Molecular Bases of PDE4D Inhibition by Memory-Enhancing GEBR

Publication Date (Web): April 13, 2018 ... Indeed, Burgin et al. demonstrated that it is possible to design allosteric modulators that are partially s...
0 downloads 4 Views 2MB Size
Subscriber access provided by UNIV OF NEW ENGLAND ARMIDALE

Molecular bases of PDE4D inhibition by GEBR-library compounds for cognitive amelioration in Alzheimer’s disease Tommaso Prosdocimi, Luca Mollica, Stefano Donini, Marta S Semrau, Anna Paola Lucarelli, Egidio Aiolfi, Andrea Cavalli, Paola Storici, Silvana Alfei, Chiara Brullo, Olga Bruno, and Emilio Parisini Biochemistry, Just Accepted Manuscript • DOI: 10.1021/acs.biochem.8b00288 • Publication Date (Web): 13 Apr 2018 Downloaded from http://pubs.acs.org on April 16, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Molecular bases of PDE4D inhibition by memory-enhancing GEBR-library compounds.

Tommaso Prosdocimi,1 Luca Mollica,2 Stefano Donini,1 Marta S. Semrau,3 Anna Paola Lucarelli,1 Egidio Aiolfi,1 Andrea Cavalli,2,4 Paola Storici,3 Silvana Alfei,5 Chiara Brullo,5 Olga Bruno,5 Emilio Parisini1,*

1

Center for Nano Science and Technology @ PoliMi, Istituto Italiano di Tecnologia, via Giovanni Pascoli 70/3,

20133 Milano, Italy 2

Computational Sciences, Istituto Italiano di Tecnologia, via Morego, 30, 16163 Genova, Italy

3

Elettra-Sincrotrone Trieste S.C.p.A., SS 14 - km 163,5 in AREA Science Park, 34149 Trieste, Italy.

4

Department of Pharmacy and Biotechnology, Alma Mater Studiorum, University of Bologna, via Belmeloro 6,

40126 Bologna, Italy. 5

Department of Pharmacy, School of Medical and Pharmaceutical Sciences, University of Genova, Viale

Benedetto XV 3, 16132 Genova, Italy.

*To whom correspondence should be addressed: Dr. Emilio Parisini, Center for Nano Science and Technology @ PoliMi, Istituto Italiano di Tecnologia, via G. Pascoli 70/3, 20133 Milano, Italy. Phone: +39-02-23999870. Fax: +39-02-23999866. E-mail: [email protected]

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 34

Abstract Selected members of the large rolipram-related GEBR family of phosphodiesterase-4 (PDE4) inhibitors have been shown to facilitate long term potentiation (LTP) and improve memory functions without causing emeticlike behavior in rodents. Despite their micromolar-range binding affinities and their promising pharmacological and toxicological profiles, little if any structure-activity relationship studies have so far been carried out in order to elucidate the molecular bases of their action. Here, we report the crystal structure of a number of GEBR library compounds in complex with the catalytic domain of PDE4D as well as their inhibitory profiles for both the long PDE4D3 isoform and the catalytic domain alone. Furthermore, we assessed the stability of the observed ligand conformations in the context of the intact enzyme using molecular dynamics simulations. The longer and more flexible ligands appear to be capable of forming contacts with the regulatory portion of the enzyme, thus possibly allowing some degree of selectivity between the different PDE4 isoforms.

Abbreviations: PDE4: Phosphodiesterase 4; PDE4D3: Phosphodiesterase 4, D3 isoform Keywords: PDE4, Alzheimer, GEBR, Phosphodiesterase, Inhibitor

ACS Paragon Plus Environment

Page 3 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Introduction

Long term potentiation (LTP), the neurological mechanism governing the learning and memory processes, depends critically on the cerebral levels of the second messenger cyclic adenosine monophosphate (cAMP) and on the correct functioning of the cAMP/PKA/CREB pathway. In the brain, cAMP levels are regulated by the activity of the type 4 phosphodiesterase (PDE4), an enzyme that hydrolyzes cAMP to 5’-AMP.1–4 Consistently, PDE4 is considered an important pharmaceutical target due to its crucial involvement in the signaling of the central nervous system.5,6 Indeed, some PDE4 inhibitors (PDE4Is) developed over time have been shown to improve memory and cognitive functions both under physiological and pathological conditions.7,8 Moreover, some PDE4Is are also used for the treatment of inflammatory diseases, such as for instance respiratory disorders (COPD, asthma) and autoimmune diseases (rheumatoid arthritis, multiple sclerosis, atopic dermatitis).9–15 PDE4Is have mostly been designed to interact with the catalytic pocket of the enzyme and compete with the cAMP hydrolysis process. As a result, albeit providing interesting pro-cognitive and anti-depressant properties, current standard inhibitors such as rolipram16 show also severe side effects due to their lack of isoform-specific binding properties.17–20 In fact, most PDE4-selective inhibitors show less than a tenfold difference in potency for the different PDE4 isoforms .11,21,22 While the four naturally-occurring PDE4 isoforms (A, B, C and D) show a high sequence similarity (78%) in their catalytic domain, they differ in two important regions: the Upstream Conserved Region (containing the UCR1 and UCR2 domains), which affect enzyme activity and dimerization, and the C-terminal sequence, which contains the Control Region 3 (CR3) (Fig. 1A).2,3,5,6

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 34

Fig. 1. A) Schematic representation of a PDE4 long isoform. B) Schematic structural representation of the high HARBS (left) and the LARBS (right) in long and in super-short isoforms, respectively.

As many as 20 PDE4D isoforms that are generated by alternative mRNA splicing have so far been identified, each of them having a specific pattern of expression. While the long isoforms of the gene include both UCR1 and UCR2 regulatory domains, shorter isoforms are characterized by either the absence of UCR1 or the absence of both UCR1 and a portion of UCR2. Further variability can be observed at the level of the N-terminal sequence, which is responsible for subcellular localization and protein-protein interactions.5 Both the UCR2 and the CR3 domains have been shown to be involved in the capping of the catalytic pocket of the enzyme. Indeed, Burgin et al. demonstrated that it is possible to design allosteric modulators that are partially specific for PDE4D and that are capable of stabilizing the closed conformation of the UCR2 domain of PDE4D over the catalytic pocket through the clamping of a central phenylalanine (Tyr in other PDE4 isoforms) by two conserved hydrophobic amino acids near the active site (Ile502 and Phe538).22 Likewise, a similar concept has been suggested also for the design of PDE4B inhibitors that stabilize the capping of the catalytic domain by CR3.23 An X-ray crystal structure of the PDE4B isoform that includes the UCR1, the UCR2, and the catalytic domain has been recently reported.24 The study provides important details of the interaction between the UCR2 and the catalytic domain and establishes that this interaction regulates the catalytic activity of PDE4. The structure reveals that, in long isoforms, UCR1 and UCR2 mediate enzyme dimerization and that the UCR2 of one monomer interacts with the catalytic domain of the other monomer, partially blocking access of either cAMP or PDE4 inhibitors to the catalytic pocket. As a result, long dimeric isoforms, and short monomeric ones feature substantially different enzymatic and pharmacologic properties. In the dimer, an α-helical portion of the UCR2 caps the catalytic pocket, giving rise to a more extended binding site relative to the short monomeric isoforms, originating the distinction between high and low affinity rolipram binding site (HARBS and LARBS) (Fig. 1B). Indeed, in the literature the possibility of designing inhibitors that are capable of docking inside the

catalytic pocket of the enzyme and at the same time interact with the capping domain has been explored with the aim of providing selectivity for the long isoforms. ACS Paragon Plus Environment

