A Study of the SERCA-Phospholamban Regulatory Interaction Using Time-Resolved Fluorescence
A DISSERTATION SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF THE UNIVERSITY OF MINNESOTA BY
Daniel Richard Stroik
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
David D. Thomas, Advisor
August 2018
Daniel Richard Stroik 2018 ©
Acknowledgements
It is a difficult task to thank everyone in a laboratory as expansive and layered as the
DDT group; but I will begin by thanking my advisor, Dr. David Thomas. I always appreciated his willingness to help others without prejudice or profit, and it is no surprise that this has afforded him many friends and allies over the years. He has always pushed me and for this I am grateful. My time here has made me excited for a future career in science and ready to go after any discovery.
I would like to thank my committee Dr. John Lipscomb, Dr. Wendy Gordon, and Dr.
Sivaraj Sivaramakrishnan. They have provided me with guidance and support, and I have always found useful knowledge and inspiration through our interactions.
Because of the directions that my thesis project took, I worked very closely with Dr.
Razvan Cornea and Dr. Mike Autry. I have always enjoyed their feedback and am very lucky to be surrounded by positive role models. I aspire to be as strong a writer and as clear-thinking a scientist as Razvan. Mike has always been supportive and is one of the most knowledgeable scientists I have ever met. The time he takes to train his students impresses me and he is truly a man of integrity. Many thanks to the lab managers Sarah
Blakely, Octavian Cornea and Destiny Ziebol for their assistance in all matters. There is no doubt that the lab would collapse without you.
To Dr. Robyn Rebbeck and Samantha Yuen, without a doubt the lab would collapse and/or be ablaze if not for your presence as well. Thanks for the conversations and friendship! I have always been impressed by your work ethic. All the extra deeds (e.g., cakes, Town Hall meetings, get-well cards, farewell parties) make the DDT lab a very positive environment to work in. Thanks to Dr. Simon Gruber for taking the time to train me during my rotation. Lab members Dr. Tory Schaaf, Peter Martin, Yahor Savich, Mark
i Rustad, and Evan Kleinbohl have made the overall experience interesting and enjoyable, and I am grateful for that. I have had many interesting and impactful conversations with Dr. Ben Binder, Mike Fealey, and Prachi Bawaskar.
Three undergraduates, Skylar Faussner, Kevyn Janicek, and Paul Thanel, have assisted me over the years, and each showed a dedication to lab that is inspirational. I have always been passionate about teaching and they gave me a chance to put the passion into practice. I wish them the best in their future endeavors!
ii Dedication To Susanna Jane Huggenberger
iii Abstract
Cardiac muscle contraction and relaxation is controlled by changes in intracellular
Ca, indicating that Ca transport is a fundamental regulator of proper muscle function in the heart. The primary cardiac Ca transporter is the sarcoendoplasmic reticulum Ca-
ATPase 2a (SERCA2a), a transmembrane protein pump embedded in the sarcoplasmic reticulum (SR). SERCA2a translocates two Ca ions from the cytosol into the SR at the cost of one ATP molecule, effectively lowering the cytosolic Ca concentration and inducing cardiomyocyte relaxation. Ca transport is regulated by a second transmembrane protein, phospholamban (PLB), which binds to SERCA2a and inhibits its activity by reducing the Ca affinity of the Ca pump. The significance of PLB-dependent regulation in muscle function is highlighted by the existence of hereditary mutations in
PLB linked to cardiomyopathy. Further, the membrane protein complex between
SERCA2a and PLB is a validated therapeutic target for reversing cardiac contractile dysfunction in more common diseases (e.g., heart failure) caused by aberrant calcium handling. However, efforts to develop compounds with SERCA2a-PLB specificity have yet to yield an effective drug.
The work presented in this thesis focuses on the structure and function of the
SERCA2a-PLB complex and its connection to several chronic cardiac diseases. In the first study (Chapter 4), we developed a structure-based high-throughput screening (HTS) method to discover compounds that disrupt the SERCA2a-PLB interaction. We identified ten compounds that reproducibly alter SERCA2a-PLB structure and function, including one compound that increases SERCA2a calcium affinity in cardiac SR membranes but not in skeletal. These results suggest that the compound is acting specifically on SERCA2a-
PLB, as needed for a drug to mitigate deficient Ca transport in heart failure. In the second study (Chapter 5), we studied the effects of the disease-causing R9C-PLB mutation in a iv human induced pluripotent stem cell (hiPSC) line to understand the mechanistic details of hereditary cardiomyopathy progression. In addition to blunted lusitropic response to β- adrenergic stimulation, we found that the R9C mutation results in an altered metabolic state and significant gene expression changes. Ongoing studies (Chapter 6) are presented, focusing on recent work to develop PLB-based gene therapy constructs and elucidate the structural mechanism by which a third transmembrane protein, dwarf opening reading frame (DWORF), regulates Ca transport in the heart.
v Contents List of Figures ...... vii List of Equations ...... viii List of Tables ...... ix Chapter 1 – Calcium Signaling in Cardiac Muscle ...... 1 1.1 The Role of Calcium in Excitation-Contraction Coupling ...... 1 1.2 The SR Ca-ATPase ...... 4 1.3 Phospholamban and DWORF ...... 8 1.4 SERCA-PLB and SERCA-DWORF Regulatory Models ...... 12 Chapter 2 – SR Ca-ATPase Activity in Healthy and Diseased States ...... 16 2.1 Altered SR Ca Cycling in Cardiac Disease ...... 16 2.2 Therapeutic Approaches for Heart Failure ...... 17 2.3 Inherited Cardiomyopathies Caused by PLB Mutations ...... 24 Chapter 3 – Principles of Fluorescence ...... 25 3.1 Phenomena of Fluorescence and Phosphorescence ...... 25 3.2 Fluorescence Resonance Energy Transfer...... 27 3.3 Steady-State and Time-Resolved Fluorescence ...... 30 Chapter 4 – Targeting Protein-Protein Interactions for Therapeutic Discovery via FRET- Based High-Throughput Screening in Living Cells ...... 34 4.1 Outline ...... 35 4.2 Introduction ...... 35 4.3 Methods ...... 39 4.4 Results ...... 46 4.5 Discussion ...... 55 4.6 Supplementary Information ...... 61 Chapter 5 – Functional and Transcriptomic Insights into Pathogenesis of R9C Phospholamban Mutation using Human Induced Pluripotent Stem Cell-Derived Cardiomyocytes ...... 70 5.1 Outline ...... 71 5.2 Introduction ...... 71 5.3 Methods ...... 74 5.4 Results ...... 76 5.5 Discussion ...... 86 5.6 Supplementary Information ...... 91 Chapter 6 – Summary and Future Directions ...... 106 References ...... 108
vi List of Figures
Figure 1. Ca transport in ventricular myocytes...... 2 Figure 2. Structure of SERCA in E2 state...... 5 Figure 3. SERCA enzymatic cycle...... 7 Figure 4. Functional effects of PLB and pS16-PLB on SERCA function...... 9 Figure 5. Structures of two disparate PLB conformers...... 10 Figure 6. Homo-oligomers and hetero-oligomers of PLB/SERCA...... 11 Figure 7. Model depicting the heptad repeats for PLB monomer and pentamer...... 12 Figure 8. Ca-ATPase enzymatic scheme ...... 13 Figure 9. Oligomeric interactions of SERCA and PLB...... 14 Figure 10. Structural model of DWORF-SERCA interaction...... 15 Figure 11. Contraction and Ca transient measurements in cardiomyocytes...... 17 Figure 12. Rodent heart failure model overexpressing SERCA2a...... 19 Figure 13. SERCA2a gene therapy trial...... 20 Figure 14. Effects of PLB mutations on the Ca-affinity of SERCA...... 23 Figure 15. Jablonski diagram displaying transitions between electronic states...... 26 Figure 16. Illustration of Stokes shift...... 27 Figure 17. Jablonski diagram depicting FRET...... 28 Figure 18. Effect of relative orientation of donor and acceptor dipoles...... 29 Figure 19. Relationship between FRET efficiency (E) and distance (R/R0)...... 30 Figure 20. Steady-state and time-resolved fluorescence measurements...... 31 Figure 21. Differences in precision between fluorescence measurements...... 33 Figure 22. Structure-based HTS to target SERCA2a-PLB complex...... 38 Figure 23. Characterization of SERCA2a-PLB biosensor...... 47 Figure 24. Effects of Ca and Tg on SERCA2a-PLB biosensor...... 48 Figure 25. High-throughput screen results...... 50 Figure 26. Effects of SERCA activators identified in this screen...... 51 Figure 27. Functional effects of SERCA activators varying Ca concentration...... 52 Figure 28. Ca-ATPase activity for SERCA inhibitors ...... 54 Figure S29. Model testing for structural state model...... 64 Figure S30. Chemical structures of SERCA2a-PLB biosensor hits ...... 65 Figure S31. FRET concentration response for SERCA inhibitors ...... 66 Figure S32. SSR Ca-ATPase activity for SERCA inhibitors ...... 67 Figure S33. Uncropped blot images of SERCA2a-PLB biosensor...... 68 Figure S34. Reproducible hits assessed across triplicate screens...... 69 Figure 35. R9C hiPSC-CMs show a blunted response to β-agonists ...... 77 Figure 36. R9C hiPSC-CMs show abnormal calcium handling ...... 79 Figure 37. R9C PLN hECTs have a blunted β-agonist response...... 80 Figure 38. R9C hiPSC-CMs exhibit a hypertrophic phenotype...... 82 Figure 39. Transcriptional response to R9C PLN of induced cardiomyocytes ...... 83 Figure 40. Profibrotic phenotype of R9C PLN of induced cardiomyocytes ...... 84 Figure 41. R9C PLN results in perturbation of miRNAs linked to cardiac metabolism ....85 Figure S42. CRISPR-insertion of the R9C PLB mutation in hiPSCs...... 103 Figure S43. Quantitative PCR analysis...... 104 Figure S44. Heatmap showing perturbation of metabolic pathways in R9C PLB...... 105 Figure 45. DWORF purification...... 107
vii List of Equations
Equation (1) ...... 25 Equation (2) ...... 26 Equation (3) ...... 27 Equation (4) ...... 28 Equation (5) ...... 29 Equation (6) ...... 29 Equation (7) ...... 30 Equation (8) ...... 31 Equation (9) ...... 32 Equation (10) ...... 32 Equation (11) ...... 42 Equation (12) ...... 43 Equation (13) ...... 43 Equation (14) ...... 43 Equation (15) ...... 44 Equation (16) ...... 44 Equation (17) ...... 44 Equation (18) ...... 44 Equation (19) ...... 44 Equation (20) ...... 44
viii List of Tables
Table S1. Concentration response for SERCA2a-PLB FRET……...... 61 Table S2. Concentration response for CSR Ca-ATPase Assay……………………. 62 Table S3. Concentration response for SSR Ca-ATPase Assay……………………. 63
ix Chapter 1 – Calcium Signaling in Cardiac Muscle
1.1 The Role of Calcium in Excitation-Contraction Coupling
Cardiac muscle contraction on the cellular level is a complex and coordinated series of events, initiated by electrical excitation at the sarcolemma of the myocyte from an associated neuron (e.g., pacemaker cells), which depolarizes the outer membrane (label
1 in Figure 1).1,2 In cardiac myocytes, deep invaginations in the sarcolemma bring the outer membrane into close proximity to the sarcoplasmic reticulum (SR), providing a localization interface between these two membrane and their associated proteins.3
Depolarization of the sarcolemma triggers several voltage-gated channels to open, the predominant species being the dihydropyridine receptor (DHPR), which allows for passive transport of calcium ions into the myocyte (label 2 in Figure 1).4 The rapid efflux of calcium (Ca) ions into the cleft between the sarcolemma and SR initiates opening of the ryanodine receptor (RyR), a Ca channel embedded in the SR, in a process known as
Ca-dependent Ca release (label 3 in Figure 1).5,6 The released Ca diffuses into the bundles of interlacing thick and thin filaments that form the myofilaments. Although Ca acts as the switch for myofilaments, contraction is graded and depends on the free Ca concentration in the cytosol, myofilament Ca sensitivity, and other factors. It is also worth noting that the 100:1 ratio of bound:free Ca is indicative of powerful cytosolic Ca buffering.7
The myofilaments, made up of actin, myosin, and the troponin/tropomyosin complex, are responsible for the mechanical force generation needed for simultaneous contraction of the myocyte and cardiac muscle. Actin and myosin readily associate to form a complex in the absence of other regulatory factors; however, in the relaxed muscle state 1 1 2 3 4
SERCA 4
Myofilament
Figure 1. Ca transport in ventricular myocytes. As the sarcolemma is depolarized (1), Ca influx occurs through DHPR channel (2). The rise in local [Ca] in the cleft leads to Ca-dependent Ca release from the SR through the RyR channel (3). Ca will then bind to troponin C, allowing contraction. Increases in cytosolic Ca activate SERCA2a, which transports Ca into the SR (4). Na/Ca exchanger (Na-CaX) also transports Ca out of the cytosol (4). Decreased cytosolic Ca causes relaxation as the actin-myosin binding is altered by troponin I. SERCA function is regulated by phospholamban (PLB). Adapted from Bers, D.M. Nature. 2002. the myosin-binding interface on actin is sterically blocked by the troponin/tropomyosin complex on the thin filament.8,9 Ca binding to the troponin/tropomyosin complex triggers a series of structural changes that alter the position of the tropomyosin relative to the actin filaments to expose the myosin-binding surface.10 This initiates mechanical contraction as myosin in the thick filaments successively binds and generates force onto the thin filaments, drawing opposing ends of the sarcomere together.11,12 Further, contraction force as a function of the free Ca is nonlinear, due to strong myofilament cooperativity with respect to Ca.7
2 As the heart fills with blood (diastole), myofilament Ca sensitivity is significantly enhanced by myofilament stretching, an important regulatory mechanism by which the heart adjusts to varying diastolic filling (i.e., the Frank-Starling response).13
Muscle relaxation requires the depletion of Ca from the cytosol to the SR
(sequestration) or out of the cell (efflux), a process that is driven primarily by the sarcoendoplasmic reticulum ATPase (SERCA) and the Na-Ca exchanger (NCX), respectively (label 4 in Figure 1).14 The contribution of these two Ca-removal mechanisms has been quantified and 70% of Ca ions removed is attributable to SERCA, compared to 28% by NCX in human, rabbit, dog, cat, ferret and guinea pig.7 The activity of SERCA is higher and NCX activity is lower in mouse and rat relative to the aforementioned species, resulting in a balance of 92% for SERCA and 7% for NCX.7 As
SERCA is the major driver of Ca sequestration in cardiomyocytes, its activity is a key determinant of the rate of relaxation of the heart, and influences cardiac contractility by regulating the size of the luminal Ca store that is available for release in the next beat.
The depletion of calcium as a result of the actions of SERCA and NCX decreases the ratio of Ca bound: unbound troponin/tropomyosin complex and revert the myofilament to its blocked state. This drives the contracted muscle towards relaxation as sarcomere length increases.
Lastly, key factors regulating calcium during cardiac excitation-contraction coupling include phospholamban (PLB) and dwarf open reading frame (DWORF), both single pass transmembrane proteins abundantly expressed and localized in the SR of cardiac myocytes.15,16 Monomeric PLB binds SERCA to form a hetero-dimer, and reversibly inhibits Ca-ATPase activity by decreasing the apparent Ca-affinity of SERCA.17 DWORF is proposed to compete with PLB with respect to SERCA binding, though this model is controversial.16
3
1.2 The SR Ca-ATPase
SERCA is part of the P-type ATPase family, named for the autophosphorylation intermediate present during the enzymatic cycle of all members of the family.18 This autophosphorylation event occurs on a conserved aspartate residue. The majority of P- type ATPases catalyze cation transport, and the translocation process is based on cyclical changes between two conformational states (E1 and E2). SERCA in particular has been the focus of structural and functional studies (reviewed extensively by
Toyoshima19 and MacLennan20,21), and serves as the paradigm for the P-type family of cation transporters.
With respect to the SR Ca-ATPase, there are three paralogs (SERCA1-3) in mammals encoded in disparate genes, which are differentially expressed according to developmental stage and tissue-type.21 The gene products are also subject to alternative splicing producing multiple isoforms. The best characterized isoform is SERCA1a, encoded by ATP2A1, which is expressed in adult skeletal muscle. The most abundant species expressed in cardiac tissue is SERCA2a, which shares 84% sequence identity with SERCA1a, and possesses a similar Ca affinity and specific activity. SERCA2b, is a splice variant of the ATP2A2 gene and is highly expressed in smooth muscle. SERCA2a and SERCA2b (splice variants) are identical for the N-terminal 993 amino acids, while
SERCA2b has a different C-terminus encoding an additional transmembrane helix.
Despite its large size (997 residues), SERCA2a is highly conserved across mammalian species, with less than nine residue differences between humans as compared to pigs, dogs, rabbits or mice.
4
Figure 2. Structure of SERCA in E2 state. The cytoplasmic actuator (A, yellow), nucleotide- binding (N, red) and phosphorylation (P, blue) domains form the headpiece. The autophosphorylation site (teal space filling) is found in the P domain. M1-M3 (orange) and M4-10 (tan) make of the transmembrane (TM) domain. Adapted from Olesen, C. Science. 2004.
SERCA2a is composed of a ten transmembrane domain, which controls Ca binding and translocation, and a cytosolic ‘headpiece’ that regulates ATP binding, autophosphorylation and dephosphorylation (Figure 2). The headpiece contains three domains: the nucleotide (N) binding domain, the phosphorylation (P) domain and the actuator (A) domain. In the context of the E1/E2 model, SERCA cycles between high Ca affinity (E1) and low Ca affinity (E2) conformations. For each ATP hydrolytic event two
Ca ions are transported into the SR lumen, and two or three protons into the cytosol
(antiport).22 5 The enzymatic cycle begins as SERCA exchanges bound protons for two cytoplasmic Ca ions in the non-phosphorylated E2 state (Figure 3). Ca binds in two disparate, proximal Ca binding sites. ATP binds in the N domain to form Ca2E1·ATP.
After ATP binding, rearrangement of the headpiece brings an aspartate residue in the P domain (Asp351 in SERCA2a) in close proximity to the ATP binding pocket. Further, a shift in the TM domain occludes cytoplasmic access to the Ca binding sites, preventing release of the bound Ca back into the cytosol. The bound ATP is hydrolyzed and the aspartate is phosphorylated to form the phosphoenzyme Ca2E1P·ADP. The ADP is released as SERCA transitions to the low-affinity Ca2E2P. The Ca binding sites are then exposed to the SR lumen, and the ions are exchanged for protons. Lastly, dephosphorylation of HnE2P (catalyzed by phosphatase activity possessed by the A domain) returns SERCA to the high-affinity E2 state and recycles the Ca-ATPase for the next enzymatic cycle.
SERCA structures resolved by x-ray crystallography include a number of SERCA1a, structures assigned as distinct intermediates during the enzymatic cycle,23-27 SERCA1a in complex with regulators (e.g., PLB),28 and the cardiac isoform SERCA2a.29 The structures of SERCA1a were resolved by the Toyoshima laboratory in a Ca-bound
(E1Ca)30 and a Ca-free state (E2),31 and represent the first high-resolution structures of a P-type ATPase. Subsequent structures trap SERCA in complex with nucleotide analogs mimicking ATP or ADP. Taken together, these static structures suggest that there is significant repositioning of several TM helices and the positions of the cytoplasmic domains during the E2-E1 transition.
Evidence of the putative binding surface of PLB is suggested in a crystal structure of
SERCA1a in complex with PLB without Ca bound (E2-PLB).28 The TM helix of PLB is positioned near the M2, M4, M6 and M9 helices of TM domain of SERCA, and this is
6
Figure 3. SERCA enzymatic cycle. SERCA transitions from low calcium affinity (E2) to high calcium affinity (E1) conformers, and involves large structural rearrangements of the four domains: Nucleotide-biding (red), Phosphorylation (blue), Actuator (yellow), and Transmembrane (gray, beige, green, purple). ATP is hydrolyzed to autophosphorylate aspartate 351, reducing the SR-ATPase calcium affinity. Calcium is then released into the SR, ADP is release, and dephosphorylated. Adapted from Moller, J.V. Q Rev Biophys. 2010.
7 corroborated by co-immunoprecipitation-based binding studies mapping the interaction.32
Lastly, the SERCA2a crystals shows few to no significant structural differences between SERCA2a and SERCA1a (despite ~160 amino acid differences in mammals), and this is consistent with models of a highly conserved Ca transport mechanism between the two isoforms.29 Further, the SERCA2a structure suggests that the binding of PLB is similar to the SERCA1a-PLB complex.29
1.3 Phospholamban and DWORF
PLB is a regulatory peptide consisting of 52 residues and three structural domains.33
Domain 1a (residues 1-16) is an amphipathic helix that associates with both the cytosol and the polar headgroups of the SR membrane. A loop (residues 17-22) connects domain 1a to a single transmembrane helix that peaks above the membrane surface
(domain 1b, residues 22-30) and transverses the SR membrane (domain 2, residues 30-
52). PLB is expressed almost exclusively in ventricular cardiomyocytes,34 and regulates
Ca transport by binding and inhibiting the SERCA2a pump. During diastole and systole, the intracellular Ca concentration oscillates between 0.1 and 1 µM (pCa 7 to pCa 6).7
PLB does not alter basal or maximal SERCA2a activity, rather it inhibits ATPase activity within this range of Ca concentration by reducing the apparent calcium affinity (pKCa)
(Figure 4).15 PLB has three phosphorylation sites at S10, S16, and T17 that are phosphorylated by protein kinase C (PKC),35 protein kinase A (PKA),36 and calmodulin kinase II (CamKII),37 respectively. Phosphorylation at S16 reduces PLB inhibitory potency and is considered the most physiologically relevant as it is directly under adrenergic control.15
8 pS16-PLB
Figure 4. Functional effects of PLB and pS16-PLB on SERCA function. SERCA activity in the absence of PLB (black, -PLB) is dependent on [Ca]. Inhibition by PLB (red, +PLB) results in rightward shift of the Ca-dependence curve, indicating a decrease in Ca-affinity. This effect is reversed through phosphorylation at S16 (blue, pPLB).
NMR structures of PLB show that the phosphorylation site at S16 is located at the end of the cytoplasmic helix (domain 1a) near the loop, allowing for access for PKA.33,38
PLB exists in at least two structural states, a tense (T) structure where the cytoplasmic helix is ordered and associated with the SR membrane surface and a relaxed (R) structure where the cytoplasmic helix becomes disordered (Figure 5).39,40 The TM domain of the protein remains essentially unchanged between the T and R states.41 S16 phosphorylation engenders an ordered-to-disordered transition of the cytoplasmic helix, and shifts the equilibrium between the T/R state towards the R structure.42,43
Phosphomimetic (S16E) PLB with a lipid anchor attached to the N-terminus
(effectively forcing PLB into T state) has similar inhibitory potency as non- phosphorylated PLB, demonstrating uncoupling of S16 phosphorylation from inhibition relief.41 Lipid bilayer composition also plays a role in the T/R equilibrium and PLB inhibition potency.44,45 Introduction of lipids with anionic head groups increases the T population and SERCA inhibition, while cationic lipids have the opposite effects.45
9 Cytoplasm
Dephosphorylation SR membrane Phosphorylation
SR lumen T State R State
Figure 5. Structures of two disparate PLB conformers. PLB – an integral membrane protein – is inserted in the SR via a single-pass TM helix comprised of domain Ib (beige) and domain II (blue). PKA-dependent phosphorylation induces structural changes including increased disorder of domain Ia (green) of the cytoplasmic helix and partial unfolding of domain Ib. The R state has a lower inhibitory potency. High-resolution structure determined by NMR.38
Self-association of PLB monomers leads to the formation of PLB pentamers (Figure
6).36 The helical structure of the TM domain of PLB generates heptad repeats that place
leucines or isoleucines on one ‘face’ of the helix (Figure 7).46 Mutation of these
hydrophobic residues destabilizes the pentamer suggesting a model where a leucine
zipper motif drives the self-interaction.47 PLB is in dynamic equilibrium between
homopentamers and monomers, where the oligomeric state is proposed to act as an
inactive reservoir.48,49 Also, there is a strong correlation between pentamer-destabilizing
mutations replacing leucines and isoleucines along the zipper and PLB inhibition
potency, indicating that the monomeric form is regulating SERCA function.46
10
Figure 6. Homo-oligomers and hetero-oligomers of PLB/SERCA. PLB is in equilibrium between a pentameric species (PLB5), monomeric (PLB1), and SERCA-bound (PLB:SERCA). The transitions are described as KD1 and KD2. Adapted from Robia et. al., Circ. Res. 2007.
A monomeric mutant of PLB can be generated by mutating 3 cysteines (C36A,
C41F, and C46A) in the transmembrane domain.50 The structure and function of this
PLB variant is indistinguishable from those of monomeric wild-type PLB.51 Disulfide bridges have not been observed involving these cysteines and replacement of the cysteines with an isosteric amino acid (α-amin-n-butyric) does not affect pentamer stability, suggesting that PLB oligomerization requires the steric properties of these 3 cysteines rather than their reactivity.50
DWORF is also a Ca-transport regulatory peptide comprised of 35 residues.
Although its structure is not determined, DWORF is predicted to contain a TM helix based on its sequence.16 DWORF is expressed in cardiac tissue, localizes to the SR, and interacts with SERCA2a.16 Overexpression in mouse models increases peak Ca transient amplitude and Ca clearance, while ablation of DWORF expression leads to decreased Ca clearance consistent with a model where DWORF activates SERCA2a function.16
11 A B
Figure 7. Model depicting the heptad repeats for PLB monomer and pentamer. (A) The transmembrane helix of PLB (depicting residues 37-52) is structured with 3.5 residues per helix turn with positions a through g of the heptad repeat circled. Isoleucines and leuicines are localized to positions a and d. (B) Five PLB monomers are arranged with intermolecular contacts at positions a and d. Adapted from Simmerman et. al., J Biol. Chem. 1996.
