination, the roles of the biological actions of morphine, demonstrating that both therapeutic and receptor is necessary to mediate morphine action on pain classic opioidadverse effects receptors of the prototypic were opioid first result fromvisualized its interaction .with The mostpathways notable and suggests effect thatof the␦- and initial ␬-receptors KO do not a single gene product. Data from Refs 24–29, 32, 33. participate in morphine analgesia under standard experi- studies was that transgenic mice lacking the MOR nomental longer conditions. exhibi Thisted is morphine corroborated- induced by the finding that morphine analgesia is preserved in mice deficient in analgesiahomologous (Matthes recombination. et al., 1996). It is The the purposegenetically of this alteredthe KOR animals (Ref. 30) orare DOR viable, (Ref. 31) fertile, genes. and review to summarize what has been learnt from the very Other major pharmacological actions of morphine show noearly obvious analysis developmentalof these mutant mice, deficits. and to discuss This thesuggestshave beenthat studiedthe opioid in MOR-deficient system is mice. not One of the future outcome expected from the detailed observation most common adverse side-effects of morphine treat- essentialof thesefor survivalexquisite ‘tools’. (Kieffer Issues &that E arevans, being 2009). addressed Morement in -isdepth respiratory functions depression, of thea biological opioid action that currently in several different laboratories are as follows: can be seen after acute administration and which receptors(1) How are dodiscussed opioids act below. at the molecular level, and what requires tight control in the clinic. Matthes et al. showed are the therapeutic implications? (2) What is the molecu- that an analgesic dose of morphine (6 mg kg–1, s.c.) lar basis for the hypothesized heterogeneity of ␮!, ␦! and decreased respiratory frequency and increased res- ␬!opioid receptor subtypes and the postulated interac- piration time in wild-type mice. However, no change in Mu Opioidtions between Receptor opioid (MOR) receptors? (3) How crucial is the respiratory parameters could be measured in similarly opioid system for survival and how are its various com- treated MOR-deficient mice24. Respiratory depression is ponentsThe MOR implicated is expressed in responses primarilyto threatening in environ- the cortex,the primary limbic factor system in the lethaland braintoxicity stem of morphine. of Loh mental stimuli? et al. showed that an extremely high dose of morphine the central nervous system (CNS) (Merrer et al., 2009).(1600 mg The kg –1MOR, s.c.) was was required shown to kill initially the mutant to mice The mode of action of opioid drugs: which and that death occurred without any of the typical mor- have anmolecular inarguable target? effect on nociception and addiction,phine effectsby mediating29. Another undesirableboth the actionbeneficial of morphine and adverseMorphine effects is the of prototypic the most opioid. broadly In contrast used to many opioidsis constipation:(Matthes etRoy al., et al 1996;. demonstrated Sora et that al., a single s.c. synthetic opioids that have been developed in the past 20 injection of 15 mg kg–1 morphine greatly inhibits gastro- 2001). years,Tail morphineimmersion, is an opioidtail flick compound and hotwith platelow recep- tests intestinalhighlighted motility that in wild-type the MOR mice, KO whereas animals no change tor selectivity. Binding studies performed on rodent in gastrointestinal transit was seen in mutant mice at were insensitivebrain membranes to the22,24 analgesicor recombinant effects receptor of morphine prepar- doses (Matthes up to 35 mget al.,kg–1 (Ref.1996). 32). CondAcute morphineitioned treat- ations23 have shown that morphine exhibits a preference ment also induces modification of locomotor activity. place preferencefor ␮-"#$#%&'"( experiments, with Ki values in demonstratedthe nanomolar range, that but KOTian animals et al. showed were thatalso horizontal immune locomotor to the hyperac- also binds to ␦- and ␬-receptors with submicromolar tivity and the inhibition of vertical locomotion, observed affinities. This two-order-of magnitude selectivity factor in wild-type mice, was absent in MOR-deficient mice26. might be sufficient to examine ␮!"#$#%&'" responses8 using Morphine also induces euphoria, a response that can be in vitro assays, but is rather low to ensure ␮-receptor evaluated in animal models using the place-preference selectivity under in vivo testing conditions. Thus, it has paradigm. Matthes et al. showed that morphine-

20 TiPS – January 1999 (Vol. 20)

rewarding effects of morphine (Sora et al., 2001). These results encouraged researched to mainly focus on the MOR subtype, especially throughout the 1990’s and into the turn of the century. Knocking out the MOR caused morphine’s analgesic and addictive properties to be eradicated, which suggested that the therapeutic and adverse effects of opioids are linked to activation of the MOR (Contet et al., 2004). However, more recently, light has been shed on the importance of the remaining members of the classic opioid receptor family. DOR and KOR are now being examined as possible analgesic drug targets. DOR and KOR ligands have the potential to induce analgesic benefits without the severe addictive side-effects which are often seen with MOR agonists (Vanderah, 2010).

Kappa Opioid Receptor (KOR) Kappa opioid receptors are expressed widely throughout the CNS. They are specifically activated by endogenous opioids derived from prodynorphin (Chavkin et al., 1982). Activation of KORs following stress has been shown to induce dysphoria and increase drug-seeking behaviors (Bruchas & Chavkin, 2010). This was seen when dynorphin (an endogenous kappa ligand) released during stress produced immobility in a forced swim test (Bruchas & Chavkin, 2010). It is hypothesized that KOR agonists could be used as beneficial analgesics for visceral pain, as KOR-deficient animals show increased sensitivity to peritoneal injections of acetic acid (Waldhoer et al., 2004; Gebhart et al., 2000). KOR agonists have been suggested to cause less tolerance and dependence than mu agonists, yet some groups have shown that neuropathic pain activates the KOR system and induces receptor tolerance (Clayton et al., 2009). KORs also do not produce the severe respiratory depression (Reisine, 1995). This potential for KOR analgesics to cause fewer or less severe side effects needs to be more fully studied. Knockout studies have shown that KORs mediate the dysphoric activities of opioids, as well as cannabinoids (Simonin et al., 1998; Ghozland et al., 2002). Expression levels of prodynorphin, the precursor of endogenous kappa ligands, are altered in the brains of drug abusers and psychiatric patients (Merg et al., 2006). Thus, while KOR agonists exhibit numerous of benefits, they also have negative side effects. This reiterates

the need for directed therapeutics to select for the analgesic effects of KOR agonists while avoiding the dysphoric effects. An interesting phenomenon is the ability of KOR agonists to functionally antagonize MOR agonists (Fields, 2004; Ackley et al., 2001; Pan et al., 1997). This supports the theory that opioid receptors are highly co-dependent, and the necessity to understand all three of the classic opioid receptors in order to properly derive targeted drugs.

Delta Opioid Receptor (DOR) The DOR was the first opioid receptor to be cloned, which was completed in 1992 (Waldhoer et al., 2004). It was identified simultaneously by two groups and isolated from NG-108 neuroblastoma cells (Evans et al., 1992; Kieffer et al., 1992). Early knockout studies showed that DORs were involved with nociception by interacting with MORs (Matthes et al., 1998) but since then, it has been seen that DOR activation leads to analgesic effects individually. Researchers saw that treatment with DPDPE, a full DOR agonist, led to spinal analgesia, which was seen to be abolished in DOR deficient mice (Kieffer, 1999). Along with pain-reliving properties, delta receptors are more distinct from KOR and MOR receptors functionally. KO studies and in vivo pharmacology studies have shown that they regulate emotional responses and exhibit anxiolytic and antidepressant activity (Filliol et al., 2000; Kieffer & Gaveriaux-Ruff, 2002; Jutkiewicz, 2006). Also, DOR agonists have the ability to act as analgesics with limited side effects (Vanderah, 2010; Waldhoer et al., 2004). Thus, there has been a focus on drug development for compounds that target the DOR (Vanderah, 2010; Wei et al., 2000; Dondio et al., 1997). Studying each of the three classic opioid receptors separately tends to lead to broad generalizations. MOR activation results in the most potent analgesic activity but also results in the highest levels of tolerance and dependence. The DOR agonists seem to exhibit less addictive potential, but are not necessarily as potent pain relievers as MOR agonists. The KOR has some medicinal benefits, but is researched cautiously due to its strong dysphoric properties. Individually, each of the opioid receptors has a number of

negative characteristics that need to be selected out deliberately, so that a combination of the best characteristics of each may be combined for maximal results.

Opioid Receptor-Like 1 (ORL-1) A fourth, orphan, opioid receptor was discovered shortly after the cloning of the classic opioid receptors. The nociceptin receptor (also known as opioid receptor-like 1) was first identified in 1994. Early studies showed low binding affinity to known opioid ligands, yet the protein exhibited greater than 90% sequence identity and approximately 60% homology to the classic receptors (Waldhoer et al., 2004). This structural similarity led to its classification as the fourth opioid receptor. However, the ORL-1 receptor is not studied as extensively as the classic opioid receptors for analgesic properties as it has been seen that this receptor is similar in structure, but not in function to the classic receptor subtypes (Chan et al., 2002; Meunier et al., 2002). ORL-1 deficient animals have unaffected basal nociceptive responses and analgesic responses to morphine (Mogil & Pasternak, 2001). Due to its variance in functionality from the classic opioid receptors, little focus will be placed on the ORL-1 receptor in this dissertation.

Cellular and Anatomical Distribution Opioid receptors are broadly expressed throughout the central and peripheral nervous systems. Generally, the distribution of the receptors correlates with the reported functions of the opioid system (Dhawan et al., 1996). Thus, it is unsurprising that the classic opioid receptors are seen in all brain areas that are part of the “pleasure-seeking pathway” or the circuits of addiction (Merrer et al., 2009; Fields, 2007). Binding sites for the receptors overlap in most brain regions, but there are some distinct differences in expression levels in various areas, as seen in Fig. 1.3 (Merrer et al., 2009). The implications of this widespread circuitry for the opioid receptors, which has a large amount of overlap and redundancy between brain areas, are that the circuitry of the opioid system is highly integrative. This idea will greatly affect future drug development, as the opioid system again reiterates the necessity for integrative studies of opioid signaling.

Figure 1.3 Anatomical distribution of opioid receptors in the rodent brain, from Merrer et al., 2009. Color of marker distinguishes receptor type, while densities of receptors are represented by size of marker. Source: Merrer et al., 2009. It is important to note that opioid receptors are widely distributed throughout the CNS and PNS, in multiple cell types. Thus, it is feasible that receptors expressed in various cells types experience varying regulatory mechanisms (Raehal & Bohn, 2005), forwarding the complexity of the opioid receptor system. Receptor signaling is greatly dependent on location and cellular context, which plays a vital role when carrying out in vitro and in vivo studies of opioid receptor signaling.

G-protein Coupling and Signaling

There are four known coupling classes of GPCRs (Gi /Go, Gs, Gq and G12/G13) (Schröder et al. 2010). It is generally accepted that the opioid receptors are inhibitory

(Gαi) coupled GPCRs (Reisine & Bell, 1993). When activated, opioid receptors modulate

intracellular effects and signaling through Gi /Go proteins. The binding of agonists to opioid receptors causes a conformational change of the cytoplasmic domain of the receptor that binds to the G-proteins. This modifies the helical arrangement of the receptor, and the alteration of structure has been suggested to drive the transition from the inactive to active state of the receptor (Kieffer & Evans, 2009). Along with recruitment of the active state, activation of opioid receptors leads to inhibition of adenylyl cyclase, activation of potassium conductance by stimulation of inwardly rectifying K+ channels (GIRKs), inhibition of calcium channels, and inhibition of neurotransmitter release (Williams et al., 2001; Chen et al, 1993). The modulation of potassium and calcium 12

conductance (increase and decrease, respectively) serves to reduce the membrane excitability, which leads to the decrease in secretion of neurotransmitters (Chen et al., 1993). These effects are produced by both exogenous and endogenous ligands, and they function to modulate nociception in the central nervous system.

The most convincing evidence that opioid receptors are Gαi-coupled GPCRs comes from their sensitivity to pertussis toxin (PTx). PTx has the ability to inactivate the stimulatory function of the Gαi subunit of the signaling complex, thus removing the inhibitory component of the pathway (Jordan & Devi, 1998). Opioid evoked responses are blockedby PTx, thus suggesting Gi /Go receptor coupling (Wong et al., 1988). Yet, it has recently been suggested that other G proteins also play a role in mediating opioid receptor signaling (Neve, 2009; Waldhoer et al., 2004). Current research has shown evidence that opioid receptors are pleiotropic, thus capable of binding to and activating more than one G-protein subtype (Neve, 2009; Kenakin, 2002). Recent evidence also suggests a PTx-insensitive or PTx-resistant G-protein transduction pathway can be activated by certain MOR agonists (Mostany et al., 2008). It is hypothesized that pleiotropy of G proteins can delay the development of morphine tolerance and thus may represent a possibly therapeutic target for improving the clinical uses of opioids (Hendry et al., 2000). While the pleiotropic nature of opioid receptors has become well-accepted, there remains a debate in the scientific community about the possibility of opioid receptors being linked to Gαs. It is hypothesized that a Gαs-coupled component of opioid signaling is related to tolerance and the phenomenon of addition, but the mechanisms are still mostly unknown (Chakrabarti et al., 2010; 2005). Yet, researchers have recently observed opioid activation which was unaffected by PTx, but was sensitive to pretreatement with

Cholera toxin (CTx), a Gαs protein activator (Chakrabarti et al., 2005; Shen et al., 1990).

These pieces of information suggest that a Gαs signaling pathway is evident. However, some past researchers were unable to immunoprecipitate Gαs with MOR, despite being able to visualize other G proteins binding with MOR (Chalecka-Franaszek et al., 2000). Other groups were able to visualize the association and showed that the magnitude of this coupling is enhanced by morphine treatment, which may be relevant in the future to understanding opioid tolerant mechanisms (Chakrabarti et al., 2005). It should be noted

that this physical association which was pulled out in stably-expressing cells does not necessarily imply that there is a functional interaction. Yet, this biochemical evidence, combined with inhibitor data, forwards the hypothesis that opioid receptors activate a Gαs signaling pathway. While research has focused heavily on the MOR, groups have also looked at the impact of Gαs signaling on the other receptor subtypes. Not much is yet known about this phenomenon. Yet, evidence of Gαs activation was seen by KOR activation in cultured hippocampal neurons (Hampson et al., 2000), suggesting that this is a phenomenon that should be studied in greater detail.

Opioid Receptor Specificity and Functional Selectivity The old dogma of GPCR signaling, known as classic receptor theory, claimed that the efficacy of a drug was independent of the receptor system (Kenakin, 2004; Kenakin 1997). It argued that there was a single, ligand bound, active conformation of a receptor that led to a fixed signaling response (Neve, 2009). Classic receptor theory has been thoroughly disproven recently, on many fronts (Fig. 1.4). Various opioid ligands have different effects on internalization, desensitization, ligand efficacy and ultimately addiction (Yu et al., 1997). Also, ligands can be classified as full agonists, partial agonists, neutral antagonists, or inverse agonists, based on their efficacy on various receptor subtypes (Zheng et al., 2010). Opioid receptor ligands show “pathway-selective signaling” where one group of agonists will activate one downstream pathway, while another will only activate a separate signaling pathway (Zheng et al., 2010). This idea of functional selectivity has been suggested to be crucial for drug development, as this phenomenon raises the possibility of selecting or designing novel ligands that differentially activate only a subset of functions of a single receptor, thereby optimizing therapeutic action (Urban et al., 2007; Mailman, 2007). One of the earliest hints that the classic receptor theory was not sufficient to explain receptor behavior was seen when the dopamine agonist dihydrexidine exhibited agonist and antagonist effects through the same receptor (Mailman, 2007). This was part of an undeniable stream of evidence disproving the dogma of receptor pharmacology. While the existence of functional selectivity has been documented in the literature (Urban et al., 2007; Schroder et al.,

2010), the correlation of agonist-selective signaling and the functionality of the opioid receptors need to be fully discerned to utilize functional selectivity for design directed therapeutics.

Figure 1.4 Discoveries and advancements in GPCR behavior understanding in the past 25 years. Source: Kenakin, 2004.

Receptor Internalization: First Implication for Functional Selectivity GPCR function is mediated through two main mechanisms: G-protein signaling and β-arrestin function (Violin & Lefkowitz, 2007). Agonists bind to the receptor and stabilize the receptor in a conformation that couples to and activates heterotrimeric G proteins. This leads to activation of downstream secondary-messenger signaling, which mediate many cellular responses and functionality. Activated receptors also stimulate GRKs, which bind β-arrestins. β-arrestins mediate three main properties of GPCR activation: desensitization, internalization and signaling. Along with arguing for intrinstic efficacies of receptors, classic receptor theory also believed that the ligand efficiency for each of these functions was directly proportional to a ligand’s efficacy for G-protein

activity (Benovic et al., 1988). It is now clear that the functional selectivity of GPCR signaling is seen in both G-protein and β-arrestin pathways, which is visualized in Fig. 1.5, and there is no formula for the amount of activity seen by any pathway after a receptor is activated (Urban et al., 2007; Violin & Lefkowitz, 2007).

Figure 1.5 GPCRs signal through two distinct mechanisms. Heterotrimeric G proteins activative one pathway, while β-arrestins are phosphorylated by GRKs and activate the separate, second downstream pathways. Receptor agonists stimulate both pathways, though to different degrees. Source: Violin & Lefkowitz, 2007.

The MOR was one of the first GPCRs shown to support the theory of functional selectivity (Urban et al., 2007). It was seen that different MOR ligands exhibit various levels of endocytotic efficacies and turnover rates of the MOR (Urban et al., 2007). These differences in the endocytotic activity of the MOR did not correlate to classic measures of GPCR activity (such as cAMP activation) and could not be explained by the classic receptor theory. After the activation of GPCRs, receptors can be desensitized by being phosphorylated by G protein-coupled receptor kinases (GRKs) (Gainetdinov et al., 2004). Desensitization is the loss of response subsequent to prolonged or repeated administration of an agonist (Hausdorff et al., 1990; Kelly et al, 2007). Desensitization is typically divided into three main steps: receptor phosphorylation by GRKs, endocytosis, and sorting (recycling or downregulation) (Aguila et al., 2007). MOR desensitization is mediated by GRKs and β-arrestin recruitment. Opioid receptor desensitization and G protein uncoupling can also occur via GRK and β-arrestin independent events (Waldhoer et al, 2004). Evidence shows that morphine and DAMGO (a selective and potent MOR agonist known to cause rapid turnover of the receptor) induce very different degrees of internalization when bound to the MOR (Keith et al., 1996). While receptor activation of

the MOR typically causes internalization and endocytosis of the receptor, treatment with morphine does not lead to rapid internalization of the receptor (Keith et al., 1996). This can be explained by morphine’s very weak ability to stimulate MOR phosphorylation and recruit β-arrestin (Violin & Lefkowitz, 2007). This phenomenon was first visualized in the mid-1990’s through imaging of radioligand binding and was validated with flow cytometic analysis of MORs in stably-transfected HEK293 cells. These experiments demonstrated that ligand functional selectivity does occur in the opioid system. They also led to the hypothesis that functional selectivity plays a vital role in opioid dependence and tolerance, and could hold the key to future directed opioid development. Internalization is a crucial first step in resensitization. The normal turnover of activated GPCRs appears to be necessary for proper functionality, as morphine, an agonist with pronounced negative side effects, does not induce normal turnover. This can lead to exacerbation of tolerance, due to the fact that desensitized receptors cannot internalize and resensitize (Schulz et al., 2004). Therefore, it can be theorized that functional selectivity plays a role in opioid dependence and tolerance. While most of the research surrounding functional selectivity has focused around the MOR thus far, there has also been a notable amount of research done on both the DOR and KOR. It has been seen that agonists of DOR receptor vary in their desensitization rate, and that desensitization and trafficking were able to partially predict the ability of the DOR agonist to promote tolerance (Aguila et al., 2007). While examining the DOR as a target for anti-depressants and anxiolytics, researchers noted the importance of being able to tease out the biological effects of selective agonists. KOR has also been identified as a player in the search to understand functional selectivity. Differences in KOR trafficking have been noted for different agonist in the KOR signaling cascade: one study in particular highlighted that Salvinorin A, a full KOR agonist, was seen to be 40-fold less potent inducing internalization and down regulation of KOR than another full KOR agonist, U50-488 (Wang et al., 2005b). Thus, the opioid system supports the existence of functional selectivity on numerous levels. But most importantly the idea that the opioid system is an ideal representation of functional selectivity at work is the belief that functional selectivity holds the key to directed therapeutics (Mailman, 2007).

GPCR activation leads to a cascade of secondary, downstream signaling events. These events are triggered by G-protein signaling and β-arrestin signaling. It is believed that the various effects and side effects of opioid receptor activation are due to the functional selectivity of ligands targeting different signaling pathways (Polakiewicz et al., 1998; Urban et al., 2007). Kenakin elegantly describes this as “ligands causing receptors to exercise only portions of their often vast repertoires of behaviors” (Kenakin, 2007b). Deconvolution of downstream signaling is vital in order to understand the mechanisms and implications of functional selectivity (Zheng et al., 2010). Among the most influential pathways is the mitogen-activated protein kinase (MAPK) signaling cascades (Gutstein et al, 1997; Fukuda et al., 1996). Kinases are enzymes that phosphorylate target proteins and change their biological activity. They are involved in nearly all physiological functions, and the protein kinase cascades are important mediators of signaling responses. The kinases (specifically ERK, JNK, p38, and PI3K for this study) have been defined as having such roles as proliferation, plasticity, long-term potentiation, cellular survival, and differentiation (Belcheva & Coscia, 2002). In some neuronal cell lines, it has been shown ERK activation is sufficient to induce morphological differentiation (Cobb, 1999). These functions of MAP kinases provide a possible connection for neuronal adaptations to decrease plasticity with accompanying opioid abuse (Nestler et al., 1993). Also, protein kinases modulate internalization and desensitization in cellular signaling-pathways. These processes are crucial for opioid receptor activation as they appear to play critical roles in opioid tolerance and addiction (Chen & Sommer, 2009). The activation of these pathways is ligand-dependent, and this selectivity is believed to play a significant role in the functional selectivity of opioid ligands (Urban et al., 2007). Kinase pathway selectivity has been examined following the activation of all of the classic opioid receptors. In a recent review of the KOR, it was noted that while we know KOR activation triggers p38 MAPK phosphorylation, the mechanisms for this mediation are completely unknown (Bruchas & Chavkin, 2010). DOR agonists have also been shown to differentially activate ERK signaling: specifically, DOR agonist DPDPE activates ERK through the G protein pathways, while DOR agonist TIPP activates through the β-arrestin pathway (Xu et al., 2010).

Opioid Receptor Dimerization Understanding the functionality of GPCRs has recently become even more complex, due to the recognition of the importance of protein-protein interactions (Alfaras-Melainis et al., 2010; Milligan, 2004; Brady & Limbird, 2001). There are numerous proteins involved in trafficking and modulation of opioid receptor signaling (Traynor, 2010). Another level of complexity arose when it was discovered that opioids not only interact with regulatory proteins, they can homo- and heterodimerize with other GPCRs (Waldhoer et al., 2004; Filizola & Weinstein, 2002; Wang et al., 2005a). Unfortunately, little is currently known about the affect that dimerization has on functionality, though it is an active area of research. Dimerization of opioid receptors is an increasingly evident phenomenon. The first evidence of opioid receptor oligomers came from Devi’s group in 1997, when they visualized homodimerization of the DOR (Cvejic & Devi, 1997). Western blotting of CHO cells stably expressing FLAG-tagged DOR exhibited a band of about 120 kDa, which is twice the size of DOR. This was the first suggestion that the receptors homodimerized when overexpressed. This result quickly changed many hypotheses surrounding opioid receptor signaling, as dimerization effects the regulation of biological activity: modulation of ligand binding, alteration in G protein activation, and novel signaling pathways are all documented results (Devi, 2001; Terrillon & Bouvier, 2004; Waldhoer et al., 2004). Studies clearly show that opioid receptors exist as homodimers; there is also evidence for opioid heterodimerization between subtypes (George et al., 2000). The identification of heterodimers started with co-immunoprecipitation assays done in transfected cell lines (Gomes et al., 2002). These studies showed the potential for receptors to physically associate with each other to form dimers or oligomers. These higher order organizations form in the absence of agonist treatment (Gomes et al., 2002). However, the results were taken with a grain of salt, as the receptors in transfected cells can reach non-physiological levels, which left the critics skeptical that oligomers would function in native environments. In order to increase understanding of the functional role of these dimers, assays in live cells and biophysical assays have been utilized to gather

information about proximity and likelihood of interactions of proteins in the context of live cells. Gomes’ group used BRET, a proximity based biophysical method to examine the likelihood of interactions of proteins in the context of live cells (Gomes et al., 2002). While these and other similar experiments reiterated the likelihood of dimerization, they still involved tagging and overexpression of receptors, two very common limitations of the current technology. All three types of classic opioid receptors are known to form heterodimers (Decaillot et al., 2008). This is crucial information, as it is known that endogenous cell types contain more than one opioid receptor (Kazmi & Mishra, 1986). The study of receptors in artificial environments may lead to erroneous conclusions. Therefore, as researchers, we must understand the limitations of transfected systems and be prepared to validate information in physiologically correct situations. We must not only study opioid receptors individually, but also in systems which mimic native cells with multiple receptor types. The overriding goal of opioid research remains development of a drug with the ability to induce analgesia without causing tolerance and dependence (Kieffer & Evans, 2002; Corbett et al., 2006). Researchers continue this quest by taking a step back. It has become apparent that we need to understand the big picture of opioid receptor signaling in order to be able to understand the functional roles that each of these pieces play. Fully breaking down the system is the only way to guarantee the development of therapeutics without the attached side effects. This has led us to the conclusion that we need to adapt the technology we use to study GPCR signaling to be able to examine full, integrative responses. In order to forward the drug discovery process we need to continue to evolve our methodology to parallel the growing base of knowledge about this transmembrane receptor family. Understanding the entire signaling cascade will allow us to exploit and manipulate functional selectivity to create truly targeted therapeutics. We hypothesize that this approach will allow us to find the missing links of opioid receptor signaling, and apply new knowledge to future drug development. As Dr. Mailman eloquently states, functional selectivity has the potential to provide a route to truly novel drugs that could not have been conceptualized from the classic principals of pharmacology (Mailman,

2007). Combining this breakthrough knowledge with innovative technology will change the future of opioid drug design.

Introduction to Biosensors Label-free biosensors are rapidly changing the way cell biology and cellular signaling events are studied. Traditional cell-based assays rely on the measurement of a discrete event. Some of the most common examples of this include the monitoring of levels of second-messengers (cAMP and Ca2+) and translocation of β-arrenstin (Fang et al., 2005a; Kenakin, 2009). Even though these methodologies only measure independent events, they have been successful in the past allowing the scientific community to forward the knowledge of GPCR signaling. Recently it has become apparent that assays studying individual signaling events lack flexibility. Changes in drug design in the past few decades have emphasized the use of technology like biosensor systems, which record integrative responses (Schroder et al., 2010; Codd et al., 2011). Where assays showing individual signaling events only reveal a piece of the puzzle, integrative responses allow the study of the entire picture of receptor signaling. In addition, biosensors are label-free. Using a label-free system has many advantages, including faster assay times than conventional protocols, accurate and high information content, and reduced interference from labels or other manipulations (Cooper, 2006). Optical biosensor technology has recently been lauded for it usefulness in measuring integrated cellular responses which can be readily transferred to downstream signaling data (Schröder et al., 2010; Ferrie et al., 2011). These advances in methodology and their application to studying GPCR signaling have the potential to revolutionize drug discovery. Current directed drug discovery leaves much to be desired, as there is a disconnect between recent advancements in the understanding of GPCR signaling and therapeutic design. This gap will hopefully be lessened by novel, breakthrough methodology.

Theory behind Resonance Biosensors Optical biosensors use a transducer to convert a stimulus-induced cellular response into a quantifiable signal (Fang, 2010a). A biological material (i.e. a protein,

ligand, or cell line) is placed on the surface of a biosensor and a baseline index of refracted light, which is shone through the bottom of the biosensor into the biological material, is measured. Then, when stimulation of the biological material occurs, this produces a change in the material itself that is transformed into a change in the refracted light from the baseline, which can be measured by a transducer (Fang, 2006; Schröder et al., 2011). Previously, this technology was reserved for the study of biomolecular interactions, which focused on determining kinetic and equilibrium constants of molecular interactions (Rich & Myszka, 2000; Homola, 2003). Yet, advancements in instrumentation have shown that resonance biosensors can also be used to look at live, whole cell integrated responses to cellular stimulation. Optical biosensors can convert a molecular recognition or signaling event in live cells event into a quantifiable signal (Fang, 2006). There are several types of optical biosensors. The two most commonly used are surface plasma resonance (SPR) and resonant waveguide grate (RWG) biosensors (see Fig 1.6). Both of these biosensors use a surface bound evanescent wave to characterize changes in a local refractive index at the sensor surface.

a Biacore T200 GWC SPRimager®II

Gold layer θ ( Glass Fig. 1.6 Principles of two types of label-free Prism biosensors. (a) Surface plasmon resonance (SPR),

Polarized light Reflected light which uses light excited surface plasmon polaritons to sense whole cells (b) Resonant waveguide grating (RWG), which uses leaky mode nanograting waveguide structure to generate an b Epicevanescent® EnSpire wave™ to sensorBIND™ whole cell responses.

Waveguide Source: Fang, submitted. Glass

Broadband light Reflected light

c ECIS™ xCELLigence™ CellKey™ SPR biosensors measure the change in refractive index of a solvent, which occurs during a complex formation or disassociation (Rich & Myszka, 2000). They are capable of characterizing binding reactions in real time without manipulation or labeling. SPR biosensors rely on a prism to direct polarized light into a planar glass substrate. One of

the main limitations of SPR biosensors is the speed of the assays, as they are currently only able to process low-throughput samples (Fang, 2010a). The RWG biosensor is a label-free technology sensitive to changes in the local refractive index within the detection zone of the sensor surface (Tiefenthaler & Lukosz, 1989). The measured change in the refractive index of a polarized light is believed to correspond directly to a change in the biological material. Live cells are cultured directly onto waveguide glass. A polarized light, which covers a range of incident waves, is used to illuminate the waveguide. When target molecules bind to the immobilized cells, the resonant wavelength shifts due to an internal change in cellular componentry, referred to as dynamic mass redistribution (DMR). This change is measured by a transducer (Fang, 2006). The RWG biosensor is composed of a glass substrate and a rating embedded waveguide film of high refractive index. The recent large-scale fabrication of RWG biosensors and instruments, as well as advanced, simplified assay protocols, have allowed the RWG system to become the first commercial platform for high throughput biochemical and cell-based assays (Fang, 2006). The Epic® system is based on the use of RWG. Epic® uses a detector system, centered on integrated fiber optics, which measures the ligand-induced shift of reflected light.