Page 5 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

The association of PDE4D with cognition processes has been demonstrated in knock-out mice, which have shown memory improvements similar to those shown by mice treated with rolipram.25 Moreover, activating missense mutations affecting PDE4D activity were recently discovered to be associated with acrodysostosis, a rare genetic disease that brings also a phenotype of mental retardation.26,27 On the other hand, PDE4B inhibition has shown important antidepressant-like properties in mice, where PDE4D inhibition failed, but no pro-cognitive properties.28 In an effort to develop a drug that selectively targets the PDE4 isoforms without side effects, a number of compounds, commonly referred to as the GEBR library, that show partial selectivity for the PDE4D isoform have been synthesized. Selective inhibition of the PDE4D isoform has been shown to represent a possible strategy for achieving memory improvements along with reduced side effects (mostly emesis and sedation).17,29 The reference inhibitors of the GEBR library are GEBR-7b30 and GEBR-32a31, which have been shown to improve spatial and objects recognition memory and to increase hippocampal levels of cAMP in transgenic mice.17,29,30 By addressing the biochemical behavior of these compounds both at a structural and functional level, we aim at identifying key structural features that may provide the molecules with the ability to bind stably into the catalytic pocket while selectively interacting with the regulatory domains of the enzyme. The relatively high number of compounds synthesized to date, some of which have interesting pharmacological and toxicological profiles, have so far provided useful, albeit not definitive, information about the inhibitors-enzyme interaction. Therefore, we set out to investigate the structure-function relationship for some selected GEBR compounds and contribute to the elucidation of the molecular bases of PDE4 inhibition. Experimental methods PDE4D catalytic domain expression and purification The pET15b vector (Novagen) containing the cDNA encoding the human full length PDE4D3 fused at its C terminus with a 6His-tag, was purchased from Eurofins MWG. The fragment of DNA encoding for the catalytic domain (cat), from amino acids 244-578, was PCR amplified from the full length PDE4D3 construct and cloned by conventional methods into a pET3a vector (Novagen) maintaining the C-terminal 6His-tag, thus generating the vector pET3a-PDE4D cat. The primers used are shown in Table S1 of the Supplementary Information. The ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 34

resulting construct was sequence-verified and transformed into E. coli BL21(DE3) pLysS cells (Invitrogen) for overexpression. The E. coli cells transformed with pET3a-PDE4D cat were cultured at 37°C in Luria-Bertani (LB) broth, supplemented with 50 mg/liter ampicillin, until the OD600 value reached 0.6. Then, overnight protein expression at 25°C was induced by adding isopropyl 1-thio-β-D-galactopyranoside (IPTG) to a final concentration of 0.5 mM. Cells were then harvested by centrifugation and resuspended in buffer A (20 mM Tris HCl pH 7.5, 150 mM NaCl) supplemented with 0.5 mM PMSF and DNase (Sigma-Aldrich), and lysed by sonication on ice. The lysate was clarified by centrifugation and the soluble fraction was loaded onto a Ni-NTA (Qiagen) column equilibrated with buffer A. The Ni-NTA beads were first washed using 5 column volumes of the same buffer supplemented with 40 mM imidazole, then the bound C-terminal His-tagged enzyme was eluted with same buffer supplemented with 400mM imidazole. The protein was then further purified by gel filtration on a Sephacryl 100 HR HiPrep 26/60 column (GE Healthcare), followed by an anion exchange purification step on a HiPrep Q HP 16/10 column (GE Healthcare); finally, the purified protein was dialyzed against buffer A. The final sample was brought to two different concentrations: 10 mg/ml for crystallization experiments and 0.025 mg/ml for inhibition assays. Protein concentration was assessed using a Bradford assay32 kit (Bio-Rad) and bovine serum albumin (Sigma) as the standard, and sample purity was assessed by 10% acrylamide SDS-PAGE.

PDE4D3 expression and purification The codon-optimized gene encoding for human PDE4D3 with a C-terminal 6His-tag inserted into the pFastBac Dual vector was purchased from GenScript. The construct carries both the PKA phosphomimetic mutation Ser54Asp and the Ser579Ala mutation, which prevents an inactivating phosphorylation.22 The plasmid was transformed into E.coli DH10EMBacY cells (kindly provided by I. Berger, University of Bristol, UK) to obtain the recombinant bacmid-DNA33. Transfection was performed in Sf9 cells (Expression Systems) plated in 6-well

ACS Paragon Plus Environment

Page 7 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

plates (1.5 x 106 cells/well) using FuGENE HT reagent (Promega). The recombinant baculovirus was harvested at 55h post-transfection (P0 stock). High-titer virus stock was generated by two rounds of amplification and used for protein expression. The Sf9 cells were infected at initial density of 1.5 x 106 / ml and harvested by centrifugation 72h post-infection. Cells were resuspended in buffer B (50 mM Hepes pH 7.5, 500 mM NaCl, 10 mM MgCl2, 10% glycerol, TCEP 1 mM). The buffer was supplemented with a protease inhibitor cocktail (Roche) and cells were lysed by sonication. The lysate was incubated for 5 minutes with benzonase nuclease and then clarified by centrifugation. Ni-NTA resin (Qiagen), pre-equilibrated with buffer C (Hepes 50 mM pH 7.5, 150 mM NaCl, 50 mM arginine, 10% glycerol, 1 mM TCEP), was added to the supernatant and incubated for 30 minutes. The protein was washed (in batch) with the same buffer supplemented with 40 mM imidazole and then loaded onto the column for gravity elution with the same buffer plus 400 mM imidazole. The flow-through of the first Ni-NTA chromatography step was then utilized to perform a second gravity Ni-NTA affinity chromatography step. The protein was then dialyzed overnight to buffer D (50 mM Hepes pH 7.5, 150 mM NaCl, 1 mM DTT, 10% glycerol) and further purified by ion-exchange chromatography buffer E (50 mM Hepes pH 7.5, 1M NaCl, 1 mM DTT, 10% glycerol). Protein concentration and sample purity were assessed as described above for the PDE4D catalytic domain. Crystallization and X-ray crystallography Apo crystals of PDE4D catalytic domain were grown by the hanging drop diffusion technique mixing 1 µL of PDE4 catalytic domain at a concentration of circa 10 mg/ml with an equal volume of a crystallization buffer containing 10-20% PEG 3350, 10-100 mM Hepes pH 7.5, 100-200 mM MgCl2. Crystals were soaked for 1-3 days in solutions equal to half concentration of all the components of the crystallization buffer plus 1-6 mM inhibitor. The soaked crystals were finally cryo-protected in the soaking buffer plus 20% v/v glycerol or 20% ethylene glycol and flash-frozen in liquid nitrogen. Diffraction datasets were collected at a wavelength of 1 Å at the X06DA-PXIII beamline at Swiss Light Source, Paul Scherrer Institute, Villigen, Switzerland. Diffractions images were processed with XDS. Datasets were phased by molecular replacement using the crystal structure of the PDE4D catalytic domain (PDB code: 2PW3) and all the refinements were carried out with Phenix.34 Crystallographic data are shown in Table 2. ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 34