1.4 SERCA-PLB and SERCA-DWORF Regulatory Models
The molecular mechanism by which PLB regulates Ca-ATPase activity hinges on
several questions. Upon binding to SERCA2a, how does PLB disrupt or alter its
enzymatic cycle? Does it slow the rate of Ca or ATP binding, ATP hydrolysis and
phosphorylation, ADP release, Ca release, dephosphorylation and Pi release, or
structural transitions and how is this manifested in the three-dimensional structural
models? Does PLB remain bound post-phosphorylation and how does phosphorylation
lead to relief of SERCA inhibition? These questions have been approached by a number
of studies measuring the effect of PLB on the partial reactions of the Ca- ATPase cycle
(reviewed extensively by Inesi14 and Jones52). Evidence has been provided suggesting
that PLB controls Ca-ATPase by decreasing the rate of phosphoenzyme decomposition
(step 7 in Figure 8).53-55 However, conflicting results have been reported demonstrating 12 • •
Figure 8. Ca-ATPase enzymatic scheme phosphate release is unchanged by PLB and that PLB decreases the Ca binding to the
E1 intermediate and/or a conformational change following binding of the first Ca ion
(step 1 in Figure 8).17,56,57 The high-resolution structure of PLB bound to SERCA1a shows that the Ca binding sites are partially collapsed, relative to the unbound structure.28 This may provide a structural explanation in support of the kinetics data where PLB decreases Ca binding to the E1 intermediate
Some cross-linking studies suggest that high [Ca] and PKA-dependent PLB phosphorylation increases SERCA activity via dissociation of the complex.58-60 However, several lines of evidence, based on FRET and co-immunoprecipiation, demonstrate that the SERCA-PLB complex is intact following PKA phosphorylation58,61-63 and micromolar
[Ca].63-65 This is consistent with an allosteric regulation or ‘Subunit’ model, wherein phosphorylated PLB bind to SERCA2a in a non-inhibitory mode (Figure 9). According to this model, PLB remains SERCA-bound throughout the Ca-ATPase cycle, and adopts disparate inhibitory and non-inhibitory conformations. Further, the equilibrium is shifted toward the non-inhibitory structure following phosphorylation. These confirmations may correspond to the T and R states observed for PLB via EPR and fluorescence.66
13 Active Inhibited
Dissociation P SERCA Model oligomers
P
P PLB oligomers Subunit Model
Figure 9. Oligomeric interactions of SERCA and PLB. SERCA and PLB can form homo- oligomers (left), and bind to form an inhibited heterodimer (center). Phosphorylation of PLB relieves inhibition (right), which may occur through dissaocation of the SERCA-PLB complex (Dissociation) model, or structural rearrangements within the complex (Subunit Model).
DWORF is recently discovered by the Olson laboratory to be a putative regulator of
Ca homeostasis. DWORF is expressed in the heart and soleus, and co-localizes with
SERCA in the SR.16 The two proteins form a complex as evidenced by co-
immunoprecipitation experiments overexpressing tagged versions of the proteins in cell
culture models.16 This led to the hypothesis that DWORF can affect Ca homeostasis by
regulating the Ca-ATPase pump. This hypothesis is supported by in vivo data showing
that ablation of the DWORF gene in mouse models causes a modest decrease in Ca
transients and SERCA activity at physiological calcium concentrations, while
overexpression of DWORF has the opposite effects.16 These results suggest that
DWORF activates SERCA2a function. With respect to the mechanism, the Olson group
proposes that DWORF displaces inhibitory PLB to increase SERCA2a activity. This and
other models are evaluated in Chapter 6 (Figure 10).
14 Inhibited Competitive Active Model
Inhibited Allosteric Active Model
Direct Enhanced Active Activation Activity Model
Figure 10. Structural model of DWORF-SERCA interaction. (top) DWORF and inhibitory PLB compete for the same binding interface of SERCA2a. (middle) The inhibitory potency of PLB is reduced by an allosteric effect upon DWORF binding to the complex. (bottom) DWORF directly activates SERCA2a Ca-ATPase activity.
15
Chapter 2 – SR Ca-ATPase Activity in Healthy and Diseased States
2.1 Altered SR Ca Cycling in Cardiac Disease
During the turn of the century, there was accumulating evidence that changes in the excitation-contraction coupling (see Section 1.1) play a significant role in the pathophysiology of heart failure (reviewed extensively by Hajjar67 and Hasenfuss68). The failing heart exhibits perturbation in both systolic and diastolic function. Specifically, myopathic human hearts removed during cardiac transplantation have a prolonged Ca transient, an elevated diastolic intracellular Ca, and decreased systolic [Ca] and contractility.69 This finding corroborated previous results in animal models.70 This experiment was repeated with recordings from patient-derived single isolated cardiomyocytes loaded with the fluorescent Ca indicator Fura-2 with similar results
(Figure 11).71-73
Because these changes in calcium handling could result from altered protein expression, levels of relevant proteins were quantitated via western blot in failing and nonfailing human myocardium. Expression of the DHPR channel,74-76 RyR channel,77-79 and calcium storage proteins (i.e., calsequestrin80 and calreticulin77) are similar in diseased and healthy heart tissue. However, steady-state levels of proteins involved in removal of cytosolic Ca were significantly altered in patients with heart failure. In particular, expression of SERCA2a was decreased along with the SERCA2a:PLB ratio.81-83 Further, NCX expression was increased, possibly as a compensatory mechanism caused by increased diastolic Ca seen in heart failure (Figure 11).84-86
Increased NCX activity may also compensate for the reduction in SR Ca-ATPase activity
16
Figure 11. Contraction and Ca transient measurements in heart failure and healthy cardiomyocytes. Recordings from human cardiomyocytes isolated from failing heart (left) or healthy donor (right) stimulated at 1 Hz. Failing heart myocytes have a characteristic decrease in contraction (top) and prolonged Ca transient (bottom). Figure adapted from del Monte, F.et al., Circulation, 1999. but at the cost of reduced Ca release from the SR during systole. This is often tested in cells and tissues by higher rates of stimulation (i.e., pacing) and manifests as a blunted frequency response, which is commonly seen in failing myocardium.87 Taken together, this evidence supports the hypothesis that altered calcium handling is present in heart failure and contributes to disease phenotypes (e.g., decreased force generation).
2.2 Therapeutic Approaches for Heart Failure
Ca uptake abnormalities have been demonstrated both in animal models of heart failure and in human patients, and this may account for the impairments of contraction and relaxation rates. Given that Ca transport into the SR is the predominant pathway for
Ca removal (section 1.1), it is proposed that increasing Ca uptake – by enhancing the 17 function of the SR-Ca ATPase – represents a strategy to reverse these defects. Two approaches have been initiated to increase SERCA2a activity: SERCA2a overexpression and PLB inhibition.
In a cultured neonatal rat myocyte model of Ca uptake dysfunction, overexpression of SERCA2a shortened the relaxation phase of the Ca transient, increased SR Ca release, and decreased resting (diastole) Ca.88 More importantly, in human cardiomyocytes isolated from patients with end-stage heart failure, SERCA2a gene transfer caused an increase in protein expression and activity, and faster contraction and relaxation velocity.73 With respect to Ca handling, cardiomyocytes overexpressing
SERCA2a showed increased systole Ca levels and decreased diastole levels as well as a normalized frequency response upon increasing rates of stimulation.73 These encouraging in vitro results prompted in vivo studies that reflect the more complex behavior of the intact heart.
Aortic constriction (banding) of rats serves as a robust animal model for heart failure.89 In this model severe contractile dysfunction is present and moribundity or morbidity occur within 30 days due to insufficient heart function. Overexpression of
SERCA2a via gene transfer in vivo restored both systolic and diastolic function, and survival was increased significantly (63% vs 9% after 30 days) in animals which received gene transfer relative to control groups (Figure 12).90 Restoration of cardiac function via
SERCA2a overexpression was recapitulated in a large animal volume-overload heart failure model.91,92 Collectively, these studies provide strong evidence that SERCA2a overexpression rescues altered Ca handling and cardiomyocyte dysfunction seen in heart failure, and suggests the feasibility of this approach as a therapeutic modality.
18 Sham
Failing +SERCA2a
Percentage Survival Percentage Failing Failing +GFP
Days Figure 12. Kaplan-Meier survival curve for rodent heart failure model testing SERCA2a overexpression. By 26 weeks after aortic banding, all animals develop heart failure. Sham- operated rats, or aortic banded rats uninfected or infected with either Ad.GFP or Ad.SERCA2a shown in this survival plot. SERCA2a overexpression increased survival significantly (63% vs 9%) as compared to controls. Figure adapted from del Monte, F.et al., Circulation, 2001.
The substantial beneficial results seen in animal model testing led to development and administration of clinical trials in human patients experiencing end-stage heart failure targeting SR Ca-ATPase activity. Despite efforts from several groups (e.g., Merck
Inc.), pharmacological targeting of SERCA2a or PLB had yet to yield agents of high specificity for either of these targets at this time; thus, gene therapy remained the most viable option for the trial.
In the clinical trial – Calcium Upregulation by Percutaneous Administration of Gene
Therapy in Cardiac Disease (CUPID) – patients with advanced heart failure were administered gene therapy using an adeno-associated type 1 (AAV) vector encoding full- length SERCA2a.93 The safety profile in these patients was positive and led to initiation of a phase 2 trial.94 The primary endpoint was defined as improvement of several efficacy parameters over a 6-month period, and a total of 39 patients with advanced HF were randomized to intracoronary infusions of AAV1.SERCA2a at three different doses.
19
1.6 Placebo )
1.2
cumulative 0.8 High dose AAV1-SERCA2a
Events ( Events 0.4 Rate Cardiovascular of
0.0 6 12 18 24 30 Months
Figure 13. SERCA2a gene therapy trial. Phase 2 cumulative clinical event rates during first three years. Shown is the cumulative rate of recurrent cardiovascular events estimated using the joint frailty model for patientsin AAV1/SERCA2a high-dose and placebo groups at three years. Figure adapted from Zsebo, M. et al., Circulation Res., 2014.
There was a trend towards a reduction of clinical events in the AAV1.SERCA2a treated groups,95 as well as a significant reduction in cardiovascular events during this period in the high-dose group relative to the placebo group (Figure 13).96 Subsequently, a larger study was designed to assess effects of AAV1.SERCA2a at the high dose in patients with advanced HF. The study included 250 patients randomized 1 treatment: 1 placebo in >50 centers in the US and Europe. The primary endpoint (time to recurrent HF-related events) was not statistically different between the experimental and control group, and the study failed to demonstrate the efficacy of the SERCA2a gene transfer at that particular dosage.97 Failure to meet its primary end goals was partially explained by dosage constraints and fundamental limitations of rAAV gene therapy; e.g., in patients with pre-existence of neutralizing antibodies.98
An alternative strategy is to administer an agent (e.g., small-molecule compound) that increases Ca transport in heart failure patients. The development of this type of 20 agent has recently become a focus in our lab. The initial approach was to measure changes in FRET between purified SERCA and PLB labeled with fluorescent dyes in a reconstituted membrane system.99 A 20,000 compound library was screen against this sample using a fluorescence intensity-based plate reader. These efforts yielded compounds that increased SERCA activity at high Ca, although the mechanism appeared to be independent of PLB. We also developed an intramolecular FRET biosensor to measure structural changes of SERCA specifically by attaching GFP-RFP to the N-terminus and an unstructured loop in the N-domain (2-color SERCA).100 This biosensor was also applied to high-throughput screening of compounds.101,102 Here, work will be presented on a novel drug discovery platform that leverages high-precision measurements made possible by time-resolved fluorescence plate readers and a FRET- based biosensor sensitive to structural perturbation or binding dynamics of SERCA-PLB
(Chapter 4).
Efforts have also been made to target PLB inhibition in order to increase Ca uptake in cardiomyopathies (reviewed extensively by MacLennan and Kranias).15 The strategies include expression of PLB antisense RNA103, aptmers104, and non-inhibitory PLB mutants64,105 to decrease PLB steady-state levels and/or inhibitory potency. The most striking example is an experiment in which the non-inhibitory, phosphomimetic PLB mutant (S16E) is expressed in a cardiomyopathic hamster model.106 In the model, there is a mutation in the δ-sarcoglycan gene resulting in myocardial degeneration and progressive heart failure.107 Introduction of the S16E-PLB mutant significantly improves cardiac function and attenuates heart failure progression.106
The potential benefits of S16E-PLB gene transfer was also tested in a large animal model.108 In this case heat failure was induced in sheep via rapid ventricular pacing (180 beats/min) for a period of 4 weeks. After induction, animals were randomly selected for
21 delivery of S16E-PLB or LacZ vector via adenovirus. Like the small animal model, animals that received the S16E-PLB gene transfer showed significant improvements in cardiac performance and SR Ca-ATPase activity as compared to the control groups.108
Despite encouraging results, concerns remain that introduction of a phosphomimetic mutant of PLB will abrogate -adrenergic regulation (upstream of PKA-dependent phosphorylation) of cardiac function, leading to detrimental effects upon chronic application. Long-term effects of this therapy remains to be tested in a comprehensive study.
Much of our knowledge of PLB mutations and their effects stems from early studies by the MacLennan group. An example of an experiment is shown in Figure 14. PLB mutants in the TM domain were generated via alanine scanning and co-expressed (with
SERCA2a) in HEK293 cells. The inhibitory potency of these mutant was measured by assaying Ca-ATPase activity and changes in Ca affinity, and the ability to form PLB oligomers was measured by western blot.47
22
Figure 14. Effects of mutations in PLB on the Ca-affinity of SERCA and oligomerization. (top) Kca is the [Ca] at which the rate of Ca uptake is half of the maximal rate. The left dashed line indicates Kca in the absence of PLB (uninhibited) and right dashed line indicates in the presence of PLB (inhibited). (bottom) Immunoblot assaying stability of PLB pentamers. PLBp and PLBm refere to pentameric and monomeric forms of PLB, respectively. Figure adapted from Kimura, Y. et al., J Biol. Chem., 1997.
Some mutations confer loss-of-inhibitory potency (e.g., L31A, N34A), while others confer gain-of-inhibitory potency (e.g., I40A). The gain-of-function mutations can be partially explained an increase in the monomer:pentamer ratio as the monomer is the inhibitory species. Our lab has recently developed FRET-based assays that directly measure the ability of PLB mutants to compete with wild-type PLB for SERCA2a binding in live cells.109 PLB mutants with low inhibitory potency that displace the endogenous
23 PLB could effectively increase SR Ca-ATPase activity. Effects and therapeutic potentials of these PLB mutants are presented in Chapter 6.
2.3 Inherited Cardiomyopathies Caused by PLB Mutations
Several human mutations in the PLB gene have been identified and proffer additional insights in PLB regulation of SERCA2a. A homozygous loss-of-function mutation, L39stop-PLB, results in dilated cardiomyopathy and premature death.110 This mutation produces a truncated form of PLB, resulting in an unstable protein and leaving
SERCA2a in an unregulated state. Two mutations (R9C-PLB and R14del-PLB) are associated with increased PLB inhibition (decreased SERCA2a Ca-affinity).111,112 A recent study investigating the effects of R9C-PLB mutation will be discussed in Chapter
5. The chronic inhibition of SERCA2a by these PLB mutants led to dilated cardiomyopathy and premature death in heterozygous individuals.109,110 These findings were recapitulated in mouse models carrying the same mutations.109,110
24 Chapter 3 – Principles of Fluorescence
3.1 Phenomena of Fluorescence and Phosphorescence
Spectroscopy – in simplest terms – can be defined as the study of the interaction between matter and light. Electrons excited by radiated energy can return to ground state through several disparate pathways, most frequently through vibrational motion or through heat energy expenditure. Certain molecules can emit the absorbed energy as a photon in a process known as luminescence (illustrated by the Jablonski diagram; Figure
15). Luminescence is divided into two types – fluorescence and phosphorescence – depending on the nature of the electronically excited state. In excited singlet states, the electron in the excited orbital (S1 in Figure 15) is paired, meaning it possesses the opposite spin as the second electron in the ground-state orbital (S0 in Figure 15). As a result, the return to ground state occurs rapidly by emission of a photon (fluorescence).
The emission rate of fluorescence (kF in Figure 15) occurs on the nanosecond timescale. Phosphorescence is the emission of light from the triplet excited states, where the electron in the excited orbital (S1) has the same spin orientation as the ground-state electron (S0). This transition to the ground state is forbidden, and requires intersystem crossing resulting in the slow emission rate of phosphorescence (kP).
The probability that an excited electron will emit a photon is determined by the
quantum yield (F ; Equation 1), where kF is the fluorescence emission rate and kNR is the non-radiative transition rate.
kF (1) F . (kF kNR )
25 Vibrational 12 -1 (kV 10 s ) States T1
S1
kA 1015 s-1 k NR kF 108 s-1 108 s-1 Energy Input Phosphorescence S0 -3 2 -1 kP=10 to 10 s
Figure 15. Jablonski diagram displaying transitions between electronic states. Light is absorbed by an electron (blue arrow), causing a transition from the ground state (S0) to the excited state (S1). The excited electron rapidly relaxes (kv) to the lowest S1 vibrational level. Relaxation to the ground state can occur through non-radiative transitions (grey curved arrow), fluorescence (green arrow) or intersystem crossing (yellow curved arrow) followed by phosphorescence (red arrow).
The quantum yield reflects the number of photons emitted (kF) over the number of photons absorbed (kF + kNR), and is dependent on the rates of these relaxation transitions.
The energy of the emission is less than that of absorption (Figure 16). The causes of the energy loss are two-fold: i) the rapid decay (kV) to the lowest vibrational level of S1 and ii) the fact that the excited electron begins at the lowest vibrational level of S0 but the electron will transition to higher vibration levels of S0 as via fluorescence (Figure 15). The energy emitted by a photon (E) is related to both the frequency (ν) and wavelength (λ) of the emitted light through Planck’s equation and Equation 3, where h is Planck’s constant and c is the speed of light:
E hv . (2)
26 Absorption Emission
Red Shift Intensity
Wavelength
Figure 16. Illustration of Stokes shift. The energy of the fluorescence emission is less than that of absorption and at a longer wavelength (Stokes shift).
c (3) . v The decrease in energy results in a lower frequency (v) and longer wavelength (λ). Thus, the emitted light is shifted toward the red end of the visible spectrum relative to the absorbed light in a phenomenon known as the Stokes shift.
3.2 Fluorescence Resonance Energy Transfer
Fluorescence resonance energy transfer (FRET) is an electrodynamic phenomenon that occurs between a donor (D) molecule in the excited state and an acceptor (A) molecule in the ground state (Figure 17). FRET is a non-radiative process caused by long-range dipole-dipole interactions between the electronic systems of the donor and acceptor molecules. The efficiency of energy transfer (E) is defined by the rate of energy transfer (kT) over the sum of kT and the rate of fluorescent decay (kD):
27 S 1 Donor Acceptor Donor Emission k Emission Excitation T
(EmA) (ExD) (FRET)
S0 Donor Accepter Excitation Donor Excitation Accepter
Emission Emission Intensity Wavelength
Figure 17. Jablonski diagram depicting FRET. (top) An excited electron transitions to ground state by transferring energy (gray curved arrow) in a non-radiative process to an acceptor electron. The excited acceptor electron can transition to the ground state through non-radiative procceses (e.g., vibration, heat; not depicted) or by emitting a photon (red arrow). (bottom) The spectrum of the donor excitation and emission, and the acceptor excitation and emission depicted.
kT E (4) (kT kD )
The rate of energy transfer (kT) depends upon four factors. First, the extent of spectral overlap between the donor emission spectrum and the acceptor absorption spectrum. This is defined as J(λ) and is the area under the curve (integral) of overlap between the donor emission and acceptor excitation; illustrated in the bottom section of
Figure 17. Second, the quantum yield of the donor (ΦD; defined in Section 3.1). Third, the relative orientation of the donor and acceptor transition dipoles (κ). For FRET to occur, the emission dipole of the donor and the absorption dipole of the acceptor must
28
Figure 18. Effect of relative orientation of donor and acceptor dipoles. If the emission dipole (red arrow) of a donor (blue cylinder) forms a 90° angle with the absorption dipole (green arrow) of the acceptor (yellow cylinder), FRET cannot occur. Dipole-dipole angles other than 90° will support FRET with a range of efficacies. Adapted from Vogel, S., et al., Sci. Signal., 2006. not be oriented perpendicular to each other (Figure 18). The range for κ2 is between 0 and 4. κ2 is generally assumed to equal 2/3, which is appropriate for dynamic random averaging of the donor and acceptor dipoles. Lastly, the distance between the donor and acceptor molecules (R). This dependence on these factors is described in the following two equations:
6 1 R 0 (5) kT D R
1 R 9,790 J 2 4 6 0 D (6) where τD is the fluorescence decay time of the donor in the absence of the acceptor, R0 is the Fӧrster distance, and η is the index of refraction of the solvent.
If R = Ro (meaning the donor and acceptor are separated by the Fӧrster distance), half of the excited donor electrons will energy transfer to the acceptor. Thus, distance (R) affects the efficiency of energy transfer (E) in the following manner:
29 (E) 6
E = 1/[1 + (R/R0) ]
Efficiency Efficiency FRET FRET
R/R0
Figure 19. Relationship between FRET efficiency (E) and distance (R/R0). The plot shows the dependence of energy transfer on the donor-acceptor distance (R).