Dynamic Mass Redistribution (DMR) It is well known that the activation of GPCRs, including the opioid receptors, leads to a cascade of downstream events (Williams et al., 2001). This internal cascade triggered by GPCR activation can be described as the DMR of the contents of a cell. The Epic® system utilizes optical biosensors to monitor this stimulation-mediated dynamic mass redistribution (Fang et al., 2006). Optical biosensors can be used to monitor the DMR of cells because the local refractive index in a cell is proportional to the mass density at the sensor surface. Therefore, a change in the refractive index (the detected signal output) reflects the redistribution of cellular matter within the sensing volume of the biosensors. The DMR signal is thus representative of the cellular activity. It is a novel and quantifiable cellular readout.

Fig 1.7 Schematic representation of DMR response in a cell, on the Epic® platform. Source: Perkinelmer.com

The Epic® platform uses 384-well cell plates with RWG biosensors in each well. The biosensor for cell sensing in Epic® can be pictured as three layers: the substrate, the waveguide, and the adherent cell layer (Fang et al., 2006). Cells are cultured directly onto the adherent cell layer, and grown to form a confluent monolayer. The Epic® microplates insert into the Epic® platform and the on-board liquid handling system adds the test compounds of choice directly onto the cells. These exogenous signals and stimulations (such as known ligands of GPCRs) mediate the activation of specific cellular signaling events, often resulting in the DMR of cellular contents. When this occurs within the sensing volume of the cells (see in Fig. 1.7) the DMR can be measured, due to a change in the angle of refraction of the evanescent wave. This is monitored in real time. A DMR signal is recorded as a shift in resonant wavelength (picometer, pm), and is a real-time kinetic response with high temporal resolution (15 sec per data point) and long duration (~ hrs). The DMR signal can be considered to be a poly-dimensional coordinate at discrete time points due to the fact that it is a measurement of the sum of all the DMR signaling events that are taking place within the sensing volume of the cells (Ferrie et al., 2011; Fang, 2010a). 24

The ability to study living cells in their native and physiologically relevant context is crucial for understanding their activity and thereby forwarding the drug discovery process. RWG biosensors, and Epic® technology in particular, have produced novel insights regarding the complexities of receptor signaling pathways, agonist functional selectivity, and the modes of actions of ligand molecules. This is due to the flexibility of assays, specifically the ability to add pathway specific inhibitors directly to live cells. High resolution makes it possible for DMR assays to quantify ligand pharmacology at different time domains, while the non-invasive nature allows researchers to probe ligand pharmacology in various conditions (Ferrie et al., 2011). Using DMR analysis, the Epic® platform forwards the field of research by expanding on the benefits of label-free real-time analytical technology.

Biosensors: A revolutionary way to study the opioid receptors? Biosensors, and Epic®, have facilitated a novel study of opioid receptors. Understanding the signaling of opioid ligands, and more specifically the signaling which leads to the addictive side effects of opioid agonists, has remained enigmatic to researchers. It is known that activation of opioid receptor sites leads to activation of a number of downstream signaling pathways. Biosensor allow for the increased flexibility and efficiency of testing which will help researchers understand the impacts of various opioid signaling pathways. The nature of the DMR output allows a broad view of signaling which is not typically provided by traditional cell assays (Codd et al., 2011). DMR assays also represent the global response of a receptor, not just a single endpoint from a conventional assay. The integrative, expansive output of DMR assays, along with the flexibility of assay design and non-invasive nature of the system make it a clear choice for the future of pharmacological research. We believe that using biosensors, paired with DMR technology, we will be able to explore the opioid receptor signaling system in a novel way. This will allow us to both fill in the current gaps in understanding and characterize the opioid receptors for future drug development. The potential of DMR assay to be used in drug discovery was recently highlighted by Ye Fang’s group, through testing the β2-adrenergic receptor (Ferrie et al., 2011). The study characterized all FDA-approved β2-adrenergic drugs. It was observed that DMR

profiles obtained from the drugs on A341 cells (which contain endogenous β2-adrenergic receptors) can be used to establish an effective link between label-free pharmacology and the in vivo therapeutic functions of these drugs. If the system can characterize known compounds, it stands to reason that templates can be made of characterized ligands for each receptor. These activation templates would be vital for future work, as they could be used to screen novel compounds for similar characteristics. Full characterization of known receptors may hold the key for identifying novel compounds which act on the receptors and could exhibit directed therapeutic functionality. The ability to examine living cells in the native state is crucial for understanding the biological functions of cellular targets and receptor signaling, which is vital to the success of drug development and discovery (Fang et al., 2006). The future of drug discovery will revolve around analyzing multi-tonal integrated signaling responses to understand the true affects of biased agonism and functional selectivity. Yet, the full potential of this methodology, and the technology that is required has yet to be determined. We believe this study helps to establish the large-scale potential of biosensor technology in GPCR research.

Rationale and Hypothesis The medicinal benefits of opioid analgesics are greatly inhibited by numerous negative side effects paired with opioid receptor activation. Much of the focus of the field of opioid research currently surrounds the search for the holy grail of opioid agonists. The overriding hypothesis of this research revolves around the idea that opioids have functional selectivity and act on multiple opioid receptors (Gharagozlou et al., 2003). The idea that GPCR signaling is linear and simplistic is no longer accepted. The ability to selectively choose the activation and downstream effects of opioids would lead to a novel direction of the therapeutic drug development. It is the ligand-receptor complex that ultimately determines the physiological cellular response of an opioid receptor (Kieffer & Evans, 2009). The complete signaling data for known opioid agonists and how they specifically react with receptors will allow us to derive ligands which possess the analgesic properties without unwanted side effects. The opioid receptor system is complex, and in order to understand all the nuances of multiple pathways activated and

numerous functionalities, we must re-evaluate the way research is being conducted. While much has been discovered about the opioid receptors using the current technology, even more remains enigmatic. We hypothesize that by using integrative DMR assays, we will be able to survey a library of opioid compounds to determine their integrative response on the family of opioid receptors. Further, we hypothesize that we will be able to use known GPCR downstream pathway inhibitors to visualize the functional selectivity of the library of opioid receptor ligands.

Chapter 2 Ligand-directed functional selectivity at the mu opioid receptor revealed by label-free integrative pharmacology on-target

Introduction Drugs that act on opioid receptors have long been recognized and used for their potent analgesic effects. The mu opioid receptor (MOR) is the subtype most commonly believed to control the analgesic and addictive effects of opioid drugs (Matthes et al., 1996). Opioid research currently focuses on understanding the signaling and functionality of clinical opioids, in order to derive drugs that will induce the beneficial effects of opioid activation without activating the harmful side effects, to create the holy grail of analgesics (Corbett et al., 2006). GPCRs are a super-family of cell membrane signaling proteins, specifically activated by a diverse array of extracullular stimuli, and their activation leads to an intracellular cascade of events (Lee et al., 2008). However, there is a gap in knowledge between the understanding of the general mechanisms of GPCR signaling (Urban et al., 2007; Kenakin, 2004) and the technology that is able to fully decipher the complexities of functional selectivity which occurs downstream of these receptors. Fortunately, a recent growth in knowledge of receptor kinetics has led to developments in cellular assay technology since the turn of the century. Conventional cell-based assays focus around the measurement of a single event, which is part of a much larger cell-signaling cascade. The most prevalent assays are based around quantification of secondary messengers (i.e. cAMP and Ca2+), use of fluorescent tagging, or genetic manipulation with reporter genes. There has been great demand for an assay system that does not require tagging or manipulation of a living cell system, as any treatment can lead to false results (William, 2004). GPCR signaling is a cell-specific and environment specific process, which can be affected by the smallest of factors. Any manipulation to the cells can affect the results, leading to false positives that are merely artifact (Elgen, 2005). It is necessary to study GPCR activation as a compilation of dynamic and integrative events. Ligand binding initiates changes in GPCR conformation, which leads 28

to the activation of numerous downstream signaling cascades. This results in a dynamic mass redistribution (DMR) of intracellular components. DMR is a measurable response that studies the shift in wavelength of refracted light indicative of a redistribution of cellular components upon activation of a GPCR (Fang, 2010; Schroder et al., 2010). DMR technology allows for GPCR signaling to be monitored and evaluated label-free. While DMR technology has been recently published and validated by multiple groups, in-depth signaling deconvolution experiments are very limited at this time (Codd et al., 2011; Ferrie et al., 2011). This study will highlight the potentials of using DMR technology and demonstrate how it can be utilized to change the course of drug development. Understanding nuances of GPCR signaling pathways activated by the MOR is the next crucial step in creating an opioid analgesics without negative side effects. To help reach this goal, we have developed a high-resolution, label-free integrative pharmacology on-target (iPOT) assay system to characterize the integrated response of cells to receptor activating ligands, and used this methodology to characterize a library of opioid receptor ligands. Key to this analysis is DMR assay, which uses a label-free optical biosensor to non-invasively report ligand-induced responses in cells. The resulting DMR signal is a reliable readout of GPCR functionality in various cell systems, wherein the dynamic redistribution of cellular contents is recorded in real-time with high sensitivity (Fang et al., 2005a). The DMR assay represents a powerful tool to delineate receptor signaling (Fang et al., 2005b; Schröder et al., 2010) and ligand pharmacology at the whole cell level (Fang & Ferrie, 2008). In this study we have characterized a library of 42 opioid receptor ligands in HEK-293 cells stably expressing the MOR (HEK-MOR cells). By measuring DMR and cAMP production, we showed that at least 29 ligands in the library were agonists for the MOR and activate MOR sites and activate distinct downstream signaling cascades. Our data indicate that the iPOT provides an integrated display of ligand-mediated receptor pharmacology and allows for a more effective prioritization of lead compounds for drug development. We hypothesize that the iPOT DMR assays will be effective in identifying known MOR ligands, and signaling deconvolution will allow for novel insights into the opioid signaling pathways.

Experimental Procedures Materials and reagents Pertussis toxin, cholera toxin, forskolin and dimethylsulfoxide (DMSO) were purchased from Sigma Chemical Co. (St. Louis, MO). DAMGO, DPDPE, BRL-53527, CTOP, naltrindole hydrochloride, norbinaltorphimine, U0126, SB202190, SP600125, and LY294002 were purchased from Tocris Biosciences. The Opioid Compound Library (consisting 64 compounds of pan-specific and receptor subtype-specific agonists and antagonists, each at 10mM, diluted in DMSO) was obtained from Enzo Life Sciences. All tissue culture media and reagents were purchased from Invitrogen (Carlsbad, CA). Fibronectin-coated Epic® biosensor microplates, TCT-compatible and polypropylene compound source plates were obtained from Corning Inc. (Corning, NY).

Cell Culture The HEK293 cells were obtained from American Type Cell Culture (Manassas, VA) and were cultured in Dulbecco’s modified Eagle’s medium (DMEM GlutaMAX-I, Gibco) supplemented with 10% non-heated inactivated fetal bovine serum and 1% penicillin-streptomycin. The HEK-MOR cell line was a generous gift from Dr. Mark von Zastrow (University of California, San Francisco). HEK-MOR cells express FLAG-tagged wild type human MOR. These cells were grown in complete DMEM GlutaMAX-I containing o 400µg/ml geneticin. All cells were maintained at 37 C and 5% CO2 in a humidified incubator. All overexpressed cells including the HEK293 cell line doubled roughly every 24 hrs.

Optical Biosensor System The Epic® wavelength interrogation system (Corning Inc, Corning, NY) was used for whole cell sensing. This system consists of a temperature-controlled unit, an optical detection unit, and an on-board liquid handling unit with robotics. The detection unit is centered on pairing a RWG biosensor with integrated fiber optics, and enables kinetic measurements of cellular responses with a time interval of ~10 sec.

The RWG biosensor is capable of detecting minute changes in local index of refraction near the sensor surface (Fang & Ferrie, 2008; Fang, 2010; Schroder et al., 2010). The local index of refraction within a cell is a function of density and its distribution of biomass (e.g. proteins, molecular complexes). The biosensor uses its evanescent wave to non-invasively detect ligand-induced dynamic mass redistribution (DMR) in cells. The evanescent wave extends into the cells and exponentially decays over distance, leading to a characteristic sensing volume of ~150 nm. This implies that any optical response mediated through the receptor activation only represents an average over the portion of the cell that the evanescent wave is sampling. The combination of many cellular events downstream of the receptor activation determines the kinetics and amplitude of a ligand-induced DMR response.

Cell Assay Methods Cells were seeded at the pre-determined densities. HEK-MOR cells were seeded at a density of 20,000 cells in 40 µL of complete media with geneticin (DMEM GlutaMAX I with 10% non-heat inactivated fetal bovine serum, 1% penicillin- streptomycin and 400 µg/mL of Geneticin) per well. The cells were seeded on fibronectin coated 384-well Epic® plates (available commercially). HEK293 cells were seeded at 16,000 cells per 40 µL of complete media (DMEM GlutaMAX I with 10% non-heat inactivated fetal bovine serum, 1% penicillin-streptomycin) per well, also on fibronectin coated 384-well Epic® plates. After seeding, the Epic® plates were left in the cell hood for 30 min to allow the cells to settle to the well bottom and to minimize edge effects in assay results. The plates were then transferred to an incubator at 37oC and 5% CO2 for 24 hrs. One hr prior to starting the DMR assays, cells were washed twice (50 µL each) with 1xHBSS containing 20 mM HEPES (pH 7.1), and then maintained in 30 µL of the HBSS in the commercial instrument for 1 hr. While incubating, compound plates were prepared by diluting stock solutions (typically 10-100 mM in DMSO) of each compound in the HBSS buffer, and then transferring to 384-well polypropylene plates (Corning). For 1-step or step 1 assays, compound plates were prepared at 4X the final concentration. For 2-step assays, step 2 compound plates were prepared at 5X the final concentration.

The compound plates were typically loaded inside the Epic® instrument min before the start of the assay. All cell assays were run using the standard 50-min protocol. This method consists of a 2-min baseline read, followed by a 10 µL compound addition (which was carried out by an on-board liquid handler) and then a final 50-min assay read. For 2-step assays, the process was repeated exactly. The resonant wavelengths of all biosensors in the microplate were normalized again to establish a second baseline, which was followed by compound addition of 10 µL of the second compound of interest and a 50-min assay read. All assays were carried out at a controlled temperature of 26.0oC, due to the fact that the DMR is temperature sensitive. Each data point represents an average of at least two replicates performed in the same assay. The assay coefficient of variation was found to be <10%.

Experimental Design A total of 13 experimental conditions were carried out, in an attempt to discern selectivity and downstream signaling of the MOR as completely as possible. These 13 conditions were grouped into three main types of assays: receptor specific assays, receptor level assays and downstream assays. There are listed in more details in the following pages. A number of dose response assays were also carried out, to more fully characterize known agonists and antagonists on the MOR.

Receptor Specificity Assays The first part of this study was completed to establish baseline efficacy of the MOR. A crucial step to understanding the deconvolution of signaling pathways is to have well-defined baseline information. It is well known that opioid receptor trafficking and opioid signaling is highly agonist dependent (Zhang et al., 2009; Melief et al., 2010). The parental line for the over-expressed HEK-MOR cells is HEK293 cells, which do not contain opioid receptors. Therefore, the first set of background studies examined ligand activation of HEK293 cells, to assess for any off-target signaling of the opioid ligand library. Baseline studies were also done on the HEK-MOR cells to visualize the efficacies

of the compounds when acting alone on the MOR. It is important to examine ligand individually in this platform and these cells lines, as the classifications of efficacies can be affected by many factors including the level of receptor expression and strength of stimulus-response coupling (Zheng et al., 2010). For the initial baseline studies of HEK293 cells, the cells were seeded normally (seeded at optimized concentration, left in hood for 30 min then stored in the incubator for 24 hrs). Cells were then washed one hr prior to the experiment and then incubated in commercial Epic® platform. During the hr incubation period, the compound plates of the opioid compound library were prepared. 1 µL of each compound (starting concentration of 10mM) was diluted in 200 mL of 1x HBSS containing 20 mM HEPES, then transferred into a 384-well polypropylene plate (Corning), resulting in a final dilution of 10 µM. The assay was carried out as a one-step assay, in which the cultured cells were introduced to a compound in the opioid receptor library (compound may be pan-specific or receptor subtype specific, and either an agonist or an antagonist). An on-board liquid handling system added 10 µL of compound to each well. The assay was run for 50 min, during which the cellular responses were monitored to determine if there was off-site activation induced by any library ligands. Baseline activity was also determined for all compounds in the opioid compound library on the MOR. The cells were prepared as before (seeded at an optimized concentration of 20,000 cells/well, left in hood for 30 min then stored in the incubator for 24 hrs). Cells were then washed one hr prior to the experiment and then incubate in commercial Epic® platform. This baseline assay was carried out as a two-step assay, in which the cultured cells were treated with 10 µL of HBSS with 20 mM HEPES buffer and then standard 50-min assay was completed to monitor the cellular responses. Second, the buffer-pretreated cells were introduced to a compound in the opioid receptor library (compound may be pan-specific or receptor subtype specific, and either an agonist or an antagonist). An on-board liquid handling system added 10 µL of compound to each well. The assay was run for 50 min, during which time efficacy of each ligand was monitored. This was run in a two-step assay so that it would serve also as a control for the other assays, all of which were carried out in a two-step format. At the conclusion of this

baseline study, activity of all of the compounds in the opioid library was characterized, in order to use as controls for later studies. In order to fully tease out specific activity of the compounds on the MOR, receptor specificity assays were also completed. Known agonists and antagonists (DAMGO and CTOP were utilized as the agonist and antagonist for the MOR, respectively) were used in sequence with the opioid library compounds to study the effects of pretreatment on cellular signaling. Cells were either pretreated with a high concentration of DAMGO or CTOP in step one and the opioid compound library in step two, or by introduced to the opioid receptor library in step one and treated with DAMGO or CTOP in assay two (see table 2.1, assays C-F). High concentrations of the agonists and antagonists were used, in order to be sure that the receptors were fully saturated. Receptor specificity studies were completed as a set of four studies for each of the MOR. Cells were prepared normally, washed one hr prior to the experiment, and then incubate in commercial Epic® platform. All four of the assays were carried out using a two-step protocol. The first two assays were done to study the effects of pretreatment with opioid receptor library compounds on the DMR induced by known agonists and antagonists. This was carried out as a two-step assay in which the cultured cells were introduced to a compound in the opioid receptor library (compound may be pan-specific or receptor subtype-specific agonist or antagonist) for step one. A 50-min assay was completed, to monitor the cellular responses. Second, the compound-stimulated cells were introduced to a known agonist or antagonist (10µM of DAMGO as the agonist for the MOR cells, and 10µM of CTOP as the antagonist). Results were used to compare the differences between the opioid agonist/antagonist-induced cellular responses in the absence and presence of the ligand pretreatment. The third and fouth assays used the same methodology, carried out in the reverse order, to determine if the known ligands (DAMGO and CTOP) fully desensitized or blocked the MOR from the ligands. Step one introduced the cultured cells to DAMGO or CTOP and step two introduced the DAMGO or CTOP-stimulated cells to the opioid library ligands.

Receptor-level Signaling Deconvolution Assays Opioid receptors are GPCRs, which activate signaling at the receptor level via G- protein signaling pathways. It has long been accepted that opioid receptors are coupled to the Gαi signaling pathway. The literature shows that activation of opioid receptors leads to an inhibition of adenylyl cyclase, an increase in potassium conductance, an inhibition of calcium channels, and an inhibition of neurotransmitter release (Fuduka et al., 1996; Williams et al., 2001) All of these responses are indicative of Gαi coupling, and are blocked by pretreatment of pertussis toxin (PTx). PTx permanently inhibits Gαi- coupled signaling via ADP ribosylation of a Cys residue and decoupling of the G protein from the receptor (Barbieri & Cortina, 1988). However, while it has been shown that Gαi -coupling is predominant in opioid signaling, more recent reports have suggested that there is a potentially an aspect of Gαs coupled signaling paired with opioid receptor activation (Chakrabarti et al., 2005; Chakrabarti et al., 2010). In order to examine the effects of the Gαs pathways in opioid signaling, cholera toxin (CTx) is utilized. CTx inhibits GTPase activity (thus activating the Gαs pathway) along with enhancing the GTP- dependent adenylyl cyclase activity (Cassel & Selinger, 1977). Also playing an important role in receptor level signaling is adenylyl cyclase. Opioid inhibition of adenylyl cyclase is suggested to be a mechanism by which opioids inhibit primary afferent excitability and relieve pain (Ingram & Williams, 2004). One of the most commonly used modulators of adenylyl cyclase activity is forskolin. Pretreatment with forskolin stimulates adenylyl cyclase activity (Seamon, 1985). To deconvolute receptor-level signaling, assays were completed in which cells underwent pretreatments with PTx, CTx, or forskolin. These assays were carried in order to visualize the effects of Gαi, Gαs, and cAMP on opioid signaling. DMRs which were completely abolished by PTx, but unaffected by CTx were identified as Gαi -coupled receptors. Ligands that were affected by CTx pretreatment were suggested to have a Gαs component involved with their signaling events. Ligands which exhibited DMRs affected by forskolin pretreatment were suggested to be Gαi-coupled as forskolin pretreatment led to the potentiation of Gαi-coupled signaling via heterologous sensitization. For all three assays, HEK-MOR cells were seeded at 20,000 cells per well on

fibronectin-coated Epic® plates. To break down the Gαi and Gαs signaling components,

Gαi and Gαs pathway inhibitors, PTx and CTx, were added to the 384-well plates 2 hrs after seeding the cells. PTx was added at concentration of 50 ng/mL, while CTX treatment was 10 ng/mL. After PTx and CTx addition, the cells were put back into the incubator for the remainder of the 24 hrs. On day 2, the cells were washed twice (50 µL each) with 1X HBSS containing 20 mM HEPES (pH 7.1), and then maintained in 30 µL of the HBSS in the commercial instrument for 1 hr. During the hr, the compound plate of the opioid compound library was prepared, as before. Then assay was run, introducing the pretreated cells to a compound in the opioid receptor library. An on-board liquid handling system added 10 µL of compound was added to each well. The assay was run for 50 min, and at the conclusion the differences between the opioid-compound induced cellular responses in the absence or presence of pretreatment with a modulator (PTx/CTx) were studied. For the forskolin pretreatment assay, both steps of the assay were carried out on day 2, after culturing in the incubator for 24 hrs. The forskolin assay was carried out as a two-step assay, in which the HEK-MOR cells were first introduced to the pathway modulator forskolin. Forskolin, at a concentration of 10 µM, was added to the cells and a standard 50-min assay was completed to monitor the cellular responses. Second, the forskolin-pretreated cells were introduced to a compound in the opioid receptor library. Specifically, 10 µL of each compound was added to each well, by an on-board liquid handling system. Step 2 of the assay was run for 50 min, and at the conclusion the differences between the opioid-compound induced cellular responses in the absence or presence of pretreatment with the cAMP modulator were studied.

Downstream Signaling Deconvolution Assays There are multiple levels at which to study pathway deconvolution of GPCR signaling. The previous set of experiments evaluated functional selectivity at the receptor level. It is also crucial to study downstream signaling in order to understand the mechanisms and implications of agonist-selective signaling (Zheng et al., 2010). The different pathways activated can have profound effects on the physiological effects of the ligands.

Kinases, and specifically the family of mitogen-activated protein kinases (MAPKs), are activated by downstream GPCR signaling (Fukuda et al., 1996). These kinases (ERK, JNK, p38, and PI3K for this study) have been defined as having roles in proliferation, plasticity, long-term potentiation and survival, and differentiaton (Belcheva & Coscia, 2002). The functions of MAP kinases provide a possible connection for neuronal adaptations which decrease plasticity accompanying opioid abuse (Nestler et al., 1993). It is also widely cited in the literature that protein kinases modulate internalization and desensitization in cellular signaling pathways. These processes appear to play critical roles in opioid tolerance and addiction (Yu, 1996; Yu et al., 1997). Adding another layer of complexity to the deconvolution of the kinase phosphorylation cascade is the idea that functional selectivity does not only affect the specific kinase pathway activated, but also the duration of activity. It has been suggested that different agonists will induce either transient or chronic activation of MAPKs (Belcheva et al., 2005). For these studies, assays singly and specifically inhibited parts of the kinase signaling cascade in order to more fully understand the downstream effects of opioid receptor activation and agonist-selective signaling. The full set of experiments was comprised of four assays, each utilizing pretreatment with a specific pathway modulator: U0126 (ERK inhibitor), SP600125 (JNK inhibitor), SB202190 (p38 MAPK inhibitor) and LY294002 (PI3K inhibitor). HEK-MOR cells were prepared normally, washed one hr prior to the experiment, and then incubated in commercial Epic® platform. All four of the assays were carried out using the same two-step protocol. They were done to study the effects known kinase inhibitors on the ligand-induced DMR response of HEK-MOR cells. For step one, the cultured cells were introduced to a kinase specific inhibitor. All ligands were used at a final concentration of 10 µM. A 50-min assay was completed, to monitor the cellular responses. Second, the inhibitor-pretreated cells were introduced to a compound in the opioid compound library. Step two assays were run for 50 min. At the conclusion, differences were compared between the opioid compound-induced DMR responses in the absence and presence of pretreatment with a kinase inhibitor. The full list of assays is shown in detail below in Table 2.1.

Assay Probe, # Cell pretreatment duration DMR readout Labels used in clustering A HEK293 none Ligand, 10 µM HEK-3, 9, 30 B HEK-MOR 0.1% DMSO in buffer, 1hr Ligand, 10 µM DMSO - Ligand-3, 9, 30 C HEK-MOR 10 µM CTOP, 1hr Ligand, 10 µM CTOP - Ligand-3, 9, 30 D HEK-MOR 10 µM DAMGO, 1hr Ligand, 10 µM DAMGO - Ligand-3, 9, 30 E HEK-MOR 10 µM ligand, 1hr DAMGO,10 µM Ligand - DAMGO-3, 9, 30 F HEK-MOR 10 µM ligand, 1hr CTOP, 10 µM Ligand - CTOP-3, 9, 30 G HEK-MOR 100ng/ml PTx, 20hr Ligand, 10 µM PTx - Ligand-3, 9, 30 H HEK-MOR 400ng/ml CTx, 20hr Ligand, 10 µM CTx - Ligand-3, 9, 30 I HEK-MOR 10 µM forskolin, 1hr Ligand, 10 µM FSK - Ligand-3, 9, 30 J HEK-MOR 10 µM U0126, 1hr Ligand, 10 µM U0126 - Ligand-3, 9, 30 K HEK-MOR 10 µM SB202190, 1hr Ligand, 10 µM SB202190 - Ligand-3, 9, 30 L HEK-MOR 10 µM SP100625, 1hr Ligand, 10 µM SP100625- Ligand-3, 9, 30 M HEK-MOR 10 µM LY294002, 1hr Ligand, 10 µM LY294002 - Ligand -3, 9, 30 Table 2.1: Assay protocols and DMR signals used for similarity analysis

cAMP Assay Cyclic AMP regulation by opioids has become a classic measurement of opioid activity and cAMP assays have long been a benchmark assay of cell signaling activity (Jordan & Devi, 1998). Monitoring adenylyl cyclase inhibition by opioid receptor activation is one of the most widely utilized assays, thus making it a well-characterized and valid marker. Comparing our DMR results to this accepted assay allows us to not only further validate the DMR system, but also look for differences between the two in order to note sensitivity of either system. For completion of the cAMP assay, cells and reagents were taken to Johns Hopkins medical campus in Baltimore MD. The cAMP high-throughput assay was completed using cAMP-Glo assay (Promega, Cat#V1502). HEK293 and HEK-MOR cells were grown up at HMC PSU in Hershey PA, and then transported in media to Hopkins University. When confluent, the cells were collected and seeded onto 384-well white Poly-D-lysine biocoat plates (BD). The cells were counted using a hemocytometer and seeded at 15,000 cells/well. All cells were

seeded in 50 µL of media/well. When complete, the plates were put back in the incubator o at 37 C and 5% CO2 overnight. The next days, plates were checked and assay materials were prepared. First, induction buffer was made by adding 500 µM IBMX and 100 µM Ro-201724 to 50 mL of 1x PBS. The opioid compound library (identical to the one used in Epic studies) was diluted to a 5x concentration of 50µM, then diluted down in induction buffer + forskolin for cAMP assays. The ligands were used at a final concentration of 10µM during experiments. The cells were then retrieved from incubator and the media was removed from the wells. A 384-well automated pipettor was used to add 7.5 µL of compounds to the cell plates. Cells were centrifuged, shaken and incubated for 10 min (total time) then 7.5 µL of lysis buffer was added to cells to stop the reactions. At this step, cells were centrifuged and shaken, and then incubated for 10 min, until all the cells were lysed. After lysis was complete, 15 µL of PKA in reaction buffer was added to each well and cells were centrifuged, shaken, and incubated for 20 min. Lastly, 30 µL of kinase Glo reagent was added to each well and the plates were covered in foil, centrifuged, shaken, and incubated for 20 min. When the reaction was complete we used a Tecan plate reader to extract results. The program was set to luminescence mode, and the results read every 1000 ms. Results were transposed into excel file, and graphed.