Table 2. Crystallographic table. Data in parentheses refer to the outer resolution shell. PDE4 activity and inhibition assays Enzyme activity was determined using a coupled assay where the AMP formed after the hydrolysis of cAMP was detected by a NADH-coupled enzyme system35. The assay was adapted to support high throughput format in a 96 transparent polystyrene plates from Grainer Bio. The 250 µL reaction mixture contained 100 mM Hepes pH 7.5, 10 mM MgCl2, 10 mM KCl, 0.3 mM ATP, 0.6 mM PEP, 0.9 U of Pyruvate Kinase, 0.9 U of Lactate Dehydrogenase, 0.8 U Myokinase and the PDE4 enzyme at an estimated final concentration of 10 nM. After 20 minutes of pre-incubation at 25°C, the reaction was started by adding cAMP at a final concentration of 0.004 mM, and the oxidation of NADH to NAD+ was monitored at 340nm (NADH ε342= 6220 M-1cm-1) in a Spark10M (Tecan). One unit (U) is defined as the amount of enzyme required to hydrolyze 1 µmol of cAMP to 5’-AMP per minute, and specific activity was expressed as U mg-1 of enzyme. Inhibition assays were performed as described above varying the compound concentrations ranging from 0.01 μM to 300 μM. Compounds were prepared in DMSO, and diluted in assay buffer to yield in a final concentration of 1%. Finally, the concentration of compound required to reduce the fractional enzyme activity of half of its initial value (IC50) was determined by ACS Paragon Plus Environment

Page 9 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

plotting the enzyme fractional activity against the logarithm of compound concentration. Curve fitting was performed with a dose response curve from Sigma Plot (Systat Software), with Eq. 1:

(Eq.1)

y = min + ((max - min) / (1+10(logIC50-x))

In which y is the fractional activity of the enzyme in the presence of inhibitor at concentration [I], max is the maximum value of y observed at [I]=0, and min is the minimum limiting value of y at higher inhibitor concentration. In all the listed experiments, data points were obtained from three independent experiments.

Molecular dynamics simulations The full length PDE4D structure was modeled using the automatic procedure implemented in the Swiss-model server.36 The regulatory domains were modeled based on the PDE4B crystal structure (PDB ID: 4WZI).24 Loops were further modeled using Scaled Molecular Dynamics (SMD).37,38Simulations on PDE4D in its holo form were done with λ = 0.7 in order to enhance the sampling of the dihedral angles belonging to the loop region: 3 simulations of 100 ns each were performed as previously reported in literature39 in order to efficiently flood the conformational space for the loop.38 This was done by keeping the protein backbone atoms restrained (i.e. using an isotropically-applied force constant of 500 kJmol-1nm-1) except those of the loops (i.e., residues 202 to 253 and 106 to 136). For each loop, the average starting structure for the subsequent MD simulations was obtained from the cluster analysis performed on all these trajectories, which was based on the backbone RMSD distance between local conformers of the unrestrained region only (“single linkage” method, as implemented in the GROMACS suite40). For each loop a single conformer has been selected on the basis of the structure corresponding to the center of the most populated structure (in all the cases with a poupation higher than 75% ), whereas the rest of the structure has been maintained as reported in the original experimental structure. The structure of the full length PDE4D3 in complex with all the ligands reported in the present work was reconstructed merging the experimental ligand pose of the catalytic domain-ligand complex with the model structure of the protein holo form derived as explained above. The metal coordination was preserved by means

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 34

of harmonic restrains40 (with a 1000 kJ mol-1 nm-1 force constant) applied between the metal ions and the coordinating atoms as determined by the crystallographic structures. Each compound was geometrically optimized prior to MD simulations via a quantum mechanical approach: electron density calculations were performed in Gaussian 0941 using the basis sets 6-31G* or 6-31g++ at the Hartree-Fock level of theory. Partial charges were derived using RESP method42 as implemented in Antechamber, leading via a GAFF parameterization to a complete topological description of each ligand to be used for classical simulations. The experimental protein−ligand complexes were then used as a starting point for MD simulations performed in a GROMACS 4.6.140 version implemented in BiKi Life Sciences 1.3 (BiKi Technologies, Genoa, Italy).43 The force field used for all the classical simulations reported in the present work was the Amber ff99SB*-ILDN.44All the complexes were first placed in the geometrical center of parallelepiped-shaped boxes of ∼2500 nm3 volume, then solvated using tLeap, with ∼80000 TIP3P water molecules.45 Some water molecules were replaced with sodium ions in order to preserve the electro neutrality of the system according to need, i.e., the charge of the protein plus the charge of the ligand, which varied. The system was minimized with the steepest descent method, followed by equilibration of the restrained protein (isotropic 1000 kJ mol-1 nm-1 force applied to each heavy atom of the protein backbone) in NPT (up to 400 ps, pressure = 1 atm) and NVT (up to 400 ps) ensembles at 300 K via a standard MD procedure. Electrostatics were treated with the cutoff method for short-range interactions and with the Particle Mesh Ewald method for the long-range ones (rlist = 1 nm, cutoff distance = 0.9 nm, vdW distance = 0.9 nm, PME order = 4)46. The constant temperature conditions were provided by using the V-rescale thermostat47, a modification from Berendsen’s coupling algorithm. The production run was performed for 200 ns for each system. All the simulations were set up using the BiKi software package and performed on a set of inhouse machines, equipped with two esacore Intel Xeon processors and 2 NVIDIA GTX 780 GPUs, for a total of ~800 CPU days.

ACS Paragon Plus Environment

Page 11 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Result and discussion As part of our ongoing investigation of a large library of PDE4D inhibitors, we report the structural characterization of the interaction of the catalytic domain of the enzyme with a number of GEBR family inhibitors (Table 1).

Table 1. The GEBR library compounds analyzed in this study. In this table, molecules are grouped based on the different chemical nature of their linker moiety.