6 R0 E (7) 6 6 R0 R
As the donor and acceptor become closer (R 3.3 Steady-State and Time-Resolved Fluorescence When making steady-state fluorescence measurements the phenomenon of FRET manifests in the following way: The intensity of the donor emission will decrease and the intensity of the acceptor emission will increase. The efficiency of energy transfer (E) can be calculated by measuring the fluorescence intensity of a sample that contains the donor molecule (FD) and another sample that contains the donor and acceptor molecule (FDA), and applying equation (8): 30 Steady-state Time-resolved FD 1.0 ) 0.8 R ( ρ FDA 0.6 , 0.4 0.2 D Fluorescence DA Fluorescence Distribution Fluorescence 0.0 0 10 200 30 5 40 1050 1560 2070 1 2 3 4 5 6 7 Time (ns) t (ns) R (nm) Figure 20. Steady-state (intensity) and time-resolved (lifetime) fluorescence measurements. (left) Simulated data depicting changes in fluorescence intensity in Donor-only (FD) sample and Donor-Acceptor (FDA) sample due to FRET. (middle) Simulated data depicting changes in fluorescence lifetime in Donor-only (FD) sample and Donor-Acceptor (FDA) sample due to FRET. (left) Gaussian distribution of distances resolved by fitting the fluorescence lifetime data. F E 1 DA (8) FD If FRET is occurring in the DA sample, there will be a decrease in FDA relative to FD (Figure 20). There are several limitations to this method to consider. This calculation requires that the number and fluorophore environment of donor molecules is unchanged between samples. In practical terms, this is difficult to achieve due to incomplete fluorophore labeling of the donor and acceptor, and the inherent lack of precision in pipet technique.113 This precision can be improved by simultaneously measuring the intensity ratio of the donor emission and acceptor emission.114 However, this ratiometric measurement only reports relative changes in E, and calculation of the actual value of E requires significant calibrations.115 Lastly, the steady-state fluorescence measurement can only report the average distance ( 31 Fluorescence lifetime is the average time an electron exists in the excited state. Multiple lifetimes may be present in the fluorescence decay of a fluorophore. Generally, this is determined empirically and described using the following equation: n F(t) = Ai exp(-t/i) , (9) i1 where F(t) is a function of the fluorescence intensity with respect to time, n is the number of lifetimes, Ai is the amplitude, and τi is the fluorescence lifetime. The fluorescence lifetime is altered by the phenomenon of FRET and this is described in equation (10). E 1 DA (10) D Where τDA is the average lifetime of donor in the presence of the acceptor and τD is the average lifetime of the donor. Thus, if FRET is occurring the τDA < τD (Figure 20). Multiple species can contribute to the fluorescence decay of the FRETing (DA) sample. These species are the decay of the donor-only and multiple structures/distances (Figure 20). The details of this type of analysis is described in the Methods section of Chapter 4. There are several key advantages to using time-resolved fluorescence. Variation in signal intensity does not affect the fluorescence lifetime as it would change fluorescence intensity-based measurement value.113 This is an important consideration for high- throughput screening as many potential therapeutic compounds cause subtle changes in structures that require a high degree of precision. This is illustrated nicely in Figure 21, 32 Figure 21. Differences in precision between lifetime-based and intensity-based fluorescence measurements. Results from an identical plate containing compounds from the LOPAC library using a fluorescence lifetime plate reader and fluorescence intensity plate reader. Two hits were identified (red symbols) that would not be identified by intensity-based screening. Red bar indicates the 3 SD hit selection window. Adapted from Gruber et al., J Biomol Screen, 2014. where differences in precision are compared between the two methodologies on an identical plate of compounds. More importantly, time-resolved fluorescence enables the resolution of multiple structural states and mole fractions within a givensample.113 These measurements can be made in live cells or in solution and more accurately reflect the dynamic behavior of molecules than static structures available through crystallography. 33 Chapter 4 – Targeting Protein-Protein Interactions for Therapeutic Discovery via FRET-Based High-Throughput Screening in Living Cells Daniel R. Stroik1, Samantha L. Yuen1, Kevyn A. Janicek1, Tory M. Schaaf1, Ji Li1, Delaine K. Ceholski2, Roger J. Hajjar2, Razvan L. Cornea1, and David D. Thomas1 1Department of Biochemistry, Molecular Biology, and Biophysics, University of Minnesota, Minneapolis, Minnesota 55455 2Cardiovascular Research Center, Icahn School of Medicine at Mount Sinai, New York City, New York 10029 Originally published in Nature Scientific Reports Vol 8, Article 12560. Reprinted with permission. Author Contribution DRS, DDT, RLC, JL, and TMS conceived, directed and analyzed all experimental research. DRS, DDT, RLC, RJH, and DKC prepared the manuscript. DRS, SLY, and KAJ performed FRET spectroscopy experiments and activity assays. All authors reviewed the manuscript. 34 4.1 Outline We have developed a structure-based high-throughput screening (HTS) method, using time-resolved fluorescence resonance energy transfer (TR-FRET) that is sensitive to protein-protein interactions in living cells. The membrane protein complex between the cardiac sarcoplasmic reticulum Ca-ATPase (SERCA2a) and phospholamban (PLB), its Ca-dependent regulator, is a validated therapeutic target for reversing cardiac contractile dysfunction caused by aberrant calcium handling. However, efforts to develop compounds with SERCA2a-PLB specificity have yet to yield an effective drug. We co-expressed GFP- SERCA2a (donor) in the endoplasmic reticulum membrane of HEK293 cells with RFP-PLB (acceptor), and measured FRET using a fluorescence lifetime microplate reader. We screened a small-molecule library and identified 21 compounds (Hits) that changed FRET by >3SD. 10 of these Hits reproducibly alter SERCA2a-PLB structure and function. One compound increases SERCA2a calcium affinity in cardiac membranes but not in skeletal, suggesting that the compound is acting specifically on the SERCA2a-PLB complex, as needed for a drug to mitigate deficient calcium transport in heart failure. The excellent assay quality and correlation between structural and functional assays validate this method for large-scale HTS campaigns. This approach offers a powerful pathway to drug discovery for a wide range of protein-protein interaction targets that were previously considered “undruggable.” 4.2 Introduction A major goal of drug discovery in recent years is the development of small molecules that target specific protein-protein interactions,116 but there is a growing consensus that such targets are intrinsically difficult to perturb specifically with small molecules.117 In the present study, we focus on the interaction of phospholamban (PLB), a 52-residue single- 35 pass transmembrane protein expressed in the sarcoplasmic reticulum (SR) of cardiac muscle, and its regulatory target SERCA2a, the cardiac SR Ca-ATPase. SERCA2a is responsible for removing Ca from the cytosol into the SR, inducing muscle relaxation.118 In its enzymatic cycle, the Ca-ATPase undergoes a transition from a high Ca affinity (E1) to a low Ca affinity (E2) conformation, with ATP binding and autophosphorylation powering the calcium transport process. PLB is in dynamic equilibrium between homopentamers and monomers, where the oligomeric state is proposed to act as a reservoir.119 Monomeric PLB reduces the apparent Ca affinity of SERCA2a, and this inhibitory effect is relieved by -adrenergic stimulation of PLB phosphorylation, thus providing a Ca-transport reserve to enhance cardiac performance. In heart failure (HF), however, there is a Ca-transport deficit that leads to elevated sarcoplasmic Ca (cytotoxic) and incomplete relaxation and filling of the ventricle (diastolic dysfunction), as well as incomplete SR re-filling with Ca, which blunts Ca release (systolic dysfunction).7 The market is saturated with preload and afterload reducers that provide symptomatic relief, but there is an urgent need for an effective and safe cardiotonic therapy that directly targets deteriorated diastolic and systolic function. Since HF depends on multiple factors and causes, numerous strategies have been developed to mitigate or reverse cardiac dysfunction. A promising approach is to enhance cardiac muscle contractility by modulating Ca transport.120,121 The SERCA2a-PLB interaction is widely viewed as an attractive target for cardiovascular therapeutic discovery and development, to correct the pathophysiological myocyte state and its consequences to cardiac function. It is well established that decreased SERCA2a activity, as seen in HF animal models and human patients, results in slower and less complete muscle relaxation after each contraction.122-125 Recent efforts using gene therapy to increase SERCA2a activity accomplish this either through SERCA2a overexpression or by reducing SERCA2a 36 inhibition by PLB.126,127 SERCA2a activation is tolerated in healthy animal models and significantly enhances cardiac function in numerous models of heart disease.128,129 These results validate SERCA2a activation for HF therapy. SERCA2a overexpression via recombinant adeno-associated virus (rAAV) was achieved in patients experiencing end-stage HF in a recent phase II clinical trial.95 Despite encouraging preliminary results,130 the trial failed to meet its primary end goals, due to dosage constraints and fundamental limitations of rAAV gene therapy; e.g., in patients with pre-existence of neutralizing antibodies.131 We have pursued an alternative approach to activate SERCA2a using small- molecule drugs that decrease SERCA2a inhibition by PLB. This small-molecule approach is designed to overcome limitations associated with gene therapy131 and is amenable to acute, non-invasive hospital intervention with the potential for chronic usage to improve cardiac contractility. As PLB is almost exclusively expressed in the heart,119 compounds that specifically target the SERCA2a-PLB complex will be inherently tissue-specific, thus reducing the risk of adverse side effects. Some previous attempts to discover compounds that activate SERCA have been largely unsuccessful, due to reliance on low-throughput, low-precision assays of ATPase activity.132 We previously reported a fluorescence resonance energy transfer (FRET)-based high- throughput screening (HTS) method using purified SERCA and PLB labeled with fluorescent dyes in a reconstituted membrane system.133 A 20,000-compound screen was performed on this in vitro sample using steady-state (intensity) fluorescence detection and identified the first SERCA activators. Surprisingly, none of these compounds directly affected the SERCA2a-PLB interaction.133 We hypothesize that detection of compounds that disrupt the SERCA2a-PLB requires a more precise detection technique. 37 a b Fluorescence SERCA2a-PLB Biosensor Decay Waveforms CMV GFP SERCA2a τDA CMV RFP PLB FRET = 1 - τD GFP-SERCA only No FRET GFP-SERCA + Fluorescence RFP-PLB k Cytosol 1 k SR -1 SR lumen 0 5 10 15 20 Time (ns) Figure 22. Structure-based high-throughput screening to target SERCA2a-PLB complex. (a) Schematic diagram illustrating expression vectors and structural model of GFP-SERCA2a free or bound to RFP-PLB. (b) Time-resolved fluorescence waveforms as a readout to measure FRET between GFP-SERCA2a and RFP-PLB. FRET is calculated as the change between lifetimes (exponential decay times) of donor-only and donor-acceptor samples (τD and τDA). Here, we introduce an HTS platform that directly monitors the SERCA2a-PLB interaction using time-resolved FRET (TR-FRET) between GFP-tagged SERCA2a and RFP-tagged PLB constructs expressed in a transformed human cell line (Figure 22a). When RFP-PLB (acceptor) is bound to GFP-SERCA2a (donor), FRET is detected as a decrease in donor fluorescence lifetime (FLT), which is calculated from fluorescence decay waveforms (Figure 22b). The FRET measurement is a direct readout of changes in PLB binding to SERCA2a and/or the structure of the SERCA2a-PLB complex. The R-6 distance dependence of FRET makes it sensitive to protein-protein interactions and subtle structural changes. We applied this structure-based SERCA2a-PLB biosensor to drug discovery, by performing a triplicate screen of the Library of Pharmacologically Active Compounds (LOPAC, 1280 compounds), using an FLT plate reader provided by Fluorescence Innovations, Inc.134,135 The FLT measurement yields a 30-fold decrease in 38 the coefficient of variation (CV = standard deviation/mean) compared with intensity detection used in our previous screen,133 resulting in an HTS platform of excellent quality. To assess the effectiveness of this HTS platform, we performed FRET concentration-response and ATPase activity assays, revealing ten Hit compounds that reproducibly affect SERCA2a-PLB structure and function. This strong correlation between structural and functional assays validates this method to discover SERCA2a-PLB effectors for large-scale HTS campaigns. 4.3 Methods 4.3.1 Molecular Biology. eGFP and tagRFP were fused to the N-terminus of human SERCA2a and human PLB, respectively, as described previously.109,134,136 We have previously demonstrated that attachment of the fluorescent proteins at these sites does not interfere with the activity of SERCA.134,137 Native PLB equilibrates between monomers and homopentamers.138 To simplify the system and ensure that we are specifically measuring the SERCA2a-PLB interaction, a monomeric mutant of PLB39 was used, with three mutations (C36A, C41A, and C46A) in the transmembrane domain. All mutations were introduced by Quikchange mutagenesis (Agilent Technologies, Santa Clara, CA) and sequenced for confirmation. 4.3.2 Cell Culture Human embryonic kidney cells 293 (HEK293, ATCC, Manassas, VA) were cultured in phenol red–free Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Waltham, MA) supplemented with 2 mM l-glutamine (Invitrogen, Waltham, MA), heat-inactivated 10% fetal bovine serum (FBS HI, Gibco). Cell cultures were maintained in an incubator with 5% CO2 (Forma Series II Water Jacket CO2 Incubator, Thermo Fisher Scientific, Waltham, 39 MA) at 37 °C. To generate the one color human SERCA2a stable cell line, HEK293 cells were transiently transfected using Lipofectamine 3000 (Invitrogen). Transiently transfected cells were treated with G418 (Sigma-Aldrich, St. Louis, MO) for two weeks to select for expressing cells. Stable cell lines expressing 2C-SERCA were screened for uniform population by flow cytometry and fluorescence microscopy. To generate SERCA2a-PLB biosensor, HEK293 cells were transiently transfected using Lipofectamine 3000 with GFP- SERCA2a and RFP-PLB in a 1:7 molar ratio for the screen. Cells were then assayed 48 hours post-transfection. For cell permeabilization experiments, cells were trypsinized (TrypLE, Thermo Fischer Scientific), PBS-washed (three times), and resuspended in a homogenization buffer (20 mM MOPS, 0.5 MgCl2, protease inhibitor cocktail (Roche, Basel, Switzerland)). Saponin (Sigma-Aldrich) was added to a final concentration of 10 µg/mL and cells were incubated for 4 min at 37ºC water bath. Saponin was removed by centrifugation for 3 min, cells were washed twice with homogenization buffer, and resuspended in homogenization buffer. 4.3.3 Western Blot Samples were separated on a 4-20% polyacrylamide gradient gel (Bio-Rad, Hercules, CA) and transferred to polyvinylidene difluoride membrane. The membrane was blocked in Odyssey Blocking Buffer (LI-COR Biosciences, Lincoln, NE) followed by overnight incubation of the primary antibody rabbit anti-GFP (1:1000; ab290, Abcam, Cambridge, United Kingdom)139, mouse anti-SERCA2 (1:1000; 2A7-A1, Abcam)134, rabbit anti-tagRFP (1:1000; ab233, Evrogen)140, mouse anti-PLB (1:1000, 2D12, Abcam)109 or rabbit anti-β- actin (1:5000, ab8227, Abcam)141 in the cold room. Blots were incubated with anti-mouse or anti-rabbit secondary antibodies conjugated to IRDye 680RD or IRDye 800CW, respectively, for 1 h at room temperature (1:20,000; LI-COR Biosciences). Blots were 40 quantified on the Odyssey scanner (LI-COR Biosciences). Full-length blots are presented in Figure S33. 4.3.4 Compound Handling and Preparation of 384-Well Assay Plates The LOPAC compounds (Thermo Fisher Scientific) were received in 96-well plates and reformatted into 384-well polypropylene intermediate plates (Greiner Bio-One, Kremsmunster, Austria) using a multichannel liquid handler, BioMek FX (Beckman Coulter, Miami, FL), and then transferred to 384-well Echo Qualified source plates (Labcyte, Inc., Sunnyvale, CA). Assay plates were prepared by transferring 50 nL of the 10 mM compound stocks or DMSO from the source plates to 384-well black polypropylene plates (Greiner), using an Echo 550 acoustic dispenser (Labcyte). LOPAC compounds were formatted in four plates, with the first two and last two columns loaded with 50 nL of DMSO and used for compound-free controls. These assay plates were then sealed and stored at −20 °C prior to usage. Cells were dispensed (50 µL) using a Multidrop Combi liquid dispenser from Thermo (Pittsburg, PA), at a density of 106 cells/mL in PBS. 4.3.5 Fluorescence Data Acquisition FLT measurements were conducted in a prototype top-read FLT-PR designed and built by Fluorescence Innovations, which reads each 384-well plate in ~3 min. GFP donor fluorescence was excited with a 473 nm microchip laser from Concepts Research Corporation (Belgium, WI), and emission was acquired with 490 nm long-pass and 520/17 nm band-pass filters (Semrock, Rochester, NY). This instrument uses a unique direct waveform recording technology that enables high-throughput FLT detection at high precision.142 We have previously demonstrated the performance of this plate reader with known fluorescence standards, as well as with a FRET-based HTS method that targets 41 SERCA.135,136,143 4.3.6 HTS Data Analysis TR-fluorescence waveforms for each well were fitted based on a two-exponential decay function using least-squares minimization global analysis software135. FRET efficiency (E) was determined as the fractional decrease of donor FLT (τD), due to the presence of acceptor fluorophore (τDA) based on Equation (10). Assay quality was determined based on controls (DMSO-only samples) on each plate, as indexed by the Z′ factor:144 3(휎퐷퐴 + 휎퐷) 푍′ = 1 − (11) |휇퐷퐴 − 휇퐷| where σD and σDA are the standard deviations (SDs) of the controls τD and τDA, respectively, and μD and μDA are the means of the controls τD and τDA, respectively. A compound was considered a Hit if it changed E by >3 SD relative to that of control samples (E0) that were exposed to 0.1% DMSO. The 3 SD Hit selection threshold is typical for normally distributed HTS data, whereby 0.27% of the readings are expected to fall outside this limit. The threshold may be further adjusted to constrain the number of Hits according to the resources available for evaluation via secondary (orthogonal) assays. 4.3.7 Time-Resolved FRET 42 TRF waveforms from donor and FRET-labeled samples were analyzed as described in our previous publications.142,145,146 The measured time-resolved fluorescence waveform, ∞ 퐼(푡) = ∫ 퐼푅퐹(푡 − 푡′) ∙ 퐹(푡′)푑푡′ (12) −∞ is a function of the nanosecond decay time t, and is modeled as the convolution integral of the measured instrument response function, IRF(t), and the fluorescence decay model, F(t). The fluorescence decay model 퐹 (푡) = 푥 퐹 (푡) + (1 − 푥 ) 퐹 (푡) 퐷+퐴 퐷 퐷 퐷 퐷퐴 (13) is a linear combination of a donor-only fluorescence decay function FD(t) and an energy transfer-affected donor fluorescence decay FDA(t). The donor decay FD(t) is a sum of n exponentials with discrete FLT species i and pre-exponential mole fractions Ai (Equation (9). For the GFP donor, two exponentials (n = 2) were required to fit the observed fluorescence. The energy transfer-affected donor decay function, FDA(t), 푁 퐹퐷퐴(푡) = ∑ 푋푗 ∙ 푇푗(푡) (14) 푗=1 is a sum over multiple structural states (j) with mole fractions Xj, represented by FRET- affected donor fluorescence decays Tj(t). The increase in the donor decay rate (inverse donor FLT) due to FRET is given by the Förster equation 43 푘 = 푘 (푅⁄푅 )−6, where 푇푖 퐷푖 0푖 (15) 푘 = 푘 + 푘 , and DAi Di Ti (16) 푘 = 1/휏 퐷푖 푖 (17) We modeled TR-FRET assuming that each structural state j ((14) corresponds to a Gaussian distribution of interprobe distances, j(R): ∞ 푛 6 −푡 푅0푖 (18) 푇푗(푡) = ∫ 휌푗(푅) ∙ ∑ 퐴푖 푒푥푝 ( ∙ [1 + ( ) ]) 푑푅 −∞ 푖=1 휏푖 푅 2 1 −[푅 − 푅j] (19) 휌j(푅) = exp ( 2 ) 휎j√2휋 2휎j 휎푗 = 퐹푊퐻푀푗⁄(2√2 푙푛 2) (20) 4.3.8 Isolation of Sarcoplasmic Reticulum Vesicles Skeletal muscle SR membrane vesicles were isolated from longissimus dorsi obtained from New Zealand white rabbits, as previously described.63 Cardiac SR membrane vesicles were isolated from ventricular myocardium obtained from fresh pig hearts.147 All experimental protocols were reviewed and approved by University of Minnesota’s Institutional Animal Care and Use Committee, an Association for Assessment and Acreditation of Laboratory Animal Care institution. 4.3.9 Ca-ATPase Activity An enzyme-coupled, NADH-linked ATPase assay was used to measure SERCA ATPase activity in 96-well microplates. Each well contained 0.8 μg (skeletal) or 2 μg (cardiac) of SR vesicles adjusted for the different SERCA contents of skeletal and cardiac SR, 50 mM MOPS (pH 7.0), 100 mM KCl, 5 mM MgCl2, 1 mM EGTA, 0.2 mM NADH, 1 mM phosphoenol pyruvate, 10 IU/mL of pyruvate kinase, 10 IU/mL of lactate 44 dehydrogenase, 3.5 μg/mL of the Ca ionophore A23187, and CaCl2 added to set free [Ca] to the desired values. The assay was started upon the addition of ATP at a final concentration of 5 mM and read in a SpectraMax Plus microplate spectrophotometer rom Molecular Devices (Sunnyvale, CA). The Ca-ATPase assays were conducted over a range of [Ca], and the ATPase activities were fitted using the Hill function. 4.3.10 Statistical Analysis Errors are reported as the standard deviation of the mean, except when noted, and statistical significance was evaluated using Student’s T test, where p < 0.05 was considered significant. 45 4.4 Results 4.4.1 Expression, localization, and FRET readout of SERCA2a-PLB biosensor. To monitor the SERCA2a-PLB interaction in a natural membrane environment, we developed a biosensor system by fusing GFP and RFP to the N-termini of SERCA2a and PLB, respectively, and measured FRET changes as a readout of changes in the SERCA2a-PLB complex structure and binding. Attaching fluorescent proteins (FP) at these sites does not interfere with the activity of SERCA or PLB.148 HEK293 cells were transfected with GFP-SERCA2a and RFP-PLB constructs, and expression was verified by SDS-PAGE and immunoblot (Figure 23a). The GFP-tagged SERCA protein was resolved from untagged SERCA, as verified by a GFP-specific antibody. There were no other bands of lower mobility, and there was no aggregation apparent in intact cells. To specifically detect the SERCA2a-PLB interaction, a monomeric mutant of PLB39 was used, with three mutations (C36A, C41F, and C46A) in the transmembrane domain. The structure and function of this PLB variant is indistinguishable from those of monomeric wild-type PLB.33,149 Transfection of this RFP-PLB variant produced a single band recognized by PLB- and RFP-specific antibodies. The SERCA2a-PLB biosensor showed the expected endoplasmic reticulum localization by confocal microscopy (Figure 23b) with no bright puncta (which would have indicated aggregation) or other non- uniformities. FRET between GFP-SERCA2a and RFP-PLB in live cells showed hyperbolic dependence on protein concentration, with a maximum energy transfer efficiency E (fractional decrease of FLT or intensity) of 0.103 ± 0.004 (Figure 23c, Figure 23d). The observed nanosecond time-resolved fluorescence decay was best fitted by a two-Gaussian distance distribution, centered at R1 = 5.6 ± 1.6 nm and R2 = 9.8 ± 1.9 nm (Figure S29).41,45,150 The measured distances are in agreement with previous TR-FRET measurements.151 46 a b c d 2.7 * .10 2.6 2.5 (ns) .05 2.4 E (FRET) E 2.3 .00 2.2 10 20 30 RFP Intensity (a.u.) Figure 23. Biochemical and spectroscopic characterization of SERCA2a-PLB biosensor. (a) Immunoblots of homogenates from untransfected HEK293 cells (lane 1), cells expressing GFP-SERCA2a (lanes 2-3), or cells expressing GFP-SERCA2a and increasing amounts of RFP-PLB (lanes 4-8). Antibodies from top to bottom are anti-GFP, anti-SERCA2, anti-RFP, anti-PLB and anti--actin, and the image has been cropped. b) Confocal fluorescence imaging of HEK293 cells expressing GFP-SERCA2a (left) or GFP-SERCA2a and RFP-PLB (right). (c) FRET efficiency E from GFP-SERCA2a (donor) to RFP-PLB (acceptor) shows hyperbolic dependence on acceptor concentration. (d) The mean FLT in the presence of saturating acceptor is τDA = 2.35 ± .02 ns, corresponding to E = 0.103 ± 0.004. Error bars indicate SD (n=3). Statistics performed by using the Student’s t test: *P ≤ 0.01. PLB inhibition of SERCA is relieved by elevated Ca.119 The most likely explanation, supported by in vitro measurements of FRET between SERCA and PLB, is that relief of inhibition occurs primarily through a structural rearrangement of the bound complex and not through disassociation.63,148,152 Specifically, SERCA-PLB FRET is reduced in the presence of a high Ca concentration but is not ablated.63,148 To test this, we quantified 47 N.S. Low Ca a * b 2.6 0.15 0.03 Low Ca, + Tg 2.5 High Ca 0.02 High Ca, +Tg 2.4 0.10 (ns) 2.3 E (FRET) E 0.01 0.05 R R 2.2 Distribution 1 2 2.1 0.00 0.00 Ca - + - + - + Ca - + - + 0 2 4 6 8 10 12 Tg - - + + - - Tg - - + + Interprobe distance, r (nm) SERCA2a-PLB (DA) (D)-only SERCA2a-PLB (DA) Figure 24. Effects of Ca and Tg on SERCA2a-PLB biosensor. (a) left- FLT measurements in saponin-permeabilized cells under conditions of low Ca (100 nM) and high Ca (10 µM), and in the absence and presence (100 nM) of the SERCA inhibitor thapsigargin (Tg). right- FRET readout calculated from (a) shows a Ca-dependent decrease in FRET, that is ablated by addition of Tg. Error bars indicate SD (n=3). Make colors and lines in a consistent with those in b Statistics performed Blackby using = no Tg, the red Student’s= Tg t test: *P ≤ 0.01 and N.S. indicating P ≥ 0.05. (b) Distance distributionsSolid = low between Ca, short dash donor = high on Ca SERCA and acceptor on PLB, based on (18). Or whatever you choose. FRET changes in permeabilized cells under conditions of low Ca (100 nM) and high Ca (10 µM). In agreement with previous studies, high Ca resulted in a modest FRET decrease (from E=0.14 to E=0.11), confirming that the binding interface is altered by structural changes but that the SERCA2a-PLB complex remains intact (Figure 24a). Treatment with thapsigargin (Tg), a SERCA inhibitor that shifts the transition between the E1/E2 conformations toward E2, prevents the Ca-dependent decreases in FRET (Figure 24a). We consistently observe a decreased DA lifetime in saponin-treated cells and cell homogenates relative to the live cell assay (Figure 23d); this may reflect changes in membrane dynamics or the fluorophore environment.153 As both Tg and low Ca promote the E2 conformation, we hypothesized that we would be able to resolve the distance distribution associated with this conformational state by TR-FRET. Indeed, we see a shift in the equilibrium of the structural states from the short distance (R1) to the long distance (R2) under high-Ca conditions, and this is ablated by Tg (Figure 24b). This demonstrates 48 that the SERCA2a-PLB biosensor is sensitive to changes in the protein complex interactions, and thus suitable for our HTS platform. 