Results DMR characterization of mu opioid receptor Agonism To characterize the MOR, we first performed DMR agonism assays, in a dose-dependent manner. This assay monitors the DMR signal upon stimulation with a MOR ligand itself. We selected two endogenous opioid agonists (endomorphin-1 and endomorphin-2) and three exogenous opioid agonists (DAMGO, morphine and fentanyl) to gain a full perspective of agonist activity at the MOR. In addition, we characterized the inhibition of DMR response in HEK-MOR cells utilizing known MOR selective (CTOP and β-funaltrexamine), as well as opioid non-selective (naloxone) antagonists. We performed a dose response analysis and determined the EC50 for each ligand tested and compared our results to the potency previously reported in the literature (Reisine, 1995; Amiche et al., 1989; Fichna et al., 2008). 39

DMR assays showed that both DAMGO and β-funaltrexamine yielded a dose-dependent DMR response in HEK-MOR cells (Fig. 2.1a, and Fig. 2.1b, respectively). Similarly, endomorphin-1, endomorphin-2, morphine, and fentanyl also yielded dose dependent DMR responses (Fig. 2.1c and Fig. 2.1d). β-Funaltrexamine, a selective MOR antagonist (Ward et al., 1982), led to a smaller maximum response than DAMGO (Fig. 2.1b), suggesting that this compound may be a partial agonist for the MOR. In contrast, both CTOP and naloxone led to a net-zero DMR in HEK-MOR cells (data not shown).

a b 1000nM 200 200 500nM 250nM 150 150 125nM 63nM 32nM 100 100 16nM 32nM 8nM 50 50 2nM Response(pm) 0.5nM Response(pm) 0 0.12nM 0 0.03nM -50 -50 0 10 20 30 40 50 0 10 20 30 40 50 Time (min) Time (min) c d 250 Endomorphin-2 250 Fentanyl DAMGO 200 Morphine 200 Endomorphin-1 β-Funaltrexamine 150 150

100 100

50 50 Response(pm) Response(pm) 0 0

-50 -50 -12 -11 -10 -9 -8 -7 -6 -5 -12 -11 -10 -9 -8 -7 -6 [Compound], log M [Compound], log M

Figure 2.1. DMR characteristics of opioid ligands in HEK-MOR cells. (a) Real time kinetic responses of DAMGO at different doses; each graph represents the mean ± s.d. of 2 independent measurements (n = 6); (b) Real time kinetic responses of β-funaltrexamine at different doses; each graph represents the mean ± s.d. of 2 independent measurements (n = 4); (c) The maximal DMR amplitudes as a function of ligand doses for endomorphin-2, DAMGO, morphine and β-funaltrexamine; (d) The maximal DMR amplitudes as a function of ligand doses for fentanyl and endomorphin-1. All dose responses represent the mean ± s.d. of 2 independent measurements (n = 4), except for DAMGO (n = 6)

Non-linear regression analysis revealed that DAMGO, morphine, endomorphin-2, and β-funaltrexamine exhibited monophasic dose responses, all of which were best fit by

a single-phase sigmoidal curve (Fig. 2.1c). This analysis revealed EC50 values of 0.93±0.12 nM (n =6), 6.0±0.5 nM (n = 6), 0.86±0.07 nM (n =4), and 0.10±0.02 nM (n

=4) for DAMGO, morphine, endomorphin-2, and β-funaltrexamine, respectively. However, fentanyl and endomorphin-1 both induced biphasic dose responses (Fig. 2.1d).

Fentanyl’s EC50 values were calculated to be 0.28±0.06 nM and 111±15 nM (n = 4), while endomorphin-1 produced EC50 values of 1.0±0.1 nM and 13.8±0.43 nM (n =4).

The EC50 values obtained using DMR assays were mostly consistent with their respective

Ki values reported in literature (Appendix B). The notable exception was fentanyl, whose

EC50 was found to be 0.28 nM in DMR assays, ~ 7x more potent than the EC50 value (1.8 nM) reported previously (James et al., 1991).

Antagonism We next examined the ability of opioid antagonists to block the activity of MOR agonists using a two-step DMR antagonism assay, in which the cells were pretreated with an antagonist at different doses for 1 hr, followed by the stimulation with an agonist for the MOR. Results showed that the DMR signals produced by endomorphin-2, DAMGO, and morphine were inhibited in a dose-dependent fashion by naloxone (Fig. 2.2). Naloxone is a well-characterized opioid antagonist with a high affinity for the MOR (Takemori & Portoghese, 1984). All three agonists were assayed at their respective

EC100, 64 nM. However, naloxone most potently inhibited the morphine response with an

IC50 value of 30.2±2.1 nM (n =4). Naloxone exhibited IC50 values of 5.13±0.47 µM (n = 4), and 5.77±0.35 µM (n = 4) on endomorphin-2 and DAMGO, respectively. Taken together, the dose-dependent activation of the MOR by known MOR agonists, as well as the dose-dependent inhibition of these agonist responses by opioid antagonists, suggest that the real-time DMR assays accurately reflect the behavior of these ligands at the MOR.

200 Figure 2.2. The inhibition of MOR agonist- induced DMR by naloxone. The maximal 150 amplitudes of the MOR agonist-induced DMR as a function of naloxone doses. All 100 three agonists, endomorphin-2, DAMGO and 50 morphine, were assayed at 64 nM. The data represents the mean s.d. of 2 independent Response(pm) Endomorphin 2 ± 0 DAMGO measurements, each in duplicate (n = 4). Morphine -50 -9 -8 -7 -6 -5 -4 Naloxone, log M

Label-free on-target pharmacology profiling of opioid ligands To examine the functional selectivity of opioid ligands, we used DMR assays to analyze the pharmacological properties of 42 opioid ligands in HEK-MOR cells. All ligands were profiled at 10 µM to achieve maximal signaling. The known potency of these ligands for the MOR is presented in supplemental material (Appendix B). We designed a set of 13 assays to characterize MOR-mediated signaling produced by each of the ligands (Table 2.1). Besides the agonism profiles in parental HEK293 cells, all other 12 assays were performed in HEK-MOR cells after pretreatment with a variety of probe molecules. The choice of probe molecules was based on known signaling pathways downstream of MOR activation. The HEK-MOR cells pretreated with 0.1% DMSO were used as a control. These assays allowed us to discern receptor specificity, G-protein coupling, downstream kinase pathway selectivity, as well as any potential off-target effects. A DMR signal is recorded as a shift in resonant wavelength (picometer, pm), and is a real-time kinetic response with high temporal resolution (15 sec per data point) and long duration (~ hrs). As was previously mentioned, the DMR signal can be considered to be a poly-dimensional coordinate (Ferrie et al., 2011; Fang, 2010a). Therefore, the DMR responses can be directly used as a basis for similarity analysis. Similarity analysis is a powerful means to determine the relationships (i.e., similarity or distances) among different biological responses, particularly for large sets of biological data (Eisen et al., 1998). However, due to the large dimensions of DMR signals obtained under the thirteen assay conditions for each specific ligand, a DMR numerical descriptor needed to be developed for effective similarity analysis. To do so, we reduced the DMR dimensions to 42

three distinct time points (3, 9, and 30 min poststimulation) (Fig. 2.3). This dimensional reduction is based on clustering of time domains of the DMR of all opioid ligands in the DMSO-treated cells. Results showed that the ligand DMR generally propagated with three distinct time periods: short (1 to 8 min), intermediate (9 to 14 min), and late (15 to 50 min). Similar patterns were obtained for all DMR signals under all thirteen conditions. Thus, we selected one time point from each period to represent each DMR (as indicated in Fig. 2.3), and found that these time points adequately represent the key features of the ligand responses. Thus, the thirteen DMR profiles of each ligand can be translated into a 39 dimensional coordinate.

a b 300

2 200 1 3 100

Response(pm) 0

t = 0 -100 0 10 20 30 40 50 Time (sec) Time (min)

Fig. 2.3 The numerical descriptor of opioid ligand pharmacology. The DAMGO DMR signal in the native HEK-MOR cells and its three representative time point descriptors (3 min, 9 min and 30 min poststimulation, as indicated by 1, 2, and 3, respectively). The black arrow indicates the time (t = 0) when DAMGO was added. The graph represents the mean ± s.d. of 2 independent measurements, each in sixteen replicates (n = 32).

Application of an unsupervised Ward hierarchical clustering algorithm and Euclidean distance metrics led to a high resolution heat map of opioid ligands (Fig. 2.4). For visualization, the real-time responses were color coded to illustrate relative differences in DMR signal strength. The red color refers to a positive value, the black a value near zero, and the green color represents a negative value. Differences in color

intensity illustrate differences in signal strength. For each assay, the three responses for each assay were grouped together (i.e., for all ligands there are three adjacent columns, termed column group (A to M), for each assay in the heat map). Our data indicates that the opioid ligands can be grouped into two super clusters: antagonists (1) and agonists (2). Each super cluster can be further divided into several subclusters (1.1 to 1.3, 2.1 to 2.3 in Fig. 2.4). All ligands in the agonist cluster led to a positive DMR (P-DMR) greater than 60 pm in the HEK-MOR cells in at least one of the three time point. Conversely, all ligands in the antagonist cluster led to a DMR, in which all of the responses at the three time points are smaller than 60 pm in the HEK-MOR cells. We postulate that similarity analysis allows us to derive hypotheses regarding the mechanisms of action of the opioid ligands tested, and to explore the relationships between function and pharmacological activity of the different opioid ligands.

3 3 3 9 30 9 30 - - 9 30 ------3 9 30 3 9 30 ------3 9 30 3 3 9 30 3 9 30 9 30 ------3 9 30 3 9 30 3 9 30 ------Ligand Ligand Ligand Ligand Ligand Ligand Ligand Ligand Ligand – – Ligand Ligand Ligand – – – – – – – DAMGO DAMGO DAMGO CTOP CTOP CTOP Ligand Ligand Ligand Ligand – – – Ligand Ligand Ligand Ligand Ligand – – – – – – – – – – Ligand Ligand Ligand Ligand Ligand Ligand – – – Ligand Ligand Ligand – – 3 9 30 – – – – – – – – – - - - HEK HEK HEK DMSO DMSO DMSO CTOP CTOP CTOP DAMGO DAMGO DAMGO Ligand Ligand Ligand Ligand Ligand Ligand PTX PTX PTX CTX CTX CTX FSK FSK FSK U0126 U0126 U0126 SB202190 SB202190 SB202190 SP100625 SP100625 SP100625 LY294002 LY294002 LY294002

A B C D E F G H I J K L M Endomorphin-2 Met5-Enkephalin Dermorphin DAMGO DADLE Dynorphin A 1-8 Dynorphin B 2 DSLET 2.1 Leu5-Enkephalin Syndyphalin SD-25 Endompphin-1 TAPP DAMME ICI 199,441 DIPPA BRL-52537 BUBUC Dynorphin A 1-13 2.2 NNC 63-0532 DPDPE DALDA Morphiceptin Norcodeine U-62066 U-69593 2.3 Tramadol (-)U-50488 Deltorphin II Nociceptin 1-13 Naltrexone 1.1 β-Funaltrexamine Nalbuphine Levallorphan SKF10047 Naltrindole 1.2 Naloxone Methionone 1 Naloxonazine Naloxone HCl Salvinorin A 1.3 Nor-Binaltorphine N-Benzylnaltrindole DMSO BNTX

Figure 2.4: A false colored heat map of all 42 opioid ligands. The heat map was generated using similarity analysis of the DMR of all ligands under 13 assay conditions. 0.1% DMSO in the assay buffer was included as a control. In order to visualize the characteristics of a DMR for each assay, the three responses (3min, 9min, and 30min) for each assay were grouped together. For each assay there are three adjacent columns to form a column group (A to M).

Receptor Specificity To determine the specificity of opioid ligands for the MOR, we used four distinct two-step DMR assays. (1) Cells were pretreated with the MOR antagonist CTOP, followed by stimulation with an opioid ligand. The ability of CTOP to block the ligand- induced DMR indicates that the agonism profile of a ligand is due to the activation of the

MOR. (2) Cells were pretreated with the MOR agonist DAMGO, followed by stimulation with an opioid ligand. The ability of the opioid ligand to alter the DAMGO response indicates that the ligand deactivates the DAMGO-activated cells. (3) Cells were pretreated with an opioid ligand, followed by stimulation with CTOP. The CTOP response indicates the MOR specific agonism of the ligands, since CTOP specifically reversed a MOR agonist-induced DMR. (4) Cells were pretreated with an opioid ligand, followed by stimulation with DAMGO, in which the DAMGO response indicates MOR specific agonism or antagonism of the ligand. Together with the agonism profiles in both parental HEK293 and HEK-MOR cells, these assays allowed us to determine the specificity, relative potency, efficacy, and signaling properties of each of the opioid ligands. The results of this analysis are shown in column groups A to F of the heat map (Fig. 2.4). Of all the compounds tested, the DOR-selective antagonist BNTX appeared to exhibit the most significant off-target activity in parental HEK293 cells. BNTX produced a negative DMR (N-DMR) in parental HEK 293 cells (Fig. 2.5a), suggesting that BNTX activated an endogenous cellular target. Further, BNTX also led to an N-DMR in HEK- MOR cells greater than that in the parental cell line (Fig. 2.5b), suggesting that the BNTX response is sensitive to the expression of the MOR. However, CTOP only slightly altered the BTNX-induced DMR in HEK-MOR, confirming that the BNTX response is largely due to the activation of an endogenous target. In addition, BNTX pretreatment inhibited the DAMGO-induced response in HEK-MOR cells (Fig. 2.5c), suggesting that BNTX is also an antagonist for the MOR. Finally, BNTX produced a N-DMR in the DAMGO pretreated HEK-MOR cells that is greater than that in the untreated cells, suggesting that BNTX blocked the DAMGO-induced MOR activation. Taken together, these results suggest that BNTX has a previously unrecognized antagonistic effect against the MOR, and also interacts with an unknown endogenous target as described by others (Gaveriaux-Ruff et al., 2001). In control experiments, we found that the assay buffer containing 0.1% DMSO, a concentration identical to that in all ligand solutions, did not lead to any obvious DMR in parental HEK293 cells (Fig. 2.5d), in HEK-MOR cells or in HEK-MOR cells with CTOP-pretreatment (Fig. 2.5e). In addition, DMSO alone did not have an effect on

DAMGO-mediated signaling whether DMSO was added before or after DAMGO stimulation (Fig. 2.5f). a HEK293 b c 200 200 200 Figure 2.5. The DMR characteristics of BNTX and the negative control (0.1% DMSO). (a) The 0 0 0 BNTX - DAMGO BNTX BNTX DMR in HEK-293 cells; (b) the BNTX -200 -200 -200 DAMGO - BNTX DMR in the buffer treated (BNTX) and CTOP Response(pm) Response(pm) Response(pm) CTOP-BNTX -400 -400 -400 pretreated (CTOP – BNTX) HEK-MOR cells; 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 Time (min) Time (min) Time (min) and (c) the DAMGO DMR in the BNTX pretreated cells (BNTX – DAMGO) in d HEK293 e f 300 300 300 comparison with the BNTX DMR in the

200 200 200 DMSO - DAMGO DAMGO pretreated cells (DAMGO – BNTX); 100 100 100 (d) the DMR induced by DMSO in HEK293; (e) DMSO 0 0 0 the DMSO DMR in HEK-MOR or the CTOP DAMGO - DMSO Response(pm) Response(pm) Response(pm) CTOP-DMSO -100 -100 -100 pretreated HEK-MOR cells; (f) the DAMGO 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 DMR after pretreatment with DMSO (DMSO – Time (min) Time (min) Time (min) DAMGO) and the DMSO DMR after pretreatment with DAMGO (DAMGO –DMSO) in HEK-MOR cells. Each curve represents the average of duplicates.

Antagonist supercluster The predominant feature of the antagonist supercluster is that all ligands in this group, except for salvinorin A, produced a negative DMR (N-DMR) response in HEK- MOR after pretreatment with DAMGO (Column group D in Fig. 2.4). The full N-DMR responses exhibited by three representative antagonists (levallorphan, β-funaltrexamine, and naltrindole) are shown in details in Fig. 2.6.

50 levallorphan Figure 2.6. The DMR signals induced by β-funaltrexamine naltrindole levallorphan, β-funaltrexamine, and naltrindole 0 in the DAMGO-activated HEK-MOR cells. Each curve represents the average of duplicates. -50

Response(pm) -100 0 10 20 30 40 50 Time (min)

The antagonist supercluster can be further subdivided into three subclusters. The subcluster 1.1 consists of naltrexone, β-funaltrexamine, nalbuphine, levallorphan, and SKF10047. Naltrexone appeared to be a neutral antagonist lacking partial agonist activity

at the MOR. The four remaining ligands produced a noticeable P-DMR in HEK-MOR cells, suggesting that the four ligands are partial agonists for the MOR. The partial agonism of these ligands was further confirmed by the fact that the CTOP pretreatment blocked their DMR. The fact that pretreatment with all ligands in the subcluster completely blocked the DAMGO response suggests that these ligands are potent antagonists. The antagonist subcluster 1.2 consists of naltrindole, naloxone methionone, naloxonazine, and naloxone HCl. All ligands in this subgroup were inactive in parental HEK293 cells, as well as in HEK-MOR cells, suggesting that the common feature of these ligands, when compared to subcluster 1.1, is the lack of partial agonist activity. Naloxone HCl and naloxonazine appeared to be more potent inhibitors of the DAMGO response than the other two ligands in this subcluster. The antagonist subcluster 1.3 consists of salvinorin A, nor-binaltorphine, and N- benzylnaltrindole. The common feature of the antagonists in this subcluster is that all three ligands appeared to be less potent inhibitors of the DAMGO response than those in subclusters 1.1 and 1.2. However, salvinorin A is distinct in that it appears to induce partial agonism activity at the MOR. Treatment of HEK-MOR cells with salvinorin A produced a noticeable P-DMR signal, while pretreatment of HEK-MOR cells with salvinorin A followed by exposure to CTOP resulted in an N-DMR response.

Agonist supercluster All ligands in the agonist supercluster produced a P-DMR in HEK-MOR cells, indicating that they all exhibit some degree of agonism at the MOR. When the HEK- MOR cells were pretreated with either DAMGO or CTOP, addition of supercluster 2 agonists produced an attenuation of the DMR response compared to their respective DMR in the DMSO-treated HEK-MOR cells. When HEK-MOR cells were pretreated with ligands in this supercluster, followed by exposure to CTOP, CTOP produced an N- DMR. These results suggest that the ligands in the agonist supercluster are acting predominantly at the MOR, although it is possible that these ligands may also act allosterically or have some off-target activity.

The agonist supercluster can be further divided into three subclusters. The agonist subcluster 2.1 is composed of endomorphin-2, (met5)-enkephalin, dermorphin, DAMGO, (D-Ala2, D-Leu5) enkephalin, dynorphin A 1-8, dynorphin B, DSLET, (Leu5)-enkephalin, syndyphalin SD-25, endomorphin-1, TAPP, DAMME and ICI 199,441. Except for ICI 199,441, all ligands in this subcluster were inactive in parental HEK-293 cells, but exhibited agonist activity in the HEK-MOR cells. Pretreatment of HEK-MOR cells with ligands in this subcluster caused the MOR to become completely desensitized, since cells failed to respond to repeated stimulation with DAMGO. Similarly, when HEK-MOR cells were pretreated with DAGMO, they failed to respond to repeated stimulation with the ligands in this subcluster. The cross-desensitization between DAMGO and ligands in the subcluster 2.1 is consistent with the idea that ligands in this group act specifically at MOR sites and that they possess sufficient potency to desensitize MORs. Further, pretreatment of HEK-MOR cells with CTOP blocked the DMR arising from these ligands, while CTOP produced an N-DMR response in the HEK-MOR cells after pretreatment with any of these ligands. Together, these results support the view that ligands in this subcluster are the potent agonists at MOR sites. The agonist subcluster 2.2 consists of DIPPA, BRL-52537, BUBUC, dynorphin A 1-13, NNC 53-0532, DPDPE, DALDA, morphiceptin, (-)-norcodeine, and U-62066. Pretreatment of HEK-MOR cells with DAMGO fully inhibited the DMR response produced by addition of ligands in this subcluster. In contrast with the ligands in the subcluster 2.1, all ligands in this subcluster produced only a partial attenuation of the DAMGO-induced DMR response when the cells were pretreated with the subcluster 2.2 ligands. These results suggest that the ligands in the subcluster 2.2 act at MOR sites, but are not as potent agonists as the ligands in subcluster 2.1. Notably, DIPPA and U-62066, previously classified as KOR-specific (Chang et al., 1994; Waldenberg, 2003), were slightly distinct from other ligands in this subcluster based on cluster analysis of all 13 assay results. The agonist subcluster 2.3 consists of U-69593, tramadol, U-50488, deltophin II and nociceptin 1-13. The members of this group are a mixture of KOR, DOR, MOR, and ORL-1 agonists. All ligands in this subcluster produced a P-DMR smaller than the DMR exhibited by agonists in the subcluster 2.1. This suggests that the ligands in this group act

as partial or weak agonists at MOR sites. Pretreatment with DAMGO or CTOP completely blocked the DMR response induced by ligands in this subcluster, supporting the view that subcluster 2.3 ligands do in fact act primarily at the MOR sites. In contrast, pretreatment of HEK-MOR cells with ligands in subcluster 2.3 did not fully desensitize MOR sites to subsequent treatment with DAMGO. These results suggest that the ligands in this subcluster are weak MOR agonists. All of the ligands tested in the agonist supercluster were inactive in parental HEK-293 cells, except for ICI 199,441. ICI 199,441 produced a P-DMR in HEK-MOR cells, but a small N-DMR in the parental cell line and in CTOP- or DAMGO-pretreated

300

200 HEK-MOR Figure 2.7 The ICI 199,441-induced DMR in distinct molecule-treated cells: HEK-293 100 cells (HEK293), HEK-MOR (HEK-MOR), or CTOP- (CTOP), or DAMGO- (DAMGO) 0 HEK293 pretreated HEK-MOR cells. Each curve represents the average of duplicates. Response(pm) -100 CTOP DAMGO

-200 0 10 20 30 40 50 Time (min) cells (Fig. 2.7). Similar to BNTX, the ability of ICI 199,441 to produce an N-DMR response in multiple assays strongly suggests that it interacts with an unknown endogenous receptor.

G-protein signaling To determine whether DMR assay is reflective of traditional assays that measure the biochemical response of cells to opioid agonist-induced MOR activation, we examined the relationship between the DMR response and whole cell cAMP changes produced by exposure of HEK-MOR cells to the library of opioid ligands. Results showed that there is a good linear correlation between cAMP and DMR assay results with an R2 of 0.81 (Fig. 2.8). The slope of 1.17 suggests that the whole cell cAMP signals are more easily saturated than the DMR signals, making it difficult to resolve strong partial agonists from full agonists. Notably, a subset of ligands including nociceptin 1-13,

deltorphin II, tramadol, endomorphin-1, and dynorphin A 1-13 were biased towards DMR, while others including SFK10047, levallorphan, nalbuphine, U-62066, BRL- 52537, norcodeine, and ICI-199441 were biased towards cAMP. Nonetheless, these results support the validity of DMR assay when compared to conventional GPCR cell- based assays.

It is well established that the MOR signaling is primarily coupled to the Gαi class of heterotrimeric G-proteins (Thompson et al., 1993). However, recent work has

challenged this dogma, and suggested that in addition to Gαi, the MOR may also signal through other G-protein subtypes (Neve, 2009). It has been shown that the DMR assay is capable of recognizing all G-protein coupled pathways (Fang et al., 2005; Schröder et al., 2010). In order to deconvolute the signaling pathways, we analyzed the effect of G- protein inhibitors on MOR-mediated signaling induced by opioid ligands.

120 y = 1.17x R2 = 0.81 ICI 199,441 100 (-)-Norcodeine DAMGO BRL-52537 80 U-62066 60 Dynorphin1-13 Nalbuphine Levallorphan 40 Endomorphin-1 SFK10047 20 Tramadol % DAMGO(cAMP)% Deltorphin II 0 BNTX Nociceptin1-13

-20 -20 0 20 40 60 80 100 120 % DAMGO (DMR)

Figure. 2.8. The comparison between the DMR responses and the whole cAMP responses induced by opioid ligands. Both were obtained in HEK-MOR cells. All data represent the mean ± s.d. of 2 independent measurements (n = 4).

First, we used pretreated the cells with PTx. Results showed that masking of Gαi by PTx inhibited the DMR induced by all opioid agonists tested, except for DAMGO and ICI-199,441 (Fig. 2.9a). ICI-199,441 produced a small N-DMR in the PTx-treated cells, while DAMGO still led to a small P-DMR (Fig. 2.9b). These results suggest that virtually all of the MOR activity induced by the library ligands is Gαi -coupled. However,

DAMGO may also contain a Gαi -independent signaling component, while ICI-199,441 is likely to activate an endogenous target and/or may also mediate Gαi –independent signaling.

a b 300 100 PTx Buffer DAMGO 200 50

100 0 DAMGO ICI 199441 0 -50 Response(pm) Response(pm) ICI 199441 -100 -100 0 10 20 30 40 50 0 10 20 30 40 50 Compound Time (min)

Figure 2.9 The sensitivity of opioid ligand-induced DMR to the pertussis toxin (PTx) pretreatment. (a) the comparison between the DMR induced by each of the 42 opioid ligands in the untreated and PTx-pretreated HEK-MOR cells, in which the maximal amplitudes were plotted.; (b) The DMR arising from ICI 199,441 and DAMGO in the PTx-pretreated HEK-MOR cells. All data represent the mean ± s.d. of 2 independent measurements (n = 4).

Next, we used CTx to permanently activate Gαs proteins by ADP ribosylation of an Arg residue of the protein. The CTx treatment is known to lead to cAMP production (Wess, 1998). The CTx treatment had little effect on the DMR response induced by most of the agonists tested, except for DAMME, ICI-199441, endomorphin-2 and BRL-51537 (Fig. 2.10a). CTx was found to reduce the DMR response induced by DAMME, ICI-199441, endomorphin-2 or BRL-51537, suggesting that these four ligands may also signal through a Gαs component. Interestingly, except for naltroxone and naltrindole, all ligands in the antagonist cluster generally produced more pronounced DMR in the CTx-pretreated cells than in the untreated cells. This suggests that the increased basal Gαs activity after the CTx pretreatment rendered these antagonists to act as partial agonists.

Lastly, we used forskolin to increase the basal cAMP level before stimulation with opioid ligands. Forskolin is an adenylate cyclase activator and is widely used to increase intracellular cAMP levels (Seamon & Daly, 1981), thus leading to the potentiation of Gαi-coupled signaling via heterologous sensitization (Tran & Fang, 2009). We found that, with the exception of U69593, pretreatment of HEK-MOR cells with forskolin up-regulated the DMR response normally produced by opioid ligands (Fig. 2.10b). Together, these results suggest that the opioid agonist-induced MOR DMR signals are predominantly the result of Gαi-mediated signaling.

a b 300 300

200 200 DAMME ICI119441

100 Endomorphin-2 100 U59593 BRL51537 Response(pm) Response(pm) 0 0 FSK CTx Buffer Buffer -100 -100 0 10 20 30 40 50 0 10 20 30 40 50 Compound Compound

Figure 2.10 The sensitivity of opioid ligand induced DMR to the pretreatment with forskolin (FSK) and cholera toxin (CTx). The maximal DMR amplitudes of all opioid ligands were compared in HEK-MOR after pretreatment with: (a) buffer versus CTx; and (b) buffer versus forskolin.

Kinase Pathways We used a panel of specific kinase inhibitors to determine the potential pathway biased activity of opioid ligands (Johnson & Lapadat, 2002). We chose U0126 to inhibit the ERK1/2 pathway (Favata et al., 1998), SB202190 to inhibit the p38 pathway (Nemoto et al., 1998), SP600125 to block JNK-mediated signaling (Hancock et al., 1998) and LY294002 to inhibit signaling through PI3K (Vhalos et al., 1994). Here, HEK-MOR cells were first pretreated with one of the kinase inhibitors, followed by exposure to an opioid ligand. For each ligand we compared the maximal DMR response within 15min poststimulation, and the response at 30 min poststimulation. First, we compared the maximal response for each ligand-induced DMR in HEK-MOR cells with that in HEK-MOR cells after pretreatment with each of the kinase inhibitors.

As shown in Fig. 11, the maximal DMR response for each ligand in cells pretreated with kinase inhibitors was, in general, comparable to that in untreated cells. However, the DMR signals induced by tramadol and DIPPA were sensitive to pretreatment with U0126 (Fig. 2.11a). Further, DMR signals produced by a subset of agonists including DIPPA, tramadol, BRL-52537, ICI 199,441, Leu5-enkephalin, DSLET, DAMME, and dynorphin

A 1-13 were significantly reduced in cells pretreated with the p38 MAPK inhibitor SB202190 compared to controls (Fig. 2.11b). The DMR response induced by β- funaltrexamine, ICI 199,441 and dynorphin A 1-13 was enhanced by SP100625 pretreatment (Fig. 2.11c). LY294002 pretreatment appeared to have little, if any, effect the maximal DMR response induced by any of the ligands (Fig. 2.11d). a b 300 300 DAMGO DAMGO

200 200

100 100 BRL-52537 DIPPA DIPPA Tramadol Tramadol Response(pm) Response(pm) 0 U0126 0 SB202190 Buffer Buffer

-100 -100 0 10 20 30 40 50 0 10 20 30 40 50 Compound Compound c d

Dynorphin A 1-13 300 ICI 119441 300 DAMGO

200 200 -funaltrexamine 100 β 100 Response(pm) 0 SP100625 Response(pm) 0 LY294002 Buffer Buffer

-100 -100 0 10 20 30 40 50 0 10 20 30 40 50 Compound Compound

Figure 2.11 The comparison of the maximum responses of opioid ligands between the buffer- and inhibitor-pretreated HEK-MOR cells. (a) U0126; (b) SB202190; (c) SP100625 and (d) LY294002. The broken circle refers to ICI 199,441, Leu5-enkephalin, DSLET, DAMME, and dynorphin A 1-13 from left to right, respectively.

We next compared the late response (i.e., the DMR amplitude at 30 min poststimulation) of each ligand-induced DMR in HEK-MOR cells with that in HEK- MOR cells after pretreatment with kinase inhibitors. The results indicate distinct sensitivity of the late responses for many ligands to the pretreatment with distinct kinase inhibitors. The late response induced by DIPPA was significantly suppressed by U0126 (Fig. 2.12a), while the late responses induced by most agonists were significantly suppressed by SB202190 (Fig. 2.12b). However, SP100625, in general, selectively potentiated the DMR induced by strong partial and full agonists (Fig. 2.12c). SP100625 also potentiated both BUBUC- and β-funaltrexamine-induced DMR. LY294002 impacted the late responses induced by most ligands, but no clear trend was observed (Fig.2.12d). These results suggest that MOR ligands exhibit functional selectivity in their ability to trigger pathway biased agonism.

a b 200 200

150 150

100 100

50 DIPPA 50 Response(pm) U0126 Response(pm) SB202190 0 0 Buffer Buffer

-50 -50 0 10 20 30 40 50 0 10 20 30 40 50 Compound Compound c d

200 Dermorphin 200

150 150 BUBUC 100 100 -funaltrexamine β 50 50 Response(pm) Response(pm) SP100625 LY294002 0 0 Buffer Buffer

-50 -50 0 10 20 30 40 50 0 10 20 30 40 50 Compound Compound

Figure 2.12 The comparison of the DMR responses (30min poststimulation) of opioid ligands between the buffer- and inhibitor-pretreated HEK-MOR cells. (a) U0126; (b) SB202190; (c) SP100625 and (d) LY294002.