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 34

This subset of inhibitors consists of rolipram-like derivatives that are based on three fundamentally different scaffolds, one represented by the parent compound GEBR-7b, one by the parent compound GEBR-54 and one by GEBR-26g alone. These compounds feature a common catecholic (rolipram-like) portion and different linker and amino terminal groups. Interestingly, within our selection of ligands, there are three pairs of very similar compounds differing only by the fluorination of the methoxy group in the catecholic moiety of the molecule (GEBR-7b vs. GEBR-20b, GEBR-54 vs. GEBR-32a and GEBR-4a vs. GEBR-11b). In the following, we will refer to the portion of the ligand exceeding the catechol moiety as the tail of the ligand, while the aliphatic/heterocyclic portion of the tail will be referred to as the linker. Within the PDE catalytic domain, the active site consists of a 15 Å-deep pocket that is lined by highly conserved residues across the different PDE isoforms. The catalytic pocket can be subdivided into three contiguous sectors, commonly referred to as the metal binding pocket (M), the Q switch and P clamp pocket (Q) and the solventfilled side pocket (S)48. The metal binding pocket nests two metal ions (Zn and Mg), which are octahedrally coordinated by conserved His and Asp residues as well as water molecules. Next to the M pocket is the Q switch and P clamp pocket, featuring a Gln residue that acts as the purine-specific coordination site that guides the docking of cAMP into the catalytic site. Moreover, the Q pocket features two conserved hydrophobic residues (Ile502 and Phe538) that stabilize the substrate (P clamp). Finally, the S pocket is filled with an extended network of water molecules. To date, all known catechol-based inhibitors adopt a similar binding mode whereby the aromatic ring of the catecholic portion of the ligands is stabilized by the hydrophobic clamp formed by Phe538 and Ile502, the former engaging in a π-π interaction with the aromatic ring of the catechol48 while the latter forming extended van der Waals contacts with it. Most if not all of the previously characterized catecholbased ligands make stabilizing contacts with portions of the M and the Q pockets, while only recently catecholbased inhibitors that extend in the S-pocket have been described (PDB ID: 5OHJ).49 In our structures, the binding conformation of the catecholic portion is highly conserved and virtually identical to that observed in the crystal structure of the rolipram complex (PDB ID: 1OYN). However, notable structural differences can be observed when comparing the tail conformation of each ligand. Sorting our crystal structures according to the direction of the tails of the ligands, we can identify three main conformational ensembles: ACS Paragon Plus Environment

Page 13 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

protruding (GEBR-7b, GEBR-4a, GEBR-32a, GEBR-54, GEBR-20b), twisted (GEBR-18a, GEBR-18b) and extended (GEBR-26g) (Fig. 2 and Fig. S1 in the supplementary information).

Fig. 2. Binding conformation of three prototypical members of the GEBR library: A) GEBR-4a, protruding. B) GEBR-18b, twisted. C) GEBR-26g, extended. The different sectors of the catalytic cavity are shown on different

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 34

colors: M pocket (blue), Q pocket (red), S pocket (green). The corresponding 2Fo – Fc electron density maps are contoured at the 1.0 σ-level. Figures were done using Pymol53. Although all the crystal structures show two independent catalytic domain-ligand complexes in the asymmetric unit, in most cases little if any structural differences could be observed in the conformation of the same ligand in the two independent complexes, thus suggesting a relatively stable and conserved binding conformation for each inhibitor. The first ensemble (protruding) can be split into two chemically different groups, where group 1 is represented by GEBR-7b-related compounds while group 2 consists of those GEBR-54-related molecules that are characterized by a pyrazole linker carrying a 2-hydroxypropyl chain in N1 (Table 1). Within all the protruding ligands, the tails develop towards the external part of the active site, extending into the S pocket where they make water-mediated contacts with the lining residues. In particular, all these ligands contact Phe506 through both their cyclopentane moiety and their morpholine portion. Further stabilizing interactions are provided by water-mediated contacts inside the M pocket or at the interface between the M and the S pocket. It is worth mentioning that, in our structure, GEBR-54 adopts two different conformations in the two independent monomers, one common to all the protruding ligands and developing into the S pocket, the other featuring close contacts between the morpholine moiety and a hydrophobic area formed by Phe538 and Ile542, whereby the tail of the ligand is shifted towards the entrance of the catalytic pocket, suggesting an even higher degree of flexibility. Interestingly, in our high-resolution structures both GEBR-54 conformations (Fig.S1) contradict a previously hypothesized pose of the ligand50. In particular, a dramatic difference relative to the theoretical pose can be appreciated in the direction of the tail. Indeed, in both conformers the morpholine nitrogen is almost 8 Å away from the Met439 backbone nitrogen, its predicted interactor, although in different directions. The second ensemble (twisted) is represented by those GEBR-54-related ligands characterized by a pyrazole linker carrying a propionyl chain on the N1, thus likely featuring a more conformationally-constrained tail (GEBR-18b and GEBR-18a). Despite the availability of the S pocket, the twisted ligands show a sharp turn in

ACS Paragon Plus Environment

Page 15 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

their structure, resulting in the formation of stabilizing intramolecular interactions with the pyrazole ring. Indeed, the morpholine and the pyrazole moieties, which are connected through a three-atom linker, are nearly parallel to each other. Moreover, morpholine makes intermolecular hydrogen bonding contacts with water molecules in the M pocket as well as hydrophobic interactions with Phe506 and Ile502. These hydrophobic contacts are further enhanced when dimethyl morpholine is present (GEBR-18b). In contrast with the protruding ensemble, here the tails are found in the M pocket over the first solvation sphere of the catalytic ions. In GEBR-18b and GEBR-18a the nitrogen atom of the pyrazole points toward the M pocket, forming a water-mediated interaction with residues Tyr325 and Asp482. Interestingly, this hub water molecule appears to be conserved in all crystal structures including the cAMP-PDE4 complex (PDB code: 2PW3).21 The carbonyl in the linker portion is engaged in a hydrogen bonding network involving two water molecules that interact both with Glu396. One of these water molecules also belongs to the first solvation shell of Mg2+. GEBR-26g is the only member of the third conformational ensemble (extended). Its flexible isoxazolinecontaining tail extends at the interface of the M and the S pockets, untwisted and flattened into the active site. Interestingly, the heteroatoms (O, N) of the central isoxazoline ring point towards the external part of the pocket and are not involved in any stabilizing interactions. This allows the tail to lay over the M pocket, with its two branching terminal moieties forming hydrogen bonding contacts with the backbone amide of Asn375 on one side and water mediated contacts with Glu505 and His370 on the other side. The carbonyl moiety in the linker region forms a water-mediated contact with both Met439 and Glu396. It is important to note that, unlike all the other chiral compounds described in this study, this crystal structure is consistent with only one enantiomer of the GEBR-26g racemic mixture docked inside the pocket. In fact, the electron density nearby the chiral carbon of the isoxazoline ring shows a directionality that is incompatible with the other enantiomer. Within all the conformational ensembles described herein, the protruding ligands appear to be of particular interest due to the fact that they extend into the S pocket and they develop towards the entrance of the active site. Interestingly, GEBR-7b, GEBR-54 and GEBR-32a, which have been shown to have the most interesting