4.4.2 HTS of LOPAC library to identify compounds that affect SERCA2a-PLB FRET. To assess the performance of our FRET-based HTS method, we screened a library of 1280 small-molecule compounds (LOPAC) for identification of SERCA2a-PLB modulators. This library is suitable for a pilot screen, as it covers all major drug target classes and most compounds are commercially available, satisfactory for the requirements of orthogonal screening.154 Assay quality was determined based on controls (DMSO-only samples) on each plate, as indexed by the Z’ factor from Eq (11).134,144 The high precision enabled using FLT measurements provides an excellent HTS assay quality (Z’ = 0.80 ± 0.01).134 We incubated test compounds with live cells expressing the SERCA2a-PLB biosensor in a 384-well assay format,136 and selected Hits that changed FLT (>3 S.D.) of the GFP donor fused to SERCA2a in triplicate screens of the library (red points in Figure 25a). The distribution of FLT for the DMSO-control wells is normal and centers at 2.30 ns (Figure 25b). Because FLT measurements are susceptible to interference from fluorescence of the test compound itself, we also acquired fluorescence emission spectra by scanning the same 384-well plates using a recently developed spectral plate reader32 (Figure 25c) and applied a spectral similarity index to flag fluorescent compounds.136,143 49 a 2.5 Candesartan cilexetil 2.4 Calmidazolium PF 3845 CP-154526 Ivermectin SCH-202676 2.3 Retinoic (ns) Demeclocycline SC-514 K114 Acid PPADS 2.2 PD-407824 Propofol 6-Hydroxy -DOPA Ro 41-0960 2.1 0 200 400 600 800 1000 1200 1400 Compound # b 300 c DMSO µ=2.301 ns Flagged σ=0.011 ns 200 Compound 100 Intensity Distribution 0 2.26 2.28 2.30 2.32 2.34 500 600 700 (ns) Wavelength (nm) Figure 25. High-throughput screen results. Compounds were screened in triplicate at a final concentration of 10 µM. (a) FLT values from one representative screen including DMSO control wells with a Hit threshold (>3 S.D. of mean) indicated by dotted blue lines. (b) Gaussian fit of the FLT distribution. (c) SERCA2a-PLB biosensor visible emission spectrum upon excitation at 473 nm (black). Addition of a fluorescent compound greatly alters the spectrum (blue). In total, 21 compounds were identified as Hits, which altered FRET from SERCA2a to PLB in at least 2 replicates after filtering out compounds with interfering fluorescence (false positives). This equates to a ~1.6% Hit rate, which is in the acceptable range between 0.5%-3% to maintain a manageable number for post-HTS testing through orthogonal assays.155 50 a b Ro 41-0960 CSR .35 2.5 ** 2.4 .30 0.3 (i.u.) 2.3 0.2 0.1 pCa6.4 .25 Lifetime (ns) Lifetime 2.2 V 0.0 0 2 4 8 16 24 32 Concentration (µM) 2.1 .20 1 10 100 0 10 20 30 Concentration (µM) Concentration (µM) Figure 26. Effects of SERCA activators identified in this screen. (a) FRET dependence on (0.01- 100 µM) CP 154526 (circle) and Ro 41-0960 (up-triangle) under conditions similar to those in the primary HTS. (b, inset) Dependence of CSR Ca-ATPase activity at pCa 6.4 on the concentration of Ro 41-0960. Error bars indicate SD (n=3). Statistics performed by using the Student’s t test: **P ≤ 0.05 4.4.3 FRET concentration-response assays. The Hit compounds were further tested in concentration-response (0.01-100 µM compound) FRET assays under the same conditions as used in the primary screens, and 17 compounds displayed concentration-dependent changes in FRET (Table S1). Of these, 5 compounds (tenidap, FPL 64176, CP 135807, olprinone, and X80) did not reach saturation of FRET, as evidenced by high EC50 values (>100 µM) when fit to Eq. Error! Reference source not found.). It is possible that these compounds are not intrinsically fluorescent but are affecting the GFP FLT by changing the fluorophore environment or structure. The remaining 12 compounds had EC50 values below 100 µM (Figure 26a, Figure S31, Table S1) and cause either an increase in FRET (e.g., Ro 41-0960) or decrease in FRET (e.g., CP 154526, ivermectin, SCH-202676) of the SERCA2a-PLB biosensor. Thus, the FRET dose-response assay can function as an effective tool to refine the list of Hits and limit the number of compounds to be tested via medium- throughput activity assays in >50K compound screens. 51 a Ro 41-0960 CSR SSR 1 6.5 1 6.5 ca ca pK pK 6.0 6.0 10 100 10 100 [Cmpd] (µM) [Cmpd] (µM) ATPase Activity ATPase ATPase Activity ATPase - - Ca Ca 0 0 8 7 pCa 6 5 8 7 pCa 6 5 b CP 154526 CSR SSR 1 3.5 0.9 3 max max V V 0.8 2.5 2 0.1 1.0 10 0.1 1.0 10 [Cmpd] (µM) [Cmpd] (µM) 1 ATPase Activity Activity (i.u.) ATPase ATPase Activity Activity (i.u.) ATPase - - Ca Ca 0 0 8 7 pCa 6 5 8 7 pCa 6 5 Figure 27. Functional effects of SERCA activators varying Ca concentration. (a) Normalized Ca- ATPase activity of cardiac SR (CSR) and skeletal SR (SSR) was measured after a 20 min incubation in the presence of Ro 41-0960 (up to 64 µM) or DMSO control. Error bars indicate SEM (n=3). (inset) Concentration response of SERCA apparent Ca affinity (pKCa). (b) Ca-ATPase activity of CSR and SSR after 20 min incubation in the presence of CP 154526 (up to 5 µM) or DMSO control. Error bars indicate SEM (n=3). (inset) Concentration response for maximum velocity (V at pCa 5; Vmax). 4.4.4 Functional effects of structural (FRET) Hits on Ca-ATPase function To characterize the relationship between the structural readout (FRET) and the functional effects, in vitro Ca-ATPase assays were performed on cardiac SR (which contains both SERCA2a and PLB). During diastole and systole, the intracellular Ca concentration oscillates between 0.1 and 1 µM (pCa 7 to pCa 6).119 PLB does not alter basal or maximal SERCA2a activity but inhibits ATPase activity within this range of Ca concentration. An initial functional screen to identify compounds that disrupt PLB 52 inhibition was carried out by measuring Ca-ATPase activity at an intermediate Ca concentration (pCa = 6.4). Ro 41-0960 increased activity by 44% (at 24 µM) in cardiac SR in this Ca-ATPase assay (Figure 26b), while 8 compounds were inhibitors under these conditions (data not shown). To further examine the functional effects of Hit compounds, Ca-ATPase assays were performed on both skeletal SR (SERCA1a only) and cardiac SR (SERCA2a+PLB) at a series of Ca concentrations ranging from pCa 5 to 8. We expect two classes of activators to be useful for the treatment of HF: (1) a compound that increases SERCA’s apparent Ca affinity (pKCa) due to relief of PLB inhibition, which would display a leftward shift in the V vs. pCa plot in cardiac SR only and (2) a compound that increases Vmax (SERCA activity at saturating Ca). We tested all 21 Hits identified in the HTS-assay for Ca-ATPase activity and found 10 Hits that reproducibly affect SERCA activity (2 activators and 8 inhibitors). The activators included Ro 41-0960 and CP-154526 (chemical structures in Figure S30). Ro 41-0960 increased Ca affinity (pKCa) in cardiac SR (SERCA2a+PLB) from 6.05 to 6.45 in a concentration-dependent manner (EC50=22.98 µM), but had no effect on Ca affinity in skeletal SR (SERCA1a only), even at the highest compound concentrations (Figure 27a), indicating that the functional effects are specific to the SERCA2a-PLB complex. Notably, SERCA’s Ca affinity at saturating [Ro 41-0960] in cardiac SR is equivalent to the Ca affinity in skeletal SR, suggesting that the compound fully reverses PLB inhibition. The second activator discovered, CP-154526, increased Vmax by 34% in cardiac SR (EC50=1.0 µM) and 35% in skeletal SR (EC50= 2.5 µM) (Figure 27b). 53 a SCH 202676 Ivermectin CSR b .) CSR 0.8 .) 0.6 .50 .70 i.u i.u max max V V .35 .25 0.1 1.0 10 0.3 0.1 1.0 10 0.4 [Cmpd] (µM) [Cmpd] (µM) ATPase Activity Activity ( ATPase Activity ( ATPase - 0.0 - 0.0 Ca 8 7 pCa 6 5 Ca 8 7 pCa 6 5 Figure 28. Ca-ATPase activity for SERCA inhibitors (a) ATPase activity from CSR) was measured after 20 min incubation with SCH 202676 (up to 2 µM) or DMSO control. (inset) Concentration response for limiting activity at high Ca (Vmax). (b) ATPase activity from CSR after 20 min incubation in the presence of ivermectin (up to 64 µM) or DMSO control. (inset) Concentration response for limiting activity at high Ca (Vmax). In addition to the SERCA activators, 8 inhibitor compounds were found in our screen. Based on the design of the assay to detect the SERCA-PLB interaction, we expected that the functional effects of Hit compounds would be to increase the apparent Ca affinity (similar to Ro 41-0960). Surprisingly, the majority (80%) of Hit compounds that had functional effects were found to inhibit SERCA. The most likely explanation is that these compounds bind to the SERCA2a-PLB complex and induce inhibitory alterations in the structure of the complex, consistent with changes in FRET between fluorophores attached to SERCA2a and PLB. These compounds inhibit SERCA activity in both skeletal (SSR, SERCA1a in the absence of PLB) and cardiac SR (CSR, SERCA2a in the presence of PLB). From the FRET measurements alone, we cannot determine potential binding interfaces of the compound to SERCA, but we speculate that these compounds do not bind to the SERCA2a-PLB interface, but rather affect FRET through allosteric changes in SERCA itself. Also, we cannot rule out the possibility of isoform-dependent activities (SERCA1a vs SERCA2a); although recent crystal structures (PDB ID=5MPM) demonstrate strong similarity between the isoforms. Figure 28 highlights the results from the two most effective SERCA inhibitors (chemical structures in Figure S30). SCH 54 202676 decreased Vmax by 68% (EC50=1.4 µM) in cardiac SR and 97% in skeletal SR (EC50=0.9 µM) (Figure 28a, Figure S32). The antiparasitic ivermectin decreased Vmax by 78% (EC50=6.2 µM) in cardiac SR and 88% in skeletal SR (EC50=5.8 µM) (Figure 28b, Figure S32). These results agree with a previous report showing inhibition of skeletal muscle SR Ca-ATPase by ivermectin,156 with 90% of the activity being inhibited at high concentrations (50 µM). Overall, there is strong correlation between structural (FRET- based) and functional (Ca-ATPase-based) results. Moreover, we found a diversity of chemotypes (Figure S30) and functional outcomes of the tested Hit compounds. 4.5 Discussion The Ca re-uptake process is deficient in cardiomyocytes in both experimental and human heart failure.82,157-160 Therapies targeting the underlying molecular causes have the potential to significantly restore heart function.120,121 SERCA2a sequesters Ca into the SR of cardiomyocytes, and its function is regulated through protein-protein interactions with PLB.118 A decrease in expression of SERCA2a is a hallmark of heart failure, and gene therapy approaches to restore its levels back to normal have been successful in experimental models but not so far in clinical trials.90,92,95,127,130,161 The relevance of the SERCA2a-PLB complex in human heart function is underscored by the existence of patients who develop cardiomyopathies due to mutations in the PLN (PLB- encoding) gene.111,162,163 Some of the PLN mutations have opposite sub-cellular mechanisms of action. One of them, R9C, results in chronic inhibition of SERCA2a by mutant PLB and early death, which is consistent with the known mechanisms on PLB regulation and the findings in mice.111,164 The other mutation is associated with loss of PLB function (L39stop) and results in dilated cardiomyopathy and premature death in the homozygous, opposite to what is seen in mice.163 The interaction between SERCA2a 55 and PLB is complex and may be species dependent. Despite great importance, the development of SERCA2a-PLB specific therapies has been limited, largely due to the lack of sensitive and rapid methods for screening compound libraries that disrupt or alter the protein-protein interface. The present report addresses this problem through the development of a structure- based HTS method using the GFP-RFP FRET pair fused to the human isoforms of SERCA2a and PLB, expressed in live cells. Of the 10 Hits that reproducibly altered SERCA2a-PLB structure and function in this report, most of the compounds (9) have not been previously identified to affect Ca-ATPase activity. The one exception is ivermectin, a broad-spectrum antiparasitic compound. Although it is non-toxic to animals at applied doses, it has been reported that ivermectin, along with other macrocyclic lactones, inhibits SERCA activity at higher doses.156 Ivermectin appears to inhibit all SERCA isoforms (1a156, 2a165, and 2b156), as well other mammalian P-type ATPases (Na+, K+ and H+/K+). Our results support these findings, as we see inhibition of skeletal SR (containing SERCA1a) and cardiac SR (containing SERCA2a and PLB) with EC50 values of 5.8 µM and 6.2 µM, respectively. Most importantly, two of the Hit compounds identified are SERCA activators. These include Ro 41-0960, which increases the apparent Ca affinity of SERCA2a, as seen in a leftward shift in the Ca-ATPase assay data, and CP 154526, which increases the enzyme’s maximal activity at high [Ca]. This confirms that this live-cell HTS platform can detect compounds that act to alter the SERCA2a-PLB interaction and also compounds that act on SERCA independently of PLB. SERCA1a is also subject to regulation by sarcolipin (SLN) in skeletal muscle. SLN functions by decreasing Ca affinity of SERCA and partially uncoupling Ca transport from ATP hydrolysis, and is proposed to bind to a similar transmembrane interface of SERCA as PLB (groove formed by M2, M6, and M9 helices) with distinct orientations of the 56 PLB/SLN cytoplasmic domains. Of the compounds tested in this pilot screen, no compound had effects on SSR exclusively that would be expected upon changes in the SERCA1a/SLN interaction, such as increased Ca affinity. Future studies with RFP- labeled SLN may be needed to discover such compounds. The advancement of fluorescent-protein biosensors as tools for drug discovery has generated new classes of reagents that report specific molecular processes within the intracellular milieu that have been previously inaccessible to HTS.166 FRET techniques are sensitive to changes in intra- and inter- molecular distances and can thus report relatively small changes in protein structures and protein-protein interactions. As a result, cell-based FRET assays have been developed to monitor a range of biological activities including membrane and cytoplasmic processes.167,168 The application of FLT measurements and analysis allows the detection of subtle FRET changes that would be missed by intensity-based technology.133,134,169 Indeed, the majority of our Hit compounds caused small FRET efficiency changes, ranging 0.02 to 0.06. The sensitivity of the assay to compounds of the SERCA2a-PLB functional interaction is indicated by the observation that ~50% of compounds tested that altered SERCA2a-PLB FRET in a concentration-dependent manner also directly affected SERCA function, as measured by an in vitro ATPase assay. The position of the fluorophore tagged to a protein can alter its function and may not report the exact position of certain domains of the tagged protein. This concern is especially relevant for FPs because of their large size. On the other hand, the fact that the fluorophore of GFP and RFP is buried within the barrel allows the fluorophore to maintain a homogenous chemical environment. Attachment of FPs to SERCA and PLB did not alter catalytic activity nor its ability to localize to the membrane,148 including the GFP-RFP FRET pair used in this study.151 Still, it is necessary to understand that FRET 57 measurements report the proximity between fluorophore tags and only indirectly on the connected protein domain arrangement. Because it is possible that the position of the GFP and/or RFP molecules are affected by assay conditions or the presence of test compounds, addition controls are needed. It is also possible that compounds may be able to quench the donor fluorescence without emitting detectable fluorescence themselves, as would be detected in the visible spectrum (Figure 25c). As a reference for this possibility, we have screened the LOPAC library under similar conditions (e.g., final [compound] = 10 µM) using HEK cells expressing the GFP-RFP FRET pair connected by a flexible linker or GFP-SERCA2a (donor) alone (data not shown). Compounds that altered the FLT greater than 3 S.D. in these controls were excluded from further consideration, and not included in the 21 Hit compounds reported here. Lastly, separation distance changes or modulation in dipole-dipole interactions can engender changes in FRET. Because the rotational correlation time of FPs is longer than observed fluorescent decays, this brings into question whether to assume dynamically averaged dipole-dipole orientations (e.g., 2 = 2/3) or a static isotropic model.170 Vogel et al., compares the dependence of CFP on the assumption of dynamic and static models, and demonstrates that the apparent FRET values observed (between 0.10 - 0.15) for the SERCA2a-PLB biosensor are similar assuming either model.171 Further, the 2 parameter was calculated by simulation for a SERCA molecule with CFP attached at the N-terminus using a similar length linker as used in this study.172 The orientation factor is 0.6, which is slightly lower than what is predicted by a dynamic model.172 Considering this, the Hit compounds identified cause relatively small changes in the FRET readout and do not eliminate the FRET signal, implying that subtle domain movement of SERCA2a and/or PLB is sufficient to influence enzymatic function. This also suggests that these compound-dependent conformational changes occur while PLB 58 is still bound to SERCA2a. This is consistent with reports detecting at least two distinct PLB conformations exist while bound to SERCA, each with opposing functional outcomes (Subunit Model).62,63,66,150,173 All these compounds are different from those identified in an earlier screen, where we used a 2-color SERCA construct.134,136,137 Thus, this SERCA2a-PLB HTS assay could be used in combination with the SERCA-only assays, to enrich the spectrum of SERCA- modulator chemotypes. More generally, the approach illustrated here is applicable to any protein-protein interaction target for which there is sufficient information about structure and function to design FRET biosensors that are sensitive to changes in the protein complex interactions. 59 Acknowledgements Joseph M. Autry provided helpful discussion, and Octavian Cornea, Destiny Ziebol and Sarah Blakely Anderson provided administrative support. Fluorescence measurements were performed using facilities provided by the Biophysical Technology Center, University of Minnesota, and by Fluorescence Innovations, Inc. (Minneapolis, MN), where assistance was provided by Benjamin D. Grant and Kurt C. Peterson. DRS was supported by NIH Training Grant T32 AR07612. TMS was supported by NIH Institutional Research and Academic Career Development Award K12 GM119955. This work was supported by grants from NIH to DDT (R01 GM27906, R01 HL129814, and R37 AG26160), RJH (R01 HL129814), and to RLC (R01 HL092097 and R01 HL138539). Data Availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Declaration of Competing Interests The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: DDT and RLC hold equity in, and serve as executive officers for, Photonic Pharma LLC. These relationships have been reviewed and managed by the University of Minnesota. Photonic Pharma had no role in this study. 60 4.6 Supplementary Information Table S1. Concentration response for SERCA2a-PLB FRET a b Compound ΔFLT (ps) ΔFLT (ps) EC50 FRET (µM) Tenidap 8 -68 NDc d Rotenone 11 20 ND Candesartan 70 181 20.1 cilexetil 31 63 73.2 TBB SCH 202676 36 35 1.8 CP 154526 48 185 9.3 c FPL 64176 19 47 ND d 10058-F4 17 -11 ND 30 38 3.2 PF3845 Palmitoyl-DL- 50 189 24.9 carnitine chloride 53 56 5.9 Ivermectin PD-407824 -103 -186 8.5 K114 -16 -102 41.3 c CP-135807 -70 -390 ND Retinoic Acid 27 -31 65.5 36 254 NDc Olprinone c X80 -40 -865 ND Ro 41-0960 -26 -152 22.5 d FG7142 4 10 ND d AMG9810 6 -1 ND Calmidazolium 45 -192 26.3 a Following compound incubation for 20 min at a final concentration of 10 µM, with an uncertainty (1 S.D.) of the measurement of 11 ps. b Following compound incubation for 20 min at a final concentration of 100 µM, , with an uncertainty (1 S.D.) of the measurement of 11 ps. c Change in FRET did not reach saturation at 100 µM. d Change in FRET was below the 3 S.D. threshold of the triplicate screen. 61 Table S2. Concentration response for CSR Ca-ATPase Assay Vmax @ ΔVmax @ pKCa @30 µM Compound 30 µM (i.u.) 30 µM (%) EC50 Vmax(µM) (µM) Tenidap 0.66 -1.7 NDa 6.09 a Rotenone 0.53 -22.0 ND 6.07 Candesartan 0.17 -75.3 11.7 NDb cilexetil TBB 0.49 -27.2 NDa 6.07 b SCH 202676 0.22 -68.2 1.4 ND 0.91 34.4 1.0 6.06 CP 154526 a FPL 64176 0.52 -23.0 ND 6.11 10058-F4 0.63 -7.5 NDa 6.08 PF3845 0.48 -28.3 NDa 6.11 Palmitoyl-DL- 0.22 -67.1 16.1 NDb carnitine chloride Ivermectin 0.15 -77.9 6.2 6.10 PD-407824 0.32 -53.4 17.3 6.04 K114 0.11 -83.7 6.7 6.10 0.65 -3.8 NDa 6.05 CP-135807 b Retinoic Acid 0.42 -37.8 11.2 ND Olprinone 0.68 -0.1 NDa 6.04 X80 0.66 -2.1 NDa 6.09 Ro 41-0960 0.53 -20.9 NDa 6.45 0.67 -0.7 NDa 6.03 FG7142 0.68 0.1 NDa 6.03 AMG9810 b Calmidazolium 0.08 -88.7 15.9 ND a Change in Vmax was <30% of DMSO control b Did not show Ca-dependent activity due to inhibition. 62 Table S3. Concentration response for SSR Ca-ATPase Assay Vmax @ ΔVmax @ pKCa @30 µM Compound 30 µM (i.u.) 30 µM (%) EC50 Vmax(µM) (µM) Tenidap 2.66 1.8 NDa 6.52 Rotenone 2.64 0.7 NDa 6.50 Candesartan 0.41 -84.5 7.42 cilexetil 6.45 TBB 2.66 1.6 NDa 6.50 SCH 202676 0.07 -97.3 0.90 NDb CP 154526 3.40 35.0 2.47 6.49 FPL 64176 2.64 0.7 NDa 6.48 10058-F4 2.63 0.5 NDa 6.48 PF3845 2.61 -0.2 NDa 6.44 Palmitoyl-DL- 0.57 -78.1 12.23 carnitine chloride 5.51 Ivermectin 0.32 -87.6 5.75 NDb PD-407824 1.31 -50.0 18.77 6.46 b K114 0.32 -87.9 3.53 ND CP-135807 2.67 2.1 NDa 6.46 Retinoic Acid 2.34 -10.6 NDa 6.48 Olprinone 2.70 3.0 NDa 6.50 X80 2.46 -6.1 NDa 6.46 Ro 41-0960 1.44 -45.1 9.78 6.47 FG7142 2.63 0.5 NDa 6.47 AMG9810 2.62 0.1 NDa 6.46 Calmidazolium 0.01 -99.5 7.43 NDb a Change in Vmax was <30% of DMSO control b Did not show Ca-dependent activity due to inhibition. 63 a 6000 b 200 2 2 4000 χ χ 100 2000 0 0 1 2 3 4 1 2 3 Number of Exponentials Fit Number of Structural States Fit Figure S29. Model testing to determine the time-resolved fluorescence decay function for GFP donor used in this study and the structural state model. Analysis was first performed on the data from the donor-only sample FD(t) (GFP-SERCA2a), which was fit by a multi-exponential function (Eq. 5); two exponentials (n = 2 in Eq. 3) were necessary 2 and sufficient for an optimum fit, as shown by χ optimization in (a). 1 = 1.38 ns, amplitude1 = 251; 2 = 2.59 ns, amplitude2 = 2188. Then the donor + acceptor data FD+A(t) was analyzed according to Eq. 6-12, assuming multiple Gaussian components (Eq. 10); two components (N = 2 in Eq. 10) were necessary and sufficient for an optimum fit, as shown by χ2 optimization in (b). Distance distribution of best fit two- structural state model corresponds to R1 = 5.6 nm, R2 = 9.8 nm, FWHM1 = 1.6 nm, FWHM2 = 1.9 nm, and XD = 0.47. 