Discussion We have used a label-free integrative pharmacology on-target (iPOT) approach to characterize a library of opioid ligands at the MOR. The HEK-MOR cell line facilitated these studies, since this cell line stably expresses the MOR and is devoid of kappa or delta receptors. We have also taken advantage of the ability of the label-free Epic® system to monitor the DMR responses with high throughput. This system has allowed us to perform multiple assay formats for studying the activity of the library of opioid ligands tested, and obtain an integrated picture of their functional selectivity. This approach differs significantly from traditional pharmacological methods in several important aspects. First, the DMR assay utilizes a whole cell readout to probe the the integrated cellular response to drugs (Fang, 2010a). Second, DMR assays are performed in real time on live cells and do not require radioactive labeling or other types of manipulations that might skew or influence the behavior of cells in response to agonist or antagonist drugs. Further, the combinations of different assay formats enabled by DMR allow one to distinguish between various ligands based on their specificity, relative potency and mechanisms of action at specific GPCR sites. DMR assays are highly sensitive, thus making it possible to examine the ligand-directed functional selectivity at the MOR. The results obtained using DMR assays are, in general, consistent with those obtained using classical radioligand-based approaches (Appendix B). Compounds previously identified as MOR agonists exhibited MOR selectivity and activated Gαi signaling pathways as measured by DMR. Compounds previously identified as antagonists behaved accordingly, and were found to inhibit the DMR response induced by known MOR agonists. These studies provide compelling support for the idea that the DMR assay can be utilized to analyze the effect of opioid ligands on the MOR. The DMR assays are sensitive, and have allowed us to uncover several unexpected and atypical characteristics of known opioid ligands. First, of all the compounds tested BNTX gave the most anomalous DMR response. BNTX has previously been classified as a selective DOR antagonist. Cluster analysis further suggests that BNTX is distinct from all other ligands tested (Fig. 2.4). The DMR profiling of BNTX shows that beside its antagonist activity at the MOR, BNTX also

produced a negative DMR response in parental HEK-293 cells and an even greater negative DMR in HEK-MOR cells, suggesting that BNTX interacts with another target endogenously expressed in HEK293 cells. The unique DMR profiles of BNTX highlight the power of the DMR assay to identify the pharmacological activity of any non-selective ligands that act on both endogenous targets and the engineered receptors. Second, DMR profiling suggests that ICI 199,441 is not only an agonist for the MOR, but also activates another endogenous target. ICI 199,441 was previously classified as a full agonist for KOR. Third, for several DOR-specific ligands tested, DSLET was the only putative delta agonist that produced a DMR response similar to those induced by the MOR full agonists in HEK-MOR cells. DMR profiling, confirmed by cAMP assay results, suggests that DSLET is a full agonist for the MOR. Fourth, nalbuphine, β-funaltrexamine and levallorphan were found to be partial agonists for the MOR. Nalbuphine and β-funaltrexamine are often considered to be MOR-specific antagonists, while levallorphan is classified as a pan-opioid receptor antagonist (Chen et al., 1993; Young et al., 1984; Belcheva & Coscia, 2002). Taken together, the label-free on-target pharmacology profiling has revealed that several of the ligands in the library exhibit novel activities that are undetectable using traditional pharmacological techniques. We have utilized DMR to examine the functional selectivity of opioid ligands at the MOR. We first used DMR assays to examine the functional selectivity of the ligands for coupling to heterotrimeric G-proteins. Our results are consistent with the view that all

MOR agonists examined signal predominantly through Gαi. However, blocking Gαi signaling with PTx revealed that both DAMGO and ICI 199,441 also signal through a Gαi -independent signaling component. Pretreatment with CTx showed that DAMME, ICI

199,441, endomorphin-2 or BRL-52537 may also activated MOR signaling through a Gαs component, as these ligands demonstrated an inhibition of DMR response when pretreated with CTx. These results suggest that specific opioid ligands may confer functional selectivity by inducing the MOR to couple to alternate subsets of heterotrimeric G-proteins. Opioid receptor activation leads to the downstream activation of numerous kinase pathways. These pathways are differentially activated by various opioid ligands (Urban et

al. 2007; Belcheva & Coscia, 2002). It has been proposed that the specification of kinase signaling pathways is crucial to understanding the underlying physiology of addiction, but little is currently known about the functional selectivity of specific opioid ligands to activate distinct downstream signaling pathways (Neve, 2009). We believe that the application of DMR technology to MOR signaling will help bridge this gap in knowledge by allowing us to understand the intricate network of modulations among opioid ligands, receptors, and downstream protein kinase signaling pathways. DMR assays showed that specific opioid ligands appear to exhibit functional selectivity towards distinct kinase signaling pathways. For example, the kinase inhibitor SP100625 appeared to potentiate the DMR signal induced by β-funaltrexamine, ICI 199,441 and dynorphin A1-13, suggesting that these ligands selectively activate the JNK kinase signaling pathway. However, since these ligands also gave a small N-DMR in parental HEK-293 cells, it is possible that these ligands can activate the JNK signaling pathway via a non-MOR receptor signaling pathway. Interestingly, DIPPA, tramadol, BRL-52537, ICI 199,441, Leu5-enkephalin, DSLET, DAMME, and dynorphin A 1-13 all produced a DMR sensitive to the p38 MAPK inhibitor SB202190, suggesting that these ligands are biased towards the p38 MAPK pathway. DAMGO has previously been implicated as signaling through this pathway using conventional molecular assays (Mace et al., 2005). However, we found that the DAMGO-induced DMR response was not completely blunted by the p38 inhibitor SB202190 using whole cell DMR assays. Nonetheless, this is the first report implicating DIPPA, tramadol and BRL-52537 in biased activation of the p38 kinase signaling pathway. Our results suggest that the p38 pathway may play a more vital role in MOR-mediated signaling than has been previously appreciated (Liu & Wong, 2005). This initial study demonstrates our ability to study MOR signaling with DMR technology. It has been seen that the sensitivity and non-invasive nature of the DMR technology, paired with the high-throughput capability of the Epic® system, allows for a novel means to elucidate the signaling of opioid receptors. Therefore, this study represents the first step in forwarding our view of how we study ligand pharmacology in help revolutionize the drug discovery process.

Chapter 3 Label-free integrative pharmacology on-target of opioid ligands at the opioid receptor family

Introduction Currently, most clinically relevant opioid agonists target the MOR (Corbett et al., 2006; Fields, 2004). Yet, the therapeutic potentials of DOR and KOR agonists are currently gaining ground in the scientific community. It had been suggested that DOR and KOR agonist have the potential to become powerful analgesics with few side effects and limited abuse potential (Vanderah, 2010; Yan & Roth, 2004). More specifically, DOR and KOR activation produce antinociceptive effects with lower respiratory depression, gastrointestinal side effects, and minimal potential for physical dependence (Neve, 2009; Yan & Roth, 2004). This suggests that it is vital to examine and characterize the entire opioid system in order to fully understand the roles and impacts of each of the classic opioid receptors. The search for the holy grail has been unsuccessful thus far, and it is possible that the answer lies in DOR or KOR activation. Opioid receptors dimerize, which greatly affects functionality of ligands (Milligan, 2004). Over the past few decades, the presence and importance of receptor dimerization has been accepted by the scientific community (Ferré et al., 2009). Yet, the absolute effects of these receptor-receptor interactions is still being researched. It is difficult to distinguish whether the biological effects of dimerization is based changes in functionality due to heterodimerization or simply the convergence of circuits (Van Rijn et al., 2010). Whatever the reason, the existence of dimers and the correlation of changes in functionality of signaling pairing with dimerization is undeniable (Devi, 2001; Terrillon & Bouvier, 2004; Waldhoer et al., 2004). SH-SY5Y cells are known to express MOR and DOR receptors endogenously (Yu et al., 2006). Also, DOR and MOR opioid receptors have been shown to form heterodimers, which appear to effect subsequent functionality of the receptors (Wang et al., 2005a). SH-SY5Y cells also express the orphanin receptor (ORL-1), a receptor that is similar in structure to the classic opioid receptors but not in function (Chan et al., 2002; Meunier et al., 2002). The existence of numerous opioid receptors in one cell line has the 59

possibility to significantly affect DMR signaling when compared to the activity of isolated opioid receptors. Also notable is the fact that biosensors give an integrated signal output, which will be impacted by changes caused by receptor dimerization. We hypothesize that all three of the classic opioid receptors are involved in the analgesic functions of opioid drugs, thus a full understanding of opioid receptor signaling from all of the classic opioid receptors is necessary to forward the field of directed opioid drug development. Therefore, we must place significant emphasis on the DOR and KOR and study them in parallel to the MOR. We also hypothesize that DMR assay technology will enable us to compare the signaling deconvolution and functional selectivity of the DOR and KOR. Similarly, examining the changes in DMR responses from endogenous receptors in SH-SY5Y cells is the first key step in directing this technology toward use in primary cells. Understanding the functional role of the various signaling pathways activated by opioid receptors, and the potential impact of dimerization of those receptors represents a crucial step in understanding the mechanisms of opioid signaling.

Experimental procedure Materials and reagents Pertussis toxin, cholera toxin, forskolin and dimethyl sulfoxide (DMSO) were purchased from Sigma-Aldrich (St. Louis, MO). DAMGO, DPDPE, BRL-53527, CTOP, naltrindole hydrochloride, norbinaltorphimine, U0126, SB202190, SP600125, and LY294002 were purchased from Tocris Biosciences (Ellisville, MO). The Opioid Compound Library (consisting of 64 compounds of pan-specific and receptor subtype- specific agonists and antagonists, each at 10mM in DMSO) was obtained from Enzo Life Sciences (Plymouth Meeting, PA). All tissue culture media and reagents were purchased from Invitrogen (Carlsbad, CA). Fibronectin-coated Epic® biosensor microplates, TCT- compatible and polypropylene compound source plates were obtained from Corning Inc. (Corning, NY).

Cell culture and dynamic mass redistribution (DMR) assays HEK293 cells and SH-SY5Y cells were obtained from American Type Tissue Culture (Manassas, VA) and cultured in Dulbecco’s modified Eagle’s medium (DMEM

GlutaMAX-I) supplemented with 10% non-heated inactivated fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. The HEK-MOR and HEK-DOR cells line were a generous gift from Dr. Mark von Zastrow (University of California, San Francisco), and the HEK-KOR cell line was donated from Dr. Lui Chen (Temple University). HEK-MOR cells express FLAG-tagged wild type human mu opioid receptor, HEK-DOR cells express FLAG-tagged delta opioid receptors, and the HEK- KOR cells expressed FLAG-tagged kappa opioid receptors. These cells were grown in complete DMEM GlutaMAX-I containing 400ug/ml geneticin. Whole cell pharmacological assays were performed using the Corning Epic® system as was previously described in detail in the chapter 2. One day prior to performing the DMR assay, cells were seeded onto fibronectin-coated Epic® microplates at a density of 16,000 cells/40 µL/well for HEK293 cells and 20,000 cells/40 µL/well for HEK-DOR and HEK-KOR cells. After seeding, the Epic® microplates were incubated for 30 min at room temperature, and then transferred to a humidified incubator (37oC, 5%

CO2) for 20-24 hrs. Human neuroblastoma SH-SY5Y cells were cultured differently than the HEK based cell lines. Cells were seeded at 15,000 cells/40 µL/well on Epic® TCT-compatible microplates. The cells were incubated for 30 min at room temperature and then o transferred to a humidified incubator (37 C, 5% CO2) for 48 hrs. This was due to slower doubling time for the neuroblastoma cells, compared to HEK-based cells.

DMR assay DMR assays were performed using Epic® system as previously described (Fang and Ferrie, 2008). Epic® system from Corning is a wavelength interrogation reader system tailored for resonant waveguide grating biosensors in microtiter plates. This system consists of a temperature-control unit (26oC), an optical detection unit, and an on- board liquid handling unit with robotics. The detection unit is centered on integrated fiber optics, and enables kinetic measures of cellular responses with a time interval of ~15 sec. Once the cells reached high confluency (~95%), they were washed twice with assay buffer (1x Hank’s balanced salt solution with 20 mM HEPES, pH7.1) and transferred to the Epic® reader for 1 hr at 26oC so a steady baseline was reached. DMR

was monitored in real time with a temporal resolution of ~15 sec throughout the assays. A typical DMR proceeded with a 2-min baseline, followed by a real time kinetic response after the compound additions using the onboard liquid handler. The DMR was recorded as a shift in resonant wavelength (picometer, pm). Different DMR assay formats were used for profiling opioid ligands. Assays were described in detail in chapter 2 of this dissertation. Briefly, DMR agonist assays were used to directly record the DMR signal arising from a ligand itself. DMR two-step assays (here referred to as antagonist assays) were used to record the DMR arising from an agonist at a fixed dose (usually its EC100) after pretreatment with an inhibitor or a ligand. The pretreatment of cells with probe molecules was achieved by incubating the cells with a probe molecule at the indicated dose for the indicated period of time (Table 3.1). The probe molecules were used to achieve a wide range of chemical environments for each cell line, which, in turn, manifest the specificity, relative potency and efficacy, and modes of action of the drugs. Specifically, cells were pretreated with either 0.1% DMSO (the positive control), 10 µM opioid ligand in the library, 100 ng/ml PTx, 400 ng/ml CTx, 10 µM forskolin, 10 µM U0126, 10 µM SB202190, 10 µM SP100625, or 10 µM LY294002 for the times indicated (Table 3.1). Cells were then stimulated with an opioid ligand, whose responses were recorded in real time and used for similarity and correlation analysis. We screened a library of 64 opioid ligands. Literature mining revealed that fifty- five of the opioid ligands in the library had previously been shown to possess binding affinity for at least one member of the classic opioid receptor family (Appendix B-D), and thus chosen for analysis in this study.

Table 3.1. Assay protocols and DMR signals used for similarity analysis

Probe, Labels used in

Cell pretreatment duration DMR readout clustering Figures

HEK293 HEK-3, 9, 30 Figure 3.1

Figures 3.6 to

Opioid* 0.1% DMSO in buffer, 1hr Ligand, 10 µM Buffer-3, 9, 30 3.8

HEK-MOR DAMGO, 10 µM MOR-3, 9, 30

HEK-DOR 10 µM ligand, 1hr DPDPE, 10 µM DOR-3, 9, 30 Figure 3.3

HEK-KOR BRL-57532, 10 µM KOR-3, 9, 30

SH-SY5Y DAMGO, 10 µM 5Y-3, 9, 30

Opioid* 100ng/ml PTX, 20hr Ligand, 10 µM PTX-3, 9, 30

Opioid* 400ng/ml CTX, 20hr Ligand, 10 µM CTX-3, 9, 30 Figures 3.6 to Opioid* 10 µM forskolin, 1hr Ligand, 10 µM FSK-3, 9, 30 3.8 Opioid* 10 µM U0126, 1hr Ligand, 10 µM U0126-3, 9, 30

Opioid* 10 µM SB202190, 1hr Ligand, 10 µM SB-3, 9, 30

Opioid* 10 µM SP100625, 1hr Ligand, 10 µM SP-3, 9, 30

Opioid* 10 µM LY294002, 1hr Ligand, 10 µM LY-3, 9, 30

* Opioid receptor expressing cell lines wherein the same assay protocol was applied.

Data visualization and clustering For each opioid ligand in a cell line, ten DMR assays were performed that measured receptor specificity, G-protein coupling, and downstream kinase pathway selectivity. The real responses at three distinct time points (3min, 9 min, and 30 min poststimulation) were extracted from each kinetic DMR signal and used to rewrite the DMR pharmacology of each ligand. We used the DMR output to form a numerical descriptor containing multi-dimensional coordinates for each ligand, which was then subject to similarity analysis. At least duplicate data for each assay were collected to generate an averaged response. For visualization, the real-time responses were color coded to

illustrate relative differences in DMR signal strength. The red color refers to a positive DMR value, the black a value near zero, and the green color represents a negative DMR value. Differences in color intensity illustrate differences in signal strength, i.e. the stronger the DMR response, the more intense the color. In the ligand matrix, each column represents one DMR response at a particular time under a specific assay condition, and each row represents one ligand. Every row and column carries equal weight. The Ward hierarchical clustering algorithm and Euclidean distance metrics (Fang, 2010a; Eisen et al., 1998) were used for generating heat maps and clustering the DMR profiles. Each assay was arranged in three consecutive columns to form a column group for clear understanding of the key characteristics of a DMR response.

Statistical analysis. For profiling, two independent measurements, each done in duplicate, were performed. All replicates passed the 2 sigma coefficient of variation test in order to be included in the analysis. Drugs whose DMR responses failed the statistical test were re- screened. At least two replicates were included for the final analysis. For dose responses, at least two independent measurements, each done at least in duplicate, were performed to calculate the mean responses and the standard deviations (s.d.).

Results Selectivity of opioid agonists at the opioid receptor family We first used DMR agonist assays to determine the selectivity of a library of opioid ligands at different opioid receptors. The DMR agonist assay measures the ability of ligands to trigger DMR signals in live-cells. The library consists of fifty-five opioid ligands (Appendix A), and all ligands were examined at 10µM in order to be compatible with high throughput screening. Five distinct cell lines were profiled, including human neuroblastoma cell line SH-SY5Y, human embryonic kidney HEK293 cells, and three engineered HEK 293 cell lines. Each of the engineered cell lines expresses a specific

a b 3 min 9 min min 30 SNC80 200 Deltorphin II 3 9 30 Endomorphin-2 Dynorphin A1-8 DADLE 0 Leu5-enkephalin DSLET Dynorphin A2-13 100 BUBUC α-neoendorphin DAMGO Met5-enkephalin Syndyphalin SD25 Endomorphin-1 DTLET Dynorphin B 0 SNC121 DAMME Naloxone HCl DPDPE DAMGO t = 0 DALDA Morphiceptin NNC63-0532 Response(pm) Dynorphin A1-13 -100 BNTX TAPP GR89696 ICI204448 Nociceptin1-13 DIPPA Nociceptin Norcodeine U-54494A -200 N-MPPP Etonitazenyl 0 10 20 30 40 50 DIPPA ICI199441 Time (min) BNTX c 8 7 6 5 1 3 2 11 10 9 12 13 14 15 39 41 44 43 42 36 35 37 34 33 32 31 30 29 28 27 26 25 18 20 19 17 16 23 22 24 21

SNC80 Met5-enkephalin DTLET Endomorphin-1 Dynorphin B α-neoendorphin Syndyphalin SD25 Leu5-enkephalin DSLET Deltorphin II SNC121 Endomorphin-2 DADLE Dynorphin A1-8 Dynorphin A1-13 NNC63-0532 GR89696 ICI204448 DAMME Dynorphin A2-13 BUBUC DPDPE DAMGO Morphiceptin DALDA TAPP Nociceptin1-13 Nociceptin Norcodeine U-54494A N-MPPP Etonitazenyl DIPPA ICI199441 BNTX

Fig. 3.1 Extracting DMR parameters for effective similarity analysis. (a) Representative DMR signals of opioid ligands in native SH-SY5Y cells. The data represents the mean ± s.d. of 2 independent measurements, each in duplicate (n = 4). Responses at three time points (3 min, 9 min, and 30min post-stimulation) were extracted to represent each DMR signal. The solid arrow indicates the time when ligands were added (t =0). (b) A false colored heat map of opioid ligand- induced DMR in native SH-SY5Y cells based on the responses at the three time points. (c) A false colored heat map based on the real time DMR signals of all opioid ligands that gave rise to a detectable DMR. The real time responses showed was reduced to every min. Three time domains were evident. 65

opioid receptor, i.e., µ- (MOR), δ- (DOR), or κ- (KOR). The stably transfected cell lines are referred to as HEK-MOR, HEK-DOR or HEK-KOR cells, respectively. Interrogating SH-SY5Y cells with the library of opioid ligands identified three types of DMR signals (Fig. 3.1a). Twenty out of the fifty-five ligands tested, including antagonists such as naloxone-HCl, were silent in the cells and produced a net-zero DMR. Thirty-one out of the fifty-five ligands tested, including DAMGO, produced a biphasic DMR response which was seen as an initial positive DMR (P-DMR) followed by a negative DMR (N-DMR) that eventually decayed above the baseline within 1 hr of stimulation. The remaining four ligands including DIPPA, etonitazenyl isothiocyanate, BNTX, and ICI 199,441, produced a biphasic DMR response whose late N-DMR component eventually decayed below the baseline. Since a DMR signal is a kinetic response and can be considered to be a poly- dimensional coordinate, i.e., a temporal series of biosensor signals (Fang et al., 2006; Fang, 2010a), it is possible to use the real time response as the basis for similarity analysis (Fang, 2010b). Similarity analysis is a powerful means to cluster molecules based on a large set of data (Gehlenborg et al., 2010). However, since a DMR response consists of over 200 dimensions due to high temporal resolution, we first reduced the DMR dimensions to three distinct time points (3, 9, and 30 min post-stimulation) for effective similarity analysis (Fig. 3.1b). This dimensional reduction is based on clustering of time domains of the DMR responses from all opioid ligands in each of the five cell lines examined. For SH-SY5Y cells, similarity analysis using the unsupervised Ward hierarchical clustering algorithm and Euclidean distance metrics (Eisen et al., 1998) showed that all DMR signals with an amplitude greater than 30 pm generally propagate with three distinct time periods: immediate (2-8 min), early (9-15 min), and late responses (15-50 min post-stimulation) (Fig. 3.1c). Clustering based on the entire kinetic response or the reduced three time-points led to similar clusters of ligands in SH-SY5Y cells (comparing Fig. 3.1c with Fig. 3.1b), although clustering based on the entire kinetic response gave rise to better resolution than by using the reduced time-points. Considering that multiple assays were performed for each ligand, we chose to limit our parameters. Thus, we used the three time-points (3, 9, and 30 min) for each DMR response to represent the DMR pharmacology obtained for all of the ligands tested.

To depict the cell type-dependent DMR responses of the library ligands, we created a heat map illustrating the selective agonism of all ligands (Fig. 3.2). Only ligands that induce a DMR response greater than 50 pm (opposed to the 30 pm cutoff which was used for Fig 3.1) at any time-point in at least one of the cell lines were included in this analysis. To clearly visualize the characteristics of a ligand-induced DMR in each cell line, the DMR responses at the three time-points were grouped together; thus, for all ligands there are three adjacent columns for each cell line in the heat map. This analysis produced three significant observations. First, six ligands gave rise to a noticeable DMR in parental HEK293 cells (Appendix A). BNTX, β- funaltrexamine, etonitazenyl isothiocyanate, and ICI 199,441 all led to an N-DMR, possibly via the interaction with an unknown endogenous target(s). Conversely, both dynorphin A 2-13 and nocicepin 1-13 led to a small P-DMR, possibly via endogenous opioid-like receptor-1 (ORL1) (see below). Except for BNTX, which led to an N-DMR in all five cell lines, each of the other five ligands produced a biphasic DMR in the four opioid receptor expressing cell lines. Second, fifty-one of the library ligands showed agonist activity in at least one of the four opioid receptor-expressing cell lines. Four ligands, naloxone methiodide, naltrindole, nor-binaltorphimine, and naltriben, were silent in all five cell lines. Interestingly, naloxone-HCl led to a small yet specific DMR in HEK-DOR cells, suggesting that this ligand may act as a partial agonist for the DOR. Several ligands that are believed to be opioid receptor antagonists also produced noticeable DMR in at least one of the opioid receptor expressing cell lines. Specifically, nalbuphine and β- funaltrexamine acted as partial agonists at MOR, DOR, and KOR sites, but were inactive in SH-SY5Y cells. Levallorphan, SKF10047, N-benzylnaltrindole and naloxonazine appeared to be partial agonists at DOR and KOR sites, while naltrexone was a partial agonist specific to the KOR. Third, the pattern of agonist activity in SH-SY5Y cells cannot be explained solely by activation of the MOR by opioid agonist ligands. This conclusion is based on the following results. First, SNC 121 was inactive in HEK-MOR cells, but active in SH- SY5Y cells. Second, tramadol and levallorphan were active in HEK-MOR cells, inactive

in SH-SY5Y cells. Third, when compared to the DAMGO-induced DMR responses, dynorphin A 1-13 produced a similar DMR in HEK-MOR cells, but a smaller DMR in 3 9 30 3 9 30 3 9 30 3 9 30 ------3 9 30 - - - HEK HEK HEK MOR MOR MOR DOR DOR DOR KOR KOR KOR 5Y 5Y 5Y SNC 80 SNC 121 DPDPE Met5-Enkephalin DTLET DOR Leu5-Enkephalin DADLE DSLET BUBUC Deltorphin II α-Neoendorphin Dynorphin A 1-8 Dynorphin B Dynorphin A 1-13 DAMME Opioid NNC 63-0532 ICI 204,448 DIPPA ICI 199,441 Dynorphin A 2-13 GR 89696 Levallorphan SKF10047 DOR/KOR N-Benzylnaltrindole Tramadol Endomorphin-1 DALDA Morphiceptin Endomorphin-2 (-)-Norcodeine MOR Syndyphalin SD-25 DAMGO TAPP Etonitazenyl Naltrexone Naloxonazine Salvinorin A U-62066 (+)-U-50488 (-)-U-50488 U-50,488H β-Funaltrexamine KOR BRL-52537 Naloxone HCl Nalbuphine U-69593 N-MPPP Nociceptin 1-13 U-54494A Nociceptin BNTX Fig. 3.2 A false colored heat map based on the DMR responses of opioid ligands in five different cell lines. The cell lines were parental HEK-293, HEK-MOR, HEK-DOR, HEK-KOR and SH-SY5Y. Only ligands that resulted in a DMR with at least one response at the three time points greater than 50 pm were included in this analysis. The negative control (DMSO) was also included.

SH-SY5Y cells. Fourth, DIPPA induced a P-DMR that decayed above the baseline within the assay duration in each of the three stably transfected cell lines, but a P-DMR response that decayed below the baseline in SH-SY5Y cells (also see Fig. 3.1a). Lastly,

nalbuphine and U-69593 were active in the three transfected cell lines, but inactive in SH-SY5Y cells.

Selectivity of opioid ligands to block the DMR response produced by the activation of opioid receptors We used a two-step DMR assay (i.e., an antagonist assays) to determine the ability of opioid ligands to block the DMR responses resulting from the activation of opioid receptors. The antagonist assay was performed in two sequential steps. Cells were pretreated for one hr with a ligand from the opioid library, followed by treatment with a fixed dose of a known opioid agonist, in order to determine the effect of the ligand on the known agonist-induced response. A ligand that does not trigger a DMR but blocks the DMR of the known agonist is termed an antagonist. Conversely, a ligand that induces a noticeable DMR response but desensitizes the cells responding to the succeeding agonist is termed an agonist. We first determined the DMR potency of a known agonist for each cell line: DAMGO for HEK-MOR, DPDPE for HEK-DOR, BRL-52537 for HEK-KOR, and DAMGO for SH-SY5Y cells. We have previously shown that DAMGO produces a monophasic dose response in HEK-MOR cells with an EC50 of 0.93±0.12 nM (Morse et al., 2011). In HEK-DOR cells, DPDPE produced biphasic dose response with two distinct

EC50’s of 0.15±0.03 nM, and 2.8±0.09 nM (2 independent measurements, n =4) (Fig. 3.3a and b). In HEK-KOR cells, BRL-52537 also produced a biphasic dose response with two distinct EC50’s of 35.6±3.1 pM, and 26.0±1.9 nM (2 independent measurements, n =4) (Fig. 3.3c and d). Conversely, in SH-SY5Y cells DAMGO produced a monophasic dose response with an EC50 of 4.5±0.3 nM (2 independent measurements, n =4) (Fig. 3.3e and f). These results indicate that each selective agonist was potent on their respective cells lines and were therefore chosen for comparative studies. We then studied the ability of ligands in the opioid ligand library to block the DMR response elicited by a specific opioid receptor agonist in the corresponding cell line. In order to achieve high resolution to differentiate the relative potency of opioid ligands to block the agonist DMR response at each receptor site, we employed a high dose for each agonist tested (10µM DAMGO for HEK-MOR cells, 10µM DPDPE for 69

HEK-DOR cells, 10µM BRL-52537 for HEK-KOR cells, and 10µM DAMGO for SH- SY5Y cells). Ward hierarchical clustering algorithm and Euclidean distance metrics were a c e 500 250 100

400 200 256nM 75 64nM 128nM 64nM 8nM 300 150 64nM 32nM 4nM 32nM 50 16nM 2nM 200 100 4nM 8nM 1nM 1nM 25 4nM 500pM 100 50 0.25nM 2nM Response(pm) 250pM Response(pm) 0.06nM Response(pm) 1nM 125pM 0 0.01nM 0 0.5nM 0 63pM 32pM -100 -50 -25 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 Time (min) Time (min) Time (min) b d f 500 250 125

400 200 100 300 75 150 200 50 100 100 25 Response(pm) Response(pm) Response(pm) 0 50 0

-100 0 -25 -13 -12 -11 -10 -9 -8 -7 -6 -5 -12 -11 -10 -9 -8 -7 -6 -5 -10 -9 -8 -7 -6 -5 DPDPE, log M BRL-52537, log M DAMGO, log M

Fig. 3.3 Dose-dependent responses of agonists in distinct cell lines. (a, b) Dose responses of DPDPE in HEK-DOR; (c, d) Dose responses of BRL- 52537 in HEK-KOR; and (e, f) Dose responses of DAMGO in SH-SY5Y cells. (a, c, e) Real time DMR, each curve represents the mean ± s.d. of 2 independent measurements, each in duplicate (n = 4). (b, d, f) The maximal amplitudes as a function of agonist doses (n=4).

used to create a high resolution map that separated these ligands into different clusters (Fig. 3.4). Close examination of ligands within each subcluster led to the following conclusions. Most of the ligands grouped into each subcluster exhibited DMR characteristics that were in general agreement with their previously described pharmacology and classifications (Appendix B-D). The KOR subcluster contains a set of KOR-specific agonists, all of which not only produced a P-DMR in HEK-KOR cells, but also caused the HEK-KOR cells to be desensitized to the repeated stimulation with BRL- 52537.