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 34

pharmacological profile29,31,50 belong to this subset of protruding ligands. Potentially, this feature may allow them to interact with the portion of the regulatory domain that caps the catalytic pocket in the intact protein, thus possibly contributing to their inhibitory mechanism. Therefore, to investigate this further, we carried out inhibition assays on both the full length PDE4D3 protein and the catalytic domain alone in order to assess any differential inhibitory behavior for all the studied ligands. Interestingly, all the ligands described herein are endowed with higher IC50s than that reported in the literature for rolipram, both towards the full PDE4D3 isoform and towards the catalytic domain alone. Moreover, unlike rolipram, they don’t show full inhibition in the range of concentrations tested, suggesting that despite the presence of the catecholic portion and the conserved binding mode, the steric hindrance provided by the long tails may impact on the ligand binding efficiency. It is worth stressing that, as pointed out by Burgin et al.22, full inhibition is not a strict requirement as partial inhibition may reduce target-based toxicity while maintaining basal levels of cAMP signaling. The complex PDE4 activation mechanism involves phosphorylation at several different sites. It has been demonstrated that the phosphorylation of Ser54 alters interactions of the UCR1 and UCR2 domains and causes UCR2 to adopt the open active conformation while the phosphorylation of Ser579 stabilizes the closed conformation. Therefore, we performed inhibition assays on a PDE4D3 construct carrying the phosphomimicking mutations Ser54Glu and the phospho-abolishing mutation Ser579Ala to produce a constitutively active form of enzyme (see the materials and methods section). IC50 values against the PDE4D3 isoform and against the catalytic domain only are shown in Table 3 and in Fig. 3. In all cases, there is a > 2.2 fold decrease in IC50 when the compounds are tested on the full length system, the only exception being GEBR-7b and its fluorinated analogue GEBR-20b, for which a lower than 1.5-fold decrease in IC50 is observed. Interestingly, in the context of type 1 scaffold (see Table 1), these are the compounds that feature a shorter and less flexible tail. For all the other compounds, the fold increase in potency in going from the catalytic domain to the PDE4D3 isoform ranges from 2.3 to 4.8. GEBR-26g, which displays an extended conformation at the interface between the M and the S pocket and therefore does not protrude towards

ACS Paragon Plus Environment

Page 17 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

the outside of the catalytic pocket, shows a 4.8-fold increase in potency when tested against the PDE4D3 isoform. It therefore appears that, within our subset of GEBR compounds, no clear correlation can be drawn between the binding conformation of the ligand in the pocket and its increase in potency on the full length protein, although longer compounds appear to achieve a better inhibition of the full length protein. Bruno et al.30 showed that GEBR-4a is able to inhibit the long PDE4D3 isoform three time more than the short isoform PDE4D2 (IC50 2.9 vs 9.2 µM). Also, it has been shown that, owing to the lack of specific dimerization motifs at the interface between the UCR2 and the UCR1, the short isoform PDE4D2 is unable to dimerize. Consistently, we observed that the same compound inhibits PDE4D3 three times more that the catalytic domain alone (IC50 2.1 ± 0.3 vs. 6.9 ± 2 µM). Likewise, GEBR-7b has been shown by Bruno et al. to inhibit both PDE4D2 and PDE4D3 similarly30. In our tests, GEBR-7b as well as its fluorinated counterpart GEBR-20b also show no significant differential inhibition. This may suggest that the extent by which the tail of the ligands protrude does impact on the ability to provide an interaction with the trans-capping portion of the dimeric enzyme, the compounds with longer protruding tails being able to provide a better chemical environment for the capping helix.

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 34

Fig. 3. IC50 curves for the different compounds relative to the PDE4D3 isoform (black dots) and the PDE4 catalytic domain only (red dots). Experimental conditions are described in the materials and methods section. The activity is normalized to the maximum activity for each enzyme, in the presence of 1% DMSO. The reported data are the mean values of three replicates ± SD (standard deviation).

Inside the GEBR7b-like compounds: role of the longer chain and of the hydroxyl moiety In the context of GEBR-7b scaffold, we have evaluated the effect of a longer and more flexible ligand. Indeed, as the UCR2 domain is considered to be flexible and subject to conformational variations, a longer and more adaptable ligand tail could fit and better adapt to the target. Our analysis is based on the crystal structure of the

ACS Paragon Plus Environment

Page 19 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

GEBR-7b and GEBR-4a complexes, the latter featuring one carbon atom-longer chain connecting the morpholine ring to the cathecholic phenyl and a hydroxyl group instead of the carbonyl. As expected, while the catechol portion of the two compounds are perfectly superimposable in the two crystal structures, the tails protrude from the catalytic pocket in a slightly different conformation. In fact, GEBR-4a appears to better fit the pocket: the portion between the nitrogen of linker region and the carbon carrying the hydroxyl group is nearly 2 Å lower into the pocket than the one of GEBR-7b. This lowering inside the catalytic pocket results in an augmented set of interactions of the ligand with the local polar chemical environment. This picture is also supported by the inhibition assays that were done against the PDE4D catalytic domain for these two compounds. Indeed, GEBR-7b shows an IC50 of 16 ± 2 µM while GEBR-4a has an IC50 of 7 ± 2 µM. This effect is also more pronounced in PDE4D3 where GEBR-7b shows an IC50 of 11 ± 2 µM while it is 2 ± 0.3 µM for GEBR-4a. As previously pointed out, with the longer and more flexible chain the appearing of a differential inhibition profile between the PDE4D catalytic domain and PDE4D3 can be appreciated. While GEBR-7b provides virtually the same level of inhibition to the two forms of the enzyme, GEBR-4a shows a 3-fold increase in the inhibition of the full length PDE4D3 compared to the PDE4D catalytic domain. The whole picture suggests that a longer and more flexible chain may efficiently adapt to the chemical environment of the active site, in particular when a regulatory domain of the enzyme is capping the catalytic pocket.

Table 3. IC50 (μM) values measured against the catalytic domain only (PDE4-CD) and against the PDE4D3 isoform. Values refer to enantiomeric/tautomeric mixtures, where present.

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 34

Role of the fluorine atoms Our library accounts for several compounds that differ for just the presence of a CH3 or a CHF2 group in the catecholic portion. The CHF2 group is present also in the PDE4 approved inhibitor roflumilast, which is utilized for the treatment of chronic obstructive pulmonary disease (COPD). In our compounds, the net effect of fluorination appears to be significant. Of the three fluorinated/non-fluorinated pairs that we screened (GEBR7b/20b, GEBR-4a/11b, GEBR-54/GEBR-32a) the fluorinated inhibitors always show a lower IC50 than the nonfluorinated ones, both when tested against the PDE4 catalytic domain alone and against the full length PDE4D3 isoform (Fig. 4). As previously reported for the PDE4-roflumilast complex structure48, also in the GEBR compounds the fluorine atoms make hydrophobic interactions with the protein inside the Q1 pocket (Fig. 5).