64 a CP-154526 Ro 41-0960 F O O N N N+ O- N OH N OH b O HO O Ivermectin O O H O O O H O SCH-202676 H O O H S N OH N N O H OH Figure S30. Chemical structures of SERCA2a-PLB biosensor hits (a) activators (b) inhibitors 65 2.40 (ns) 2.35 Lifetime 2.30 1 10 100 Concentration (µM) Figure S31. FRET concentration response for SERCA inhibitors FRET dependence on (0.01-100 µM) SCH 202676 (circle) and Ivermectin (up-triangle) under conditions similar to those in the primary HTS. 66 SCH 202676 Ivermectin SSR SSR 3 3 3 max 3 2 max 2 V V 1 1 2 2 0.1 1.0 10 0.1 1.0 10 [Cmpd] (µM) [Cmpd] (µM) 1 1 ATPase Activity (i.u.) Activity ATPase ATPase Activity (i.u.) Activity ATPase - - Ca Ca 8 7 pCa 6 5 8 7 pCa 6 5 Figure S32. SSR Ca-ATPase activity for SERCA inhibitors ATPase activity from SSR was measured after 20 min incubation with ivermectin (up to 2 µM), SCH 202676 (up to 64 µM) or DMSO control. (inset) Concentration response for limiting activity at high Ca (Vmax). 67 anti-SERCA2 anti-PLB * anti-actin anti-RFP # Figure S33. Uncropped blot images of SERCA2a-PLB biosensor Immunoblots of homogenates from untransfected HEK293 cells (lane 1), cells expressing GFP-SERCA2a (lanes 2-3), or cells expressing GFP-SERCA2a and increasing amounts of RFP-PLB (lanes 4-8). Antibodies are anti-SERCA2, anti-RFP, anti-PLB and anti--actin. * indicates non-specific band. # indicates degradation products. 68 .15 .10 | (ns) | Δ | .05 Figure S34. Reproducible hits assessed across triplicate screens FLT hits were reproducible (triplicate) using a 3 S.D. threshold (red line). Compounds shown: Ro 41-0960, SCH 202676, CP 154526, and Ivermectin. Each bar (individual screen) indicates the difference in FLT value compared to the mean value of the corresponding screen. 69 Chapter 5 – Functional and Transcriptomic Insights into Pathogenesis of R9C Phospholamban Mutation using Human Induced Pluripotent Stem Cell-Derived Cardiomyocytes Delaine K. Ceholski1, Irene C. Turnbull1, Chi-Wing Kong2, Simon Koplev3, Joshua Mayourian1, Przemek A. Gorski1, Francesca Stillitano1, Angelos A. Skodras4, Mathieu Nonnenmacher1, Ninette Cohen3, Johan L.M. Bjorkegren3, Daniel R. Stroik5, Razvan L. Cornea5, David D. Thomas5, Ronald A. Li2, 6, Kevin D. Costa1, Roger J. Hajjar1 1Cardiovascular Research Center, Icahn School of Medicine at Mount Sinai, New York City, New York 10029 2Department of Pediastrics and Adolescent Medicine, Hong Kong University, Pokfulam, Hong Kong 3Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, NY 10029 4Microscopy Core, Icahn School of Medicine at Mount Sinai, New York, NY 10029 5Department of Biochemistry, Molecular Biology, and Biophysics, University of Minnesota, Minneapolis, Minnesota 55455 6Ming Wai Lau Centre for Reparative Medicine, Karolinska Institutet, Solna SE-171, Sweden Originally published in Journal of Molecular and Cellular Cardiology. Vol. 119, pp 147-154. Reprinted with permission from Elsevier. Author Contribution DRS, RLC, and DDT reviewed and aided manuscript preparation. 70 5.1 Outline Dilated cardiomyopathy (DCM) can be caused by mutations in the cardiac protein phospholamban (PLB). We used CRISPR/Cas9 to insert the R9C PLB mutation at its endogenous locus into a human induced pluripotent stem cell (hiPSC) line from an individual with no cardiovascular disease. R9C PLB hiPSC-CMs display a blunted β- agonist response and defective calcium handling. In 3D human engineered cardiac tissues (hECTs), a blunted lusitropic response to β-adrenergic stimulation was observed with R9C PLB. hiPSC-CMs harboring the R9C PLB mutation showed activation of a hypertrophic phenotype, as evidenced by expression of hypertrophic markers and increased cell size and capacitance of cardiomyocytes. RNA-seq suggests that R9C PLB results in an altered metabolic state and profibrotic signaling, which was confirmed by gene expression analysis and picrosirius staining of R9C PLB hECTs. The expression of several miRNAs involved in fibrosis, hypertrophy, and cardiac metabolism were also perturbed in R9C PLB hiPSC-CMs. This study contributes to better understanding of the pathogenic mechanisms of the hereditary R9C PLB mutation, potentially identifying novel pathways to target in the treatment of dilated cardiomyopathy. 5.2 Introduction Dilated cardiomyopathy (DCM) is a common cause of heart failure and is characterized by biventricular dilation and progressive cardiac dysfunction.174 Approximately one-third of DCM cases are of hereditary origin, and mutations that depress force generation and alter calcium handling are prominent, particularly in 71 contractile or cytoskeletal proteins.175 Phospholamban (PLB) is a sarcoplasmic reticulum (SR) membrane protein that regulates the activity of the cardiac SR calcium pump (SERCA2a) by reducing its affinity for calcium, resulting in a decreased transport of calcium into the SR.176 This inhibition is alleviated through the phosphorylation of PLB by protein kinase A at Ser16 or by calcium/calmodulin-dependent protein kinase II at Thr17.15 Several hereditary mutations in PLB have been linked to dilated cardiomyopathy (DCM), including Arg9Cys (R9C),111 Arg9Leu,177 Arg9His,177 deletion of Arg14 (R14del),112,178,179 Arg25Cys (R25C),180 and Leu39stop.110 All mutations are autosomal dominant, as they have only been identified in heterozygous patients, except for Leu39stop, which has been found in both heterozygous and homozygous patients of which the heterozygous patients have no cardiovascular phenotype.110 While each mutation has distinct effects on PLB function and SR calcium homeostasis, all affected patients present with cardiac hypertrophy, decreased ejection fraction, and, in the case of R14del and R25C, ventricular arrhythmias.181 R9C PLB was first identified in a large American family and, in a transgenic mouse model (TgR9C), was found to abolish PLB-mediated inhibition of SERCA2a and “trap” protein kinase A (PKA), preventing it from phosphorylating wild-type PLB and eliciting a typical β-adrenergic response111,182.58,84 In a separate study, a dose-dependent effect was observed with wild-type and R9C PLB with a homozygous mouse model of R9C (PLB-/- + TgR9C) having significantly increased survival, accelerated SR calcium uptake rates, and improved hemodynamics compared to a heterozygous mouse model of R9C (PLB+/- + TgR9C).183 Further studies have identified other potential mechanisms, including disulfide bridging of the cysteines at position 9 in PLB leading to a compact pentamer structure which is exacerbated by oxidative stress.184,185 Molecular profiling of TgR9C PLB mice showed activation of profibrotic and proinflammatory signaling with 72 marked shifts in metabolic gene transcription.186 In a purified membrane system, R9C PLB had a dominant negative effect on SERCA1a, resulting in loss of function even in the presence of wild-type PLB.182 This is consistent with a recent nuclear magnetic resonance study on purified PLB in lipid bilayers, which showed that the R9C mutation shifts the conformational equilibrium of PLB toward the inhibitory T state of the protein,164 where the cytoplasmic helix of PLB is helically ordered and associated with the membrane surface.41 Thus, the mechanism by which R9C PLB causes DCM appears to be multifaceted and its elucidation confounded by the model systems used, which may not be representative of the human cardiomyocyte.181,182,187 In this study, we used genome engineering in human induced pluripotent stem cell- derived cardiomyocytes (hiPSC-CMs) to elucidate the disease-causing mechanisms of the hereditary R9C PLB mutation. The R9C hiPSC-CMs display a blunted response to β- agonists, defective calcium handling, and a hypertrophic phenotype. Human engineered cardiac tissues (hECTs) derived from the hiPSC-CMs indicate that R9C PLB causes an abnormal lusitropic response following β-adrenergic stimulation. Transcriptomic analysis provided additional mechanistic insight, revealing that R9C PLB activates fibrosis, proinflammatory pathways, and cardiac stress in hiPSC-CMs. Excess fibrosis via collagen deposition was also validated in R9C hECTs. Furthermore, small RNA-seq identified multiple miRNAs involved in cardiac metabolism and which are upstream of metabolic genes and transcription factors that are differentially regulated in R9C PLB hiPSC-CMs, indicating an altered metabolic state. These functional and transcriptomic studies represent the first insights into R9C PLB pathology in human cardiomyocytes. 73 5.3 Methods An expanded material and methods can be found in Supplementary Data. 5.3.1 CRISPR transfection of hiPSCs and cardiac differentiation CRISPR transfection of hiPSCs (SKiPS-31.3 line) was performed by electroporation as described.188 Once positive clones were obtained, cardiac differentiation was done as described189,190.91,92 All hiPSC lines were karyotyped and top ten CRISPR off-target sites as determined by COSMID were screened.191 5.3.2 Calcium transient measurements hiPSC-CMs were loaded with a calcium-sensitive fluorescent dye (Fura-2 AM, cell permeant) and the ratios of fluorescence intensities (excitation ratio of 340/380 nm) were recorded using the IonOptix system (Ionoptix, Milton, MA). The electrically-induced calcium transients were triggered by pulses from a MyoPacer (IonOptix, Milton, MA) at 40V and 0.5 Hz and measurements were obtained at room temperature. Calcium traces were analyzed using IonWizard software (IonOptix) to calculate the amplitude (peak height relative to baseline), and tau (time of relaxation). 5.3.3 Single cell calcium measurements Calcium imaging for calcium firing measurements was carried out with a spinning disk confocal microscope using the calcium-sensitive dye, X-Rhod 1, with pacing at 0.25 Hz. Diastolic calcium levels were estimated as the background-corrected fluorescence intensity of X-Rhod 1 without any stimulation. 74 5.3.4 Immunocytochemistry hiPSC-CMs were stained with antibodies against human PLB, cardiac Troponin T, and DAPI using established methods that have been previously published.178 5.3.5 Gene expression RNA was extracted from hiPSC-CMs and cDNA was derived as previously described.192 Gene expression compared to the housekeeping gene β2-microglobulin was determined using the ΔΔCt method. 5.3.6 hECT generation, measurement of developed force, and picrosirius staining hECTs were generated by combining hiPSC-CMs and collagen as previously described.193 Contractile function of the hECT was evaluated using real-time noninvasive optical tracking of the integrated flexible endposts. The measured post deflection was used to calculate developed force by applying a beam-bending equation from elasticity theory as previously described.194,195 Frozen tissue sections were fixed in 4% paraformaldehyde then incubated in picrosirius red. Images of entire sections were obtained as a large tile scan using a Nikon PlanApo x10/0.45 objective on a Nikon Eclipse Ti2 microscope. 5.3.7 RNA sequencing RNA-seq was performed on RNA extracted from hiPSC-CMs at day 37 of differentiation. Poly-A selection and mRNA-SEQ library preparation were performed at the Mount Sinai Genomics Core Facility. The RiboZero Prep was performed for differential gene expression and the Small RNA Prep was performed for differential expression of small RNAs. 75 5.3.8 Statistical analysis All statistics were performed using Prism (GraphPad, La Jolla, CA) and analysis was done by one-way ANOVA followed by the Bonferroni post hoc test; analysis of two group comparisons was done by Student’s t-test (*P < 0.05, **P < 0.01, ***P < 0.001, #P < 0.0001, and ‘ns’ is not significant). Data is presented as mean ± standard error of the mean (SEM) and all values were obtained from a minimum of three separate cardiac differentiations for each clone. 5.4 Results 5.4.1 Derivation of hiPSCs with CRISPR-inserted R9C PLB mutation We generated and characterized hiPSCs (SKiPS-31.3 line) from dermal fibroblasts from a healthy 45-year-old male volunteer with no evidence of cardiovascular disease, as previously described.196 Following transfection with plasmids containing CRISPR/Cas9 + PLB-targeting gRNA and the donor matrix (Figure S42A), positive clones were identified by PCR (Figure S42B) and confirmed by Sanger sequencing (Figure S42C). Positive clones heterozygous for R9C PLB were selected by allelic discrimination using TaqMan probes (Figure S42D), and two isogenic clones were selected for further characterization (R9C2 and R9C3). All clones, including the parental line, were karyotypically normal (Figure S42E). To assess off-target activity of the gRNA, the ten most homologous genomic regions were determined by COSMID191 and sequenced in both R9C clones; no off-target mutations were observed at any locus (Supplementary Table 1). 76 Basal A B ** Isoproterenol Forskolin WT hiPSC-CMs R9C hiPSC-CMs 20 Basal Basal 15 10 O O as % of baseline) of % as Amplitude Amplitude 5 ht F/F F/F Δ Δ Time (s) Time (s) 0 (peak (peak 1.5 + 500nM Iso + 500nM Iso ** /Forskolin /Forskolin 1.0 O O 0.5 F/F F/F (Tau; seconds) (Tau; Δ Δ Time for relaxation for Time Time (s) Time (s) 0.0 Figure 35. R9C hiPSC-CMs demonstrate a blunted response to β-agonist and adenylate cyclase activators. (A) Representative calcium transients of R9C hiPSC-CMs at day 35 of differentiation (stimulated at 40V, paced at 0.5Hz) compared to isogenic wild-type controls under basal conditions (top) and following treatment with 500nM isoproterenol or forskolin (bottom). (B) Quantification of calcium amplitude and time for relaxation (tau) were performed for wild-type and both inserted R9C clones in hiPSC-CMs at day 35. White bar = basal conditions; red bar = 500nM isoproterenol; blue bar = 500nM forskolin (N=10-12). (C) Two representative examples of calcium firing in inserted R9C hiPSC-CMs at day 40 of differentiation (top) and total quantitation of calcium firings of R9C compared to isogenic wild-type hiPSC-CMs (bottom). 5.4.2 R9C PLB impairs calcium handling and β-adrenergic response without relocalization From the edited hiPSCs, we derived cardiomyocytes (hiPSC-CMs) heterozygous for R9C PLB using established protocols189,190 for functional and transcriptomic characterization. PLB plays a critical role in modulating calcium homeostasis through its regulation of SERCA2a calcium sensitivity. We first examined functional effects of the R9C PLB mutation on calcium handling properties of the hiPSC-CMs. Intracellular calcium transients were recorded for wild-type and R9C hiPSC-CMs with electrical stimulation using fura2-AM, a cell permeable fluorescent calcium indicator. We observed 77 that calcium transients were regular in appearance for both wild-type and R9C hiPSC- CMs but R9C hiPSC-CMs exhibited an increased amplitude and decreased time to relaxation under basal conditions compared to wild-type hiPSC-CMs (Figure 35A). Since R9C mutation impacts proper PLB phosphorylation by protein kinase A,111,187 we treated the hiPSC-CMs with either 500nM isoproterenol (a β-agonist) or 500nM forskolin (an adenylate cyclase activator), both of which would result in phosphorylation of PLB and an increase in SERCA2a activity culminating in increased calcium uptake and release in the hiPSC-CMs. Treatment of wild-type hiPSC-CMs with either compound resulted in the expected increase in calcium amplitude and decreased time to relaxation; however, this response was absent in R9C hiPSC-CMs (Figure 35A and B). In fact, R9C hiPSC-CMs appear to be hyper-stimulated, with amplitude and tau values similar to those from wild- type hiPSC-CMs treated with isoproterenol or forskolin. This suggests that R9C PLB prevents proper basal inhibition of SERCA2a by PLB and consequently, calcium transport activation via the β-agonist pathway. 78 Figure 2 A B 1900 ns 2500 2400 1 .5 1850 ) 2300 1800 2200 2100 1 .0 1750 2000 1900 (Hz firings 1700 1800 0 .5 1650 1700 1600 Calcium Calcium Fluorescence Intensity Fluorescence 1600 Intensity Fluorescence 1500 0 .0 0 10000 20000 0 10000 20000 30000 40000 ms ms C D WT-CMs R9C-CMs 6 * ** 4 fluorescence 2 0 Intracellular Intracellular DNA/cTNT/PLB Figure 36. R9C hiPSC-CMs show abnormal calcium handling without PLB relocalization. (A) Two representative examples of raw calcium firings in R9C hiPSC- CMs loaded with X-Rhod1 at day 40 of differentiation (paced at 0.25 Hz). (B) Total quantitation of calcium firings of R9C hiPSC-CMs (day 40) compared to isogenic wild- type hiPSC-CMs paced at 0.25 Hz. Samples that captured correctly at 0.25 Hz are shown as a single point. (C) Diastolic calcium concentration of wild-type and R9C single hiPSC-CMs. (N=12-20). (D) Representative immunofluorescence images showing the intracellular protein distribution of phospholamban (PLB) and cardiac troponin T (cTNT) in wild-type and R9C hiPSC-CMs at day 35. Nuclei were counterstained with DAPI. Calcium transients in single R9C PLB hiPSC-CMs captured the field stimulation irregularly, varying between high frequency and slow, irregular calcium transients (two raw tracings of separate single CMs stimulated at 0.25 Hz are shown in Figure 36A), while most wild-type hiPSC-CMs fired at the expected10 frequencyµm (0.25 Hz; Figure 36B). 10 µm While not significantly different from wild-type hiPSC-CMs, calcium handling is clearly impaired in R9C hiPSC-CMs (Figure 36B), probably due to the highly irregular nature of the calcium firings. This hypothesis is further supported by a significant increase in diastolic calcium levels in R9C hiPSC-CMs compared to wild-type hiPSC-CMs, as evidenced by increased intracellular X-Rhod1 fluorescence (Figure 36C). 79 A WT hECT B 8 1 .5 R9C hECT 7 * 50 ** 6 * * 1 .0 5 4 3 0 .5 relaxation 2 Developed force Developed 1 (normalized to WT) to (normalized 0 WT)to (normalized 0 .0 < 35 > 35 < 35 > 35 to Time < 35 > 35 < 35 > 35 Time post-differentiation (days) Time post-differentiation (days) C D 2 .0 1 .5 *** * 1 .5 beat rate rate beat 1 .0 ns ns 1 .0 0 .5 0 .5 (normalized to WT)to (normalized (normalized to WT)to (normalized 0 .0 50 relaxation to Time Spontaneous 0 .0 Basal + ISO Basal + ISO Basal + ISO Basal + ISO Figure 37. R9C PLN human engineered cardiac tissues (hECTs) have a blunted β- agonist response. hECTs derived from wild-type or inserted R9C hiPSC-CMs were tested during stimulation at 0.5Hz for (A) developed force and (B) time to relaxation (50%) before and after day 35 of differentiation. (C) Spontaneous beat rate and (D) time to relaxation (50%) were determined on wild-type and R9C hECTs after day 35 before and after treatment with 500nM isoproterenol. Immunocytochemistry of R9C hiPSC-CMs showed typical PLB localization and was unremarkable compared to wild-type hiPSC-CMs (Figure 36D), indicating that these abnormalities in calcium handling are not attributed to aberrant PLB localization as is the case with other hereditary PLB mutations (e.g., R14del PLB).178 5.4.3 R9C PLB results in an attenuated chronotropic and lusitropic response to β- agonists in human engineered cardiac tissue Human engineered cardiac tissues (hECTs) were created using established methods194,195 with either wild-type and R9C hiPSC-CMs at Day 17 of differentiation, and functional measurements were performed either before and after day 35 of differentiation (Figure 37A and 3B) or with and without 500nM isoproterenol (Figure 37C and 3D). R9C 80 hECTs demonstrated the same progressive increase in developed force and decrease in time to relaxation 50 over time as wild-type hECTs, although the increase in developed force was not significant (Figure 37A and 3B). This implies that R9C hECTs mature in a similar fashion to wild-type hECTs over time. Upon stimulation with a β-agonist, wild-type hECTs exhibit an increased spontaneous beat rate and decreased time to relaxation. It should be noted that an increase in developed force would be also be expected with a - agonist, but this has been unsuccessfully modeled in the hECT system in our hands (not shown). Similar to what was observed in R9C hiPSC-CMs, R9C hECTs demonstrate an attenuated chronotropic and lusitropic response to the β-agonist isoproterenol (Figure 37C and 3D), and the time to relaxation indicates that R9C is already working at maximum as its basal value is similar to that of wild-type hECTs treated with isoproterenol (Figure 37D). These results further support our hypothesis that R9C PLB results in blunted β-adrenergic signaling. 5.4.4 R9C hiPSC-CMs exhibit activation of hypertrophy markers and increased cell size Gene expression analysis by qPCR revealed an increase in expression in ANF and BNP and a decrease in the α/β MHC ratio, which is indicative of cardiac hypertrophy (Figure 38A). This phenotype was present in hiPSC-CMs 35 days after differentiation but absent 21 days after differentiation (data not shown) signifying a time-dependent onset of phenotype. Cell membrane capacitance, which is an indicator of cell size as measured by patch clamp of single hiPSC-CMs, was significantly increased in single R9C hiPSC-CMs compared to wild-type hiPSC-CMs further indicating the hypertrophic nature of the R9C hiPSC-CMs (Figure 38B). Quantification of cardiomyocyte area was performed using imaging-based techniques and an approximate two-fold increase in cell 81 Figure 38. R9C hiPSC-CMs exhibit a hypertrophic phenotype. (A) Quantitative PCR analysis of gene expression of cardiac hypertrophy markers ANF, BNP, MHC6 and MHC7 in isogenic CMs at day 35 (wild-type and R9C PLN). ANF and BNP gene expression were normalized to β2-microglobulin (β2M) housekeeping gene and shown as relative expression to wild-type hiPSC-CMs. MHC6 and MHC7 expression were normalized to β2M housekeeping gene and presented as MHC6/MHC7 ratio (N=3-5). (B) Cell membrane capacitance measurements of isogenic hiPSC-CMs (measure of cell size) (N=3). (C) Cardiomyocyte size of R9C hiPSC-CMs compared to isogenic wild-type controls. (D) Transcriptomic analysis of R9C hiPSC-CMs confirms hypertrophic phenotype (NPPA=ANF, NPPB=BNP, MYH6/7=MHC6/7). size of R9C hiPSC-CMs compared to wild-type CMs was observed (Figure 38C). Lastly, transcriptomic analysis of differentially expressed genes (DEGs) showed significant perturbation of natriuretic peptides (NPPA and NPPB) and an approximate decrease in the MYH6/MYH7 ratio in R9C compared to wild-type hiPSC-CMs (Figure 38D). In accordance with our findings, biventricular dilation was observed both in patients and transgenic mice harboring the R9C mutation and transgenic R9C mice showed significant enlargement in heart size.111 82 B A CTGF TGFB2 TGFBR2 2.5 2.0 2.5 2M) ** * β ** * 2.0 * 2.0 * 1.5 1.5 1.5 1.0 1.0 1.0 0.5 0.5 0.5 Gene expression Gene 0.0 0.0 0.0 (normalized to to (normalized GDF15 ATF3 E2F1 2.5 3 2.5 * 2M) β * 2.0 * * 2.0 * * 2 1.5 1.5 1.0 1.0 1 0.5 0.5 Gene expression Gene 0.0 0.0 0 (normalized to to (normalized Figure 39. Transcriptional response to R9C PLN of induced cardiomyocytes involves altered metabolic phenotype. (A) Standardized expression of significant DEGs for R9C hiPSC-CMs compared to isogenic control. (B) Quantitative PCR analysis to confirm perturbed expression of profibrotic (TGFβ2 and TGFβR2) and proinflammatory (GDF15 and CTGF) cytokines, and transcription factors involved in fibrosis and cardiac stress (ATF3 and E2F1). 5.4.5 Transcriptomic and functional analysis identifies activation of pro-fibrotic pathways and a shift in metabolism in R9C PLB hiPSC-CMs Transcriptomic analysis was carried out in a single genetic background (R9C PLB inserted into wild-type SKiPS31.3 line by CRISPR/Cas9) so comparative analysis and significant perturbations in DEGs, due exclusively to insertion of R9C PLB, could be accurately quantified. Overall, RNA-seq revealed that R9C PLB results in activation of profibrotic signaling and a significant shift in metabolism (Figure 39A). We observed upregulation of profibrotic (TGFβ2 and TGFβR2) and proinflammatory (GDF15 and CTGF) cytokines, and an upregulation in transcription factors involved in fibrosis and cardiac stress (ATF3 and E2F1). These changes in expression were confirmed by qPCR in both clones of R9C hiPSC-CMs (Figure 39B). Picrosirius staining of hECTs revealed differential staining of type III and type I collagen in wild-type and R9C PLB hECTs (Figure 40A). Type I collagen (red-yellow under polarized light in picrosirius red stained 83 C WT hECT (D49) R9C hECT (D49) 50 µm 50 µm D 1 .5 1 .2 ** *** 1 .0 1 .0 0 .8 0 .6 0 .4 0 .5 0 .2 hue green/red Absolute ratio of ratio Absolute green/yellow hue green/yellow Absolute ratio of ratio Absolute 0 .0 0 .0 Figure 40. Transcriptional response to R9C PLN of induced cardiomyocytes involves profibrotic phenotype. (A) Human engineered cardiac tissues (hECTs) stained with picrosirius red, with images obtained by polarized light microscopy. Type I collagen appears as red and yellow fibers and type III collagen (fibrosis) appears green. (B) Quantification of fibrosis through analysis of different collagen types by their color purity (hue) in polarized light images, expressed as the absolute ratio of the mean green/yellow hue and green/red hue. sections) is intentionally added to hiPSC-CMs at the time of tissue fabrication, therefore fibrosis was evaluated by measuring the relative abundance of type III collagen (green under polarized light in picrosirius red stained sections). Quantitative analysis of images obtained with polarized light microscopy confirmed increased fibrosis in R9C PLB hECTs as exemplified by a significantly increased ratio of green-to-yellow and green-to-red hues, which is indicative of increased type III-to-type I collagen (Figure 40B). RNA-seq also revealed a metabolic shift from aerobic to anaerobic metabolism in R9C hiPSC-CMs (Figure 39A). Genes involved in glucose utilization were upregulated, including GNPNAT1, PGM3, and GFPT1, and the transcription factor Tbx15 (Figure 84 A B R9C/WT 20 miR199a-3p miR29a-3p 4 15 ** *** ** 10 3 10 ** 0 2 5 1 -10 Gene expression Gene 0 0 -20 0 2 4 6 8 10 RNU44) to (normalized Log10(Intensity) miR1-5p miR378a-3p miR378b 1.5 1.5 1.5 1.0 1.0 1.0 * * ** * * 0.5 0.5 0.5 *** Gene expression Gene 0.0 0.0 0.0 (normalized to RNU44) to (normalized Figure 41. Transcriptional response to R9C PLN results in perturbation of miRNAs linked to cardiac metabolism and fibrosis. (A) Differentially expressed miRNAs in R9C PLN hiPSC-CMs compared to the isogenic control. miRNAs expressed at least 2-fold higher or lower in R9CPLN hiPSC-CMs are shown in red. (B) Differentially expressed miRNAs involved in cardiac metabolism or fibrosis (expressed at least 2-fold higher in the PLN mutant). miRNAs in red or blue are upregulated or downregulated, respectively. (C) Quantitative PCR analysis confirming perturbed expression of miRNAs shown in (B). S43). There were also reductions in both PPARα and PPARGC1β, which are involved in fatty acid oxidation, further signaling a switch to anaerobic metabolism. GO pathways analysis confirmed the perturbation of metabolic pathways in R9C PLB (Figure S44). 