3 9 30 3 9 30 3 9 30 ------3 9 30 - - - MOR MOR MOR DOR DOR DOR KOR KOR KOR 5Y 5Y 5Y

α-Neoendorphin Dynorphin A 1-8 Dynorphin B Dynorphin A 1-13 Non-selective GR 89696 ICI 199,441 SNC 80 SNC 121 DPDPE BUBUC DOR Deltorphin II Naltriben Naltrindole N-Benzylnaltrindole Met5-Enkephalin DTLET Leu5-Enkephalin DADLE MOR/DOR DSLET DAMME BNTX Levallorphan Etonitazenyl Endomorphin-2 Nalbuphine MOR β-Funaltrexamine SKF10047 NNC 63-0532 Naloxone methiodide (-)-Norcodeine U-54494A nor-Binaltorphimine Tramadol Less potent DMSO Dynorphin A 2-13 Nociceptin 1-13 Nociceptin Syndyphalin SD-25 Endomorphin-1 DAMGO Morphiceptin DALDA MOR TAPP Naltrexone Naloxone HCl Naloxonazine DIPPA ICI 204,448 U-62066 U-69593 (-)-U-50488 KOR BRL-52537 Salvinorin A U-50,488H (+)-U-50488 N-MPPP

Fig. 3.4 A false colored heat map based on the selectivity of opioid ligands to the opioid receptor family. The DMR of a specific agonist at a fixed dose in each cell line after pretreatment with an opioid ligand in the library was used to generate the heat map. The agonists used were DAMGO (10 µM), DPDPE (10 µM), BRL-52537 (10 µM), and DAMGO (10 µM) for HEK-MOR, HEK- DOR, HEK-KOR and SH-SY5Y cells, respectively. All of the ligands in the library were assayed at 10 µM, and used to pretreat the cells for about 1 hr. All fifty-five ligands were included in the analysis. The positive controls (i.e., the agonist responses in the DMSO pretreated cells) were also included. 71

The DOR subcluster contains both DOR-specific agonists and antagonists. SNC80, SNC121, DPDPE, BUBUC and deltorphin II acted as specific agonists at DOR sites, while naltriben, naltrindole and N-benzylnaltrindole were relatively specific antagonists for the DOR. The two MOR-specific subclusters contain agonists specific to the MOR, except for three non-selective opioid antagonists. These included naltrexone, naloxone HCl, and naloxonzaine. β-funaltrexamine was also clustered with one of the two MOR-specific subclusters. β-funaltrexamine had previously been identified as a partial agonist for the MOR (Morse et al., 2011). The non-selective clusters consist of α- neoendorphin, dynorphin A 1-8, dynorphin B, dynorphin A 1-13, GR 89696, and ICI 199,441, which were potent agonists at all opioid receptor sites. All six agonists completely desensitized the HEK-MOR cells to subsequent stimulation with DAMGO. However, both HEK-DOR and HEK-KOR cells still responded to their corresponding agonists, DPDPE and BRL52537, with an N-DMR signal after pretreatment with α- neoendorphin, dynorphin A 1-8, dynorphin B, or dynorphin A 1-13. It is possible that these four agonists only activate an arm of the pathways elicited by either DPDPE or BRL52537 for their respective receptors. The MOR/DOR subcluster consists of a group of agonists with potency at both MOR and DOR sites. The exception is BNTX, which appears to be an antagonist against MOR and DOR. The remaining ligands were grouped with the negative control (i.e., DMSO), and displayed low potency to all three opioid receptors examined. Together, these results confirm that opioid ligands are divergent in their relative potency at distinct members of the opioid receptor family. We further examined the DMR response produced by DAMGO in SH-SY5Y cells with and without pretreatment with the library ligands, based on reported affinities of opioid ligands (Appendix B-D). SH-SY5Y cells endogenously express MOR and DOR, and thus may exhibit very different signaling patterns due to the potential for receptor dimerization. Only ligands whose binding affinities at specific receptor sites are known were included in this analysis. The ligands that blocked the DAMGO-elicited DMR in HEK-MOR also blocked the DAMGO-induced DMR in SH-SY5Y cells, suggesting that the DAMGO response in SH-SY5Y cells mainly originated from the activation of the MOR. However, the extent of the DAMGO-induced DMR observed after pretreatment with the library of opioid ligands in SH-SY5Y cells cannot be explained by the known 72

binding affinities of these ligands at either MOR or the DOR sites alone (Fig. 3.5a and Fig. 3.5b, respectively). It is possible that receptor dimerization or the differing cellular context accounts for these differences. Three potent antagonists, β-funaltrexamine, levallorphan and nor-binaltorphimine, appeared to be less potent blocking the DAMGO- induced DMR in SH-SY5Y cells than that would be expected at MOR binding sites. Conversely, three agonists including SKF10047, ICI 199,441 and DIPPA desensitized SH-SY5Y cells with greater potency than their reported affinities at the MOR (Fig. 3.5a). Further, the DOR-selective agonists including deltorphin II, SNC121, BUBUC, SNC80 and DPDPE desensitized SH-SY5Y cells with lower potency than expected (Fig. 3.5b). In comparison, the DAMGO-induced DMR response in HEK-MOR cells after pretreatment with library ligands correlated well with their known binding affinities. The exception was of a group of antagonists that included nor-binaltorphimine, N- benzylnatrindole, naloxone methiodide, naltrindole, and naltriben (Fig. 3.5c). Interestingly, the ability of a group of opioid antagonists to block the DPDPE-induced DMR in HEK-DOR cells was also divergent from their known binding affinities to the DOR (Fig. 3.5d), suggesting that there are more than one population of receptors in HEK-DOR cells (see below). Together, these results suggest that ligand pharmacology at the whole cell level from cells which contain multiple receptors is different from the binding profiles obtained from cell lines expressing only one receptor family member.

a b

125 Nociceptin 125

Norcodeine 100 100 Deltorphin II SNC 121 75 75 BUBUC SNC 80 Nor-binaltorphimine DPDPE 50 Levallorphan 50 β-funaltrexamine Naltriben 25 25 Levalloprhan SKF10047 Natrindole 0 ICI 199441 0 % DAMGO in SH-SH-5YDAMGOin % DIPPA SH-SH-5YDAMGOin % -25 -25 -12 -11 -10 -9 -8 -7 -6 -5 -12 -11 -10 -9 -8 -7 -6 -5 Affinity to the MOR, log M Affinity to the DOR, log M c d 125 125

Nor-binaltorphimine Naloxone methiodide 100 N-benzylnatrindole 100 Nalbuphine Naloxone methiodide Nociceptin β-Flualtrexamine Naltrindole Naltrexone 75 Naltriben 75 Naloxone HCl DAMGO 50 50 Naloxonazine Nor-Binaltorphimine 25 25 Levallorphan U-62066

0 ICI199441

% DPDPEHEK-DOR% in 0 % DAMGO in HEK-MORDAMGOin % Norcodeine Naltrindole N-benzylnatrindole -25 -25 -12 -11 -10 -9 -8 -7 -6 -5 -12 -11 -10 -9 -8 -7 -6 -5 Affinity to the MOR, log M Affinity to the DOR, log M

Figure 3.5 The inhibition pattern by opioid ligands. (a) The percentage of DAMGO responses in SH-SY5Y cells as a function of the binding affinity of opioid ligands to the MOR. (b) The percentage of DAMGO responses in SH-SY5Y cells as a function of the binding affinity of opioid ligands to the DOR. (c) The percentage of DAMGO responses in HEK-MOR cells as a function of the binding affinity of opioid ligands to the MOR. (d) The percentage of DPDPE responses in HEK-DOR cells as a function of the binding affinity of opioid ligands to the DOR. The percentage of agonist responses after pretreatment with ligands in the library were calculated based on the normalization of the agonist response in the presence of a ligand to the positive control (i.e., the agonist response after pretreatment with the vehicle buffer only). The data points in pink were calculated based on the known binding affinity of each ligand against the specific receptor using % agonist response = 1/[1+10^log (X – Ki)], wherein X is the concentration of each ligand, and Ki the binding affinity obtained in literature.

Functional selectivity of opioid ligands at the opioid receptors To examine the functional selectivity of opioid ligands, we used DMR assays to analyze the pharmacological properties of the library of opioid ligands in HEK-DOR cells, HEK-KOR cells and SH-SY5Y cells. Functional selectivity was a measurement of the sensitivity of the ligand-induced DMR response to pretreatment of cells with various probe molecules. This assay allowed us to characterize changes in opioid receptor- 74

mediated DMR signals produced by each of the opioid ligands in the library. The choice of probe molecules was based on known signaling pathway cascades, and used to block specific cellular signaling protein(s) downstream of the opioid receptors. The probe molecules include PTx, CTx, forskolin, U0126, SB202190, SP100625, and LY294002.

PTx blocks Gαi signaling (Barbieri and Cortina, 1988) while CTx activates of Gαs (Gill and Meren, 1978). Forskolin is an activator of adenylyl cyclase and is widely used for cell-based screening due to potentiation of Gαi-mediated signaling and desensitization of

Gαs-mediated signaling (Tran and Fang, 2009). U0126, SB202190, SP100625, and LY294002 are kinase pathway inhibitors for MEK1/2, p38 MAPK, JNK, and PI3K, respectively. It has been suggested that opioid ligands often exhibit functional selectivity on these pathways (Neve, 2009; Raehal et al., 2011). We excluded BNTX, β-funaltrexamine, etonitazenyl isothiocyanate, ICI 199,441, dynorphin A2-13 and nocicepin 1-13 for further analysis because of their previously determined cross-reactivity with unknown endogenous cellular targets (Morse et al., 2011). To visualize the effects of the pretreatments with probe molecules, we used the net change of the DMR response of a ligand (i.e., its DMR in DMSO treated cells minus its DMR in a probe molecule pretreated cells) for similarity analysis. This was done for all assay conditions except for PTx pretreatment wherein the raw DMR responses were used. The DMR in the DMSO treated cells were also included as references. A positive net change indicates that the probe pretreatment potentiates a ligand-induced DMR response, while a negative net change indicates a decrease in a ligand-induced DMR response by the probe pretreatment. The average responses of at least 2 experiments were used. Statistical analysis showed that for a total of 3960 DMR data points obtained (3 cell lines x 8 assay conditions x 55 ligands x 3 time points), 97.1% gave rise to an absolute difference between replicates for a ligand under one condition that is smaller than 10 pm, and the remaining 2.9% (115 parameters, all of which occurred in either HEK-DOR or HEK-KOR cells) smaller than 20 pm. Thus, a net change induced by the probe pretreatment greater than 30 pm for a ligand DMR parameter is considered to be significant for both HEK-DOR and HEK-KOR cells, while a net change greater than 20 pm is to be significant for SH-SY5Y cells.

Profiling the ligand library induced net-DMR responses in HEK-DOR cells after pretreatment with seven probe molecules produced a heat map which grouped the ligands into two large superclusters (Fig. 3.6). Notably, all ligands gave rise to a P-DMR response under at least one assay condition. The first supercluster consists of antagonists and ligands that were inactive in the untreated HEK-DOR cells, except for endomorphin- 1 which acted as a partial agonist in the control HEK-DOR cells (i.e., the cells pretreated with the vehicle only). All ligands in this supercluster exhibited a small P-DMR in the forskolin-pretreated cells, suggesting that these ligands gave rise to weak partial agonist activity when the basal cAMP level is high. The second supercluster can be further subdivided into three subclusters, one comprised of ligands such as DPDPE who appear to act as full ligands, and two others comprised of ligands that appear to act as partial agonists. The compounds in the full agonist subcluster still triggered a noticeable DMR

response in PTx pretreated cells, suggesting that these agonists also activate a Gαi protein- independent signaling pathway. CTx pretreatment potentiated the DMR response of these agonists. However, forskolin only potentiated the early DMR response but suppressed the late DMR response (i.e., 30min post-stimulation). The ERK1/2 inhibitor U0126, the JNK inhibitor SP100625, and the PI3K inhibitor LY294002 generally potentiated the DMR response of agonists in the full agonist subcluster, while the p38 MAPK inhibitor SB202190 suppressed their DMR response. The second subcluster was comprised of DIPPA, dynorphin A 1-13, NNC63-0532, N-benzylnaltrindole, and U-5449A. Both DIPPA and dynorphin A 1-13 produced a noticeable DMR response in PTx pretreated cells, and exhibited a potentiated DMR response in the CTx or forskolin treated cells, while the other three agonists did not. Further, forskolin pretreatment significantly suppressed the late DMR response of dynorphin A 1-13. All five agonists in this subcluster were insensitive to U0126, SP100625 and LY294002 pretreatment. U-5449A was insensitive to SB202190 pretreatment, but the DMR responses of the other four agonists were suppressed by SB202190. The third subcluster consists of fifteen ligands,

3 9 30 3 9 30 ------3 9 30 3 9 30 3 9 30 ------3 9 30 3 9 30 3 9 30 ------Buffer Buffer Buffer PTX PTX PTX CTX CTX CTX FSK FSK FSK U0126 U0126 U0126 SB SB SB SP SP SP LY LY LY

Endomorphin-1 Naltrexone Salvinorin A Nor-binaltorphimine Naltriben DALDA (+)-U-50488 Tramadol Nociceptin Naloxone methiodide U-62066 Morphinceptin Naloxone HCl Met5-enkephalin Dynorphin B SNC80 DPDPE α-Neoendorphin DTLET BUBUC DADLE Deltorphin II Dynorphin A 1-8 DSLET DAMME Leu5-enkephalin GR 89696 SNC121 ICI 204,448 DIPPA Dynorphin A 1-13 NNC63-0532 N-Benzylnaltrindole U-5449A Endomorphin-2 Nalbuphine N-MPPP Levallorphan Syndyphalin SD-25 SKF10047 DAMGO TAPP U-69593 BRL-52537 (-)-U-50488 U-50488H Naltrindole (-)-Norcodeine Naloxonazine DMSO

Fig. 3.6 A false colored heat map based on the functional selectivity of opioid ligands at the DOR. The DMR signals of ligands in HEK-DOR cells with and without (i.e., buffer) pretreatment with probe molecules including PTX, CTX, U0126, SB202190, SP100625 and LY294002 were used to generate the heat map. All of the ligands in the library were assayed at 10 µM. The negative control (DMSO) was also included. To effectively visualize the impact of probe molecules, the net change for each ligand after pretreatment was obtained via subtraction, except for both the positive control (i.e., DMR in cells pretreated with the buffer vehicle only) and the ligand DMR in PTX-pretreated cells for which the raw data were used.

including endomorphin-2. None of the ligands in this subcluster produced a DMR response in PTx- treated cells. The DMR responses of all ligands in this subcluster were insensitive to the pretreatment with CTx, U0126 or SB202190, but were potentiated by forskolin pretreatment. A set of agonists in this subcluster led to an increased DMR response in SP100625- or LY294002-pretreated cells. Together, these results suggest that the opioid ligands are divergent in their ability to trigger G protein-independent signaling and activate distinct kinase pathways downstream to activation of the DOR. The DMR profiles obtained in HEK-KOR cells under the seven assay conditions produced a heat map that also separated the ligands into two superclusters (Fig. 3.7). The first cluster consists of the DMSO negative control and nor-binaltorphimine. The absence of any DMR under all conditions suggests that nor-binaltorphimine behaved as a true neutral antagonist at the KOR. The second supercluster can be further subdivided into multiple subclusters, each of which produced a P-DMR signal under at least one assay condition. Agonists that produced a detectable P-DMR in the PTx pretreated cells include DIPPA, dynorphin B, α-neoendorphin, dynorphin A 1-8, dynorphin A 1-13, (-)-U-50488, U-50488H, salvinorin A, U-69595, U-62066, BRL-52537, and GR89696. Unlike the responses seen in HEK-MOR and HEK-DOR cells, the DMR responses of almost all agonists were found to be insensitive to both CTx- and forkolin-pretreatment in HEK- KOR cells. A similar pattern was observed for both SP100625- and LY294002-treatment. However, pretreatment of HEK-KOR cells with SB202190 suppressed the ligand-library DMR response induced by virtually all agonists, with U-50488H exhibiting the most significant suppression. Further, U0126 selectively suppressed the DMR of (-)-U-50488 and U-50488H. Together, these results suggest that p38 MAPK pathway plays a more significant role in signaling mediated via the KOR than any of the other kinase pathways. We next used the seven assay pretreatments to interrogate the effects of the opioid ligand library on the DMR responses elicited in SH-SY5Y cells (Fig. 3.8). Ligands in the agonist supercluster typically behaved as would be expected. However, some ligands, most notably DIPPA, produced unique DMR responses. DIPPA triggered a biphasic DMR response which eventually decayed below the baseline in the native SH-SY5Y cells, while PTx pretreatment suppressed both the early and late DMR response. CTx and

3 9 30 3 9 30 ------3 9 30 3 9 30 3 9 30 ------3 9 30 3 9 30 3 9 30 ------Buffer Buffer Buffer PTX PTX PTX CTX CTX CTX FSK FSK FSK U0126 U0126 U0126 SB SB SB SP SP SP LY LY LY

DIPPA Dynorphin B α-neoendorphin Dynorphin A 1-8 Dynorphin A 1-13 Naltriben DALDA N-benzylnaltrindole Tramadol Deltorphin II Endomorphin-1 Naloxone methiodide Morphinceptin Met5-enkephalin DADLE Endomorphin-2 Nociceptin SNC80 DTLET DPDPE (-)-norcodeine DAMGO Nalbuphine SFK10047 NNC63-0532 (-)-U-50488 U-50488H Salvinorin A U-69595 U-62066 BRL-52537 GR89696 (+)-U-50488 ICI204448 N-MPPP U-54494A Naltrindole Syndyphalin SD-25 Naloxone HCl Naltrexone Levallorphan Leu5-enkephalin TAPP Naloxonazine SNC121 DAMME BUBUC DSLET Nor-binaltorphimine DMSO

Fig. 3.7 A false colored heat map based on the functional selectivity of opioid ligands at the KOR. All of the ligands in the library were assayed at 10 µM. The negative control (DMSO) was also included. To effectively visualize the impact of probe molecules, the net change for each ligand after pretreatment was obtained via subtraction, except for both the positive control (i.e., DMR in cells pretreated with the buffer vehicle only) and the ligand DMR in PTX-pretreated cells for which the raw data were used.

forskolin potentiated the DMR response. U0126 converted the DMR response to a single phase N-DMR and SB202190 delayed the time to reach its peak. This unique pattern suggests that DIPPA activates both Gαi-dependent and independent pathways, and is biased predominantly towards the ERK1/2 and p38 MAPK pathways. Except for DAMGO and TAPP, ligands in the agonist supercluster led to little or no DMR reponse in PTx pretreated cells. Both CTx and forskolin suppressed the DMR response induced by dynorphin A 1-8, DPDPE and DALDA. Forskolin also suppressed the DMR response produced by Leu5-enkephalin, DSLET and DAMME. In general, the kinase inhibitors suppressed the same group of agonists, which included dynorphin A 1-8, DPDPE, DALDA, GR89696 and DAMME. These results suggest that ligand pharmacology in SH-SY5Y cells is distinct from those in both HEK-MOR and HEK-DOR.

Potency and efficacy of opioid ligands at distinct opioid receptors Based on the label-free integrative pharmacology on-target profiles, we further examined the dose responses of selected ligands at distinct opioid receptors. For HEK- DOR cells, seven additional ligands, besides DPDPE, were profiled using both DMR 1- step agonist and 2-step antagonist assays. The agonist DMR assays showed that five of these ligands, DPDPE, DAMGO, ICI 199,441, naltrindole and β-funaltrexamine, gave rise to dose-dependent responses in HEK-DOR cells (Fig. 3.9a) while naltriben and naloxone HCl were silent in HEK-DOR cells. BNTX, the last ligand examined, yielded a negative DMR via activation of an unknown endogenous target (Morse et al., 2011). DPDPE which resulted in a biphasic dose response (Fig. 3.3a and Fig. 3.3b), while all other agonists led to monophasic dose responses, yielding EC50 values of 281.1±21.3 nM ( n= 4), 104.8±4.9 nM (n =4), 6.1±0.9 nM (n =4), 5.5±0.7 nM (n = 4) for ICI 199,441, DAMGO, naltrindole and β-funaltrexamine, respectively (Fig. 3.9a). The maximal amplitudes were 300±23 pm (n =16), 235±13 pm (n =16), 82±9 pm (n =16), and 35±4 pm (n =16), for ICI 19944, DAMGO, β-funaltrexamine, and naltrindole, respectively. The biphasic dose-response of DPDPE led to two saturable amplitudes, 279±11 pm and 415±17 pm (n =16), respectively. The DMR antagonist assay showed that distinct ligands differentially blocked the succeeding DPDPE-induced DMR response in HEK-DOR cells (Fig. 3.9b and Fig. 3.9c). 80

3 9 30 3 9 30 ------3 9 30 3 9 30 3 9 30 ------3 9 30 3 9 30 3 9 30 ------Buffer Buffer Buffer PTX PTX PTX CTX CTX CTX FSK FSK FSK U0126 U0126 U0126 SB SB SB SP SP SP LY LY LY

SNC80 Endomorphin-1 α-neoendorphin Dynorphin B Syndyphalin SD-25 Endomorphin-2 DSLET DADLE DTLET ICI204448 Leu5-enkephalin BUBUC Deltorphin II Dynorphin A 1-8 DPDPE DALDA

Agonist GR89696 (-)-norcodeine DAMME Dynorphin A1-13 Met5-enkephalin SNC121 NNC63-0532 Morphiceptin DAMGO TAPP Naltrexone Tramadol Levalloprhan SKF10047 (-)U-50488 Salvinorin A Nalbuphine U-69593 Nociceptin U-54594A N-MPPP Naltrindole Naloxonazine (+)-U-50488 U-50488H BRL-52537 DIPPA U-62066 Nor-binaltorphimine Naloxone methiodide Naloxone HCl Naltriben N-benzylnaltrindole DMSO

Fig. 3.8 A false colored heat map based on the functional selectivity of opioid ligands at the endogenous receptors in SH-SY5Y cells. All of the ligands in the library were assayed at 10 µM. The negative control (DMSO) was also included. To effectively visualize the impact of probe molecules, the net change for each ligand after pretreatment was obtained via subtraction, except for both the positive control (i.e., DMR in cells pretreated with the buffer vehicle only) and the ligand DMR in PTX-pretreated cells for which the raw data were used.

The dose-dependent inhibition by DPDPE is best fitted with single phase sigmoidal non- linear regression, leading to an IC50 of 1.25±0.10 nM (n=4). Similar monophasic inhibitory dose-responses were obtained for naltrindole (IC50, 8.87±0.39 nM; n=4), ICI

199,441 (IC50, 753±67 nM; n=4), and naltriben (IC50, 5.30±0.36 nM; n=4). However, the other ligands led to a biphasic dose-dependent inhibition of the DPDPE-induced DMR, including the partial agonists DAMGO (IC50: 368±51 nM and 7.58±1.32 µM; n =4) and

β-funaltrexamine (IC50: 35.2±7.3 nM and 4.42±0.35 µM; n =4), and the antagonists

BNTX (IC50: 11.0±0.9 nM and 2.06±0.32 µM; n =4) and naloxone HCl (IC50: 135.1±14.9 nM and 6.41±0.75 µM; n =4).

a b c

600 DPDPE β-funaltrexamine 600 DPDPE Naltrindole 600 BNTX DAMGO DAMGO ICI 199441 Naltriben 500 ICI 199441 500 500 β-funaltrexamine Naloxone HCl 400 Naltrindole 400 400 300 300 300 200 200 200

Response(pm) 100 Response(pm) 100 Response(pm) 100 0 0 0 -100 -100 -100 -12 -11 -10 -9 -8 -7 -6 -5 -4 -12 -11 -10 -9 -8 -7 -6 -5 -4 -10 -9 -8 -7 -6 -5 -4 Compound, log M Compound, log M Compound, log M

Fig. 3.9 Dose responses of a panel of opioid ligands in HEK-DOR cells. (a) Dose dependent responses of opioid ligands obtained using DMR agonist assays. The maximal amplitudes were plotted as a function of agonist doses. (b) Dose- dependent desensitization by the DOR agonists of HEK-DOR cells to the repeated stimulation with DPDPE at 64 nM. (c) Dose-dependent inhibition by the DOR antagonists of HEK-DOR cells to the succeeding stimulation with DPDPE at 64 nM. For (b) and (c) the maximal amplitudes of the DPDPE DMR were plotted as a function of ligand doses, and data represents the mean ± s.d. for 2 independent measurements, each in duplicate (n=4).

We next characterized the KOR using six opioid ligands, including BRL-52537. The DMR agonist assay showed that all six ligands triggered dose-dependent DMR responses (Fig. 3.10a), similar to the BRL-52537 response (Fig. 3.3b). Like BRL-52537, ICI 199,441, DIPPA and DAMGO all yielded biphasic dose responses. This analysis revealed EC50 values of 3.2±0.4 nM and 45.8±3.1 nM (n =4) for ICI 199,441, 13.4±1.5 nM and 239.6±11.2 nM (n = 4) for DIPPA, and 93.8±7.4 nM and 4.5±1.1 µM (n =4) for DAMGO. The two saturable amplitudes were 125±7 pm and 205±9 pm (n =4) for ICI199441, 120±6 pm and 207±13 pm (n=4) for BRL-52537, 138±8 pm and 205±9 pm 82

(n = 4) for DIPPA, and 139±6 pm and 200±8 pm (n =4) for DAMGO. In contrast, the partial agonists naloxone HCl and β-funaltrexamine yielded monophasic dose responses with EC50 of values of 1.4±0.2 nM and 0.50±0.09 nM, respectively. The maximal responses were 69±5 pm and 151±8 pm for naloxone HCl and β-funaltrexamine, respectively. Further, the two-step DMR antagonist assay showed that distinct ligands differentially inhibited the HEK-KOR cells response to repeated stimulation with 64 nM BRL-52537 (Fig. 3.10b). ICI 199,441, DIPPA, DAMGO and BRL-52537 each inhibited the BRL-52537 response with single phase sigmoidal non-linear regression producing

IC50 values of 5.6±0.4 nM (n=4), 454.9±32.3 nM (n=4), 2.21±0.51 µM, and 4.1±0.23 nM, respectively. In contrast, the dose-dependent inhibition by the two antagonists were best fitted with a biphasic sigmoidal non-linear regressions, which exhibited biphasic

IC50’s of 643±47 nM and 3.61±0.45 µM (n=4) for β-funaltrexamine, and 67.2±5.6 nM and 2.05±0.0.54 µM (n=4) for naloxone HCl.

a 300 ICI 199441 BRL52537 250 DIPPA DAMGO 200 Naloxone HCl β-Funaltrexamine Fig. 3.10 Dose responses of a panel of 150 opioid ligands in HEK-KOR cells. (a) Dose dependent responses of opioid 100 ligands obtained using DMR agonist

Response(pm) 50 assays. The maximal amplitudes were plotted as a function of agonist doses. 0 Data represents the mean ± s.d. for 2 -50 independent measurements, each in -12 -11 -10 -9 -8 -7 -6 -5 -4 duplicate (n=4). (b) Dose-dependent inhibition of the DMR of 64 nM BRL- Compound, log M b 57532 by opioid ligands. The maximal amplitudes of the BRL-57532 DMR 200 were plotted as a function of ligand doses. Data represents the mean ± s.d. 150 for 2 independent measurements, each in duplicate (n=4). 100

50 BRL57532 ICI 199441 Response(pm) β-Funaltrexamine 0 Naloxone HCl DIPPA DAMGO -50 -12 -11 -10 -9 -8 -7 -6 -5 -4 Compound, log M 83

Lastly, we characterized the DMR response elicited by opioid receptors in SH- SY5Y cells utilizing 8 known agonists and antagonists. Results from the DMR agonist assay showed that all ligands yielded dose-dependent P-DMR responses characteristic of agonists, except for naloxone HCl and β-funaltrexamine, neither of which produced an observable DMR response in SH-SY5Y cells (Fig. 3.11a). Similar to DAMGO (Fig. 3.3e), the dose-dependent activation responses were best fitted using a single phase sigmoidal non-linear regression, revealing EC50 values of 26.5±2.1 nM (n=4), 1.4±0.2 nM (n=4), 2.4±0.2 nM (n=4), 1.2±0.1 nM (n=4), and 2.8±0.3 nM (n=4) for morphine, fentanyl, endomorphin-1, endomorphin-2 and CTOP, respectively (Fig. 3.11a). The maximal DMR responses was found to be 102±8 pm, 94±5 pm, 105±7 pm, 102±6 pm, 102±7 pm, and 31±4 pm (n=16 for all) for DAMGO, morphine, fentanyl, endomorphin-1, endomorphin-2 and CTOP, respectively. The DMR antagonist assay showed that all ligands blocked the DMR response produced by 64 nM DAMGO in a dose-dependent fashion. Single IC50 values of 1.0±0.1 nM, 115.8±14.7 nM, 4.2±0.3 nM, 10.0±0.9 nM, 5.8±0.4 nM, 475.5±39.7 nM, 231.4±21.5 nM, and 8.4±0.7 nM were obtained for DAMGO, morphine, fentanyl, endomorphin-1, endomorphin-2, CTOP, naloxone HCl, and β-funaltrexamine, respectively (Fig. 3.11b). Together, these results suggest that the family of opioid receptors exhibit complex pharmacology, possibly arising from the existence of distinct populations of receptors in disctinct cell backgrounds (Giraldo, 2008). As a result, varying agonists may activate these receptors with varying degrees of efficacy, and distinct antagonists may block the activation of these receptors with different potencies.

a Endomorphin-2 DAMGO CTOP 150 Fentanyl Morphine Naloxone Endomorphin-1 β-Funaltrexamine 125 100 Fig. 3.11 Dose responses of a panel of 75 opioid ligands in SH-SY5Y cells. (a) Dose dependent responses of opioid ligands 50 obtained using DMR agonist assays. The maximal amplitudes were plotted as a Response(pm) 25 function of agonist doses. (b) Dose- 0 dependent inhibition of the DMR of 64 -25 nM DAMGO by opioid ligands. The -11 -10 -9 -8 -7 -6 -5 -4 maximal amplitudes of the DAMGO Compound, log M DMR were plotted as a function of ligand doses. For (a) and (b) data represents the b mean ± s.d. for 2 independent Endomorphin-2 DAMGO CTOP measurements, each in duplicate (n=4). 150 Fentanyl Morphine Naxolone Endomorphin-1 β-Funaltrexamine 125 100 75 50

Response(pm) 25 0 -25 -11 -10 -9 -8 -7 -6 -5 -4 Compound, log M

Discussion Functional selectivity represents the underlying basis for drug selectivity, one of the most important pharmacological properties of drug molecules that determine their in vivo efficacy and therapeutic index. However, functional selectivity has not been fully integrated into the mainstream drug discovery and development processes. This is partly because of the simplistic nature of molecular assays used to characterize the pharmacological properties of drug molecules and partly because of unknown molecular mode(s) of actions that are critical to identify in vivo efficacy or in vivo side effects of drugs. This issue is exemplified by opioid ligands. Molecular assays have revealed a wide array of biased agonisms demonstrated by opioid ligands which appear to be cell-system and assay technology dependent (Neve, 2009; Raehal et al., 2011). Also,

multidimensional biased agonism makes it difficult to rank candidate compounds for in vivo testing and to relate a specific biased agonism of drug molecules to their in vivo profiles. Many opioid ligands often display relatively poor selectivity binding to different opioid receptor family members (Dhawan et al., 1996). This problem is exacerbated by the fact that the binding affinity profiles of opioid ligands do not directly translate into their selectivity in cellular and in vivo environments due to the expression of more than one opioid receptor in native cells, as well as the possibility that opioid receptors may present in different oligomerizational states (Jordan et al., 2001; Jordan & Devi, 1998; He et al., 2002; Gomes et al., 2002; Van Rijn et al., 2010). Thus, an effective means to differentiate drug candidate molecules based on both binding and functional selectivity in native cells would be beneficial to identify and prioritize lead compounds, and to relate in vitro results to in vivo profiles. Recently, we have developed a label-free integrative pharmacology on-target

(iPOT) approach and applied it to differentiate individual ligands in libraries for both β2- adrenergic receptor (Ferrie et al., 2011) and the MOR (Morse et al., 2011). The high resolution heat maps obtained allowed us to sort these ligands into distinct clusters based on their cellular binding profiles and pathway biased agonism. Here, we extended this approach to survey the entire classic opioid receptor family (mu-, kappa- and delta- receptors). Both recombinant and native cells expressing opioid receptors were used to generate DMR profiles of a library of opioid ligands using a battery of DMR assay formats. The DMR profiles were translated into numerical coordinates for all ligands. These coordinates were subject to similarity analysis to determine the similarity and distance between ligand pairs. The results obtained were visualized using a color-coded heat map with a distance-dendrogram. This methodology led to several notable conclusions. First, the off-target activity of a subset of ligands including BNTX, β- funaltrexamine, etonitazenyl isothiocyanate, ICI 199,441, dynorphin A 2-13 and nociceptin 1-13 was visualized in both HEK293 and SH-SY5Y cells, indicating that DMR assays are indeed capable of characterizing molecules with much wider pathway coverage than conventional pharmacological or molecular assays. The most notable finding is that almost all ligands in the library behaved as agonists in at least one opioid receptor expressing cell line with or without pretreatment

with probe molecules. Using the DMR agonist assay, we found that fifty-one out fifty- five opioid ligands displayed agonist activity in at least one opioid receptor-expressing cell line (Fig. 3.2). This is significant since as many as thirteen ligands in the library were classified as opioid receptor antagonists (Appendix B-D). The four ligands that were silent in all four opioid receptor expressing cell lines included naloxone methiodide, naltrindole, nor-binaltorphimine and naltriben. After excluding the six ligands that gave rise to unknown off-target activity, we found that all forty-nine remaining ligands exhibited agonist activity for the DOR in either native or forskolin pretreated HEK-DOR cells. In HEK-KOR cells, nor-binaltorphimine exhibited distinct behavior, as it did not trigger any DMR response under any conditions, leading us to conclude that nor- binaltorphimine was a true neutral antagonist for the KOR.