Fig. 5. Multipolar interactions in the Q pocket involving the difluoromethoxy moiety of the fluorinated ligands and residues Trp498 and Asn487. The 2F0-Fc map is shown at the 1.0 σ-level.

ACS Paragon Plus Environment

Page 21 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Interestingly, while one of the fluorine atoms interacts with the carbonyl oxygen of Trp498 (3.4-3.5 Å), the two fluorines interact simultaneously with the side chain oxygen of Asn487 (3.6 Å), forming a multipolar contact network that is commonly observed when fluorinated compounds interact with proteins.51

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 34

Fig. 4. Comparison of the IC50 curves for the three pairs of fluorinated (black dots) and non-fluorinated (red dots) compounds. IC50s are measured relative to the PDE4D catalytic domain only (left column) and to the PDE4D3 isoform (right column). The reported data are the mean values of three replicates ± SD (standard deviation).

GEBR-32a tautomerism In the GEBR-32a complex structure, one of the two monomers features the expected ligand in the protruding conformation, while the second one displays a tautomeric form in which the tail of the ligand is attached to N2 of the pyrazole ring. This tautomeric form, which also adopts a protruding conformation, is a secondary product of the synthetic process. In fact, as already reported in the literature52, N-substitutions on the pyrazole ring often provide a mixture of two regioisomers. Whereas their distribution is generally unpredictable, the N1-substituted derivative is often the major one, as in our case. The presence of two tautomers, not highlighted by preliminary investigations but clearly observed in the electron density map of the crystal structure reported here, has been confirmed by a more accurate NMR analysis (Supplementary information), which also allowed the identification of the mixture composition. NMR data provided evidence that also the oxirane derivative 1, which is the GEBR32a precursor, was obtained as an inseparable mixture 1a/1b (2.1:1), thus resulting in GEBR-32a being also obtained as a mixture GEBR-32a/GEBR-32ataut (5:1) (Scheme 1, SI). The structure of 1a was assigned to the major oxirane tautomer by comparing the more intense peaks in the 1H NMR spectrum of the analogue 3-(3cyclopentyloxy-4-methoxy-phenyl)-1-oxiranylmethyl-1H-pyrazole already reported50 and those of the analyzed mixture (Fig. S2, SI). Consequently, the isomeric structure of 1b was assigned to the minor tautomer. The ratio 1a/1b (2.1:1) was calculated by comparing the integral values of the peaks of HA of CH2O-epox. group at 2.53 ppm and 2.58 ppm respectively (Fig. S3, SI). In the case of the GEBR-32a mixture, the GEBR-32a structure was assigned to the major tautomer by comparing the chemical shift of the more intense peaks with those of the precursor 1a. The ratio (5:1) was estimated by comparing the integral values of the peaks of the CH-O group of the cyclopentyl substituent at 4.91 ppm and 4.83 ppm of the major and the minor tautomer, respectively (Fig. S4, SI).

ACS Paragon Plus Environment

Page 23 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Molecular dynamics To gain an insight into the dynamical behavior of our ligands in the context of a long PDE4D isoform that includes the regulatory domains UCR2 and part of UCR1, we did 200 ns molecular dynamics (MD) simulations of the complexes with GEBR-7b and GEBR-32a. The choice of these compounds has been guided by their very promising pharmacological profiles observed in animal models. Furthermore, they feature two different scaffolds and they are both protruding. Overall, the pose provided by the crystallographic structures is maintained in both cases during the entire course of the simulations. The relatively low (< 3Å) RMSD of the binding sites of the ligands (Fig.S4), calculated for all their atoms, reveals that the local structure of PDE4D involved in the harboring of the ligand is retained during the simulation time. The RMSD of the ligands, calculated on the backbone atoms of the binding site residues only, demonstrate the low positional variability of both ligands in the catalytic pocket. In the PDE4D dimer, while GEBR-32a displays the same conformation for both the monomers together with a low RMSD for the catalytic pockets (Fig.S5a), GEBR-7b assumes two different final conformations (Fig.S5b). One (conformation A) closely resembles the experimental structure in terms of both the tail orientation and the distance to the neighboring residues, whereas the other (conformation B, not shown) is markedly different to the experimental one. Notably, a pose similar to conformation B, had been previously hypothesized for GEBR-54 in a recent publication

50

reporting a putative

docking-based binding mode of the ligand in the catalytic domain of PDE4D. It is conceivable that GEBR-7b may be able to explore conformations that are less energetically favorable and hence experimentally inaccessible or poorly accessible due to kinetic reasons. By visual inspection, the comparison between the binding mode of GEBR-7b (Fig. 6A) and GEBR-32a (Fig. 6B) reveals for the latter an average final position closer to the UCR2 capping helix, with the compound assuming a concave conformation under Phe196 (Fig. 7), whose implication in PDE4D drug development has been previously described by Burgin and coworkers. 22

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 34

Fig. 6. Average orientations of GEBR-7b (a) and GEBR-32a (b) in the binding site of PDE4D with respect to the UCR2 domain at the end of MD simulations. The helix moiety of UCR2 is represented in ribbons (in red for GEBR-7b, in yellow for GEBR-32a) spanning residues from 98 to 102. The lower part of panel a and b report the same structure represented in the upper part of the panel turned by 90 degrees anticlockwise with respect to the observer. The average of the ligand position has been computed on the last 50ns of simulation in both cases. A superimposition of the helix moiety of UCR2 is reported (with the same color coding) in panel c, where bundles of 100 poses of GEBR-7b and GEBR-32a are reported from the MD simulations (sampled every 500 ps over the last 50 ns).

A bundle of frames sampled from the trajectory is reported in Figure 6C, showing the calculated conformational differences between GEBR-7b and GEBR-32a. In both cases, the conformation of the ligand tail is stabilized by the orientation of Phe196 which, together with Phe506 and Phe538 (Fig. 7), provides a harboring hydrophobic groove.

ACS Paragon Plus Environment

Page 25 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

To the best of our knowledge, this is the first MD simulation of a long PDE4D isoform that complements experimentally derived data. Owing to the difficulty in obtaining experimental structural data for the full length form of the protein, MD may be considered, to date, the best in silico technique that can be used to validate the stability of the crystallographic conformation in the presence of the capping regulatory domain UCR2.

Fig. 7. Theoretical model (frame taken from GEBR-32a MD simulation) of the hydrophobic groove that is formed at the interface of the catalytic domain and the UCR2. The ligand is shown in light blue. In red are enlightened Phe538 and Phe506 (from the catalytic domain) and Phe196 (from UCR2), which form stabilizing contacts with the ligand.