5.4.6 Small RNA-seq reveals perturbation of miRNAs linked to cardiac metabolism and fibrosis in R9C hiPSC-CMs Small RNA-seq identified several miRNAs that were up- or down-regulated at least 2-fold compared to wild-type in R9C PLB hiPSC-CMs (Figure 41A). In R9C hiPSC-CMs, 85 we identified several miRNAs implicated in cardiac metabolism, including miR-29a (VDAC1/VDAC2/ATP transport, apoptosis), miR-199a (PPARδ/fatty acid oxidation, glycolysis), and miR-378a/b (ATP synthesis) (Figure 41B).197 A decrease in miR-1 expression was also observed, consistent with the hypertrophic phenotype observed in R9C hiPSC-CMs. These changes in expression in miRNAs were confirmed using qPCR in both clones of R9C hiPSC-CMs (Figure 41C). Altogether, these data point to R9C PLB altering cellular metabolism in cardiomyocytes, in addition to cardiomyocyte hypertrophy, fibrosis, and aberrant calcium handling. 5.5 Discussion In this study, we used gene editing in hiPSC-CMs to study the hereditary R9C PLB mutation which has been linked to DCM. This approach alleviates the need for patient sample donation and can also be used to study molecular mechanisms linking an abnormal cardiac phenotype to a particular mutation. Using CRISPR/Cas9, we inserted R9C PLB into an isogenic hiPSC line and observed that the derived cardiomyocytes exhibited a blunted response to β-agonists, abnormal and irregular calcium firings leading to increased diastolic calcium, and a hypertrophic phenotype. R9C PLB hECTs demonstrated normal force development over time compared to wild-type, but confirmed a loss in β-agonist response. RNA-seq of R9C hiPSC-CMs showed an activation of fibrosis, which was also observed in R9C hECTs, and dysregulated metabolism, which was confirmed by perturbation of miRNAs involved in cardiac metabolism. The R9C PLB mutation is quite rare but has a high penetrance, as it is poorly tolerated in patients with DCM.111 It was originally identified in an American family but has since been identified in a cohort of DCM patients in South Africa.198 Our group 86 previously characterized patient-derived R14del PLB hiPSC-CMs;185 the R9C PLB hiPSC-CMs display abnormal calcium handling similar to R14del PLB but without arrhythmia or PLB mislocalization (Figure 35 and Figure 36). Strikingly, R9C PLB appears to blunt β-agonist responses in hiPSC-CMs, which we observed both in hiPSC- CMs and hECTs (Figure 35 and Figure 37). These defects have been previously observed in rabbit cardiomyocytes infected with R9C PLB, which show acute positive inotropic and lusitropic effects yielding negative outcomes in impaired frequency potentiation and blunted β-adrenergic response,185 and in proteoliposomes containing only SERCA1a and R9C PLB, which showed no phosphorylation of R9C PLB by protein kinase A.187 Nuclear magnetic resonance has shown that the R9C mutation shifts the conformational equilibrium of PLB toward a state in which its cytoplasmic domain is membrane-associated,164 which is sometimes associated with SERCA-inhibition. Other studies have shown that R9C PLB stabilizes the PLB pentamer, preventing its phosphorylation by protein kinase A, and that these effects may be enhanced by oxidative stress.184 PLB pentamers have also been shown to reduce phosphorylation of PLB monomers, providing another potential mechanism by which β-adrenergic signaling may be reduced in R9C hiPSC-CMs. In fact, reduced β-adrenergic signaling caused by defective calcium signaling via calsequestrin overexpression has been shown to produce overt heart failure in mouse models;199 however, how this translates to β-adrenergic signaling in human DCM is unknown. Defective calcium handling and β-adrenergic signaling are probably linked to the hypertrophic phenotype we observed in R9C hiPSC- CMs. Expression of hypertrophic markers (ANF, BNP, α/βMYHC), the same trends of which were also observed in RNA-seq, and increased cell size and cell membrane capacitance all point to a hypertrophic phenotype (Figure 38). 87 Another mechanism to explain the absence of chronotropic and lusitropic effects in R9C PLB iPSC and hECT settings, in response to -adrenergic stimulation, is indicated by the observed basal positive chronotropy and lusitropy of R9C PLB. That is, R9C PLB expressed in the WT PLB background accelerates the rhythm and relaxation rate under basal conditions. This is consistent with the reported loss-of-inhibition of R9C PLB and its ability to almost completely compete with WT PLB.187 Loss-of-inhibition PLBs that retain affinity for SERCA comparable to that of WT PLB have been reported64 and they are pursued for gene-therapy potential.106,108,200 The pathological phenotype observed in humans and mice expressing R9C PLB, and at cellular level here, suggests that this PLB mutant are involved in more complex, unresolved interactions, as highlighted by the altered metabolism and profibrotic signaling we observed. RNA-seq revealed that R9C PLB resulted in activation of fibrosis through expression of profibrotic cytokines and mediators of cardiac stress responses, including members of the TGFβ superfamily, proinflammatory cytokines (CTGF, GDF15), and transcription factors (E2F1, ATF3) (Figure 39). Functionally, this fibrotic phenotype was confirmed in hECTs by quantification of type III collagen. These results support the massive cardiac interstitial fibrosis that was observed in both patients and transgenic mice harboring the R9C mutation.111 Through RNA-seq, we also observed perturbation of metabolic pathways, which demonstrated a shift from aerobic to anaerobic metabolism (Figure 39, Figure S43, Figure S44). This shift in metabolism has been observed in the diseased heart and was also identified by single-cell RNA-seq in transgenic R9C PLB mice.186,201 However, a recent study has called into question the design of the TgR9C mouse,202 which demonstrates the need for further experiments in other models to verify previously identified mechanisms and conclusions drawn from this model. Remarkably, the characteristics of the R9C PLB mutation are reminiscent of PLB ablation in mice - the 88 major adaptive change that occurred was with regard to myocardial energetics, specifically to increase ATP synthesis and utilization in an attempt to maintain an increased contractile performance.203 Using small RNA-seq, we identified several miRNAs that have been implicated in metabolic reprogramming during heart failure, such as miR-199a-3p, -1-5p, -29a-3p, and -378a/b. Notably, upregulation of miR-199a has been shown to inhibit glycolysis and PPARδ during heart failure.197 PPARGC1B is a target of miR-199a-3p (www.targetscan.org) and we observe a decrease in PPARGC1B expression by RNA- seq (Figure 39A). miR-29a has been associated with fibrosis in the heart.204 miR-1 is a well-studied cardiac miRNA that has been implicated in dilated cardiomyopathy; decreases in miR-1 have been linked to cardiac hypertrophy.205 Decreases in miR-378 have also been observed in cardiovascular disease and are correlated with occurrence of cardiac hypertrophy.206 Aberrant calcium handling is a hallmark of heart disease. Several recent studies, including our own, have demonstrated that altered calcium signaling can lead to a multitude of functional and transcriptional changes in the heart, promoting other pathological events such as fibrosis and metabolic reprogramming. Using CRISPR/Cas9 to insert the hereditary R9C PLB mutation into an isogenic hiPSC line, we have demonstrated the link between this mutation and calcium dysregulation, β-adrenergic signaling, hypertrophy, fibrosis, and cellular metabolism at both a functional and transcriptional level. These findings may contribute to a better understanding of the pathogenic mechanisms behind the hereditary R9C PLB mutation, potentially identifying novel pathways to target in the treatment of dilated cardiomyopathy. ACKNOWLEDGMENTS 89 This work is supported by the National Institutes of Health (R01 HL117505, HL 132226, HL119046, HL129814, HL128072, HL128099, HL132684, HL131404, and HL135093; and R37 AG026160), and a Transatlantic Fondation Leducq grant (Cellular and Molecular Targets to Promote Therapeutic Cardiac Regeneration). DKC is a fellow of the American Heart Association (15POST25090116). ICT is supported by NIH/NHLBI K01 HL 133424-01. We would like to acknowledge the Microscopy Core and Black Family Stem Cell Institute at the Icahn School of Medicine at Mount Sinai. 90 5.6 Supplementary Information 5.6.1 Materials and Methods 5.6.1.1 hiPSC culture and cardiac differentiation The hiPSC line (SKiPS-31.3) was derived by the reprogramming of human dermal fibroblast obtained from a skin biopsy of a 45-year-old male volunteer with informed consent (Staten Island University Hospital, Staten Island, NY), as described 196,207. The line was propagated under feeder-independent conditions as described 208. hiPSCs were cultured in mTeSR1 media (STEMCELL Technologies, Vancouver, Canada) and maintained on tissue culture plates coated with hESC-qualified Matrigel (Corning, Fisher Scientific) in 5% CO2/5% O2/90% N2 environment at 37°C. Cardiac differentiation of hiPSCs was performed using the monolayer method 189. hiPSC-CMs were kept in basal differentiation medium for up to 6 weeks following the differentiation protocol. All cytokines were purchased from R&D Systems and small molecules were purchased from Sigma-Aldrich. All hiPSC-CM cultures were maintained in 5% CO2/air environment. 5.6.1.2 Construction of CRISPR-inserted PLN hiPSC clones Optimized CRISPR sites to target human PLN on chromosome 6 were designed using the Zhang lab website (http://crispr.mit.edu/). Two 20-nt gRNAs were designed to target the following sequences in hiPSCs: 5`-TTC TTA TAG CTG AGC GAG TG-3` (cleaves PLN at nucleotide 21) and 5`-AGC TTT TGA CGT GCT TGT TG-3` (cleaves PLN at nucleotide 61). Both gRNAs were cloned into the pX330-U6-Chimeric_BB-CBh-hSpCas9 plasmid (gift from Feng Zhang, Addgene #42230) so that they follow the U6 promoter 209. The homologous recombination (HR) donor matrix used was described previously (Karakikes et al. 2015) but was modified to contain either the R9C or R14del PLN 91 mutation and several silent mutations around nucleotide 21 or 61 to prevent cleavage of donor matrix by Cas9. Transfection of CRISPR plasmid and HR matrix and selection of clones was performed as previously described (Karakikes et al. 2015), except with 25 µg pX330 plasmid and 15 µg HR donor matrix. Both gRNAs were tested but only PLN61 gRNA resulted in positive hiPSC clones. Genomic DNA was extracted using the Quick- gDNA Micro Prep kit (Zymo Research, Irvine, CA, USA). Junctions between chromosome 6 and the HR donor matrix were screened by PCR using the forward primer 5`-GAC CAG AAA TAT GCT ATA GGA ACT TAA C-3` located on chromosome 6 upstream of the PLN matrix (PLN genomic positions 10,084–10,111) and the reverse primer 5`-CAG GGC TGC CTT GGA AA-3` located in the PGK promoter of the selection cassette. These PCR products were sequenced (Genewiz, New Jersey, USA) to verify insertion of PLN mutation. In addition, positive clones were verified to be heterozygous for PLN mutation using allele-specific TaqMan probes and primers, which were synthesized by Life Technologies (Carlsbad, CA, USA). For wild-type/R9C PLN, primers are 5`-CTG GTA TCA TGG AGA AAG TCC AAT AC-3` (forward) and 5`- CGT GCT TGT TGA GGC ATT TCA-3` (reverse), and probes are 5`-CTT ATA GCT GAG CGA GTG AG-3` (wild-type; VIC fluorophore) and 5`-TAG CTG AGC AAG TGA G-3` (R9C; FAM fluorophore). For wild-type /R14del PLN, primers are 5`-CTG GTA TCA TGG AGA AAG TCC AAT AC-3` (forward) and 5`-GCA AGA GAC ATA TTA AGA TGA GAC AGA-3` (reverse), and probes are 5`-GCT CAG CTA TAA GAA GAG CCT C-3` (wild-type; VIC fluorophore) and 5`-GCT ATA AGA GCC TCA ACC-3` (R14del; FAM fluorophore). Allelic discrimination was performed according to manufacturer specifications and protocols using an ABI Prism 7500 Fast Real-Time PCR system (Applied Biosystems, Foster City, CA, USA). “Best fit” heterozygous clones were selected by the Applied Biosystems software program. 92 5.6.1.3 CRISPR off-target effects The top 10 unique off-target sites (based on score) for gRNA PLN61 were determined with COSMID (Cradick et al., 2014) using the following criteria: NRG PAM, 3 mismatches with no indels, and 2 mismatches with 1-base deletions or insertions. Primers and targets are listed in Supplementary Table 1. Primers were synthesized by Integrated DNA Technologies and PCR was performed using Qiagen Taq Mastermix. Sequencing was performed using the forward primer for each off-target site and was done by Genewiz. 5.6.1.4 Karyotyping Cytogenetic analysis of hiPSCs was performed by the Mount Sinai Genetic Testing Laboratory (New York, NY). The staining performed was G-bands by trypsin using Giemsa (band level 450). A minimum of 2 cells were karyotyped for each clone and cytogenetic analysis was performed and reviewed by the Laboratory Director and Laboratory Medical Consultant of the Genetic Testing Laboratory. 5.6.1.5 RNA extraction and quantitative RT-PCR Total RNA was extracted from hiPSC-CMs using the Quick-RNA MiniPrep kit (Zymo Research) and quantified using a NanoDrop 2000 (ThermoFisher, Rockville, MD). Reverse transcription was performed using the iScript cDNA synthesis kit (Biorad Laboratories, Hercules, CA) for gene expression and qScript microRNA cDNA Synthesis Kit (Quanta Biosciences, Beverly, MA) for miR quantification (this includes a polyA tailing reaction and first-strand cDNA synthesis reaction). Quantitative PCR (10 ng cDNA/reaction) was performed using a two-step system with SYBR Advantage qPCR 93 Premix (Clontech Laboratories, Mountain View, CA) on the ABI Prism 7500 Fast Real- Time PCR system (Applied Biosystems, Foster City, CA) according to manufacturer recommendations. Fold changes in gene expression were determined using the comparative Ct method (ΔΔCt) with normalization to the housekeeping gene β2M. Fold changes in miR levels were calculated in the same way but normalized to the small nucleolar RNA SNORD44 (primers are included in the qScript microRNA cDNA synthesis kit). Primers used are shown in Supplementary Table 2. 5.6.1.6 Calcium transient analysis At 18 or 30 days post-differentiation, hiPSC-CMs were dissociated using the Detach kit 2 (PromoCell, Heidelberg, Germany), plated on Matrigel-coated coverslips, and cultured in basal media. After 48 hours, the CMs were loaded with a calcium-sensitive fluorescent dye (Fura-2 AM, cell permeant; ThermoFisher, Rockville, MD) and the ratios of fluorescence intensities (excitation ratio of 340/380 nm) were recorded using the IonOptix system (Ionoptix, Milton, MA). The electrically-induced calcium transients were triggered by pulses from a MyoPacer (IonOptix, Milton, MA) at 40V and 0.5 Hz and measurements were obtained at room temperature. Calcium traces were analyzed using IonWizard software (IonOptix) to calculate the baseline, amplitude (peak height), and tau (time of relaxation). For measurements with isoproterenol (DL isoproterenol hydrochloride; Sigma-Aldrich, St. Louis, MO), basal calcium transients were recorded, isoproterenol (final concentration of 500 nM) was added to the cells, and then calcium transients were measured again. These experiments were repeated a minimum of 5 times for each condition. 5.6.1.7 Immunocytochemistry and cell size measurements 94 HiPSC-CMs were dissociated and plated on Matrigel-coated coverslips, fixed in 4% paraformaldehyde and permeabilized in blocking/permeabilization buffer (2% BSA/2% FBS/0.05% NP40 in PBS) for 45 min and incubated with primary antibodies overnight at 4°C. The cells were washed in blocking buffer and incubated with Alexa-conjugated secondary antibodies (Invitrogen) diluted in blocking/permeabilization buffer (1:750). Finally, after washing in PBS the cells were counterstained with 4,6-diamidino-2- phenylindole (DAPI). Immunofluorescence images were acquired using the 63X oil immersion objective (numerical aperture = 1.4) of a Leica SP5 DM confocal microscope (Microscopy Core, Mount Sinai Hospital, NY). The following antibodies were used: The following antibodies were used: rabbit polyclonal anti-cardiac troponin T (1:400, Abcam, ab45932) and mouse monoclonal anti-PLN (1:200, Badrilla, clone A1). Images of hiPSC-CMs stained with cTnT and DAPI were taken at 40X magnification and examined with ImageJ. Cell size was estimated by manually drawing around each cell and exporting cell area for a minimum of 20 cells. Only intact CMs with stained nuclei were used. 5.6.1.8 Isolation of contracting cardiospheres into single cardiomyocytes To obtain single hiPSC-CMs, contracting cardiospheres were digested sequentially with Collagenase IV (1 mg/ml in PBS+/+, 30 min at 37C) and then trypsin-EDTA (0.05 %, 5 min at 37C). Isolated hiPSC-CMs were seeded on Matrigel-coated glass coverslips or glass bottom dish and maintained in culture. Assays were performed 5–7 days post seeding. 5.6.1.9 Single-cell electrophysiology 95 Cell capacitance (Cm) was recorded by whole-cell patch-clamp technique using the EPC10 amplifier and Pulse/PulseFit software (HEKA, Germany) at 37 °C. The patch pipettes were prepared to have a typical resistance of 3–6 MΩ. The internal pipette solution contained 110 mM potassium aspartate, 20 mM KCl, 10 mM HEPES, 1 mM MgCl2, 0.1 mM ATP (disodium salt), 5 mM ATP (magnesium salt), 5 mM phosphocreatine (disodium salt) and 1 mM EGTA (pH 7.2). The external bath solution contained 140 mM NaCl, 5 mM KCl, 1 mM MgCl2, 10 mM D-glucose, 1 mM CaCl2 and 10 mM HEPES (pH 7.4). Calcium imaging assays to measure calcium transients elicited were also performed on single hiPSC-CMs isolated from the differentiated cardiospheres and seeded on Matrigel-coated glass coverslips or glass bottom dish. Assays were performed 5–7 days post seeding (30–50 days post differentiation). Imaging was carried out with a spinning disk confocal microscope (PerkinElmer, USA) at 37 °C. Calcium- sensitive dye, X-Rhod 1 (Invitrogen, 1.5 μM, 15 min at 37°C), was loaded into the isolated CMs, followed by washing and subsequent 5 min incubation in a HEPES buffered bathing solution containing 140 mM NaCl, 5 mM KCl, 1 mM MgCl2, 10 mM D- glucose, 1 mM CaCl2 and 10 mM HEPES (pH 7.4). CMs were paced at 0.25 Hz. Raw calcium transients elicited are shown a fluorescence intensity over time and cumulative data is shown as a single dot for each CM – for example, if the CM captured correctly at the stimulated 0.25 Hz, it is shown as a single dot at 0.25 Hz. Diastolic calcium levels were estimated as the background-corrected fluorescence intensity of X-Rhod 1 without any stimulation. Calcium transient recordings were sampled at 33 Hz. 5.6.1.10 Fabrication of human engineered cardiac tissues (hECTs) On day 22 to 26 post-differentiation, hiPSC-CMs were washed with 1X PBS (no calcium) and enzymatically dissociated using Detach 2 kit (PromoCell). After centrifugation (300g 96 x 5min) cells were resuspended in RPMI+B27 culture media, followed again by centrifugation (300g x 5min) the supernatant medium was removed and the cell pellet was combined with ice cold collagen type I (Sigma-Aldrich, St. Louis, MO) and Matrigel at a ratio of 1:8:1 (v/v/v) as previously described 194,195. The combined cell-matrix suspension was pipetted into custom made polydimethylsiloxane (PDMS) elastomer molds with integrated flexible end posts and removable inserts that form a rectangular well to contain 100μL of the cell-matrix suspension; they were placed in the incubator (37°C and 5% CO2) for 2 hours to allow the collagen to polymerize; afterwards the hECTs were bathed with RPMI+B27 culture medium. After 48 hours the inserts were removed, leaving in place the hECT. For maintenance, the hECT received daily half- medium exchanges of RPMI+B27 culture medium until day five when culture medium was changed to high glucose DMEM supplemented with 10% newborn bovine serum (Atlanta Biologicals, Lawrenceville, GA), 1% penicillin/streptomycin (GIBCO, Carlsbad, CA), and 0.2% amphotericin B (Sigma-Aldrich, St. Louis, MO), and were maintained at 37°C and 5% CO2. 5.6.1.11 Analysis of hECT contractile function Evaluation of contractile function of the hECT was performed using real-time noninvasive optical tracking of the integrated flexible endposts. The measured post deflection was used to calculate developed force by applying a beam-bending equation from elasticity theory as previously described 194,195. The hECT within the PDMS mold was transferred to a laminar airflow hood while submerged in culture medium at 37°C. The post deflection was captured in real-time with a high-speed camera (100 frames/s) and a LabVIEW program (National Instruments, Austin, TX). The contractility of ECTs was evaluated without and with electrical field stimulation at 6 V/cm, at rates of 0.5 Hz - 97 1.0 Hz with 0.25 Hz increments, for spontaneous beat rate and developed force measurements respectively. Response to isoproterenol was evaluated without electrical stimulation. 5.6.1.12 RNA-seq processing and two-factor model Total RNA was extracted from hiPSC-CMs 35 days post-differentiation using a Zymo Quick RNA Mini Prep and 4µg of RNA was used to generate a RNAseq sequencing library. Poly-A selection and mRNA-SEQ library preparation were performed at the Mount Sinai Genomics Core Facility. Sequencing (50 bases, paired ends) was performed using an Illumina HiSeq2500. Annotated reads were obtained using STAR and HTSeq and normalized to full library size. RNA-seq reads from the ribozero protocol were aligned to hg38 using HISAT2 210 with the most recent “genome_snp_tran” index, which takes into account common SNPs. The alignments were assembled using StringTie 210 with default parameters and subsequently extracted as a count matrix at the gene level. The counts were normalized using DESeq2 size factors 211. The R14del data were modeled using a two-factor design in DESeq2 taking into account the background genotype and the status of R14del — irrespective of whether the mutation was naturally occurring or introduced by CRISPR. After correcting for multiple hypotheses testing using Benjamini-Hochberg at FDR < 20% 212, we thus identified differentially expressed genes (DEGs) associated with R14del across 2 background genotypes and at a total sample size of 6. Normalized and pseudo log-transformed counts were used for principal component analysis (PCA). Standardized z-scores and log fold changes where used to visualize gene expression data as heat maps. 98 RNA-seq data from the small RNA preparation protocol were trimmed using Trimmomatic removing the adaptor sequences 213. To detect and quantify both novel and known miRNAs, we used the miRDeep2 software suite 214. First, we aligned trimmed reads to hg38 using Bowtie 215, removed known miRNAs present in miRBase 21 216, and retained miRNAs that had lengths of >16bp and a miRDeep2 score >1. We considered the 77 miRNAs thus identified as novel miRNA candidates in a cardiomyocyte context. To quantify both known and candidate miRNAs we realigned the reads against a combined library of miRBase21 and candidate miRNAs using Bowtie. We analyzed the resulting miRNA counts with the above described two-factor model using DESeq2 and identified miRNAs with a mean fold change of ±2 for crR14del (n=2), pR14del (n=2) and crR9C PLN (n=2) hiPSC-CMs compared to their respective isogenic controls. 5.6.1.13 Picrosirius Red staining and polarized light image analysis Frozen tissue sections were fixed in 4% paraformaldehyde (30 min), washed in distilled water (2 min), then incubated in picrosirius red (60 min) followed by dehydration in 95% ethanol x2 (15 dips), ethanol 100% (1min) cleared in xylene x2 (3 min) and mounted. Images of entire sections were obtained as a large tile scan, using a Nikon PlanApo x10/0.45 objective on a Nikon Eclipse Ti2 microscope. Two matching color images per section were obtained, captured in brightfield and between crossed polarizers in order to visualize the birefringent collagen type I and type III fibers; images were subsequently imported and analyzed using Fiji software to quantify the hue of the colors visualized by polarized light microscopy, in order to separate type I and type III collagen. Initially, the brightfield image was opened in Fiji software and ROIs were defined; collagen types are indistinguishable in brightfield; thus, this step ensures an unbiased 99 selection of the areas for quantification. Subsequently, the corresponding polarized image was opened and the preselected ROIs were chosen and duplicated into individual smaller images. These were transformed from the RGB to the CIELAB color space (CIE 1976, CIELAB) 217. The channels a* and b* of the CIELAB designate the hue, or purity, of the perceived colors on the image, independent of their intensity (contained in the luminance channel L of the CIELAB and not used for the analysis). Negative values of the a* channel indicate the hue of green color whereas positive values of a* indicate the hue of the red color; the ratio of the two within each ROI was calculated as measure of the content between the two types of collagen. Positive values of b* channel indicate the hue of the yellow color; the ratio between the yellow/green content was additionally calculated as a measure between the overlap of the two collagen types (producing a yellow tint) related to the amount of type III collagen (appearing as green). Values of hue within the a* and b* channels were obtained as average numbers within each ROIs, larger than an arbitrary guard threshold value of 10 (or -10 in case the negative scale was used, as in the a* negative indicating the green hue); this step was performed to prevent the use of pixels of low hue values, which would encompass a great deal of uncertainty in terms of the structure producing the perceived color. 