Pathway biased agonism was also visualized for many ligands. First, Gαi - independent signaling was evident in the DMR responses produced by a subset of ligands in PTx-treated cells. Generally, full agonists and strong partial agonists for each receptor led to a detectable DMR response in PTx-treated cells, indicative of activating Gαi - independent signaling. Second, CTx and forskolin pretreatment generally increased the DMR response induced by opioid agonists in both HEK-MOR (Morse et al., 2011) and HEK-DOR cells (Fig. 3.6), but clearly suppressed the DMR of a subset of opioid ligands in HEK-KOR and SH-SY5Y cells (Fig. 3.7 and Fig. 3.8, respectively). These patterns suggest that the KOR in HEK-KOR cells and the opioid receptors in SH-SY5Y may also signal via a pathway distinct from Gαi. Third, the DMR responses produced by various agonists in different opioid receptor expressing cell lines showed clear differences in their sensitivity to pathway modulation by the panel of kinase inhibitors including U0126, SB202190, SP10065 and LY294002. The suppression of the DMR of opioid agonists in HEK-KOR cells by SB202190 suggests that the p38 MAPK pathway may play an important yet general role in the signaling downstream the activation of the KOR. The iPOT analysis of opioid ligands further indicates the presence of distinct populations of receptors (i.e., conformational or oligomerizational states). This is visualized by the differences in ligand specificity from the HEK-MOR and HEK-DOR cells lines, when compared to the SH-SY5Y cells. We hypothesize these differences in DMR signaling are due to the different populations of receptors in SH-SY5Y cells, and

changes in cellular environment. These results further the idea that opioid receptors exhibit complex pharmacology. Yet, the high resolution heat maps and pharmacological characterizations of the opioid receptor family using DMR assays suggest that the iPOT is powerful new approach for elucidating of the complex and multifaceted efficacy of GPCR ligands. The iPOT approach offers a unique platform for drug development when functional selectivity is important.

Chapter 4 Closing Discussion

This dissertation was developed around the belief that the complete and integrative characterization of GPCR signaling will help change the future of directed drug design and targeted therapeutics. Specifically, understanding the entire scope of opioid receptor signaling will provide insights for the therapeutic design of powerful analgesics, ideally resulting in nociception without addiction and other negative side effects. It is well known that opioid agonists have powerful analgesic properties (Casy & Parfitt, 1986), yet there is growing evidence that the negative side effects of opioid drugs are leading to a significant increase in opioid abuse and misuse (Raehal et al., 2011; Keiffer & Gaveriaux-Ruff, 2002). This societal problem is causing researchers to rethink their current approach to opioid drug design. It has become apparent that we need to take a step back, and understand the entire cascade of opioid receptor signaling in order to be able to tease out the beneficial functionality from the negative side effects. The use of high-throughput biosensors to survey signaling networks and functional selectivity of the opioid system has the capability to allow researchers to complete full GPCR signaling deconvolution in live cells. This dissertation focuses on testing biosensor technology on the family of opioid receptors to determine if there is potential to use DMR assays to not only comprehend signaling, but also shape the future of opioid drug design. The biochemical and pharmacological characterization of opioid receptors has proven to be a difficult problem for a number of reasons. First, native cells often express more than one opioid receptor (Dhawan et al., 1996). Yet, opioid ligands often exhibit relatively poor selectivity for different opioid receptors (Raynor et al., 1994). Second, opioid receptors, like other GPCRs, are believed to form homo- or heterodimers, which often result in signaling that is distinct from monomeric receptors (Cvejic & Devi, 1997; Jordan & Devi, 1999; Gomes et al., 2002). The dual potency that was exhibited by many of the ligands in this study is indicative of the potential existence of opioid receptor homodimers (Giraldo, 2008). Third, the plasticity of receptor conformations (Kenakin & Miller, 2010) implies that distinct ligands could result in pathway-selective activity via stabilization of a specific receptor conformation or a specific set of conformations (Keith 89

et al., 1996; Bohn et al., 2000). Fourth, differences in intracellular environments may also affect ligand-mediated pharmacology, given that biological functions are often cell context dependent (Ferguson, 2001). Downstream signaling, agonist functional selectivity and dimerization play large roles in the functionality of opioid activation, but the specifics are still being heavily researched and debated (Neve, 2009; Raehal et al., 2011; Urban et al., 2007; Ferré et al., 2009). Researchers are trying to decipher not only how these factors affect the beneficial functions, but also influence the detrimental side effects of opioid agonists. We need to understand the signaling which leads to opioid dependence and the impact that various signaling pathways have on analgesia and addiction (Raehal et al, 2011). Therefore, this research was completed to try and begin the large task of uncovering the downstream effects of known opioid ligands in order to classify their modes of action, both positive and negative. We attempted to do this by focusing greatly on the functional selectivity of opioid receptor ligands. To determine the roles of functional selectivity, it is vital to deconvolute the downstream signaling networks. The discovery of ligand-directed functional selectivity has led to new avenues for achieving desired drug selectivity. Functional selectivity describes the differential ability of drug molecules to activate one of the multiple downstream pathways to which a receptor is coupled (Kenakin, 2005; Mailman, 2007; Galandrin et al., 2007; Urban et al., 2007). Opioid receptors exemplify many aspects of functional selectivity (Neve, 2009). It has been postulated that the functional selectivity of opioid drugs are related to their clinical profiles, particularly to the progression of analgesic tolerance after extended use (Raehal et al., 2011). However, the ability to integrate functional selectivity to the drug development process remains a challenging problem. The wide spectrum of signaling events mediated by a receptor (Law et al., 2000), coupled with the differences in signaling components in distinct cell types (Kenakin and Miller, 2010), makes it extremely difficult to fully discover and quantify the functional selectivity of drug molecules using conventional molecular assays. Also, conventional molecular assays screen compounds based on a predetermined molecular hypothesis, but such a hypothesis may or may not be relevant to the pathogenesis of a disease (Swinney and Anthony, 2011). A further complication is that the existence of

signaling readout- and cell background-dependent potency and efficacy are due to the operational bias of drug molecules on the receptor (Galandrin et al., 2007). The possibility that a drug may have multidimensional efficacy makes it difficult to optimize and prioritize drug candidate molecules. In many instances, the efficacy profiles obtained from candidate drugs may not be good predictors of their in vivo therapeutic qualities. It may be difficult to sort out which molecular modes of action lead to a desired therapeutic impact. Thus, assays that are phenotypic in nature yet allow mechanistic descriptions of drug actions would be advantageous. These challenges have caused researchers to question the capability of conventional assays to discern cellular signaling events. Researchers are constantly looking for novel technologies, like DMR assays, that may be up to the task. Whole cell DMR assays are comparable to conventional assay systems and capable of studying all types of GPCR signaling (Schröder et al., 2010). Also, the integrative readout of DMR assays presents a much larger view of the downstream signaling cascade (Ferrie et al., 2011). The DMR assay system was recently recognized for correctly identifying and clustering known β-adrenergic drugs, and it has also properly visualized opioid receptor activation (Ferrie et al., 2011; Codd et al., 2011). Furthermore, DMR technology has the potential to characterize functional selectivity of ligands (Codd et al., 2011). The work presented in this body of research demonstrates the benefits of using novel, integrative technology to understand GPCR signaling. Our work is the first compilation of evidence that DMR technology will be applicable to characterize all opioid receptor signaling on a large scale, with the goal of understanding how signaling truly and completely affects functionality. When the compound pharmacology of the library of ligands was evaluated using DMR assay, the output generally confirmed findings that had been previously obtained with classical assay methods. The ligands exhibited receptor selectivity on the three classic subtypes, antagonists blocked the activity of known ligands, and PTX pretreatment blocked ligand-induced DMR signaling. This study effectively corroborated decades of opioid receptor signaling, providing undeniable support for this novel, high- throughput technology replacing the classic assay methods. Yet, along with supporting the validity of DMR assay as a more efficient and effective technology for studying

receptor signaling, the biosensors were able to also tease out information which was previously unknown or overlooked. Schröder’s laboratory, one of the group’s working to fully characterize DMR technology, recently provided evidence that integrative visualization of signaling pathways by DMR, which is not feasible through recording of defined downstream signaling events in single component functional assays, enables unexpected signaling phenomenon to be identified (Schröder et al., 2010). This thesis fully agrees with that theory, and supports it by visualizing aspects of opioid receptor signaling which were not identifiable using conventional methodology.

Functional Selectivity The opioid receptors are regulated by numerous agonist-dependent activities (Zhang et al., 1998). Two of the most important agonist-dependent activities occur at the G-protein level and downstream at the level of MAP kinase pathways. G-protein activation is one of the first and most vital steps in the intracellular cascade following opioid receptor activation. Opioid receptors interact with specific heterotrimeric G proteins, which then dictate downstream signaling. The dogma of opioid receptor signaling states that opioid receptors are Gαi coupled. However, recent studies have suggested that some opioid ligands are pleiotropic. They have the ability to couple to multiple classes of G-proteins, as well as switch between G-inhibitory and G-stimulatory activity (Raehal et al., 2011). These ideas are still fairly novel in the field, and are met with some adversity. This study clearly demonstrates that numerous opioid ligands activate Gαi-insensitive pathways, in all of our opioid receptor-expressing cell lines (Fig. 2.9; Fig. 2.10; Fig. 3.6; Fig. 3.7; Fig. 3.8). Also, a number in ligands led to a DMR response that is sensitive to CTx pretreatment, suggesting that there is a Gαs signaling component activated by ligands in the opioid library. One of the main benefits of this study, and this technology, is the ability to run numerous assays in various formats in a reasonable timeframe. This allowed us to run 13 different assays on 5 cells lines, utilizing many inhibitors and pretreatments. However, there are many more G-protein inhibitors and combinations of treatments that can be used in the future to more completely understand G-protein signaling activated by various opioid ligands. Specifically, treatment with PTx led to most DMR signaling being

abolished in most ligands, but the remaining DMR signaling seen (mostly from full DOR agonists in HEK-DOR cells) remains elusive. Follow up studies and/or combinatorial pretreatment (i.e. dual-inhibition with PTx and CTx together) will allow us to continue our quest for the crucial elements of opioid signaling. While not much is known about the functional effects of the pleiotropic nature of opioid receptors, it has been suggested that this phenomenon is related to opioid tolerance and receptor response to chronic exposure to opioids (Gintzler & Chakrabarti, 2000). Therefore, this will be a crucial area of research for future drug development of targeted opioid therapeutics. By understanding the activation patterns of opioid ligands that are known to be addictive, one can hypothesize that we will be able to manipulate that information to our advantage. We hope that being able to uncover how the biased signaling leads to activation of the negative functionality of opioid receptor activation, we will be able to specifically activate only pathways that induce beneficial functionality. Opioid receptor signaling also leads to the activation of downstream kinase pathways. We studied the effects of known kinase inhibitors on a library of opioid ligands. We hypothesized that use of inhibitors of the kinase pathways would cause significant alterations in ligand-induced DMR signaling. It has been previously suggested that opioid use and dependence can lead to alterations in MAPK signaling (Williams et al., 2001). Also, the kinase pathways are essential to the neurobiology of drug addiction (Chao & Nestler, 2004). Thus, it can be theorized that any novel information about the selective activation of downstream signaling pathways by specific ligands can be utilized to understand the connection between signaling and functionality of opioid receptor activation. It is vital to remember that integrated signaling cascades often overwhelm individual signaling patterns and regulatory events (Cobb, 1999). The kinase pathways have extensive interactions with each other, and they tend to function as nets, not nodes. They are regulated directly and indirectly by one another. This complicates the use of single inhibitors. Therefore, one must consider that a ligand-induced DMR may not be affected unless there is a profound ligand-specific effect. Small preferences for one pathway or another will likely be covered up by the network of kinase signaling, so any information gathered in a deconvolution study such as this will require follow-up studies.

Also, it is important to note that combinatorial knockdown of pathways may allow us to more thoroughly outsmart a system which is full of checks and balances in future studies. By knocking down individual MAPK pathways, we were able to visualize novel information about agonist-biased signaling (Fig. 2.11; Fig. 2.12; Fig. 3.6; Fig. 3.7; Fig. 3.8). Arguably the most interesting result seen was the impact of the p38 inhibitor SB202190. Pretreatment with SB202190 caused significant inhibitions of the ligand- induced DMR response for many ligands, especially KOR ligands acting on HEK-KOR cells. p38 is the least well-characterized kinase pathway of those utilized in this study, and its prevalent impact on DMR response suggests its potential importance. Thus, it would be a worthwhile focus for future studies examining the role of kinase pathways in opioid functionality. While this thesis focused on the impact of MAPK signaling pathways, there are many other downstream cascades that are activated through opioid receptor activation (Polakiewicz et al., 1998; Mestek et al., 1995; Law et al, 2000). Therefore, future studies will be able to expand the scope of the deconvolution assays to understand more of the extensive, integrative web of opioid receptor downstream signaling.

SH-SY5Y cells GPCR activity is determined not only by the initiation of signaling cascades but also by the regulatory mechanisms which control the extent and duration of their signals (Bohn et al., 2004). The behavior is greatly determined by cellular environment. Thus, we chose to study an endogenous opioid receptor system, and compare it to the engineered HEK-based cell lines. Looking at the human neuroblastoma cell line SH-SY5Y, it was clear that the environment of the opioid receptors had a large impact on the ligand- induced DMR responses. First, the differences in ligand specificity between HEK-MOR and SH-SY5Y cells, or between HEK-DOR and SH-SY5Y cells (Fig. 3.4 and Fig. 3.5) cannot be explained solely on the known affinity of these ligands for the MOR or for the DOR, respectively. Such differences seem to be reflective of different populations of endogenous opioid receptors in SH-SY5Y cells. Second, the dose-dependent efficacy and potency of the ligands, together with the dose-dependent desensitization/inhibition of the activation of opioid receptors, clearly shows that the same ligands produce very different

types of dose responses (Fig 2.1; Fig. 3.11). These responses may be monophasic or biphasic in a ligand- and cell-dependent manner. These results indicate that opioid ligands exhibit a complex DMR pharmacology, which is only exacerbated by changes in environmental context. Nonetheless, the present study represents the first analysis using label-free cellular assays to assess the binding and functional selectivity of opioid ligands across the entire classic opioid receptor family. Most studies only have the ability to study opioid ligands individually, or in small numbers. One of the many benefits of this DMR assay was the ability to study an entire library of known opioid ligands under identical conditions on multiple receptor types in a high-throughput setting. This allowed for a more complete and complex comparison of their selectivity and functionality. Yet, one significant constraint of this study was the limited access to restricted substances, leaving a gap in the opioid library tested. Medicinal opioid drugs and illicit substances were not available to use in these deconvolution studies. Thus, follow-up work needs to be done with clinically available and restricted opioid substances, in order to see where they fall in the spectrum of characteristics.

Benefits and Pitfalls of DMR technology Through the utilization and characterization of this methodology, it has become clear that there are a number of positive and negative aspects of DMR technology. Many of the benefits of the Epic system have already been highlighted throughout this dissertation. The high-throughput capabilities of Epic allow for an unprecedented number of ligands to be fully characterized in a timely manner. Also, the integrated nature of the responses, combined with non-invasive protocols, allows for huge amounts of highly informative data to be collected from the whole cell level (Fang, 2010a). Furthermore, drug-induced DMR is sensitive to cellular context, which can be controlled using genetic manipulation or intervention with probe molecules or pathway inhibitors, allowing for expanded non-invasive assay formats. Finally, because of the non-invasive nature of the assay combined with the absence of compounds that may affect cell viability (such as fluorescent dyes or other hazardous labels) DMR responses can be recorded for extended time periods, which could allow for more thorough analysis of a

systems biological response (Schröder et al., 2011). This adds up to a very flexible, efficient system which has the possibility to revolutionize the early steps of drug discovery. There are also a number of weaknesses associated with DMR technology as it stands. First and foremost, the integrated responses and DMR outputs often lack clarity. It is sometimes difficult to distinguish the true cause of the signaling curves. Unexpected positive DMR curves can be due to many factors because of the integrative nature of the DMR output. Some possible sources of the DMR responses would be off-target activation of an unknown GPCR or allosteric activation of receptors (Kenakin, 2007a). Also, unexpected inhibition of DMR responses can have a number of causes, inverse agonism or heterologous desensitization being two possibilities (Kenakin, 2004; Kovoor et al., 1997). Second, at this point, the Epic assay system requires a confluent monolayer of cells for proper functionality of the DMR readout (Schröder et al., 2011). This is a significant limitation as it is difficult to use many cell types, especially neurons and primary cells, in a confluent format. Researchers are still determining how to encourage primary cells to proliferate in culture, and at this point they do not reach confluency (Ray et al., 1993). Therefore, while the next clear step in this research would be to complete opioid receptor characterization on primary neuronal cell lines, this is not feasible at this point in DMR development. These limitation need to be focused on, as assay development will help to forward DMR capabilities in future studies. Elucidating the cause for unknown DMR responses requires follow-up studies. It is important to understand the causes for these ambiguous DMR responses as best we can as they could have a significant impact on future drug development. It has been suggested that allosteric modulators may have advantages over other ligands as potential drugs, as they may exhibit greater selectivity and saturability (Christopoulos, 2002). Allosteric effects are currently best verified in secondary studies by examining the radioligand binding of the compounds of interest. It is important to note that the role of allosteric activity in drug discovery is an active area of research and more direct ways of identifying allosteric activation are being developed (Christopoulous, 2002; Hardy & Wells, 2004). Along with radioligand binding, classic biochemical assays such as cAMP

activation and intracellular calcium flux will help to identify if ligands are truly activating a specific receptor, or if off-target activation is confusing the responses. Also, the cell lines used in this study can be treated with agonists, collected, lysed and used in western blot analysis to look for downstream markers of GPCR activation, like cAMP or MAPK. The studies completed in this dissertation used agonist doses that were much higher than known EC50 concentrations (see Appendix B, C & D). This protocol can be viewed as another limitation of the DMR system. High concentrations were use to simplify the protocols used in high throughput screening, and it was decided that over- activation of the opioid system would be more beneficial than missing potential targets. However, it is known that testing all of the ligands at 10µM may have led to off-target activation that may not occur at lower concentrations, and thus may skew our interpretations. The cost of reagents and instrumentation, as well as time, has thus far limited more thorough protocol including dose responses. But, future studies will greatly benefit from a full panel of dose response curves being completed for each compound tested. Running the dose responses would allow for a proper saturating dose to be determined for each compound, so that all ligands may be tested at their individual EC100 values during the signaling deconvolution assays. This would require a significant increase in the amount of assays run, as each compound would require concentration curves to be carried out on each type of receptor. Yet, the information would be invaluable in understanding the signaling effects of each ligand on their specific receptors, and should be done in the future. However, it is important to note that running the experiments at such high concentrations (as was done in this dissertation) was not unprecedented (Ferrie et al., 2011). The results are greatly informative, allowing us a first view of the activity of all the compounds in the opioid compound library on each of the classic opioid receptors. Continuing the study through dose response characterization and activation at concentrations closer to the EC100 will only expand our knowledge of the opioid receptor signaling system. Running dose responses and repeating the entire repertoire of assays at lower concentration will allow for a more directed and on-target assessment for each of the ligands. But other follow-up studies will also allow us to verify any interesting, off-target activation we may have seen using 10 µM concentrations of our ligands. A number of

classic GPCR activation assays have been shown to correspond well with the results found in Epic (Schröder et al., 2010). Thus, we could validate any inexplicable DMR activation by looking at cAMP activation (Ferrie et al., 2011), intracellular calcium levels (Conner & Henderson, 1996), or GTPγS binding (Zhu et al., 1997). Thus, there are a number of ways to verify any false-positive off-target activation we may have visualized due to high concentrations of ligands used in these studies.

Future directions For over a decade, it has been suggested that we need to exploit the complexities of the opioid receptor system in order to manipulate therapeutic design and create drugs for pain without detrimental side effects (Pan et al., 1997; Vanderah et al., 2010). The future of analgesics does not lie in MOR agonists alone, but in combinatorial therapy that can target the best functions of each of the opioid receptor systems. This highly homologous system holds the key to creating directed therapeutics when it is studied and understood in an integrative manner. This thesis demonstrates the power of utilizing novel technology to truly begin to study this problem in a rationale manner. This is merely the first step in being able to fully manipulate the system. Now that we have characterized a library of known opioid ligands, this set of assays must be carried out on other pharmacologically active or unknown compounds. It is necessary to reiterate the fact that the Epic® system is still merely the first step in changing drug discovery. As a high-throughput screening system, its current format makes DMR technology a breakthrough way to identify, characterize, and categorize compounds (Ferrie et al., 2011). This will, in turn, guide us to the best potential drug compounds for directed follow-up studies to screen and test these drugs in vitro and in vivo. Furthermore, it is important to emphasize that DMR technology was very effective in linking in vitro characterization of β-adrenergic drugs (through studying their potency, mechanisms of activation and inactivation, and pathway biased activity) to their known in vivo effects (Ferrie, 2011). This strong correlation seen in β-adrenergic drugs will hopefully be seen also in opioid drugs, when the study is expanded to include medicinally utilized opioid ligands. An effective link between in vitro profiles of opioid

ligands and their in vivo indications would be the next major step in utilizing DMR technology in drug development. DMR technology will allow for unprecedented ease in the selection of drug targets due to receptor and ligand characterization, but the identified compounds will still need to be examined by numerous cellular studies and animal models. Therefore, while the novel, high-throughput DMR studies will have a substantial impact on the future systems biology and pharmacology, it will still encompass only the first step in a long line of studies which will be carried out on each compound of interest (Schroder et al, 2011). This integrative technology will help to forward the future of drug discovery by revolutionizing the first steps of identification and characterization of drug targets. This hypothesis of integration is supported by multiple studies published recently that highlight the need for novel characterization assays and high throughput methodology for drug discovery (Ferrie et al., 2011; Xia et al., 2011; Codd et al., 2011). The literature also shows that opioid receptor desensitization and trafficking are key steps in the development of opioid tolerance and dependence (Waldhoer et al., 2004; von Zastrow et al., 2003). These processes involved with receptor turnover are dependent on pathways that are influenced by functional selectivity of ligands (Urban et al., 2007, Kenakin, 2007b; Violin and Lefkowitz, 2007). Based on this, it can be inferred that functional selectivity plays a key role in the negative side effects of opioid use and abuse. It again points to the importance of understanding all of the downstream cascades of opioid activation to enhance our knowledge of opioid receptor pharmacology for the development of analgesic drugs. This thesis demonstrates the potential for the DMR platform to support the development of receptor-specific assays for future biochemical and pharmacological studies. It is possible to see the capacity of this technology through replicating the characterization and deconvolution protocols discussed here. Using known GPCRs and specific ligands, catalogs of receptor activities can be made. This will allow for a much more complete understanding of currently available drugs and active compounds. Along with organization and classification of the activity of known compounds, DMR technology can also be used to determine the targets and effects of novel compounds. The DMR responses of novel compounds can be easily screened to

determine their molecular targets, as well as levels of potency and efficacy. In addition, these compounds can easily be run through the same pre-determined set of assays as known ligands, like the set of assays designed in this thesis, and the functional activities of novel and known ligands can be compared and analyzed. The potential ease of identifying and characterizing numerous novel or unknown ligands suggests that DMR technology will revolutionize the future of drug discovery. In addition, results from the concentration dependent assays, which were done on specific ligands for each receptor, suggest that the integrative nature and screening capabilities of this system will allow us to modulate and modify the activity of current compounds. The existence of biphasic EC50 curves for certain ligands (Fig 2.1; Fig 3.3; Fig 3.9) which are visualized when the ligands are tested at high dosages, allows us to hypothesize that further manipulation of doses and potencies of drugs will lead to changes in the resulting activity of receptors. The high-throughput ability and flexibility of this assay system will allow for extensive studies to be completed, fully characterizing signaling responses of compounds at varying potencies and dosages on a very large scale. The benefits of manipulating functional selectivity for drug discovery have been highlighted in the literature, not only in the opioid system but also through other receptor pathways. While this line of research is still in infancy, some promising studies have suggested advantages of utilizing functional selectivity for drug development. It has been seen that 5-HT2A antagonists cause differential rates or levels of internalization of receptors, and these differences can alter the therapeutic value of an antagonist drug for the 5-HT2A receptor. Thus, understanding how these antagonists lead to alterations in internalization can help manipulate the system in favor of more beneficial drugs. (Urban et al., 2007; O’Conner & Roth, 2005). Additionally, agonist-selective activation of the D2 receptor has recently been highlighted through the activity of the drug aripirazole (Abilify) (Urban et al., 2006; Urban et al., 2007). Aripirazole seems to work as both a partial agonist and full agonist but lacks typical D2 receptor antagonist properties, depending on the cellular environment. It is hypothesized that this unique activation pattern is due to functional selectivity, and the end result is an atypical antipsychotic drug with an excellent side-effect profile. These current examples and the results of this thesis

100

only support and strengthen the argument for directed therapeutics to revolve around the functionally selective properties of receptor ligands. There is also potential for DMR technology to help decipher the effects of opioid receptor dimerization. It is well known that opioid receptors form homo- and heterodimers, and that these protein-protein interactions modulate receptor functionality, through little is currently known about the specific effects (Devi, 2001; Waldhoer et al., 2004; Decaillot et al., 2008). However, recent research highlighted that the MOR-DOR heterodimer negatively regulates spinal analgesia. This suggests that the disassociation of the MOR from the DOR can be a potential strategy to improve opioid analgesic properties (He et al., 2011). The integrative nature of DMR technology may be useful in understanding signaling changes that occur in response to receptor dimerization. Receptors could be transfected into singularly and in combination into specific parental cells, to control for cellular background. Full characterization of these receptors using libraries of agonists and multiple assay formats, similar to those used in this dissertation, could be completed to note signaling differences due to dimerization. Also, it would be feasible to try and interrupt these protein-protein interactions if the binding sites were known and blocked. We could see if the signaling would revert back to DMR responses seen by individual receptors. This could help us gain information on the effects of the disassociation of dimers for future therapeutics. To fully exploit the potentials of DMR and manipulate opioid dimerization for future drug development, we do need to have a more complete understanding of (1) which opioid receptors are endogenous expressed in cell lines and (2) how they interact. Studying transfected cell lines is the first step in understanding receptor signaling, but the next step is endogenous systems. Understanding cellular context and receptor interactions is vital for future studies, and a more complete characterization of the cell lines and receptors themselves is necessary to forward DMR signaling deconvolution on endogenous receptors. While this study highlights the potential for DMR assays to be utilized for future drug development, it is only the first step of many to revolutionize directed therapeutics. Dr. Mailman, a forerunner of the functional selectivity movement, has repeatedly argued that functionally selective drugs have the promise to improve the therapeutic profile of

101

compounds acting at various receptors, and believes that the best way to reach this goal is to approach it from all angles (Mailman, 2007). Knowledge of the mechanisms involved will result in unparalleled usefulness for detecting functionally selective compounds. Yet, it is imperative to reiterate that the DMR technology will allow for revolution of the first steps of drug discovery. Characterization and identification of target compounds will not replace the need for many follow-up assays, in both in vitro and in vivo systems. We have demonstrated the ability of DMR assays to be used to screen a library of known ligands with great success identifying the signaling properties of these ligands, while teasing out novel information about specific compounds. Yet this approach needs to be greatly expanded. We touched on some of the downstream pathway inhibitors, while there are many more known GPCR pathway modulators that can be tested and used to characterize opioid receptor signaling. The high sensitivity, coupled with broad pathway coverage, makes the non-invasive, manipulation-free, and label-free biosensor cell-based assay an attractive means to elucidate the signaling of opioid receptors, and by extension, all GPCRs. Therefore, the DMR on-target pharmacological approach represents a novel and practical means to study receptor-based pharmacology in multiple dimensions, and provides a global view of ligand pharmacology which should accelerate the drug discovery process and help prioritize opioid receptor active compounds for future drug development.