Conclusions To contribute to the elucidation of the molecular bases of selective inhibition of different PDE isoforms, we provided a structural and functional characterization of a number of members of the GEBR family of PDE4D inhibitors. Indeed, when tested in animal models, some of these compounds have been previously shown to cause limited side effects relative to other PDE4 inhibitors, suggesting some degree of specificity or lack of cross-reactivity among different PDE4 isoforms. Within all the crystal structures of the PDE4 catalytic domain

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 34

shown herein, each in complex with a different member of the GEBR library of compounds, we identified three major ligand conformational classes, which we classified as protruding, twisted and extended. Whereas all the ligands display a common catechol portion that binds into the PDE catalytic pocket similarly to rolipram, they also feature different chemical scaffolds of variable lengths, thus possibly allowing some degree of interaction with the regulatory domains of the enzyme. Owing to the known intrinsic difficulty in obtaining the crystal structure of the full length enzyme, we assessed the ability of these compounds to interact with portions of the enzyme outside the catalytic pocket by examining their binding conformation inside the active site of the catalytic domain and by comparing their potency measured against the full length protein and the catalytic domain only. To further validate our conclusions based on these sets of observations, we used molecular dynamic simulations to generate a model of the full length protein in complex with two of our ligands, each representative of the two general chemical scaffolds based on which they had been designed. Even though only weak and unspecific hydrophobic interactions between the tail portion of the ligand and the capping portion of the enzyme (Phe196) can be postulated based on our simulations, which is also in agreement with the lack of a large differential inhibitory behavior, the conformations of both ligands that were derived from our crystal structure analysis were maintained over the course of the molecular dynamics simulations, suggesting a certain degree of stability of the complex. Based on the data shown herein and on the pharmacological profiles of selected GEBR library compounds previously described in the literature, the subset of protruding ligands appears to be the most promising candidate for further library development. Within this conformational ensemble, we have shown that the shorter compounds (GEBR-7b and GEBR-20b) do not inhibit differentially the PDE4 catalytic domain and PDE4D3, while with the lengthening and the enhanced flexibility of the tail a differential behavior starts to appear (GEBR4a and GEBR-11b). Moreover, within the protruding compounds, the one with the greatest differential behavior is GEBR-54. Interestingly, the GEBR-54 structure shows greater flexibility than those featuring the standard protruding conformation in the S pocket. Future work will be therefore directed toward the expansion and the chemical modification of those long-tailed compounds that show a flexible protruding conformation and in the optimization of the interaction with the capping element of the regulatory domains of the enzyme. Efforts in ACS Paragon Plus Environment

Page 27 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

designing new PDE4 inhibitors that can protrude sufficiently from the catalytic pocket to form stabilizing interactions with the pocket-capping portion of the enzyme may indeed result in enhanced specificity and lower cross-reactivity, thus likely helping to achieve the goal of reducing the severe side-effects problems that are associated with the PDE4 inhibitors currently used for cognitive amelioration in neurodegenerative disease patients.

Supporting information PCR primers for the PDE4-cat construct, Electron density images of the inhibitors in the catalytic pockets, NMR data for GEBR-32a, Molecular dynamics RMSD plots.

Accession Numbers The final crystallographic coordinates of the crystal structures shown here are available in the RCSB PDB (accession codes: 6F6U, 6F8R, 6F8T, 6F8U, 6F8V, 6F8W, 6F8X, 6FDC).

Acknowledgments The authors wish to thank Istituto Italiano di Tecnologia for funding. They also thank the X06DA-PXIII beamline personnel of the Paul Scherrer Institute, Villigen, Switzerland, for help with the data collection. The authors declare that they have no conflicts of interest with the contents of this article. O.Bruno and C.Brullo have an intellectual property interest for all cited compounds (PCT Int. Appl. (2015), WO 2015121212 A1 20150820).

References (1) Francis, S. H., Blount, M. A., and Corbin, J. D. (2011) Mammalian Cyclic Nucleotide Phosphodiesterases: Molecular Mechanisms and Physiological Functions. Physiol. Rev. 91, 651–690. (2) Houslay, M. D. (2010) Underpinning compartmentalised cAMP signalling through targeted cAMP ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 34

breakdown. Trends Biochem. Sci. 35, 91–100. (3) Conti, M., and Beavo, J. (2007) Biochemistry and Physiology of Cyclic Nucleotide Phosphodiesterases: Essential Components in Cyclic Nucleotide Signaling. Annu. Rev. Biochem. 76, 481–511. (4) Baillie, G. S. (2009) Compartmentalized signalling: Spatial regulation of cAMP by the action of compartmentalized phosphodiesterases. FEBS J. 276, 1790–1799. (5) Menniti, F. S., Faraci, W. S., and Schmidt, C. J. (2006) Phosphodiesterases in the CNS: Targets for drug development. Nat. Rev. Drug Discov. 5, 660–670. (6) Maurice, D. H., Ke, H., Ahmad, F., Wang, Y., Chung, J., and Manganiello, V. C. (2014) Advances in targeting cyclic nucleotide phosphodiesterases. Nat. Rev. Drug Discov. 13, 290–314. (7) Houslay, M. D., Schafer, P., Zhang, K. Y. J., and Houslay, M. D. (2005) Phosphodiesterase-4 as a therapeutic target. Drug Discov. Today 10, 1503–1519. (8) Prickaerts, J., Heckman, P. R. A., and Blokland, A. (2017) Investigational phosphodiesterase inhibitors in phase I and phase II clinical trials for Alzheimer’s disease. Expert Opin. Investig. Drugs 26, 1033–1048. (9) Fabbri, L. M., Calverley, P. M., Izquierdo-Alonso, J. L., Bundschuh, D. S., Brose, M., Martinez, F. J., and Rabe, K. F. (2009) Roflumilast in moderate-to-severe chronic obstructive pulmonary disease treated with longacting bronchodilators: two randomised clinical trials. Lancet 374, 695–703. (10) Calverley, P. M., Rabe, K. F., Goehring, U.-M., Kristiansen, S., Fabbri, L. M., and Martinez, F. J. (2009) Roflumilast in symptomatic chronic obstructive pulmonary disease: two randomised clinical trials. Lancet 374, 685–694. (11) Hatzelmann, A., Morcillo, E. J., Lungarella, G., Adnot, S., Sanjar, S., Beume, R., Schudt, C., and Tenor, H. (2010) The preclinical pharmacology of roflumilast - A selective, oral phosphodiesterase 4 inhibitor in development for chronic obstructive pulmonary disease. Pulm. Pharmacol. Ther. 23, 235–256. (12) Page, C. P., and Spina, D. (2012) Selective PDE inhibitors as novel treatments for respiratory diseases.