5.6.2 Supplementary References: [1] Galende E, Karakikes I, Edelmann L, Desnick RJ, Kerenyi T, Khoueiry G, et al. Amniotic fluid cells are more efficiently reprogrammed to pluripotency than adult cells. Cell Reprogram 2010; 12: 117-25. [2] Karakikes I, Senyei GD, Hansen J, Kong CW, Azeloglu EU, Stillitano F, et al. Small molecule-mediated directed differentiation of human embryonic stem cells toward ventricular cardiomyocytes. Stem cells translational medicine 2014; 3: 18-31. 100 [3] Ludwig TE, Bergendahl V, Levenstein ME, Yu J, Probasco MD, Thomson JA. Feeder-independent culture of human embryonic stem cells. Nat Methods 2006; 3: 637- 46. [4] Bhattacharya S, Burridge PW, Kropp EM, Chuppa SL, Kwok WM, Wu JC, et al. High efficiency differentiation of human pluripotent stem cells to cardiomyocytes and characterization by flow cytometry. J Vis Exp 2014; 52010. [5] Cong L, Ran FA, Cox D, Lin S, Barretto R, Habib N, et al. Multiplex genome engineering using CRISPR/Cas systems. Science 2013; 339: 819-23. [6] Serrao GW, Turnbull IC, Ancukiewicz D, Kim do E, Kao E, Cashman TJ, et al. Myocyte-depleted engineered cardiac tissues support therapeutic potential of mesenchymal stem cells. Tissue engineering. Part A 2012; 18: 1322-33. [7] Mayourian J, Cashman TJ, Ceholski DK, Johnson BV, Sachs D, Kaji DA, et al. Experimental and Computational Insight Into Human Mesenchymal Stem Cell Paracrine Signaling and Heterocellular Coupling Effects on Cardiac Contractility and Arrhythmogenicity. Circ Res 2017; 121: 411-423. [8] Pertea M, Kim D, Pertea GM, Leek JT, Salzberg SL. Transcript-level expression analysis of RNA-seq experiments with HISAT, StringTie and Ballgown. Nat Protoc 2016; 11: 1650-67. [9] Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014; 15: 550. [10] Benjamini Y, Hochberg Y. Controlling the False Discovery Rate - a Practical and Powerful Approach to Multiple Testing. J Roy Stat Soc B Met 1995; 57: 289-300. [11] Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 2014; 30: 2114-20. 101 [12] Friedlander MR, Mackowiak SD, Li N, Chen W, Rajewsky N. miRDeep2 accurately identifies known and hundreds of novel microRNA genes in seven animal clades. Nucleic Acids Res 2012; 40: 37-52. [13] Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol 2009; 10: R25. [14] Kozomara A, Griffiths-Jones S. miRBase: annotating high confidence microRNAs using deep sequencing data. Nucleic Acids Res 2014; 42: D68-73. [15] Snorradottir AO, Isaksson HJ, Kaeser SA, Skodras AA, Olafsson E, Palsdottir A, et al. Deposition of collagen IV and aggrecan in leptomeningeal arteries of hereditary brain haemorrhage with amyloidosis. Brain Res 2013; 1535: 106-14. 102 A B C cleavage WT TACCTCACTCGCTCAGC 1kb R9C TACCTCACTTGCTCAGC PLN Locus WT Exon 2 (chr. 6) Exon 1 - + 2 3 PLN R9C PGK Puro EGFP A Recombination Matrix R9C LoxP LoxP D E 4 R9 C 9 - 3 9 - 2 3 e l e l l A 2 C 9 W T R 1 NTC WT (31.3) R9C2 R9C3 0 0 1 2 3 W T A lle le Figure S42. Hereditary DCM modelled using CRISPR-insertion of the R9C PLB mutation in hiPSCs. (A) Design of the PLB recombination matrix used to insert PLB mutation. For selection of positive colonies, a puromycin resistance gene and eGFP were inserted along with a polyA tail following the human PGK promoter. Validation of positively-transfected hiPSC colonies with proper insertion of PLB recombination matrix was verified by PCR. (B) Sanger sequencing. (C) Allelic discrimination. (D) NTC = no template control, X refers to clones that were eliminated by software. (E) Cytogenetic analysis of all clones reveals a normal male karyotype (46, XY) 103 PPARA PPARGC1A PPARGC1B 2.5 ) 1.5 * 1.5 * 2M 2.0 β 1.0 1.5 1.0 * * *** 1.0 *** expression 0.5 0.5 0.5 Gene 0.0 0.0 0.0 (normalized to to (normalized GNPAT1 PGM3 GFPT1 ) 3 ** 3 4 * ** * 2M * β 3 * 2 2 2 expression 1 1 1 Gene (normalized to to (normalized 0 0 0 ) 3 RXRG TBX15 2M 5 ** β * * ** 4 2 3 expression 1 2 1 Gene (normalized to to (normalized 0 0 Figure S43. Quantitative PCR analysis to confirm perturbation of genes involved in cardiac metabolism, which were identified by RNA-seq. 104 Figure S44. Heatmap of GO pathways showing perturbation of metabolic pathways in R9C PLB. 105 Chapter 6 – Summary and Future Directions The role of Ca in excitation-contraction coupling in the heart was explored in Chapter 1. It is clear that this process is complex and highly regulated involving multiple molecular factors. Of these factors, the contribution of SERCA2a to Ca transport into SR is evident. The key regulator of SERCA2a function in the heart is the small transmembrane protein PLB and recent evidence suggests that a third protein, DWORF, may also be involved in Ca transport. The validation and development of therapies targeting SERCA2a and PLB in order to reverse cardiac contractile dysfunction was explained in Chapter 2. Despite significant progress in the past two decades, there is still an urgent need to develop an agent or therapy that effectively enhances Ca transport. In Chapter 3, the principles of fluorescence were reviewed, including a direct comparison between steady-state and time-resolved fluorescence measurements. Two studies were presented with recent developments towards drug discovery targeting Ca transport (Chapter 4) and insights into the mechanism of hereditary cardiomyopathy progression (Chapter 5). While these studies have advanced high-throughput drug screening technology and resolved questions regarding pathogenesis of hereditary PLB mutations, there are questions unaddressed: Structural model of DWORF-SERCA interaction The mechanism by which DWORF regulates Ca transport and/or SERCA2a is ambiguous (Figure 10). Developing a structural model of DWORF-SERCA2a interaction is an ongoing project that we are approaching using both EPR and TR-FRET. These spectroscopic studies require purified DWORF and we have chosen to use MBP purification strategy (Figure 45). We have successful cloned the DWORF sequence and expressed an MBP-tagged DWORF in E.Coli to be used in EPR and fluorescence studies. 106 Cloning and 32 32 min 14.5 14.5 min 34 min 41 min - - - mutagenesis - DWORF - treated treated - TestTEV rxn TestTEV Flowthrough TEV MBP Expression and Eluant Eluant 29 Eluant 13 Eluant 32 Eluant 34 Purification MBP MBP cleavage, HPLC MBP-DWORF TEV Labeling, Reconstitution Reconstitution WT-DWORF TR-FRET, EPR ATPase Figure 45. DWORF purification. (left) Schematic of purification steps. (right) Eluants collected at various time points during HPLC purification and proteins visualized via Coomassie-staining. Eluant 34-41 min contains separated DWORF. The MBP-tag can be successful excised and separated from DWORF by HPLC (Figure 45). This is crucial step forward for the project. Alignment of the SERCA regulatory peptides reveals intriguing similarities and differences between the inhibitory (PLB) and activating (DWORF) regulators. In particular, PLB has an asparagine residue at position 34, which aligns with an isoleucine residue at position 17 in DWORF. This asparagine residue disrupts key SERCA transport sites though a network of intermolecular contacts. We predict that I17N- DWORF inhibits SERCA, which would support the hypothesis that DWORF binds to the same interface as PLB. 107 References 1 Huxley, A. F. The activation of striated muscle and its mechanical response. Proc R Soc Lond B Biol Sci 178, 1-27 (1971). 2 Huxley, A. F. & Taylor, R. E. Local activation of striated muscle fibres. J Physiol 144, 426-441 (1958). 3 Ishikawa, H. Formation of elaborate networks of T-system tubules in cultured skeletal muscle with special reference to the T-system formation. J Cell Biol 38, 51-66 (1968). 4 Bean, B. P. Classes of calcium channels in vertebrate cells. Annu Rev Physiol 51, 367-384, doi:10.1146/annurev.ph.51.030189.002055 (1989). 5 Zorzato, F. et al. Molecular cloning of cDNA encoding human and rabbit forms of the Ca2+ release channel (ryanodine receptor) of skeletal muscle sarcoplasmic reticulum. J Biol Chem 265, 2244- 2256 (1990). 6 Fabiato, A. Calcium-induced release of calcium from the cardiac sarcoplasmic reticulum. Am J Physiol 245, C1-14, doi:10.1152/ajpcell.1983.245.1.C1 (1983). 7 Bers, D. M. Cardiac excitation-contraction coupling. Nature 415, 198-205, doi:10.1038/415198a (2002). 8 Spudich, J. A. & Watt, S. The regulation of rabbit skeletal muscle contraction. I. Biochemical studies of the interaction of the tropomyosin-troponin complex with actin and the proteolytic fragments of myosin. J Biol Chem 246, 4866-4871 (1971). 9 Adelstein, R. S. & Eisenberg, E. Regulation and kinetics of the actin- myosin-ATP interaction. Annu Rev Biochem 49, 921-956, doi:10.1146/annurev.bi.49.070180.004421 (1980). 10 Takeda, S., Yamashita, A., Maeda, K. & Maeda, Y. Structure of the core domain of human cardiac troponin in the Ca(2+)-saturated form. Nature 424, 35-41, doi:10.1038/nature01780 (2003). 11 Huxley, A. F. & Niedergerke, R. Structural changes in muscle during contraction; interference microscopy of living muscle fibres. Nature 173, 971-973 (1954). 12 Huxley, H. & Hanson, J. Changes in the cross-striations of muscle during contraction and stretch and their structural interpretation. Nature 173, 973-976 (1954). 13 Dobesh, D. P., Konhilas, J. P. & de Tombe, P. P. Cooperative activation in cardiac muscle: impact of sarcomere length. Am J Physiol Heart Circ Physiol 282, H1055-1062, doi:10.1152/ajpheart.00667.2001 (2002). 108 14 Inesi, G. Mechanism of calcium transport. Annu Rev Physiol 47, 573-601, doi:10.1146/annurev.ph.47.030185.003041 (1985). 15 MacLennan, D. H. & Kranias, E. G. Phospholamban: a crucial regulator of cardiac contractility. Nat Rev Mol Cell Biol 4, 566-577, doi:10.1038/nrm1151 (2003). 16 Nelson, B. R. et al. A peptide encoded by a transcript annotated as long noncoding RNA enhances SERCA activity in muscle. Science 351, 271-275, doi:10.1126/science.aad4076 (2016). 17 Cantilina, T., Sagara, Y., Inesi, G. & Jones, L. R. Comparative studies of cardiac and skeletal sarcoplasmic reticulum ATPases. Effect of a phospholamban antibody on enzyme activation by Ca2+. J Biol Chem 268, 17018-17025 (1993). 18 Palmgren, M. G. & Nissen, P. P-type ATPases. Annu Rev Biophys 40, 243-266, doi:10.1146/annurev.biophys.093008.131331 (2011). 19 Toyoshima, C. How Ca2+-ATPase pumps ions across the sarcoplasmic reticulum membrane. Biochim Biophys Acta 1793, 941-946, doi:10.1016/j.bbamcr.2008.10.008 (2009). 20 MacLennan, D. H. & Holland, P. C. Calcium transport in sarcoplasmic reticulum. Annu Rev Biophys Bioeng 4, 377-404, doi:10.1146/annurev.bb.04.060175.002113 (1975). 21 Periasamy, M. & Kalyanasundaram, A. SERCA pump isoforms: their role in calcium transport and disease. Muscle Nerve 35, 430-442, doi:10.1002/mus.20745 (2007). 22 Hasselbach, W. Atp-Driven Active Transport of Calcium in the Membranes of the Sarcoplasmic Reticulum. Proc R Soc Lond B Biol Sci 160, 501-504 (1964). 23 Toyoshima, C., Nomura, H. & Tsuda, T. Lumenal gating mechanism revealed in calcium pump crystal structures with phosphate analogues. Nature 432, 361-368, doi:10.1038/nature02981 (2004). 24 Toyoshima, C. & Mizutani, T. Crystal structure of the calcium pump with a bound ATP analogue. Nature 430, 529-535, doi:10.1038/nature02680 (2004). 25 Takahashi, M., Kondou, Y. & Toyoshima, C. Interdomain communication in calcium pump as revealed in the crystal structures with transmembrane inhibitors. Proc Natl Acad Sci U S A 104, 5800- 5805, doi:10.1073/pnas.0700979104 (2007). 26 Sorensen, T. L., Moller, J. V. & Nissen, P. Phosphoryl transfer and calcium ion occlusion in the calcium pump. Science 304, 1672-1675, doi:10.1126/science.1099366 (2004). 27 Olesen, C. et al. The structural basis of calcium transport by the calcium pump. Nature 450, 1036-1042, doi:10.1038/nature06418 (2007). 109 28 Akin, B. L., Hurley, T. D., Chen, Z. & Jones, L. R. The structural basis for phospholamban inhibition of the calcium pump in sarcoplasmic reticulum. J Biol Chem 288, 30181-30191, doi:10.1074/jbc.M113.501585 (2013). 29 Nissen, P. et al. Structures of the heart specific SERCA2a Ca2+- ATPase. bioRxiv, 344911 (2018). 30 Toyoshima, C., Nakasako, M., Nomura, H. & Ogawa, H. Crystal structure of the calcium pump of sarcoplasmic reticulum at 2.6 A resolution. Nature 405, 647-655, doi:10.1038/35015017 (2000). 31 Toyoshima, C. & Nomura, H. Structural changes in the calcium pump accompanying the dissociation of calcium. Nature 418, 605- 611, doi:10.1038/nature00944 (2002). 32 Toyoshima, C. et al. Modeling of the inhibitory interaction of phospholamban with the Ca2+ ATPase. Proc Natl Acad Sci U S A 100, 467-472, doi:10.1073/pnas.0237326100 (2003). 33 Zamoon, J., Mascioni, A., Thomas, D. D. & Veglia, G. NMR solution structure and topological orientation of monomeric phospholamban in dodecylphosphocholine micelles. Biophys J 85, 2589-2598, doi:10.1016/S0006-3495(03)74681-5 (2003). 34 Vangheluwe, P. et al. Sarcolipin and phospholamban mRNA and protein expression in cardiac and skeletal muscle of different species. Biochem J 389, 151-159, doi:10.1042/BJ20050068 (2005). 35 Movsesian, M. A., Nishikawa, M. & Adelstein, R. S. Phosphorylation of phospholamban by calcium-activated, phospholipid-dependent protein kinase. Stimulation of cardiac sarcoplasmic reticulum calcium uptake. J Biol Chem 259, 8029-8032 (1984). 36 Wegener, A. D. & Jones, L. R. Phosphorylation-induced mobility shift in phospholamban in sodium dodecyl sulfate-polyacrylamide gels. Evidence for a protein structure consisting of multiple identical phosphorylatable subunits. J Biol Chem 259, 1834-1841 (1984). 37 DeSantiago, J., Maier, L. S. & Bers, D. M. Phospholamban is required for CaMKII-dependent recovery of Ca transients and SR Ca reuptake during acidosis in cardiac myocytes. J Mol Cell Cardiol 36, 67-74 (2004). 38 Traaseth, N. J., Buffy, J. J., Zamoon, J. & Veglia, G. Structural dynamics and topology of phospholamban in oriented lipid bilayers using multidimensional solid-state NMR. Biochemistry 45, 13827- 13834, doi:10.1021/bi0607610 (2006). 39 Karim, C. B., Kirby, T. L., Zhang, Z., Nesmelov, Y. & Thomas, D. D. Phospholamban structural dynamics in lipid bilayers probed by a spin label rigidly coupled to the peptide backbone. Proc Natl Acad Sci U S A 101, 14437-14442, doi:10.1073/pnas.0402801101 (2004). 110 40 Kirby, T. L., Karim, C. B. & Thomas, D. D. Electron paramagnetic resonance reveals a large-scale conformational change in the cytoplasmic domain of phospholamban upon binding to the sarcoplasmic reticulum Ca-ATPase. Biochemistry 43, 5842-5852, doi:10.1021/bi035749b (2004). 41 Karim, C. B., Zhang, Z., Howard, E. C., Torgersen, K. D. & Thomas, D. D. Phosphorylation-dependent conformational switch in spin- labeled phospholamban bound to SERCA. J Mol Biol 358, 1032- 1040, doi:10.1016/j.jmb.2006.02.051 (2006). 42 Metcalfe, E. E., Traaseth, N. J. & Veglia, G. Serine 16 phosphorylation induces an order-to-disorder transition in monomeric phospholamban. Biochemistry 44, 4386-4396, doi:10.1021/bi047571e (2005). 43 Gustavsson, M. et al. Lipid-mediated folding/unfolding of phospholamban as a regulatory mechanism for the sarcoplasmic reticulum Ca2+-ATPase. J Mol Biol 408, 755-765, doi:10.1016/j.jmb.2011.03.015 (2011). 44 Gustavsson, M., Traaseth, N. J. & Veglia, G. Activating and deactivating roles of lipid bilayers on the Ca(2+)- ATPase/phospholamban complex. Biochemistry 50, 10367-10374, doi:10.1021/bi200759y (2011). 45 Li, J., James, Z. M., Dong, X., Karim, C. B. & Thomas, D. D. Structural and functional dynamics of an integral membrane protein complex modulated by lipid headgroup charge. J Mol Biol 418, 379- 389, doi:10.1016/j.jmb.2012.02.011 (2012). 46 Simmerman, H. K., Kobayashi, Y. M., Autry, J. M. & Jones, L. R. A leucine zipper stabilizes the pentameric membrane domain of phospholamban and forms a coiled-coil pore structure. J Biol Chem 271, 5941-5946 (1996). 47 Kimura, Y., Kurzydlowski, K., Tada, M. & MacLennan, D. H. Phospholamban inhibitory function is activated by depolymerization. J Biol Chem 272, 15061-15064 (1997). 48 Becucci, L. et al. On the function of pentameric phospholamban: ion channel or storage form? Biophys J 96, L60-62, doi:10.1016/j.bpj.2009.03.013 (2009). 49 Vostrikov, V. V., Mote, K. R., Verardi, R. & Veglia, G. Structural dynamics and topology of phosphorylated phospholamban homopentamer reveal its role in the regulation of calcium transport. Structure 21, 2119-2130, doi:10.1016/j.str.2013.09.008 (2013). 50 Karim, C. B., Stamm, J. D., Karim, J., Jones, L. R. & Thomas, D. D. Cysteine reactivity and oligomeric structures of phospholamban and 111 its mutants. Biochemistry 37, 12074-12081, doi:10.1021/bi980642n (1998). 51 Karim, C. B., Marquardt, C. G., Stamm, J. D., Barany, G. & Thomas, D. D. Synthetic null-cysteine phospholamban analogue and the corresponding transmembrane domain inhibit the Ca-ATPase. Biochemistry 39, 10892-10897 (2000). 52 Simmerman, H. K. & Jones, L. R. Phospholamban: protein structure, mechanism of action, and role in cardiac function. Physiol Rev 78, 921-947, doi:10.1152/physrev.1998.78.4.921 (1998). 53 Tada, M., Ohmori, F., Yamada, M. & Abe, H. Mechanism of the stimulation of Ca2+-dependent ATPase of cardiac sarcoplasmic reticulum by adenosine 3':5'-monophosphate-dependent protein kinase. Role of the 22,000-dalton protein. J Biol Chem 254, 319-326 (1979). 54 Antipenko, A. Y., Spielman, A. I., Sassaroli, M. & Kirchberger, M. A. Comparison of the kinetic effects of phospholamban phosphorylation and anti-phospholamban monoclonal antibody on the calcium pump in purified cardiac sarcoplasmic reticulum membranes. Biochemistry 36, 12903-12910, doi:10.1021/bi971109v (1997). 55 Antipenko, A. Y., Spielman, A. I. & Kirchberger, M. A. Comparison of the effects of phospholamban and jasmone on the calcium pump of cardiac sarcoplasmic reticulum. Evidence for modulation by phospholamban of both Ca2+ affinity and Vmax (Ca) of calcium transport. J Biol Chem 272, 2852-2860 (1997). 56 Negash, S., Chen, L. T., Bigelow, D. J. & Squier, T. C. Phosphorylation of phospholamban by cAMP-dependent protein kinase enhances interactions between Ca-ATPase polypeptide chains in cardiac sarcoplasmic reticulum membranes. Biochemistry 35, 11247-11259, doi:10.1021/bi960864q (1996). 57 Mahaney, J. E., Autry, J. M. & Jones, L. R. Kinetics studies of the cardiac Ca-ATPase expressed in Sf21 cells: new insights on Ca- ATPase regulation by phospholamban. Biophys J 78, 1306-1323, doi:10.1016/S0006-3495(00)76686-0 (2000). 58 Asahi, M., McKenna, E., Kurzydlowski, K., Tada, M. & MacLennan, D. H. Physical interactions between phospholamban and sarco(endo)plasmic reticulum Ca2+-ATPases are dissociated by elevated Ca2+, but not by phospholamban phosphorylation, vanadate, or thapsigargin, and are enhanced by ATP. J Biol Chem 275, 15034-15038 (2000). 59 James, P., Inui, M., Tada, M., Chiesi, M. & Carafoli, E. Nature and site of phospholamban regulation of the Ca2+ pump of sarcoplasmic reticulum. Nature 342, 90-92, doi:10.1038/342090a0 (1989). 112 60 Jones, L. R., Cornea, R. L. & Chen, Z. Close proximity between residue 30 of phospholamban and cysteine 318 of the cardiac Ca2+ pump revealed by intermolecular thiol cross-linking. J Biol Chem 277, 28319-28329, doi:10.1074/jbc.M204085200 (2002). 61 Li, J., Bigelow, D. J. & Squier, T. C. Conformational changes within the cytosolic portion of phospholamban upon release of Ca-ATPase inhibition. Biochemistry 43, 3870-3879, doi:10.1021/bi036183u (2004). 62 James, Z. M., McCaffrey, J. E., Torgersen, K. D., Karim, C. B. & Thomas, D. D. Protein-protein interactions in calcium transport regulation probed by saturation transfer electron paramagnetic resonance. Biophys J 103, 1370-1378, doi:10.1016/j.bpj.2012.08.032 (2012). 63 Mueller, B., Karim, C. B., Negrashov, I. V., Kutchai, H. & Thomas, D. D. Direct detection of phospholamban and sarcoplasmic reticulum Ca-ATPase interaction in membranes using fluorescence resonance energy transfer. Biochemistry 43, 8754-8765, doi:10.1021/bi049732k (2004). 64 Lockamy, E. L., Cornea, R. L., Karim, C. B. & Thomas, D. D. Functional and physical competition between phospholamban and its mutants provides insight into the molecular mechanism of gene therapy for heart failure. Biochem Biophys Res Commun 408, 388- 392, doi:10.1016/j.bbrc.2011.04.023 (2011). 65 Hou, Z., Kelly, E. M. & Robia, S. L. Phosphomimetic mutations increase phospholamban oligomerization and alter the structure of its regulatory complex. J Biol Chem 283, 28996-29003, doi:10.1074/jbc.M804782200 (2008). 66 Dong, X. & Thomas, D. D. Time-resolved FRET reveals the structural mechanism of SERCA-PLB regulation. Biochem Biophys Res Commun 449, 196-201, doi:10.1016/j.bbrc.2014.04.166 (2014). 67 Kho, C., Lee, A. & Hajjar, R. J. Altered sarcoplasmic reticulum calcium cycling--targets for heart failure therapy. Nat Rev Cardiol 9, 717-733, doi:10.1038/nrcardio.2012.145 (2012). 68 Hasenfuss, G. & Pieske, B. Calcium cycling in congestive heart failure. J Mol Cell Cardiol 34, 951-969 (2002). 69 Gwathmey, J. K. et al. Abnormal intracellular calcium handling in myocardium from patients with end-stage heart failure. Circ Res 61, 70-76 (1987). 70 Gwathmey, J. K. & Morgan, J. P. Altered calcium handling in experimental pressure-overload hypertrophy in the ferret. Circ Res 57, 836-843 (1985). 113 71 Beuckelmann, D. J., Nabauer, M. & Erdmann, E. Intracellular calcium handling in isolated ventricular myocytes from patients with terminal heart failure. Circulation 85, 1046-1055 (1992). 72 Beuckelmann, D. J. & Erdmann, E. Ca(2+)-currents and intracellular [Ca2+]i-transients in single ventricular myocytes isolated from terminally failing human myocardium. Basic Res Cardiol 87 Suppl 1, 235-243 (1992). 73 del Monte, F. et al. Restoration of contractile function in isolated cardiomyocytes from failing human hearts by gene transfer of SERCA2a. Circulation 100, 2308-2311 (1999). 74 Dixon, I. M., Lee, S. L. & Dhalla, N. S. Nitrendipine binding in congestive heart failure due to myocardial infarction. Circ Res 66, 782-788 (1990). 75 Gengo, P. J. et al. Myocardial beta adrenoceptor and voltage sensitive calcium channel changes in a canine model of chronic heart failure. J Mol Cell Cardiol 24, 1361-1369 (1992). 76 Colston, J. T., Kumar, P., Chambers, J. P. & Freeman, G. L. Altered sarcolemmal calcium channel density and Ca(2+)-pump ATPase activity in tachycardia heart failure. Cell Calcium 16, 349-356 (1994). 77 Meyer, M. et al. Alterations of sarcoplasmic reticulum proteins in failing human dilated cardiomyopathy. Circulation 92, 778-784 (1995). 78 Schillinger, W. et al. Unaltered ryanodine receptor protein levels in ischemic cardiomyopathy. Mol Cell Biochem 160-161, 297-302 (1996). 79 Sainte Beuve, C. et al. Cardiac calcium release channel (ryanodine receptor) in control and cardiomyopathic human hearts: mRNA and protein contents are differentially regulated. J Mol Cell Cardiol 29, 1237-1246 (1997). 80 Arai, M., Alpert, N. R., MacLennan, D. H., Barton, P. & Periasamy, M. Alterations in sarcoplasmic reticulum gene expression in human heart failure. A possible mechanism for alterations in systolic and diastolic properties of the failing myocardium. Circ Res 72, 463-469 (1993). 81 Studer, R. et al. Gene expression of the cardiac Na(+)-Ca2+ exchanger in end-stage human heart failure. Circ Res 75, 443-453 (1994). 82 Hasenfuss, G. et al. Relation between myocardial function and expression of sarcoplasmic reticulum Ca(2+)-ATPase in failing and nonfailing human myocardium. Circ Res 75, 434-442 (1994). 114 83 Linck, B. et al. Messenger RNA expression and immunological quantification of phospholamban and SR-Ca(2+)-ATPase in failing and nonfailing human hearts. Cardiovasc Res 31, 625-632 (1996). 84 Flesch, M. et al. Evidence for functional relevance of an enhanced expression of the Na(+)-Ca2+ exchanger in failing human myocardium. Circulation 94, 992-1002 (1996). 85 Reinecke, H., Studer, R., Vetter, R., Holtz, J. & Drexler, H. Cardiac Na+/Ca2+ exchange activity in patients with end-stage heart failure. Cardiovasc Res 31, 48-54 (1996). 86 Hasenfuss, G. et al. Relationship between Na+-Ca2+-exchanger protein levels and diastolic function of failing human myocardium. Circulation 99, 641-648 (1999). 87 Hasenfuss, G., Pieske, B., Holubarsch, C., Alpert, N. R. & Just, H. Excitation-contraction coupling and contractile protein function in failing and nonfailing human myocardium. Adv Exp Med Biol 346, 91-100 (1993). 