102

References

Ackley MA, Hurley RW, Virnich DE, Hammond DL (2001) A cellular mechanism for the antinociceptive effect of a kappa opioid receptor agonist. Pain 91(3):377-388. Aguila B, Coulbault I, Boulouard M, Liveilli F, Davis A, et al. (2007) In vitro and in vivo pharmacological profile of UFP-512, a novel selective δ-opioid receptor agonist; correlation between desensitization and tolerance. BJoP 152:1312-1324. Akil H, Watson SJ, Young E, Lewis ME, Khachaturian H, et al. (1984) Endogenous opioids: biology and function. Ann Rev of Neuro 7:223-255. Alfaras-Melainis K, Gomes A, Rozenfeld R, Zacharious V, Devi LA (2010) Modulation of opioid receptor function by protein-protein interactions. Front Biosci 14: 3594- 3607. Amiche M, Sagan A, Mor A, Delfour A, Nicolas P (1989) Dermenkephalin (Tyr-D-Met-

Phe-His-Leu-Met-Asp-NH2): A potent and fully specific agonist for the delta opioid receptor. Mol Pharmacol 35: 774-779. Bailey CP, Connor M (2005) Opioids: cellular mechanisms of tolerance and physical dependence. Curr Op in Pharm 5(1): 60-68. Barbieri JT, Cortina G (1988) ADP-ribosyltransferase mutations in the catalytic S-1 subunit of pertussis toxin. Infect Immun 56: 1934-1941. Belcheva MM, Clark AL, Haas PD, Serna JS, Hahn JW, et al. (2005) µ and κ opioid receptors activate ERK/MAPK via different protein kinase C isoforms and secondary messengers in astrocytes. JBC 280(30): 27662-27669. Belcheva MM, Coscia CJ (2002) Diversity of G protein-coupled receptor signaling pathways of ERK/MAP kinase. Neurosignals 11(1):34-44. Benovic JL, Staniszewski C, Major Jr F, Caron MG, Leftkowitz W (1988) beta- Adrenergic receptor kinase. Activity of partial agonists for stimulation of adenylate cyclase correlates with ability to promote receptor phosphorylation. JBC 263:3893-3897. Bohn LM, Gainetdinov RR, Caron MG (2004) G protein-coupled receptor kinase/β- arrestin systems and drugs of abuse: psychostimulatng and opiate studies in knockout mice. Neuromolecular Medicine 5(1): 41-50. Bohn LM, Gainetdinov RR, Lin FT, Leftowitz RJ, Caron MG (2000) µ-opioid receptor desensitization by β-arrestin-2 determines morphine tolerance by not dependence. Nature 408: 720-723. Brady AE, Limbird LE (2001) G protein-coupled receptor interacting proteins: Emerging roles in localization and signal transduction. Cellular Signaling 14:297-309. Bruchas MR, Chavkin C (2010) Kinase cascades and ligand-directed signaling at the kappa opioid receptor. Psychopharmacology 210(2):137-147.

103

Butler SF, Black RA, Serrano JM, Wood ME, Budman SH (2010) Characteristics of prescription opioid abusers in treatment: prescription opioid use history, age, use patterns and functional severity. J Opioid Manag 6(4):239-241, 246-252. Cassel D, Selinger Z (1977) Mechanism of adenylate cyclase activation by cholera toxin: inhibition of GTP hydrolysis at the regulatory site. PNAS 74: 3307-3311. Casy AF & Parfitt RT (1986) Opioid Analgesics: chemistry and receptors. Plenum Press: New York.

Chakrabarti S, Chang A, Gintzler AR (2010) Subcellular localization of µ-Opioid Gs signaling. JPET 333(1):193-200. Chakrabarti S, Regec A, Gintzler AR (2005) Biochemical demonstration of mu-opioid receptor association with Gsa: enhancement following morphine exposure. Molecular Brain Res 135:217-225. Chalecka-Franaszek E, Weems HB, Crowder AT, Cox BE, Cote TE (2000) Immunoprecipitation of high-affinity, guanine nucleotide-sensitive, solubilized mu-opioid receptors from rat brain: coimmuno-precipitation of the G proteins Gαo, Gαi1, Gαi3. J Neurochem 74: 1068-1078. Chan JSC, Yung LY, Lee JWM, Wu YL, Pei G, et al. (2002) Pertussis Toxin-insensitive

signaling of the ORL1 receptor: coupling to Gz and G16 proteins. J of Neurochem 71(5): 2203-2210. Chang AC, Takemori AE, Portoghese PS (1994) 2-(3,4-Dichlorophenyl)-N-methyl-N-[(1S)-1-(3- isothiocyanatophenyl)-2-(1-pyrrolidinyl)ethyl]acetamide: an opioid receptor affinity label that produces selective and long-lasting kappa antagonism in mice. J Med Chem 37: 1547–1549. Chao J, Nestler EJ (2003) Molecular neurobiology of drug addiction. Annu Rev Med 55:113- 132. Chavkin C, James IF, Goldstein A (1982) Dynorphin is a specific endogenous ligand of the kappa opioid receptor. Science 215(4531):413-415. Chen Y, Mestek A, Lui J, Hurley JA, Yu L (1993) Molecular cloning and functional expression of a µ-opioid receptor from rat brain. Molecular Pharm 44:8-12. Chen JC, Smith ER, Cahill M, Cohen R, Fishman JB (1993) The opioid receptor binding of dezocine, morphine, fentanyl, butorphanol, and nalbuphine. Life Sci 52: 389- 396. Chen Y, Sommer C (2009) The role of mitogen-activated protein kinase (MAPK) in morphine tolerance and dependence. Mol Neurobio 40:101-107. Christopoulos A (2002) Allosteric binding sites on cell-surface receptors: novel targets for drug discovery. Nature Rev Drug Discovery 1:198-210.

Clayton CC, Xu M, Chavkin C (2009) Tyrosine Phosphorylation of Kir3 following κ- opioid receptor activation of p38 MAPK causes heterologous desensitization. JBC 284(46):31872-31881.

104

Cobb M (1999) MAP kinase pathways. Progress in Biophysics and Molecular Biology. 71: 479-500. Codd EE, Mabus JR, Murray BS, Zhang SP, Flores CM (2011) Dynamic mass redistribution as a means to measure and differentiate signaling via opioid and cannabinoid receptors. ASSAY and drug development technologies. 9(4): 362- 372. Connor M, Henderson G (1996) δ- and µ-opioid receptor mobilization of intracellular calcium in SH-SY5Y human neuroblastoma cells. BJP 117:333-340 Contet C, Kieffer BL, Befort K (2004) Mu opioid receptor: a gateway to drug addiction. Current opinion on neurobiology 14: 370-378. Cooper MA (2006) Optical biosensors: where next and how soon? Drug discovery today 11(23/24):1061-1067. Corbett AD, Henderson G, McKnight AT, Paterson SJ (2006) 75 years of opioid research: the exciting but vain quest for the Holy Grail. BJP 147(S1):S153-S162. Cvejic S, Devi LA (1997) Dimerization of the δ opioid receptor: implications for a role in receptor internalization. JBC 272 (43):26959-26964. Decaillot FM, Rozenfeld R, Gupta A, Devi LA (2008) Cell surface targeting of µ-δ opioid receptor heterodimers by RTP4. PNAS 105(41): 16045-16050. Devi LA (2001) Heterodimerization of G-protein-coupled receptors: pharmacology, signaling and trafficking. TiPS 22:532-537. Dhawan BN, Cesselin F, Raghubir R, Reisine T, Bradley PB, et al. (1996) International Union of Pharmacology. XII. Classification of opioid receptors. Pharmacol Rev 48:567-592. Dondio G, Ronzoni S, Eggleston DS, Artico M, Petrillo P, et al. (1997) Discovery of a novel class of substituted Pyrrolooctahydroisoquinolines as potent and selective δ opioid agonists, based on an extension of the message−address concept. J Med Chem 40(20): 3192-3198. Eisen MB, Spellman PT, Brown PO, Botstein D (1998) Cluster analysis and display of genome-wide expression patterns. PNAS 95: 14863-14868. Elgen RM (2005) Emerging concepts in GPCR function – the influence of cell phenotypes on GPCR pharmacology. Proc West Pharmacol Soc 48:31-34. Eriksen J, Sjogren P (1997) Opioids in pain management. Acta Anaesthesiologica Scandinavica 41:1-3 Evans CJ, Keith DE, Morrison H, Magendzo K, Edwards RH (1992) Cloning of a delta opioid receptor by functional expression. Science 258(5090): 1952-1955. Fang Y (2006) Label-free cell-based asays with optical biosensors in drug discovery. Assay and drug development technologies 4(5): 583-595. Fang Y (2010a) Label-free receptor assays. Drug discovery today technologies. 7(1): e5- e11. Fang Y (2010b) Probing cancer signaling with resonant waveguide grating biosensors. Exp Opin Drug Discov 5: 1237-1248.

105

Fang Y, Ferrie AM (2008) Label-free optical biosensor for ligand-directed functional

selectivity of acting on β2 adrenoceptor in living cells. FEBS Lett 582: 558–564. Fang Y, Ferrie AM, Fontaine NH, Mauro J, Balakrishnan J (2006) Resonant waveguide grating for living cell sensing. Biophysical Journal 91: 1925-1940. Fang Y, Ferrie AM, Fontaine NH, Yuen PK (2005a) Characteristics of dynamic mass redistribution of epidermal growth factor receptor signaling in living cells measure with label-free optical biosensors. Anal Chem 77:5720-5725.

Fang Y, Li G, Peng J (2005b) Optical biosensor provides insights for bradykinin B2 receptor signaling in A431 cells. FEBS Lett 579: 6365-6374. Favata MF, Horiuchi KY, Manos EJ, Daulerio AJ, Stradlet DA, et al. (1998) Identification of a novel inhibitor of mitogen-activated protein kinase. JBC 273: 18623-18632. Ferguson SSG (2001) Evolving conceptrs in G protein-coupled receptor endocytosis: the role in receptor desensitization and signaling. Pharm Rev 53:1-24. Ferré S, Baler R, Bouvier M, Caron MG, Devi LA, et al. (2009) Building a new conceptual framework for receptor heteromers. Nature Chemical Biology 5: 131-134. Ferrie AM, Sun H, Fang Y (2011) Label-free integrative pharmacology on-target of drugs at the B2-adrenergic receptor. Scientific Reports. 1(33) doi:10.1038/srep00033 Fichna J, Do-Rego JC, Janecki T, Staniszewska R, Poels J, et al. (2008) Novel highly potent µ-opioid receptor antagonist based on endomorphin-2 structure. Bioorg Med Chem Lett 18: 1350-1353. Fields HL (2004) State-dependent opioid control of pain. Nature Rev Neuro. 5:565-575. Fields HL (2007) Understanding how opioids contribute to reward and analgesia. Regional Anesthesia and Pain Med 32(3):242-246. Filizola M, Villar HO, Loew GH (2001) Molecular determinants of non-specific recognition of δ, µ, and κ opioid receptors. Bioorganic & Medicinal Chem 9:69- 76. Filizola M, Weinstein H (2002) Structural model for dimerization of G-protein coupled receptors: The opioid receptor homodimers. Biopolymers 66:317-325. Filliol D, Ghozland S, Chluba J, Martin M, Matthes HWD, et al. (2000) Mice deficient for δ- and µ- opioid receptors exhibit opposing alteration of emotional responses. Nature Genetics 25:195-200. Fukuda K, Kato A, Morikawa H, Shoda T, Mori K (1996) Functional coupling of the µ, δ, and κ-opioid receptors to mitogen-activated protein kinase and arachidonate release in chinese hamster ovary cells. J of Neurochem 67:1309-1316. Gainetdinov RR, Premont RT, Bohn LM, Lefkowitz RJ, Caron MG (2004) Desensitization of G protein-coupled receptors and neuronal functions. Annu Rev Neurosci 27:107-144. Galandrin S, Oligny-Longpre G, Bouvier M (2007) The evasive nature of drug efficacy: implications for drug discovery. TiPS 8: 423-430.

106

Gaveriaux-Ruff C, Filliol D, Simonim F, Matthes HWD, Kieffer BL (2001) Immunosupression by δ-opioid antagonist naltrindole: δ- and triple µ/δ/κ-opioid receptor. JPET 298:1193-1198. Gebhart GF, Su X, Joshi S, Ozaki N, Sengupta JN (2000) Peripheral opioid modulation of visceral pain. Annals of the NY Acad of Sci 909:41-50. Gehlenborg N, O'Donoghue SI, Baliga NS, Goesmann A, Hibbs MA, et al. (2010) Visualization of omics data for systems biology. Nature Methods 7: S56–S68. George SR, Fan T, Xie Z, Tse R, Tam V, et al. (2000) Oligomerization of the µ- and δ- opioid receptors: Generalization of novel functional properties. JBC 275: 26128- 26135. Gharagozlou P, Demirci H, Clark JD, Lameh J (2003) Activity of opioid ligands in cells expressing cloned mu opioid receptors. BMC pharmacology 3(1):1-8. Ghozland S, Matthes HWD, Simonin F, Filliol D, Kieffer BL, et al. (2002) Motivational effects of cannabinoids are mediated by µ-opioid and κ-opioid receptors. J of Neuroscience. 22(3):1146-1154. Gill DMM, Meren R (1978) ADP-ribosylation of membrane proteins catalyzed by cholera toxin: basis of the activation of adenylate cyclase. PNAS 75: 3050-3054. Gintzler AR, Chakrabarti S (2000) Opioid tolerance and the emergence of new opioid receptor- coupled signaling. Molec Neurobio 21:21-33. Giraldo J (2008) On the fitting of binding data when receptor dimerization is suspected. Br J Pharmacol 155: 17-23. Gomes I, Filipovska J, Jordan BA, Devi LA (2002) Oligomerization of opioid receptors. Methods 27: 358-365. Gutstein HB, Rubie EA, Mansour A, Akil A, Woodgett (1997) Opioid effects on mitogen-activated protein kinase signaling cascades. Anesthesiology 87(5):1118- 1126. Hampson RE, Jian M, Deadwyler SA (2000) Receptors reduce potassium K current via activation on Gs proteins in cultured hippcampal neurons. J of Neurophysio 84 (5) 2356-2364. Hancock CN, Macias AT, Mackerell Jr AD, Shapiro P (2006) Mitogen activated protein (MAP) kinases: development of ATP and non-ATP dependent inhibitors. Med Chem 2: 213-222. Hardy JA, Wells JA (2004) Searching for new allosteric sites in enzymes. Current Opinion in structural biology 14:1-10. Hausdorff WP, Caron MG, Lefkowitz RJ (1990) Turning of the signal: desensitization of beta-adrenergic receptor function. FASEBJ 4:2881-2889. He L, Fong J, von Zastrow M, Whistler JL (2002) Regulation of opioid receptor trafficking and morphine tolerance by receptor oligomerization. Cell 108:271– 282.

107

He SQ, Zhang ZN, Guan JS, Liu HR, Zhao B, et al. (2011) Facilitation of µ-opioid receptor activation by preventing δ-opioid receptor-mediated codegradation. Neuron 69:120-131. Hendry IA, Kelleher KL, Bartett SE, Leck KJ, Reynolds AJ, et al. (2000) Hypertolerace

to morphine in Gza-deficient mice. Brain Research 870(1-2):10-19. Homola J (2003) Present and future of surface plasmon resonance biosensors. Anal Bioanal Chem 377:528-539. Hopkins AL, Groom CR (2002) The druggable genome. Nature Rev Drug Discovery 1: 727-730.

Ingram SL, Williams JT (1994) Opioid inhibition of Ih via adenylyl cyclase. Neuron 13:179-186. Inturrisi CE (2002) Clinical Pharmacology of Opioids for Pain. Clin J of Pain 18(4):S1- S11. James MK, Feldman PL, Schuster SV, Bilotta JM, Brackeen MF, et al. (1991) Opioid receptor activity of GI 87084B, a novel ultra-short acting analgesic, in isolated tissues. JPET 259: 712-718. Jin J, Kittanakom S, Wong V, Reyes BAS, Van Bockstaele EJ, Stagljar I et al. (2010) Interaction of the mu-opioid receptor with GPR177 (wntless) inhibits wnt secretion: potential implications for opioid dependence. BMC Neurosci 11(33): 1-15. Johnson GL, Lapadat R (2002) Mitogen-activated protein kinase pathways mediated by ERK, JNK, and p38 protein kinases. Science 298: 1191-1192. Jordan B, Devi LA (1998) Molecular mechanisms of opioid receptor signaling transduction. British J of Anesthesia 81:12-19. Jordan BA, Trapaidze N, Gomes I, Nivarthi R, Devi LA (2001) Oligomerization of

opioid receptors with β2-adrenergic receptors: a role in trafficking and mitogen- activated protein kinase activation. PNAS 98: 343-348. Jutkiewicz EM (2006) The antidepressant-like effects of delta-opioid receptor agonists. Mol Interv 6:162-169. Kazmi SM, Mishra RK (1986) Opioid receptors in human neuroblastoma SH-SY5Y cells: Evidence for distinct morphine (µ) and enkaphlin (δ) binding sites. Biochem and Biophys Research Communications 137(2):813-820. Keith DE, Murray SR, Zaki PA, Chu PC, Lissin DV, et al. (1996) Morphine activates opioid receptors without causing their rapid internalization. JBC 271(32):19021- 10924. Kelly E, Baily CP, Henderson G (2007) Agonist-selective mechanisms of GPCR desensitization. BJP 153:S379-S388. Kenakin T (1997) Pharmacologic analysis of drug-receptor interaction. Lippincott- Raven, Philadelphia. Kenakin T (2002) Efficacy at g-protein-coupled receptors. Nature Rev Drug Discovery 1:103-110.

108

Kenakin T (2004) Principles: Receptor theory in pharmacology. TiPS 25(4): 186-192. Kenakin T (2007a) Collateral efficacy in drug discovery: taking advantage of the good (allosteric) nature of 7TM receptors. TiPS 28(8):407-415. Kenakin T (2007b) Functional selectivity through protean and biased agonism: who steers the ship? Molecular Pharm 72(6):1393-1401. Kenakin T (2009) Cellular assays as portals to seven-transmembrane receptor-based drug discovery. Nature Rev Drug Discovery 3(8): 617-626. Kenakin T, Miller LJ (2010) Seven transmembrane receptors as shapeshifting proteins: the impact of allosteric modulation and functional selectivity on new drug discovery. Pharmacol Rev 62: 265-304. Kieffer BL (1995) Recent advances in molecular recognition and signal transduction of active peptides: receptors for opioid peptides. Cellular and Molec Neurobio. 15(6): 615-635. Kieffer BL (1999) Opioids: first lessons from knockout mice. TiPS 20(1):19-26. Kieffer BL, Befort K, Gavériaux-Ruff C, Hirth CG (1992) The δ-opioid receptor: Isolation of a cDNA by expression closing and pharmacological characterization. PNAS 89:12048-12052. Kieffer BL, Evans CJ (2002) Opioid tolerance – in search of the holy grail. Cell 108:587- 590. Kieffer BL, Evans CJ (2009) Opioid Receptors: from binding sites to visible molecules in vivo. Neuropharm 56:205-212. Kieffer BL, Gavériaux-Ruff C (2002) Exploring the opioid system by gene knockout. Progress in Neurobiology 66(5):285-306. Kong H, Raynor K, Yano K, Takeda J, Bell GI, et al. (1994) Agonists and antagonists bind to different domains of the cloned κ opioid receptor. PNAS 91:8042-8046. Kovoor A, Nappey V, Kieffer BL, Chavkin C (1997) µ and δ opioid receptors are differentially desensitized by the coexpression of β-adrenergic receptor kinase 2 and β-arrestin 2 in Xenopus oocytes. JBC 272:27605-27611 Law PY, Loh HH (1999) Regulation of opioid receptor activities. JPET 289(2): 607-624. Law PY, Wong YH, Loh HH (2000) Molecular mechanisms and regulation of opioid receptor signaling. Annu Rev Pharmcol Toxicol 40: 389-430. Lee PH, Gao A, van Staden C, Ly J, Salon J, et al. (2008) Evaluation of Dynamic Mass Redistribution Technology for Pharmacological Studies of Recombinant and Endogenously Expressed G Protein-Coupled Receptors. ASSAY and Drug Development Technologies. 6(1): 83-94. Liu AMF, Wong YH (2005) µ-opioid receptor-mediated phosphorylation of IkB kinase in human neuroblastoma SH-SY5Y cells. Neurosignals 14: 136-142. Lord JAH, Waterfield AA, Hughes J, Kosterlitz HW (1977) Endogenous opioid peptides: multiple agonists and receptors. Nature 267(5611): 495-499.

109

Mace G, Miaczynska M, Zerial M, Nebreda AR (2005) Phosphorylation of EEA1 by p38 MAP kinase regulates u opioid receptor endocytosis. EMBO J 24: 3235-3246. Mailman RB (2007) GPCR functional selectivity has therapeutic impact. TiPS 8(28):390-396. Mansour A, Khachaturian H, Lewis ME, Akil H, Watson SJ (1988) Anatomy of CNS opioid receptors. Trends in Neurosci 11(7):308-312. Marinissen MJ, Gutkind JS (2001) G-protein-coupled receptors and signaling networks: emerging paradigms. TiPS 22(7):368-76. Matthes HWD, Maldonado R, Simonin F, Valverde O, Slowe S, et al. (1996) Loss of morphine- induced analgesia, reward effect and withdrawal symptoms in mice lacking the µ-opioid- receptor gene. Nature 383: 819-823. Matthes HW, Smadja C, Valverde O, Vonesch JL, Foutz AS, et al. (1998) Activity of the delta- opioid receptor is partially reduced, whereas activity of the kappa-receptor is maintained in mice lacking the mu-receptor. J Neurosci 18:7285-7295. Mayer DJ, Liebskind JC (1974) Pain reduction by focal electrical stimulation of the brain: an anatomical approach. Brain Res 68:73-93. Melief EJ, Miyatake M, Bruchas MR, Chavkin C (2010) Ligand-directed c-Jun N- terminal kinase activation disrupts opioid receptor signaling. PNAS 107(25):11608-11613. Merg F, Filliol D, Usynin U, Bazov I, Bark N et al. (2006) Big dynorphin as a putative endogenous ligand for the κ-opioid receptor. J of Neurochem 97:292-301. Merrer JL, Becker JAJ, Befort K, Kieffer BL (2009) Reward processing by the opioid system in the brain. Physio Rev 89: 1379-1412. Mestek A, Hurley JH, Bye LS, Campbell AD, Chen Y et al. (1995) The human µ opioid receptor: modulation of functional desensitization by calcium/calmodulin- dependent protein kinase and protein kinase C. J of Neuro 15: 2396-2406 Meunier JC, Mollereau C, Toll L, Suaudeau C, Moisand C, et al. (2002) Isolation and structure of the endogenous agonist of opioid receptor-like ORL1 receptor. Nature 377: 532-535. Milligan G (2004) G protein-coupled receptor dimerization: function and ligand pharmacology. Molecular Pharm. 66(1):1-7. Mogil JS, Pasternak GW (2001) The molecular and behavioral pharmacology of the orphanin FQ/Nociceptin peptide and receptor family. Pharm Reviews 53(3):381- 415. Mollereau C, Parmentier M, Mallieux P, Butour JL, Moisand C, et al. (1994) ORL1, a novel member of the opioid reeptor family: cloning, functional expression and localization. FEBS letters 341(1):33-38. Morse M, Tran E, Sun H, Levenson R, Fang Y (2011) Ligand-directed functional selectivity at the mu opioid receptor revealed by label-free integrative pharmacology on-target. PLoS One doi:10.1371/journal.pone.0025643

110

Mostany R, Diaz A, Valdizan EM, Rodriguez RM, Garzon J, Hurle MA (2008) Supersensitivity to µ-opioid receptor-mediated inhibition of the adenylyl cyclase pathway involves pertussis toxin-resistant Gα protein subunits. Neuropharm 54:989-997. Nemoto S, Xiang J, Huang S, Lin A (1998) Induction of apoptosis by SB202190 through inhibition of p38B mitogen activated protein kinase. JBC 273: 16415-16420. Nestler EJ, Hope BT, Widnell KL (1993) Drug addiction: a model for the molecular basis of neural plasticity. Neuron 11:995-1006. Neubig RR, Siderovski DP (2002) Regulators of G-protein signaling as new central nervous system drug targets. Nature Rev Drug Discovery 1: 187-197. Neve KA (2009) Functional selectivity of G protein-coupled receptor ligands. Humana: New York. Okie (2010) A flood of opioids, a rising tide of deaths. NEJM. 363 (21): 1981-1985. O'Connor KA, Roth BL (2005) Finding new tricks for old drugs: an efficient route for public-sector drug discovery. Nat Rev Drug Discov 4: 1005–1014. Ossipov MH, Malan TPJ, Lai J, Porreca F (1997) Opioid pharmacology of acute and chronic pain. In: Dickenson A, Besson JM (Eds.), Handbook of Experimental Pharmacology, Vol. 130. Springer, Berlin, pp 305-327. Palos GH, Mendoza TR, Cantor SB, Aday LA, Cleeland CS (2004) Perceptions of analgesic use and side effects: what the public values in pain management. J of Pain and Symptom management 28(5):460-473. Pan ZZ, Tershner SA, Fields HL (1997) Cellular mechanism for anti-analgesic action of agonists of the κ-opioid receptors. Nature 389: 382-385. Pasternak GW (1993) Pharmacological mechanisms of opioid analgesics. Clinical Neuropharm 16(1):1-18. Pert BC, Snyder SH (1973) Opiate receptor: demonstration in nervous tissue. Science 179(4077):1011-1014. Polakiewicz RD, Schieferl S, Dorner LF, Kansra V, Comb MJ (1998) A mitogen- activated protein kinase pathway is required for µ-opioid receptor desensitization. JBC 273: 12402-12406. Polakiewicz RD, Schieferl SM, Gingras AC, Sononberg N, Comb MJ (1998) µ-opioid receptor activates signaling pathways implicated in cell survival and translational control. JBC 273: 23534-23541. Raehal KM, Bohn LM (2005) Mu opioid receptor regulation and opiate responsiveness. AAPS journal 7(3):E587-E591. Raehal KM, Schmid CL, Groer CE, Bohn LM (2011) Functional selectivity at the µ opioid receptor: implications for understanding opioid analgesia and tolerance. Pharmacol Rev 63: DOI:10.1124/pr.111.004598. Ray J, Peterson DA, Schinstine M, Gage FH (1993) Proliferation, differentiation, and long-term culture of primary hippocampal neurons. PNAS 90(8):3602-3606

111

Raynor K, Kong H, Yasuda K, Yu L, Bell GI, et al. (1994) Pharmacological characterization of the cloned kappa-, delta-, and mu-opioid receptors. Mol. Pharmacol. 45:330-334. Reisine T (1995) Opiate receptors. Neuropharm 34(4):463-472. Reisine T, Bell GI (1993) Molecular biology of opioid receptors. Trends in Neuroscience 16(12):506-510. Rich RL, Myszka DG (2000) Advances in surface Plasmon resonance biosensor analysis. Current Opinion in Biotechnology 11:54-61. Schröder R, Janssen N, Schmidt K, Kebig A, Merten N, et al. (2010) Deconvolution of complex G protein-coupled receptor signaling in live cells using dynamic mass redistribution measurements. Nature Biotech 28: 943-949. Schröder R, Schmidt J, Blättermann S, Peters L, Janssen N, et al. (2011) Applying label- free dynamic mass redistribution technology to frame signaling of G protein- coupled receptors noninvasively in living cells. Nature Protocols 6(11): 1748- 1760 Schulz S, Mayer D, Pfeiffer M, Stumm R, Koch T, et al. (2004) Morphine induces terminal mu-opioid receptor desensitization by sustained phosphorylation of serine-375. EMBOJ 23:3282-3289. Seamon KB (1985) Activation of hormone-sensitive adenylate cyclase by forskolin. Drug Development Research 6:181-192. Seamon KB, Daly JW (1981) Forskolin: a unique diterpene activator of cyclic AMP- generating systems. J Cyclic Nucl Res 7: 201-224. Shen KF, Crain SM (1990) Cholera toxin-A subunit blocks opioid excitatory effects on sensory neuron action potentials indicating mediation by Gs-linked opioid receptors. Brain Res 525: 225-231. Simonin F, Valverde O, Smadja C, Slowe S, Kitchen I, et al. (1998) Disruption of the κ– opioid receptor gene in mice enhances sensitivity to chemical visceral pain, impairs pharmacological actions of the selective κ–agonist U-50,488H and attenuates morphine withdrawal. EMBOJ. 17: 886-897. Snyder SH, Childers SR (1979) Opiate receptors and opioid peptides. Annu Rev Neurosci 2:35-64. Sora I, Elmer G, Funada M, Peiper J, Li XF, et al. (2001) µ opiate receptor gene dose effects on different morphine actions: evidence for differential in vivo µ receptor reserve. Neuropsychopharmacology 25(1):41:54. Stein C, Schäfer M, Machelska H (2003) Attacking pain at its source: new perspectives on opioids. Nature Med 9(8):1003-1008. Swinney DC and Anthony J (2011) How were new medicines discovered? Nature Rev Drug Discov 10: 507-519. Takemori AE, Portoghese PS (1984) Comparative antagonism by naltrexone and naloxone of µ, κ, and δ agonist. Eur J Pharmacol 104: 101-104.