88 Hajjar, R. J., Schmidt, U., Kang, J. X., Matsui, T. & Rosenzweig, A. Adenoviral gene transfer of phospholamban in isolated rat cardiomyocytes. Rescue effects by concomitant gene transfer of sarcoplasmic reticulum Ca(2+)-ATPase. Circ Res 81, 145-153 (1997). 89 Kagaya, Y., Hajjar, R. J., Gwathmey, J. K., Barry, W. H. & Lorell, B. H. Long-term angiotensin-converting enzyme inhibition with fosinopril improves depressed responsiveness to Ca2+ in myocytes from aortic-banded rats. Circulation 94, 2915-2922 (1996). 90 del Monte, F. et al. Improvement in survival and cardiac metabolism after gene transfer of sarcoplasmic reticulum Ca(2+)-ATPase in a rat model of heart failure. Circulation 104, 1424-1429 (2001). 91 Byrne, M. J. et al. Recirculating cardiac delivery of AAV2/1SERCA2a improves myocardial function in an experimental model of heart failure in large animals. Gene Ther 15, 1550-1557, doi:10.1038/gt.2008.120 (2008). 92 Kawase, Y. et al. Reversal of cardiac dysfunction after long-term expression of SERCA2a by gene transfer in a pre-clinical model of heart failure. J Am Coll Cardiol 51, 1112-1119, doi:10.1016/j.jacc.2007.12.014 (2008). 93 Hajjar, R. J. et al. Design of a phase 1/2 trial of intracoronary administration of AAV1/SERCA2a in patients with heart failure. J Card Fail 14, 355-367, doi:10.1016/j.cardfail.2008.02.005 (2008). 94 Jaski, B. E. et al. Calcium upregulation by percutaneous administration of gene therapy in cardiac disease (CUPID Trial), a 115 first-in-human phase 1/2 clinical trial. J Card Fail 15, 171-181, doi:10.1016/j.cardfail.2009.01.013 (2009). 95 Jessup, M. et al. Calcium Upregulation by Percutaneous Administration of Gene Therapy in Cardiac Disease (CUPID): a phase 2 trial of intracoronary gene therapy of sarcoplasmic reticulum Ca2+-ATPase in patients with advanced heart failure. Circulation 124, 304-313, doi:10.1161/CIRCULATIONAHA.111.022889 (2011). 96 Zsebo, K. et al. Long-term effects of AAV1/SERCA2a gene transfer in patients with severe heart failure: analysis of recurrent cardiovascular events and mortality. Circ Res 114, 101-108, doi:10.1161/CIRCRESAHA.113.302421 (2014). 97 Greenberg, B. et al. Calcium upregulation by percutaneous administration of gene therapy in patients with cardiac disease (CUPID 2): a randomised, multinational, double-blind, placebo- controlled, phase 2b trial. Lancet 387, 1178-1186, doi:10.1016/S0140-6736(16)00082-9 (2016). 98 Greenberg, B. et al. Prevalence of AAV1 neutralizing antibodies and consequences for a clinical trial of gene transfer for advanced heart failure. Gene Ther 23, 313-319, doi:10.1038/gt.2015.109 (2016). 99 Cornea, R. L. et al. High-throughput FRET assay yields allosteric SERCA activators. J Biomol Screen 18, 97-107, doi:10.1177/1087057112456878 (2013). 100 Hou, Z. et al. 2-Color calcium pump reveals closure of the cytoplasmic headpiece with calcium binding. PLoS One 7, e40369, doi:10.1371/journal.pone.0040369 (2012). 101 Gruber, S. J. et al. Discovery of enzyme modulators via high- throughput time-resolved FRET in living cells. J Biomol Screen 19, 215-222, doi:10.1177/1087057113510740 (2014). 102 Schaaf, T. M. et al. High-Throughput Spectral and Lifetime-Based FRET Screening in Living Cells to Identify Small-Molecule Effectors of SERCA. SLAS Discov 22, 262-273, doi:10.1177/1087057116680151 (2017). 103 Watanabe, A. et al. Phospholamban ablation by RNA interference increases Ca2+ uptake into rat cardiac myocyte sarcoplasmic reticulum. J Mol Cell Cardiol 37, 691-698, doi:10.1016/j.yjmcc.2004.06.009 (2004). 104 Soller, K. J. et al. Rheostatic Regulation of the SERCA/Phospholamban Membrane Protein Complex Using Non- Coding RNA and Single-Stranded DNA oligonucleotides. Sci Rep 5, 13000, doi:10.1038/srep13000 (2015). 116 105 He, H. et al. Effects of mutant and antisense RNA of phospholamban on SR Ca(2+)-ATPase activity and cardiac myocyte contractility. Circulation 100, 974-980 (1999). 106 Hoshijima, M. et al. Chronic suppression of heart-failure progression by a pseudophosphorylated mutant of phospholamban via in vivo cardiac rAAV gene delivery. Nat Med 8, 864-871, doi:10.1038/nm739 (2002). 107 Ikeda, Y. et al. Altered membrane proteins and permeability correlate with cardiac dysfunction in cardiomyopathic hamsters. Am J Physiol Heart Circ Physiol 278, H1362-1370, doi:10.1152/ajpheart.2000.278.4.H1362 (2000). 108 Kaye, D. M. et al. Percutaneous cardiac recirculation-mediated gene transfer of an inhibitory phospholamban peptide reverses advanced heart failure in large animals. J Am Coll Cardiol 50, 253-260, doi:10.1016/j.jacc.2007.03.047 (2007). 109 Gruber, S. J., Haydon, S. & Thomas, D. D. Phospholamban mutants compete with wild type for SERCA binding in living cells. Biochem Biophys Res Commun 420, 236-240, doi:10.1016/j.bbrc.2012.02.125 (2012). 110 Haghighi, K. et al. Human phospholamban null results in lethal dilated cardiomyopathy revealing a critical difference between mouse and human. J Clin Invest 111, 869-876, doi:10.1172/JCI17892 (2003). 111 Schmitt, J. P. et al. Dilated cardiomyopathy and heart failure caused by a mutation in phospholamban. Science 299, 1410-1413, doi:10.1126/science.1081578 (2003). 112 Haghighi, K. et al. A mutation in the human phospholamban gene, deleting arginine 14, results in lethal, hereditary cardiomyopathy. Proc Natl Acad Sci U S A 103, 1388-1393, doi:10.1073/pnas.0510519103 (2006). 113 Lakowicz, J. R. Principles of Fluorescence Spectroscopy. 3rd edn, (Springer, 2006). 114 Pollok, B. A. & Heim, R. Using GFP in FRET-based applications. Trends Cell Biol 9, 57-60 (1999). 115 Bidwell, P., Blackwell, D. J., Hou, Z., Zima, A. V. & Robia, S. L. Phospholamban binds with differential affinity to calcium pump conformers. J Biol Chem 286, 35044-35050, doi:10.1074/jbc.M111.266759 (2011). 116 Arkin, M. R. & Wells, J. A. Small-molecule inhibitors of protein- protein interactions: progressing towards the dream. Nat Rev Drug Discov 3, 301-317, doi:10.1038/nrd1343 (2004). 117 117 Wells, J. A. & McClendon, C. L. Reaching for high-hanging fruit in drug discovery at protein-protein interfaces. Nature 450, 1001-1009, doi:10.1038/nature06526 (2007). 118 Eisner, D. A., Caldwell, J. L., Kistamas, K. & Trafford, A. W. Calcium and Excitation-Contraction Coupling in the Heart. Circulation research 121, 181-195, doi:10.1161/CIRCRESAHA.117.310230 (2017). 119 MacLennan, D. H. & Kranias, E. G. Phospholamban: a crucial regulator of cardiac contractility. Nat Rev Mol Cell Biol 4, 566-577, doi:10.1038/nrm1151 (2003). 120 Lipskaia, L., Chemaly, E. R., Hadri, L., Lompre, A. M. & Hajjar, R. J. Sarcoplasmic reticulum Ca(2+) ATPase as a therapeutic target for heart failure. Expert Opin Biol Ther 10, 29-41, doi:10.1517/14712590903321462 (2010). 121 Lompre, A. M. et al. Ca2+ cycling and new therapeutic approaches for heart failure. Circulation 121, 822-830, doi:10.1161/CIRCULATIONAHA.109.890954 (2010). 122 Schwinger, R. H. et al. Unchanged protein levels of SERCA II and phospholamban but reduced Ca2+ uptake and Ca(2+)-ATPase activity of cardiac sarcoplasmic reticulum from dilated cardiomyopathy patients compared with patients with nonfailing hearts. Circulation 92, 3220-3228 (1995). 123 Gwathmey, J. K. et al. Abnormal intracellular calcium handling in myocardium from patients with end-stage heart failure. Circulation research 61, 70-76 (1987). 124 Hasenfuss, G. Alterations of calcium-regulatory proteins in heart failure. Cardiovasc Res 37, 279-289 (1998). 125 Hasenfuss, G. Animal models of human cardiovascular disease, heart failure and hypertrophy. Cardiovasc Res 39, 60-76 (1998). 126 Hoshijima, M. et al. Chronic suppression of heart-failure progression by a pseudophosphorylated mutant of phospholamban via in vivo cardiac rAAV gene delivery. Nat Med 8, 864-871, doi:10.1038/nm739 (2002). 127 Miyamoto, M. I. et al. Adenoviral gene transfer of SERCA2a improves left-ventricular function in aortic-banded rats in transition to heart failure. Proceedings of the National Academy of Sciences of the United States of America 97, 793-798 (2000). 128 Schmidt, U. et al. Restoration of diastolic function in senescent rat hearts through adenoviral gene transfer of sarcoplasmic reticulum Ca(2+)-ATPase. Circulation 101, 790-796 (2000). 129 Byrne, M. J. et al. Recirculating cardiac delivery of AAV2/1SERCA2a improves myocardial function in an experimental model of heart 118 failure in large animals. Gene Ther 15, 1550-1557, doi:10.1038/gt.2008.120 (2008). 130 Zsebo, K. et al. Long-term effects of AAV1/SERCA2a gene transfer in patients with severe heart failure: analysis of recurrent cardiovascular events and mortality. Circulation research 114, 101- 108, doi:10.1161/CIRCRESAHA.113.302421 (2014). 131 Greenberg, B. et al. Prevalence of AAV1 neutralizing antibodies and consequences for a clinical trial of gene transfer for advanced heart failure. Gene Ther 23, 313-319, doi:10.1038/gt.2015.109 (2016). 132 Johnson, R. G., Jr. Pharmacology of the cardiac sarcoplasmic reticulum calcium ATPase-phospholamban interaction. Annals of the New York Academy of Sciences 853, 380-392 (1998). 133 Cornea, R. L. et al. High-Throughput FRET Assay Yields Allosteric SERCA Activators. J Biomol Screen, doi:1087057112456878 [pii] 10.1177/1087057112456878 (2012). 134 Gruber, S. J. et al. Discovery of enzyme modulators via high- throughput time-resolved FRET in living cells. J Biomol Screen 19, 215-222, doi:10.1177/1087057113510740 (2014). 135 Petersen, K. J. et al. Fluorescence lifetime plate reader: resolution and precision meet high-throughput. Rev Sci Instrum 85, 113101, doi:10.1063/1.4900727 (2014). 136 Schaaf, T. M. et al. High-Throughput Spectral and Lifetime-Based FRET Screening in Living Cells to Identify Small-Molecule Effectors of SERCA. SLAS Discov 22, 262-273, doi:10.1177/1087057116680151 (2017). 137 Hou, Z. et al. Two-color calcium pump reveals closure of the cytoplasmic headpiece with calcium binding. PLoS ONE, accepted (2012). 138 Robia, S. L. et al. Forster transfer recovery reveals that phospholamban exchanges slowly from pentamers but rapidly from the SERCA regulatory complex. Circ Res 101, 1123-1129 (2007). 139 Pankiv, S. et al. p62/SQSTM1 binds directly to Atg8/LC3 to facilitate degradation of ubiquitinated protein aggregates by autophagy. J Biol Chem 282, 24131-24145, doi:10.1074/jbc.M702824200 (2007). 140 Nakamura, N. et al. Endosomes are specialized platforms for bacterial sensing and NOD2 signalling. Nature 509, 240-244, doi:10.1038/nature13133 (2014). 141 Aleksunes, L. M., Augustine, L. M., Scheffer, G. L., Cherrington, N. J. & Manautou, J. E. Renal xenobiotic transporters are differentially expressed in mice following cisplatin treatment. Toxicology 250, 82- 88, doi:10.1016/j.tox.2008.06.009 (2008). 119 142 Muretta, J. M. et al. High -performance time-resolved fluorescence by direct waveform recording. Rev Sci Instrum 81, 103101-103101 - 103101-103108 (2010). 143 Schaaf, T. M., Peterson, K. C., Grant, B. D., Thomas, D. D. & Gillispie, G. D. Spectral Unmixing Plate Reader: High-Throughput, High-Precision FRET Assays in Living Cells. SLAS Discov 22, 250- 261, doi:10.1177/1087057116679637 (2017). 144 Zhang, J. H., Chung, T. D. & Oldenburg, K. R. A Simple Statistical Parameter for Use in Evaluation and Validation of High Throughput Screening Assays. J Biomol Screen 4, 67-73, doi:10.1177/108705719900400206 (1999). 145 Nesmelov, Y. E. et al. Structural kinetics of myosin by transient time- resolved FRET. Proceedings of the National Academy of Sciences of the United States of America 108, 1891-1896, doi:10.1073/pnas.1012320108 (2011). 146 Muretta, J. M., Petersen, K. J. & Thomas, D. D. Direct real-time detection of the actin-activated power stroke within the myosin catalytic domain. Proceedings of the National Academy of Sciences of the United States of America 110, 7211-7216, doi:10.1073/pnas.1222257110 (2013). 147 Fruen, B. R., Bardy, J. M., Byrem, T. M., Strasburg, G. M. & Louis, C. F. Differential Ca(2+) sensitivity of skeletal and cardiac muscle ryanodine receptors in the presence of calmodulin. American journal of physiology. Cell physiology 279, C724-733 (2000). 148 Bidwell, P., Blackwell, D. J., Hou, Z., Zima, A. V. & Robia, S. L. Phospholamban binds with differential affinity to calcium pump conformers. J Biol Chem 286, 35044-35050, doi:10.1074/jbc.M111.266759 (2011). 149 Lockwood, N. A. et al. Structure and function of integral membrane protein domains resolved by peptide-amphiphiles: application to phospholamban. Biopolymers 69, 283-292, doi:10.1002/bip.10365 (2003). 150 Zamoon, J., Nitu, F., Karim, C., Thomas, D. D. & Veglia, G. Mapping the interaction surface of a membrane protein: Unveiling the conformational switch of phospholamban in calcium pump regulation. Proc Natl Acad Sci U S A 102, 4747-4752 (2005). 151 Blackwell, D. J., Zak, T. J. & Robia, S. L. Cardiac Calcium ATPase Dimerization Measured by Cross-Linking and Fluorescence Energy Transfer. Biophys J 111, 1192-1202, doi:10.1016/j.bpj.2016.08.005 (2016). 120 152 Li, J., Xiong, Y., Bigelow, D. J. & Squier, T. C. Phospholamban binds in a compact and ordered conformation to the Ca-ATPase. Biochemistry 43, 455-463, doi:10.1021/bi035424v (2004). 153 Augustin, J. M., Kuzina, V., Andersen, S. B. & Bak, S. Molecular activities, biosynthesis and evolution of triterpenoid saponins. Phytochemistry 72, 435-457, doi:10.1016/j.phytochem.2011.01.015 (2011). 154 Dandapani, S., Rosse, G., Southall, N., Salvino, J. M. & Thomas, C. J. Selecting, Acquiring, and Using Small Molecule Libraries for High- Throughput Screening. Curr Protoc Chem Biol 4, 177-191, doi:10.1002/9780470559277.ch110252 (2012). 155 Hughes, J. P., Rees, S., Kalindjian, S. B. & Philpott, K. L. Principles of early drug discovery. Br J Pharmacol 162, 1239-1249, doi:10.1111/j.1476-5381.2010.01127.x (2011). 156 Bilmen, J. G., Wootton, L. L. & Michelangeli, F. The inhibition of the sarcoplasmic/endoplasmic reticulum Ca2+-ATPase by macrocyclic lactones and cyclosporin A. Biochem J 366, 255-263, doi:10.1042/BJ20020431 (2002). 157 Morgan, J. P., Erny, R. E., Allen, P. D., Grossman, W. & Gwathmey, J. K. Abnormal intracellular calcium handling, a major cause of systolic and diastolic dysfunction in ventricular myocardium from patients with heart failure. Circulation 81, III21-32 (1990). 158 Schmidt, U. et al. Contribution of abnormal sarcoplasmic reticulum ATPase activity to systolic and diastolic dysfunction in human heart failure. J Mol Cell Cardiol 30, 1929-1937, doi:10.1006/jmcc.1998.0748 (1998). 159 Hasenfuss, G. et al. Calcium handling proteins in the failing human heart. Basic Res Cardiol 92 Suppl 1, 87-93 (1997). 160 Hasenfuss, G. et al. Calcium cycling proteins and force-frequency relationship in heart failure. Basic Res Cardiol 91 Suppl 2, 17-22 (1996). 161 del Monte, F. et al. Abrogation of ventricular arrhythmias in a model of ischemia and reperfusion by targeting myocardial calcium cycling. Proc Natl Acad Sci U S A 101, 5622-5627, doi:10.1073/pnas.0305778101 (2004). 162 Liu, G. S. et al. A novel human R25C-phospholamban mutation is associated with super-inhibition of calcium cycling and ventricular arrhythmia. Cardiovasc Res 107, 164-174, doi:10.1093/cvr/cvv127 (2015). 163 Haghighi, K. et al. Human phospholamban null results in lethal dilated cardiomyopathy revealing a critical difference between 121 mouse and human. J Clin Invest 111, 869-876, doi:10.1172/JCI17892 (2003). 164 Nelson, S. E. D. et al. Effects of the Arg9Cys and Arg25Cys mutations on phospholamban's conformational equilibrium in membrane bilayers. Biochim Biophys Acta 1860, 1335-1341, doi:10.1016/j.bbamem.2018.02.030 (2018). 165 Pimenta, P. H., Silva, C. L. & Noel, F. Ivermectin is a nonselective inhibitor of mammalian P-type ATPases. Naunyn Schmiedebergs Arch Pharmacol 381, 147-152, doi:10.1007/s00210-009-0483-z (2010). 166 Giuliano, K. A. & Taylor, D. L. Fluorescent-protein biosensors: new tools for drug discovery. Trends Biotechnol 16, 135-140, doi:10.1016/S0167-7799(97)01166-9 (1998). 167 Rohde, J. A., Thomas, D. D. & Muretta, J. M. Heart failure drug changes the mechanoenzymology of the cardiac myosin powerstroke. Proc Natl Acad Sci U S A 114, E1796-E1804, doi:10.1073/pnas.1611698114 (2017). 168 Lo, C. H. et al. An Innovative High-Throughput Screening Approach for Discovery of Small Molecules That Inhibit TNF Receptors. SLAS Discov 22, 950-961, doi:10.1177/2472555217706478 (2017). 169 Rebbeck, R. T. et al. High-Throughput Screens to Discover Small- Molecule Modulators of Ryanodine Receptor Calcium Release Channels. SLAS discovery : advancing life sciences R & D 22, 176- 186, doi:10.1177/1087057116674312 (2017). 170 Vogel, S. S., Thaler, C. & Koushik, S. V. Fanciful FRET. Sci STKE 2006, re2, doi:10.1126/stke.3312006re2 (2006). 171 Vogel, S. S., Nguyen, T. A., van der Meer, B. W. & Blank, P. S. The impact of heterogeneity and dark acceptor states on FRET: implications for using fluorescent protein donors and acceptors. PLoS One 7, e49593, doi:10.1371/journal.pone.0049593 (2012). 172 Svensson, B., Autry, J. M. & Thomas, D. D. Molecular Modeling of Fluorescent SERCA Biosensors. Methods Mol Biol 1377, 503-522, doi:10.1007/978-1-4939-3179-8_42 (2016). 173 Pallikkuth, S. et al. Phosphorylated phospholamban stabilizes a compact conformation of the cardiac calcium-ATPase. Biophys J 105, 1812-1821, doi:10.1016/j.bpj.2013.08.045 (2013). 174 Hershberger, R. E., Hedges, D. J. & Morales, A. Dilated cardiomyopathy: the complexity of a diverse genetic architecture. Nat Rev Cardiol 10, 531-547, doi:10.1038/nrcardio.2013.105 (2013). 175 Schonberger, J. & Seidman, C. E. Many roads lead to a broken heart: the genetics of dilated cardiomyopathy. Am J Hum Genet 69, 249-260, doi:S0002-9297(07)61072-6 [pii] 122 10.1086/321978 (2001). 176 Kranias, E. G. & Hajjar, R. J. Modulation of cardiac contractility by the phospholamban/SERCA2a regulatome. Circ Res 110, 1646- 1660, doi:10.1161/CIRCRESAHA.111.259754 (2012). 177 Medeiros, A. et al. Mutations in the human phospholamban gene in patients with heart failure. Am Heart J 162, 1088-1095 e1081, doi:10.1016/j.ahj.2011.07.028 S0002-8703(11)00621-1 [pii] (2011). 178 Karakikes, I. et al. Correction of human phospholamban R14del mutation associated with cardiomyopathy using targeted nucleases and combination therapy. Nat Commun 6, 6955, doi:10.1038/ncomms7955 (2015). 179 van der Heijden, J. F. & Hassink, R. J. The phospholamban p.Arg14del founder mutation in Dutch patients with arrhythmogenic cardiomyopathy. Neth Heart J 21, 284-285, doi:10.1007/s12471- 013-0413-z (2013). 180 Liu, G. S. et al. A novel human R25C-phospholamban mutation is associated with super-inhibition of calcium cycling and ventricular arrhythmia. Cardiovasc Res 107, 164-174, doi:10.1093/cvr/cvv127 (2015). 181 Young, H. S., Ceholski, D. K. & Trieber, C. A. Deception in simplicity: hereditary phospholamban mutations in dilated cardiomyopathy. Biochem Cell Biol 93, 1-7, doi:10.1139/bcb-2014-0080 (2015). 182 Ceholski, D. K., Trieber, C. A. & Young, H. S. Hydrophobic imbalance in the cytoplasmic domain of phospholamban is a determinant for lethal dilated cardiomyopathy. J Biol Chem 287, 16521-16529, doi:10.1074/jbc.M112.360859 M112.360859 [pii] (2012). 183 Schmitt, J. P. et al. Alterations of phospholamban function can exhibit cardiotoxic effects independent of excessive sarcoplasmic reticulum Ca2+-ATPase inhibition. Circulation 119, 436-444, doi:10.1161/CIRCULATIONAHA.108.783506 (2009). 184 Ha, K. N. et al. Lethal Arg9Cys phospholamban mutation hinders Ca2+-ATPase regulation and phosphorylation by protein kinase A. Proc Natl Acad Sci U S A 108, 2735-2740, doi:10.1073/pnas.1013987108 (2011). 185 Abrol, N., de Tombe, P. P. & Robia, S. L. Acute inotropic and lusitropic effects of cardiomyopathic R9C mutation of phospholamban. J Biol Chem 290, 7130-7140, doi:10.1074/jbc.M114.630319 (2015). 123 186 Burke, M. A. et al. Molecular profiling of dilated cardiomyopathy that progresses to heart failure. JCI Insight 1, doi:10.1172/jci.insight.86898 (2016). 187 Ceholski, D. K., Trieber, C. A., Holmes, C. F. & Young, H. S. Lethal, hereditary mutants of phospholamban elude phosphorylation by protein kinase A. J Biol Chem 287, 26596-26605, doi:10.1074/jbc.M112.382713 (2012). 188 Costa, M. et al. A method for genetic modification of human embryonic stem cells using electroporation. Nat Protoc 2, 792-796, doi:nprot.2007.105 [pii] 10.1038/nprot.2007.105 (2007). 189 Bhattacharya, S. et al. High efficiency differentiation of human pluripotent stem cells to cardiomyocytes and characterization by flow cytometry. J Vis Exp, 52010, doi:10.3791/52010 (2014). 190 Burridge, P. W. et al. Chemically defined generation of human cardiomyocytes. Nat Methods 11, 855-860, doi:10.1038/nmeth.2999 (2014). 191 Cradick, T. J., Qiu, P., Lee, C. M., Fine, E. J. & Bao, G. COSMID: A Web-based Tool for Identifying and Validating CRISPR/Cas Off- target Sites. Mol Ther Nucleic Acids 3, e214, doi:10.1038/mtna.2014.64 (2014). 192 Ceholski, D. K. et al. CXCR4 and CXCR7 play distinct roles in cardiac lineage specification and pharmacologic beta-adrenergic response. Stem Cell Res 23, 77-86, doi:10.1016/j.scr.2017.06.015 (2017). 193 Mayourian, J. et al. Exosomal microRNA-21-5p Mediates Mesenchymal Stem Cell Paracrine Effects on Human Cardiac Tissue Contractility. Circ Res 122, 933-944, doi:10.1161/CIRCRESAHA.118.312420 (2018). 194 Serrao, G. W. et al. Myocyte-depleted engineered cardiac tissues support therapeutic potential of mesenchymal stem cells. Tissue engineering. Part A 18, 1322-1333, doi:10.1089/ten.TEA.2011.0278 (2012). 195 Mayourian, J. et al. Experimental and Computational Insight Into Human Mesenchymal Stem Cell Paracrine Signaling and Heterocellular Coupling Effects on Cardiac Contractility and Arrhythmogenicity. Circ Res 121, 411-423, doi:10.1161/CIRCRESAHA.117.310796 (2017). 196 Karakikes, I. et al. Small molecule-mediated directed differentiation of human embryonic stem cells toward ventricular cardiomyocytes. Stem Cells Transl Med 3, 18-31, doi:10.5966/sctm.2013-0110 (2014). 124 197 Pinti, M. V., Hathaway, Q. A. & Hollander, J. M. Role of microRNA in metabolic shift during heart failure. Am J Physiol Heart Circ Physiol 312, H33-H45, doi:10.1152/ajpheart.00341.2016 (2017). 198 Fish, M. et al. Mutation analysis of the phospholamban gene in 315 South Africans with dilated, hypertrophic, peripartum and arrhythmogenic right ventricular cardiomyopathies. Sci Rep 6, 22235, doi:10.1038/srep22235 (2016). 199 Cho, M. C. et al. Defective beta-adrenergic receptor signaling precedes the development of dilated cardiomyopathy in transgenic mice with calsequestrin overexpression. J Biol Chem 274, 22251- 22256 (1999). 200 Ha, K. N. et al. Controlling the inhibition of the sarcoplasmic Ca2+- ATPase by tuning phospholamban structural dynamics. J Biol Chem 282, 37205-37214, doi:10.1074/jbc.M704056200 (2007). 201 van Bilsen, M., Smeets, P. J., Gilde, A. J. & van der Vusse, G. J. Metabolic remodelling of the failing heart: the cardiac burn-out syndrome? Cardiovasc Res 61, 218-226 (2004). 202 Kraev, A. Insertional Mutagenesis Confounds the Mechanism of the Morbid Phenotype of a PLN(R9C) Transgenic Mouse Line. J Card Fail 24, 115-125, doi:10.1016/j.cardfail.2017.12.009 (2018). 203 Chu, G. et al. Compensatory mechanisms associated with the hyperdynamic function of phospholamban-deficient mouse hearts. Circ Res 79, 1064-1076 (1996). 204 Sassi, Y. et al. Cardiac myocyte miR-29 promotes pathological remodeling of the heart by activating Wnt signaling. Nat Commun 8, 1614, doi:10.1038/s41467-017-01737-4 (2017). 205 Small, E. M., Frost, R. J. & Olson, E. N. MicroRNAs add a new dimension to cardiovascular disease. Circulation 121, 1022-1032, doi:10.1161/CIRCULATIONAHA.109.889048 (2010). 206 Ganesan, J. et al. MiR-378 controls cardiac hypertrophy by combined repression of mitogen-activated protein kinase pathway factors. Circulation 127, 2097-2106, doi:10.1161/CIRCULATIONAHA.112.000882 (2013). 207 Galende, E. et al. Amniotic fluid cells are more efficiently reprogrammed to pluripotency than adult cells. Cell Reprogram 12, 117-125, doi:10.1089/cell.2009.0077 (2010). 208 Ludwig, T. E. et al. Feeder-independent culture of human embryonic stem cells. Nat Methods 3, 637-646, doi:10.1038/nmeth902 (2006). 209 Cong, L. et al. Multiplex genome engineering using CRISPR/Cas systems. Science 339, 819-823, doi:10.1126/science.1231143 science.1231143 [pii] (2013). 125 210 Pertea, M., Kim, D., Pertea, G. M., Leek, J. T. & Salzberg, S. L. Transcript-level expression analysis of RNA-seq experiments with HISAT, StringTie and Ballgown. Nat Protoc 11, 1650-1667, doi:10.1038/nprot.2016.095 (2016). 211 Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550, doi:10.1186/s13059-014-0550-8 (2014). 212 Benjamini, Y. & Hochberg, Y. Controlling the False Discovery Rate - a Practical and Powerful Approach to Multiple Testing. J Roy Stat Soc B Met 57, 289-300 (1995). 213 Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114-2120, doi:10.1093/bioinformatics/btu170 (2014). 214 Friedlander, M. R., Mackowiak, S. D., Li, N., Chen, W. & Rajewsky, N. miRDeep2 accurately identifies known and hundreds of novel microRNA genes in seven animal clades. Nucleic Acids Res 40, 37- 52, doi:10.1093/nar/gkr688 (2012). 215 Langmead, B., Trapnell, C., Pop, M. & Salzberg, S. L. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol 10, R25, doi:10.1186/gb-2009-10-3-r25 (2009). 216 Kozomara, A. & Griffiths-Jones, S. miRBase: annotating high confidence microRNAs using deep sequencing data. Nucleic Acids Res 42, D68-73, doi:10.1093/nar/gkt1181 (2014). 217 Snorradottir, A. O. et al. Deposition of collagen IV and aggrecan in leptomeningeal arteries of hereditary brain haemorrhage with amyloidosis. Brain Res 1535, 106-114, doi:10.1016/j.brainres.2013.08.029 (2013). 126