112

Terrillon S, Bouvier M (2004) Roles of G-protein-coupled receptor dimerization. EBMO reports. 5(1):30-35. Thompson RC, Mansour A, Akil H, Watson SJ (1993) Cloning and pharmacological characterization of a rat µ opioid receptor. Neuron 11: 903-913. Tiefenthaler K, Lukosz W (1989) Sensitivity of grating couplers as integrated optical chemical sensors. J Op Soc Am B 6:209-220. Tran E, Fang Y (2009) Label-free optical biosensor for probing integrative role of adenylyl cyclase in G protein-coupled receptor signaling. J Recept Signal Transduct 29:154-162. Traynor J (2010) Regulator of G protein-signaling proteins and addictive drugs. Ann. NY Acad Sci 1187: 341-352. Urban JD, Clarke WP, von Zastrow M, Nichols DE, Kobilka B, et al. (2007) Functional selectivity and classical concepts of quantitative pharmacology. Perspectives Pharm 320: 1-13. Urban JD, Vargas GA, von Zastrow M, Mailman RB (2006) Aripiprazole has functionally

selective actions at dopamine D2 receptor-mediated signaling pathways. Neuropsychopharmacology 32:67-77. Van Rijn RM, Whistler JL, Waldhoer M (2010) Opioid-receptor-heteromer-specific trafficking and pharmacology. Current Opinion in Pharm 10(1): 73-79. Vanderah TW (2010) Delta and kappa opioid receptors as suitable drug targets for pain. Clinical J of Pain. 26: S10-S15. Vhalos CJ, Matter WF, Hui KY, Brown RF (1994) A specific inhibitor of phosphatidylinositol 3- kinase, 2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4-one (LY294002). JBC 269: 5241-5248. Violin JD, Lefkowitz RJ (2007) β-Arrestin-biased ligands at seven transmembrane receptors. TiPS 28(8):416-422. von Zastrow M (2003) Mechanisms regulating membrane trafficking of G protein-coupled receptors in the endocytic pathway. Life Sci 74:217-224. Waldenberg MLG (2003) A review of the properties of spiradoline: a potent and selective κ- opioid receptor agonist. CNS Drug Review 9: 187-198. Waldhoer M, Bartlett SE, Whistler JL (2004) Opioid Receptors. Annu. Rev Biochem 73:953-90. Wang D, Sun X, Bohn LM, Sadée W (2005a) Opioid receptor homo- and heterodimerization in living cells by quantitative bioluminescence resonance energy transfer. Mol Pharm 67(6):2173-2184. Wang WW, Shahrestanifar M, Jin J, Howells RD (1995) Studies on µ and δ opioid receptor selectivity utilizing chimeric and site mutagenized receptors. PNAS 92: 12436-12440.

113

Wang Y, Tang K, Inan S, Siebert D, Holzgrabe U, et al. (2005b) Comparison of pharmacological activities of three distinct kappa ligands (Salvinorin A, TRK-820 and 3FLB) on kappa opioid receptors in vitro and their antipruritic and antinociceptive activities in vivo. JPET 312: 220-230. Ward SJ, Portoghese PS, Takemori AE (1982) Pharmacological characterization in vivo of the novel opiate, β-funaltrexamine. JPET 220: 494-498. Wei ZY, Brown W, Takasaki B, Plobeck N, Delorme D, et al. (2000) N,N-Diethyl-4- (phenylpiperidin-4-ylidenemethyl)benzamide: A novel, exceptionally selective, potent δ opioid receptor agonist with oral bioavailability and its analogues. J Med Chem 43(21): 3895-3905. Wess J (1998) Molecular basis of receptor/G-protein-coupling selectivity. Pharmacol Ther 80: 231-264. William C (2004) cAMP detection methods in HTS: selecting the best from the rest. Nature Rev Drug Disc 2:125-135. Williams JT, MacDonald JC, Mazoni O (2001) Cellular and synaptic adaptations mediating opioid dependence. Physiological reviews 81(1):299-352. Wong YH, Demoliou-Mason CD, Barnard EA (1988) ADP-ribosylation with pertussis toxin modulates the GTP-sensitive opioid ligand binding in digitonin-soluble extracts of rat brain membranes. J Neurochem 51:114-121. Xia M, Guo V, Huang R, Shahane SA, Austin CP, et al. (2011) Inhibition of morphine- induced cAMP overshoot: a cell-based assay model in high-throughput fashion. Cell Mol Neurobiol 31:901-907. Xu C, Hong MH, Zhang LS, Hou YY, Wang YH, et al. (2010) Serine 363 of the δ-opioid receptor is critical for adopting distinct pathways to activate ERK1/2 in response to stimulation with different ligands. J of Cell Science 123:4259-4270. Yan F, Roth BL (2004) Salvinorin A: A novel and highly selective κ-opioid receptor agonist. Life Sci 75:2615-2619. Young AM, Stephens KR, Hein DW, Woods JH (1984) Reinforcing and discriminative stimulus properties of mixed agonist-antagonist opioids. JPET 229: 118-126. Yu VC, Elger S, Duan DS, Lameh J, Sadée W (2006) Regulation of cyclic AMP by the µ-opioid receptor in human neuroblastoma SH-SY5Y cells. J of Neurochem 55(4): 1390-1396. Yu Y, Zhang L, Yin X, Sun H, Uhl GR, et al. (1997) µ opioid receptor phosphorylation, desensitization, and ligand efficacy. JBC 272:28869-28874. Zhang J, Ferguson SG, Barak LS, Bodduluri SR, Laporte SA, et al. (1998) Roles for G protein- couples receptor kinase in agonist-specific regulation of µ-opioid receptor responsiveness. PNAS 95:7157-7162. Zhang Y, Xiong W, Lin X, Ma X, Yu LC (2009) Receptor Trafficking induced by µ – opioid-receptor phosphorylation. Neuroscience and Biobeh Reviews 33:1192- 1197.

114

Zheng H, Chu J, Zhang Y, Loh HH, Law PY (2011) Modulating µ-opioid receptor phosphrylation switches agonist-dependent signaling as reflected in PKCε activation and dendritic spine stability. JBC 286:12724-12733. Zheng H, Loh HH, Law PY (2010) Agonist-selective signaling of G protein-coupled receptor: mechanisms and implication. IUBMB Life. 62(2):112-119. Zhu J, Luo LY, Li JG, Chen C, Lui-Chen LY (1997) Activation of the cloned human Kappa opioid receptor by agonists enhances [35S]GTPγS binding to membranes: determination of potencies and efficacies of ligands. JPET 282(2):676-684

115

Appendix

Dynorphin A2-13

0 Nociceptin1-13 ICI199441

β-funaltrexamine -50 Etonitazenyl isothiocyanate

BNTX

Response(pm) -100

-150 0 10 20 30 40 50

Time (sec) Time (min)

Appendix A. DMR characteristics of a subset of ligands in parental HEK293 cells. The DMR was due to the activation of an unknown endogenous target(s). Data represents the mean ± s.d. for 2 independent measurements (n=4).

116

Appendix B: Opioid ligands and their affinity binding to the MOR. Ki (IC50/EC50) Compound Pharmacology (nM) Reference (-)-Norcodeine Opioid agonist 266.9 Lotsch et al., 2006 (-)-U-50488 HCl Kappa agonist 830 Tam, 1985 (+)-U-50488 HCl Kappa partial agonist SKF10047 Opioid agonist/antagonist 1900 Tam, 1985 DADLE Opioid agonist 150 Tam, 1985 DAMME Mu agonist 0.33 Leslie et al., 1982 DAMGO Mu agonist 1.1 Amiche et al., 1989 DALDA Mu agonist 1.69 Schiller et al., 1989 BUBUC Delta agonist 2980 Gacel et al., 1990 DSLET Delta agonist 120 Creese et al., 1976 DTLET Delta agonist (Leu5)-Enkephalin Mu agonist 7.4 Childers et al., 1979 (Met5)-Enkephalin Mu agonist 3 Childers et al., 1979 TAPP Mu agonist

DIPPA Kappa antagonist 1799 (IC50) Chang et al., 1994 α-Neoendorphin Kappa1 agonist β-Funaltrexamine HCl Mu antagonist 0.33 Raynor et al., 1994 BNTX maleate Delta antagonist 18 Reisine, 1995 BRL-52537 Kappa agonist 1560 Chen et al., 2004 DAMGO Mu agonist 1.1 Amiche et al., 1989 Deltorphin II Delta agonist >1000 Reisine, 1995 DPDPE Delta agonist 600 Amiche et al., 1989 Dynorphin A (1-13) Kappa agonist 31 Tam, 1985 Dynorphin A (1-8) Kappa agonist 22.3 Merg et al., 2006 Dynorphin A (2-13) Kappa agonist Dynorphin B Kappa agonist 13 Merg et al., 2006 Endomorphin-1 Mu agonist 0.67 Goldberg et al., 1998 Endomorphin-2 Mu agonist 0.43 Goldberg et al., 1998 Etonitazenyl Rios and Tephly, isothiocyanate Mu antagonist 71 2002 GR 89696 fumarate Kappa agonist ICI 199,441 HCl Kappa agonist 4500 Costello et al. 1988 ICI 204,448 HCl Kappa agonist >1000 Raynor et al., 1994 Levallorphan tartrate Opioid antagonist 1.7 Carroll et al., 1988 Mu opioid Nalbuphine HCl antagonist/kappa agonist 11 Raynor et al., 1994

117

Naloxonazine 2HCl Opioid antagonist 0.054 Raynor et al., 1994 Naloxone HCl Opioid antagonist 2.4 Titeler et al., 1989 Lewanowitsch and Naloxone methiodide Opioid antagonist 28.9 Irvine, 2003 Naltrexone HCl Opioid antagonist 0.77 Carroll et al., 1988 Naltriben mesylate Delta antagonist 80.8 Raynor et al., 1994 Naltrindole HCl Delta agonist 64 Reisine, 1995 N-Benzylnaltrindole 11.2 (IC50 HCl Delta antagonist Ratio)** Korlipara et al., 1994 Weerawarna et al., N-MPPP* Kappa agonist >1000 1994 NNC 63-0532 ORL1 agonist 140 Titeler et al., 1989 Nociceptin ORL1 agonist 270 Halab et al., 2002 Nociceptin (1-13) NH2 ORL1 agonist nor-Binaltorphimine 2HCl Kappa antagonist 2.2 Raynor et al., 1994

Salvinorin A Kappa agonist 1728 (EC50) Rothman et al., 2007 SNC 121 Delta agonist SNC 80 Delta agonist

Syndyphalin SD-25 Mu agonist 0.29 (IC50) Quirion et al., 1982 Tramadol HCl Mu agonist 2400 Gillen et al., 2000 U-50,488H mesylate Kappa agonist U-54494A HCl Kappa agonist U-62066 Kappa agonist 252 Clark et al., 1988 U-69593 Kappa agonist 1600 Titeler et al., 1989 * N-MPPP, N-Methyl-N-[(1S)-1-phenyl-2-(1-pyrrolidinyl)ethyl]phenylacetamide HCl. ** * IC50 ratio is the IC50 of agonist in the presence of 100nM antagonist.

118

Appendix C: Opioid ligands and their affinity binding to the DOR Ki (IC50) Compound Pharmacology (nM) Reference (-)-Norcodeine Opioid agonist (-)-U-50488 HCl Kappa agonist 2300 Chang, 2003 (+)-U-50488 HCl Kappa partial agonist Opioid SKF10047 agonist/antagonist 19000 Tam, 1985 DADLE Opioid agonist 0.63 Chavkin, 1982

DAMME Mu agonist 7.7 (IC50) Leslie et al., 1982 DAMGO Mu agonist 64.7 Amiche et al., 1989 DALDA Mu agonist 19200 Schiller et al., 1989 BUBUC Delta agonist 2.9 Gacel et al., 1990 DSLET Delta agonist 4.8 Clark et al., 1988 DTLET Delta agonist 22 Wang et al., 1991 (Leu5)-Enkephalin Mu agonist 6.2 Childers et al., 1979 (Met5)-Enkephalin Mu agonist 2 Childers et al., 1979 Charpentier et al., TAPP Mu agonist 695 1991 >1000 DIPPA Kappa antagonist (IC50) Chang et al., 1994 α-Neoendorphin Kappa1 agonist 1.17 Mansour et al., 1995 β-Funaltrexamine HCl Mu antagonist 48 Raynor et al., 1994 BNTX maleate Delta antagonist 0.66 Reisine, 1995 BRL-52537 Kappa agonist DAMGO Mu agonist >1000 Reisine, 1995 Deltorphin II Delta agonist 3.3 Reisine, 1995 DPDPE Delta agonist 2.15 Amiche et al., 1989 Dynorphin A (1-13) Kappa agonist >1000 Reisine, 1995 Dynorphin A (1-8) Kappa agonist 32.6 Merg et al., 2006 Dynorphin A (2-13) Kappa agonist Dynorphin B Kappa agonist 12.8 Merg et al., 2006 Endomorphin-1 Mu agonist >500 Goldberg et al., 1998 Endomorphin-2 Mu agonist >500 Goldberg et al., 1998 Etonitazenyl isothiocyanate Mu antagonist >1000 Fichna et al., 2008 GR 89696 fumarate Kappa agonist ICI 199,441 HCl Kappa agonist ICI 204,448 HCl Kappa agonist >1000 Raynor et al., 1994 Levallorphan tartrate Opioid antagonist 1 Childers et al., 1979 Nalbuphine HCl Mu antagonist/kappa 163 Tam, 1985 119

agonist Naloxonazine 2HCl Opioid antagonist 8.6 Raynor et al., 1994 Naloxone HCl Opioid antagonist 16 Tam, 1985 Lewanowitsch and Naloxone methiodide Opioid antagonist 203.5 Irvine, 2003 Naltrexone HCl Opioid antagonist 149 Reisine, 1995 Naltriben mesylate Delta antagonist 0.4 Spetea et al., 1998 Naltrindole HCl Delta agonist 0.02 Reisine, 1995 N-Benzylnaltrindole 459 (IC50 Korlipara et al., HCl Delta antagonist Ratio**) 1994 >1000 Weerawarna et al., N-MPPP* Kappa agonist (IC50) 1994 NNC 63-0532 ORL1 agonist Nociceptin ORL1 agonist >1000 Halab et al., 2002 Nociceptin (1-13) NH2 ORL1 agonist nor-Binaltorphimine 2HCl Kappa antagonist 65 Raynor et al., 1994 Salvinorin A Kappa agonist >1000 Roth et al., 2002 SNC 121 Delta agonist 3.8 Ni et al., 1994 SNC 80 Delta agonist 1.8 Bryans, 1999

Syndyphalin SD-25 Mu agonist 1250 (IC50) Quirion et al., 1982 Tramadol HCl Mu agonist 57700 Codd et al., 1995 U-50,488H mesylate Kappa agonist 2100 Tam, 1985 U-54494A HCl Kappa agonist U-62066 Kappa agonist 9400 Wadenberg, 2003 U-69593 Kappa agonist 13400 Clark et al., 1988 * N-MPPP, N-Methyl-N-[(1S)-1-phenyl-2-(1-pyrrolidinyl)ethyl]phenylacetamide HCl. ** * IC50 ratio is the IC50 of agonist in the presence of 100nM antagonist.

120

Appendix D: Opioid ligands and their affinity binding to the KOR Ki (IC50/EC50) Compound Pharmacology (nM) References (-)-Norcodeine Opioid agonist (-)-U-50488 HCl Kappa agonist 4.2 Lahti et al., 1985 (+)-U-50488 HCl Kappa partial agonist SKF10047 Opioid agonist/antagonist 1600 Tam, 1985 DADLE Opioid agonist 1900 Lahti et al., 1985 DAMME Mu agonist 280 Chang, 2003 Amiche et al., DAMGO Mu agonist >20000 1989 DALDA Mu agonist 26 Zhao et al., 2003 Gacel et al., BUBUC Delta agonist 1900 1990 DSLET Delta agonist >1000 Reisine, 1995 Besse et al., DTLET Delta agonist > 10000 1990 Childers et al., (Leu5)-Enkephalin Mu agonist 9.4 1979 Childers et al., (Met5)-Enkephalin Mu agonist 2.9 1979 TAPP Mu agonist Chang et al., DIPPA Kappa antagonist 2.21 (IC50) 1994 Meng et al., α-Neoendorphin Kappa1 agonist 1.4 1993 Raynor et al., β-Funaltrexamine HCl Mu antagonist 2.8 1994 BNTX maleate Delta antagonist 55 Reisine, 1995 BRL-52537 Kappa agonist 0.24 Chen et al., 2004 DAMGO Mu agonist >20000 Reisine, 1995 Deltorphin II Delta agonist >1000 Reisine, 1995 Mosberg et al., DPDPE Delta agonist 12000 1987 Dynorphin A (1-13) Kappa agonist 0.98 Lahti et al., 1985 Dynorphin A (1-8) Kappa agonist 0.275 Merg et al., 2006 Meng et al., Dynorphin A (2-13) Kappa agonist 680 1993 Dynorphin B Kappa agonist 3.8 Merg et al., 2006 Goldberg et al., Endomorphin-1 Mu agonist >500 1998 Endomorphin-2 Mu agonist >500 Goldberg et al., 121

1998 Etonitazenyl isothiocyanate Mu antagonist Caudle et al., GR 89696 fumarate Kappa agonist 41.7 1997 Kumar et al., ICI 199,441 HCl Kappa agonist 0.054 2000 Kumar et al., ICI 204,448 HCl Kappa agonist 2.69 2005 Childers et al., Levallorphan tartrate Opioid antagonist 0.3 1979 Mu opioid antagonist/kappa Raynor et al., Nalbuphine HCl agonist 61 1994 Raynor et al., Naloxonazine 2HCl Opioid antagonist 11 1994 Naloxone HCl Opioid antagonist 6.3 Lahti et al., 1985 Lewanowitsch Naloxone methiodide Opioid antagonist 1010 and Irvine, 2003 Naltrexone HCl Opioid antagonist 0.83 Smith, 1989 Spetea et al., Naltriben mesylate Delta antagonist >10000 1998 Naltrindole HCl* Delta agonist 66 Reisine, 1995 N-Benzylnaltrindole 1.3 (IC50 Korlipara et al., HCl Delta antagonist Ratio)** 1994 Weerawarna et N-MPPP* Kappa agonist 1.4 (IC50) al., 1994 Thomsen and NNC 63-0532 ORL1 agonist 405 Hohlweg, 2000 Halab et al., Nociceptin ORL1 agonist >1000 2002 Nociceptin (1-13) NH2 ORL1 agonist nor-Binaltorphimine Raynor et al., 2HCl Kappa antagonist 0.027 1994 0.63nM Chavkin et al., Salvinorin A Kappa agonist (EC50) 2004 SNC 121 Delta agonist SNC 80 Delta agonist 2900 Chang, 2003 Quirion et al., Syndyphalin SD-25 Mu agonist 13000 (IC50) 1982 Tramadol HCl Mu agonist 42700 Codd et al., 1995 Gairin et al., U-50,488H mesylate Kappa agonist 7 1985 Vonvoigtlander U-54494A HCl Kappa agonist 21 et al., 1987 122

U-62066 Kappa agonist 1.4 Lahti et al., 1985 U-69593 Kappa agonist 5.4 Lahti et al., 1985 * N-MPPP, N-Methyl-N-[(1S)-1-phenyl-2-(1-pyrrolidinyl)ethyl]phenylacetamide HCl. ** * IC50 ratio is the IC50 of agonist in the presence of 100nM antagonist.

123

References for Appendix

Amiche M, Sagan S, Mor A, Delfour A, Nicholas P (1989) Dermenkephalin (Tyr-D- Met-Phe-His-Leu-Met-Asp-NH2): A potent and fully specific agonist for the delta opioid receptor. Mol. Pharmacol. 35: 774-779. Besse D, Lombard MC, Zajac JM, Roques, JM Besson (1990) Pre- and postsynaptic distribution of µ, δ, and κ opioid receptors in the superficial layers of the cervical dorsal horn of the rat spinal cord. Brain Res 521:15-22. Bryans JS (1999) Novel Treatments for pain. IDrugs 2:1170-1182. Carroll JA, Shaw JS, Wickenden AD (1988) The physiological relevance of low agonist affinity binding at opioid µ-receptors. Br. J. Pharmacol. 94: 625-631. Caudle RM, Mannes AJ, Iadarola MJ (1997) GR89,696 is a kappa-2 opioid receptor agonist and a kappa-1 opioid receptor antagonist in the guinea pig hippocampus. PNAS USA 283:1342-1349. Chang AC, Takemori AE, Portoghese PS (1994) 2-(3,4-Dichlorophenyl)-N-methyl- N-[(1S)-1-(3-isothiocyanatophenyl)-2-(1- pyrrolidinyl)ethyl]acetamide: An opioid receptor affinity label that produces selective and long-lasting kappa antagonism in mice. J Med Chem 37:1547–1549. Chang KJ (2003) The delta receptor. Informa Healthcare: New York. Charpentier S, Sagan S, Delfour A, Nicolas P (1991) Dermenkephalin and deltorphin I reveal similarities within ligand-binding domains of µ- and δ-opioid receptors and an additional address subsite on the δ-receptor. Biochem Biophys Res Comm 179:1161-1168. Chavkin C (1982) Dynorphin is a specific endogenous ligand of the kappa opioid receptor. Science 215: 413-415. Chavkin C, Sud S, Jin W, Stewart J, Zjawiony JK (2004) Salvinorin A, an active component of the hallucinogenic sage salvia divinorum is a highly efficacious κ- opioid receptor agonist: structural and functional considerations. PNAS USA 308:1197-1203. Chen TY, Goyagi T, Toung TJ, Kirsch JR, Hurn PD et al (2004) Prolonged opportunity for ischemic neuroprotection with selective kappa-opioid receptor agonist in rats. Stroke 35:1180. Childers SR, Creese I, Chowman AM, Snyder SH (1979) Opiate receptor binding affected differentially by opiates and opioid peptides. Eur J Pharmacol. 55: 11- 18. Clark MJ, Carter BD, Medzihradsky F (1988) Selectivity of ligand binding to opioid receptors in brain membranes from the rat, monkey and guinea pig. Eur. J. Pharmacol. 148: 343-351. Codd EE, Shank RP, Schupsky JJ and Raffe RB (1995) Serotonin and norepinephrine uptake inhibiting activity of centrally acting analgesics: structural determinants and role in antinociception. JPET 273:1263-1270. Costello GF, Main BG, Barlow JJ, Carrol JA, Shaw JS (1988) A novel series of potent and selective agonists at the opioid kappa-receptor. Eur. J. Pharmacol. 151: 475- 478.

124

Creese I, Feinberg AP, Snyder SH (1976) Butyrophenone influences on the opiate receptor. Eur. J. Pharmacol. 36: 231-235. Fichna J, Do-Rego JC, Janecki T, Staniszewska R, Poels J et al (2008) Novel highly potent µ-opioid receptor antagonist based on endomorphin-2 structure. Bioorg Med Chem Lett 18:1350-1353. Gacel GA, Fellion E, Baamonde A, Dauge V, Roquies BP (1990) Synthesis, biochemical and pharmacological properties of BUBUC, a highly selective and systemically active agonist for in vivo studies of δ-opioid receptors. Peptides 11:983-988. Gairin JE, Gouarderes C, Mazarguil H, Alvinerie P, Cros J (1985) [D- Pro10]dynorphin-(1-11) is a highly potent and selective ligand for kappa opioid receptors. Eur. J. Pharmacol. 106: 457-458. Gillen C, Haurand M, Kobelt DJ, Wnendt S (2000) Affinity, potency and efficacy of tramadol and its metabolites at the cloned human µ-opioid receptor. Naunyn- schmiedeberg's Arch. Pharmacol. 362:116-121. Goldberg IE, Rossi GC, Letchworth SR, Mathis JP, Ryan-Moro J et al. (1998) Pharmacological characterization of endomorphin-1 and endomorphin-2 in mouse brain. JPET 286:1007-1013. Halab L, Becker JA, Darula Z, Tourwe D, Kieffer BL et al (2002) Probing opioid receptor interactions with azacycloalkane amino acids. Synthesis of a potent and selective ORL1 antagonist. J Med Chem 45:5353-7. Koch G,Wiedermann K, Teschemacher H. (1985) Opioid activities of human β- casomorphins. Naunyn-Schmiederberg's Archives Pharm. 331:351-354. Korlipara VL, Takemori AE, Portochese PS (1994) N-Benzylnaltrindoles as long acting δ-opioid receptor antagonists. J. Med. Chem. 37:1882. Kumar V, Guo D, Cassel JA, Daubert JD, DeHaven RN et al (2005) Synthesis and evaluation of novel peripherally restricted κ-opioid receptor agonists. Bioorg Med Chem Letters 15:1091-1095. Kumar V, Marella MA, Cortes-Burgos LC, Chang AC, Cassel JA et al (2000) Arylacetamides as peripherally restricted kappa opioid receptor agonists. Bioorg Med Chem Letters 10:2567-2570. Lahti RA, Mickelson MM, McCall JM, VonVoigtlander PF (1985) [3H]U-69593 a highly selective ligand for the opioid kappa receptor. Eur J Pharmacol 109: 281-284. Leslie FM, Tso A, Hurlbut DE (1982) Differential appearance of opiate receptor subtypes in neonatal rat brain. Life Sci 31:1393-1396. Lewanowitsch T, Irvine RJ (2003) Naloxone and its quaternary derivative, naloxone methiodide, have differing affinities for µ, δ, and κ opioid receptors in mouse brain homogenates. Brain Res 964:302-305. Lotsch J, Skarke C, Schmidt H, Rohrbacher M, Hofmann U et al (2006) Evidence for morphine-independent central nervous opioid effects after administration of codeine: Contribution of other codeine metabolites. Pharmacodynamics Drug Action 79:35-48. Mansour A, Hoversten MT, Taylor LP, Watson SJ, Akil H (1995) The cloned µ, δ, and κ receptors and their endogenous ligands: Evidence for two opioid peptide recognition cores. Brain Res 700: 89-98.

125

Meng F, Xie GX, Thompson RC, Mansour A, Goldstein A et al (1993) Cloning and pharmacological characterization of a rat kappa opioid receptor. PNAS USA 90:9954-9958. Merg F, Filliol D, Usynin I, Bazv I, Bark N, Hurd YL et al (2006) Big dynorphin as a putative endogenous ligand for the κ-opioid receptor. J Neurochem 97: 292-301. Mosberg HI, Omnaas JR, Goldstein A (1987) Structural requirements for delta opioid receptor binding. Mol. Pharmacol. 31: 599-602. Ni Q, Xu H, Partilla JS, Porreca F, Calderon SN et al (1994) [3H]SNC121: A novel high affinity ligand for rat brain delta receptors: Preliminary studies. Regulatory Peptides 54: 209-210. Quirion R, Kiso Y, Pert CB (1982) Syndyphalin SD-25: a highly selective ligand for mu opiate receptors. FEBS Lett 141:203-206. Raynor K, Kong H, Yasuda K, Yu L, Bell GI et al (1994) Pharmacological characterization of the cloned kappa-, delta-, and mu-opioid receptors. Mol Pharmacol 45:330-334. Reisine T (1995) Opiate receptors. Neuropharm 34:463-472. Rios GR, Tephly TR (2002) Inhbition and active sites of UDP-glocuronosyltransferases 2B7 and 1A1. Drug Metab Dispos 30: 1364-1367 Roth BL, Baner K, Westkaemper R, Siebert D, Rice KC et al (2002) Salvinorin A: A potent naturally occurring nonnitrogenous κ opioid selective agonist. PNAS USA 99: 11934–11939. Rothman RB, Murphy DL, Xu H, Godin JA, Dersch CM et al (2007) Salvinorin A: allosteric interactions at the µ-opioid receptor. JPET 320:801-810. Schiller PW, Nguyen TM, Chung NN, Lemieus C (1989) Dermorphin analogs carrying an increased positive net charge in their "message" domain display extremely high mu-opioid receptor selectivity. J Med Chem 32:698-703 Smith JA, Hunter JC, Hill RG, Hughes J (1989) A kinetic analysis of kappa-opioid agonist binding using the selective radioligand [3H]U69593. J Neurochem 53: 27- 36. Spetea M, Otvos F, Toth G, Nguyen TMD, Schiller PW et al (1998) Interaction of agonist peptide [3H]Tyr-D-Ala-Phe-Phe-NH2 with µ-opioid receptor in rat brain and CHO-µ1 cell line. Peptides 19: 1091-1098. Tam SW (1985) (+)-[3H]SKF 10,047, (+)-[3H]ethylketocyclazocine, µ, κ, δ and phencyclidine binding sites in guinea pig brain membranes. Eur J Pharmacol. 109: 33-41. Thomsen C, Hohlweg R (2000) (8-Naphthalen-1-ylmethyl-4-oxo-1-phenyl-1,3,8-triaza- spiro[4.5]-dec-3-yl)-acetic acid methyl ester (NNC 63-0532) is a novel potent nociceptin receptor agonist. Br J Pharmacol 131:903-908 Titeler M, Lyon RA, Kuhar MJ, Frost JF, Dannais RF et al (1989) µ Opiate receptors are selectively labelled by [3H]carfentanil in human and rat brain. Eur J Pharmacol. 167: 221-228. VonVoigtlander PF, Hall ED, Ochoa MC, Lewis RA, Triezenberg HJ (1987) U-54494A: A unique anticonvulsant related to kappa opioid agonists. JPET 243:542-547. Wadenberg MLG (2003) A review of the properties of spiradoline: A potent and selective κ-opioid receptor agonist. CNS Drug Review 9:187-198.

126

Wang H, Pelaprat D, Roques BP, Vanhove A, Chi ZQ et al (1991) [3H]Ohmefentanyl preferentially binds to µ-opioid receptors but also labels σ-sites in rat brain sections. Eur J Pharm 3:341-350. Weerawarna SA, Davis RD, Nelson WL (1994) Isothiocyanate-substituted kappa- selective opioid receptor ligands derived from N-methyl-N-[(1S)-1-phenyl-2-(1- pyrrolidinyl)ethyl]phenylacetamide. J Med Chem. 37:2856-2864. Zhao GM, Qian X, Schiller PW, Szeto HH (2003) Comparison of [Dmt1]DALDA and DAMGO in binding and G protein activation at µ, δ, and κ opioid receptors. JPET 307:947-954.

127

Curriculum Vitae

Megan Morse

Education

B.A. Behavioral Neuroscience (Honors) Aug 2003 to May 2007 Lehigh University Bethlehem, PA

Ph.D. Candidate (expected graduation 2012) Aug 2007 to Present Department of Neuroscience Neuroscience Program Pennsylvania State University College of Medicine Hershey, Pennsylvania

Corporate Internship June 2004 to August 2004 Life Sciences Department June 2005 to August 2005 Corning Inc. Corning, NY

Posters

“Selective goal-directed behaviors predict heroin ‘addiction-like’ behavior in rats” Tacelosky DM*, Morse M*, Levenson RG, Grigson PS Society for Neuroscience: Drug Reinforcement, Seeking, and Reinstatement. Nov 18, 2010. San Diego, CA

Publications

Jin J, Morse M, Frey C, Petko J, Levenson R (2010) Expression of GPR177 (Wnt/Evi/Sprinter), a highly conserved wnt-transport protein, in rat tissue, zebrafish embryos, and cultured human cells. Dev. Dynamics 239(9): 2426-2434

Morse M, Sun H, Tran E Levenson R, Fang Y (2012) Label-free integrative pharmacology on- target of ligands at the opioid receptor family. (Submitted)

Morse M, Tran E, Sun H, Levenson R, Fang Y (2011) Ligand-directed functional selectivity at the mu opioid receptor revealed by label-free integrative pharmacology on-target. PLoS One, In press.