ACS Paragon Plus Environment

Page 29 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Curr. Opin. Pharmacol. 12, 275–286. (13) Schafer, P. H., Parton, A., Capone, L., Cedzik, D., Brady, H., Evans, J. F., Man, H. W., Muller, G. W., Stirling, D. I., and Chopra, R. (2014) Apremilast is a selective PDE4 inhibitor with regulatory effects on innate immunity. Cell. Signal. 26, 2016–2029. (14) Kavanaugh, A., Mease, P. J., Gomez-Reino, J. J., Adebajo, A. O., Wollenhaupt, J., Gladman, D. D., Hochfeld, M., Teng, L. L., Schett, G., Lespessailles, E., and Hall, S. (2015) Longterm (52-week) results of a phase III randomized, controlled trial of apremilast in patients with psoriatic arthritis. J. Rheumatol. 42, 479– 488. (15) Papp, K., Reich, K., Leonardi, C. L., Kircik, L., Chimenti, S., Langley, R. G. B., Hu, C., Stevens, R. M., Day, R. M., Gordon, K. B., Korman, N. J., and Griffiths, C. E. M. (2015) Apremilast, an oral phosphodiesterase 4 (PDE4) inhibitor, in patients with moderate to severe plaque psoriasis: Results of a phase III, randomized, controlled trial (Efficacy and Safety Trial Evaluating the Effects of Apremilast in Psoriasis [ESTEEM] 1). J. Am. Acad. Dermatol. 73, 37–49. (16) Barad, M., Bourtchouladze, R., Winder, D. G., Golan, H., and Kandel, E. (1998) Rolipram, a type IVspecific phosphodiesterase inhibitor, facilitates the establishment of long-lasting long-term potentiation and improves memory. Proc. Natl. Acad. Sci. U. S. A. 95, 15020–15025. (17) García-Osta, A., Cuadrado-Tejedor, M., García-Barroso, C., Oyarzábal, J., and Franco, R. (2012) Phosphodiesterases as therapeutic targets for Alzheimer’s disease. ACS Chem. Neurosci. 3, 832–844. (18) Ricciarelli, R., and Fedele, E. (2015) Phosphodiesterase 4D: An enzyme to remember. Br. J. Pharmacol. 172, 4785–4789. (19) Bruno, O., Ricciarelli, R., Prickaerts, J., Parker, L., and Fedele, E. (2014) PDE4D inhibitors: A potential strategy for the treatment of memory impairment? Neuropharmacology 85, 290–292. (20) Gurney, M. E., D’Amato, E. C., and Burgin, A. B. (2014) Phosphodiesterase-4 (PDE4) Molecular Pharmacology and Alzheimer’s Disease. Neurotherapeutics 12, 49–56. ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 34

(21) Wang, H., Robinson, H., and Ke, H. (2007) The Molecular Basis for Different Recognition of Substrates by Phosphodiesterase Families 4 and 10. J. Mol. Biol. 371, 302–307. (22) Burgin, A. B., Magnusson, O. T., Singh, J., Witte, P., Staker, B. L., Bjornsson, J. M., Thorsteinsdottir, M., Hrafnsdottir, S., Hagen, T., Kiselyov, A. S., Stewart, L. J., and Gurney, M. E. (2010) Design of phosphodiesterase 4D (PDE4D) allosteric modulators for enhancing cognition with improved safety. Nat. Biotechnol. 28, 63–70. (23) Fox, D., Burgin, A. B., and Gurney, M. E. (2014) Structural basis for the design of selective phosphodiesterase 4B inhibitors. Cell. Signal. 26, 657–663. (24) Cedervall, P., Aulabaugh, A., Geoghegan, K. F., McLellan, T. J., and Pandit, J. (2015) Engineered stabilization and structural analysis of the autoinhibited conformation of PDE4. Proc. Natl. Acad. Sci. 2015, 201419906. (25) Li, Y.-F., Cheng, Y.-F., Huang, Y., Conti, M., Wilson, S. P., O’Donnell, J. M., and Zhang, H.-T. (2011) Phosphodiesterase-4D Knock-Out and RNA Interference-Mediated Knock-Down Enhance Memory and Increase Hippocampal Neurogenesis via Increased cAMP Signaling. J. Neurosci. 31, 172–183. (26) Kaname, T., Ki, C.-S., Niikawa, N., Baillie, G. S., Day, J. P., Yamamura, K., Ohta, T., Nishimura, G., Mastuura, N., Kim, O.-H., Sohn, Y. B., Kim, H. W., Cho, S. Y., Ko, A.-R., Lee, J. Y., Kim, H. W., Ryu, S. H., Rhee, H., Yang, K.-S., Joo, K., Lee, J., Kim, C. H., Cho, K.-H., Kim, D., Yanagi, K., Naritomi, K., Yoshiura, K., Kondoh, T., Nii, E., Tonoki, H., Houslay, M. D., and Jin, D.-K. (2014) Heterozygous mutations in cyclic AMP phosphodiesterase-4D (PDE4D) and protein kinase A (PKA) provide new insights into the molecular pathology of acrodysostosis. Cell. Signal. 26, 2446–2459. (27) Lynch, D. C., Dyment, D. A., Huang, L., Nikkel, S. M., Lacombe, D., Campeau, P. M., Lee, B., Bacino, C. A., Michaud, J. L., Bernier, F. P., Parboosingh, J. S., and Innes, A. M. (2013) Identification of Novel Mutations Confirms Pde4d as a Major Gene Causing Acrodysostosis. Hum. Mutat. 34, 97–102. (28) Zhang, C., Xu, Y., Zhang, H.-T., Gurney, M. E., and O’Donnell, J. M. (2017) Comparison of the

ACS Paragon Plus Environment

Page 31 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Pharmacological Profiles of Selective PDE4B and PDE4D Inhibitors in the Central Nervous System. Sci. Rep. 7, 40115. (29) Bruno, O., Fedele, E., Prickaerts, J., Parker, L. A., Canepa, E., Brullo, C., Cavallero, A., Gardella, E., Balbi, A., Domenicotti, C., Bollen, E., Gijselaers, H. J. M., Vanmierlo, T., Erb, K., Limebeer, C. L., Argellati, F., Marinari, U. M., Pronzato, M. A., and Ricciarelli, R. (2011) GEBR-7b, a novel PDE4D selective inhibitor that improves memory in rodents at non-emetic doses. Br. J. Pharmacol. 164, 2054–2063. (30) Bruno, O., Romussi, A., Spallarossa, A., Brullo, C., Schenone, S., Bondavalli, F., Vanthuyne, N., and Roussel, C. (2009) New selective phosphodiesterase 4D inhibitors differently acting on long, short, and supershort isoforms. J. Med. Chem. 52, 6546–6557. (31) Ricciarelli, R., Brullo, C., Prickaerts, J., Arancio, O., Villa, C., Rebosio, C., Calcagno, E., Balbi, M., van Hagen, B. T. J., Argyrousi, E. K., Zhang, H., Pronzato, M. A., Bruno, O., and Fedele, E. (2017) Memoryenhancing effects of GEBR-32a, a new PDE4D inhibitor holding promise for the treatment of Alzheimer’s disease. Sci. Rep. 7, 46320. (32) J.B.W., Hammond, and N.J., Kruger. (1988) The Bradford Method for Protein Quantitation. New Protein Tech. Methods Mol. Biol. 3. (33) Bieniossek, C., Imasaki, T., Takagi, Y., and Berger, I. (2012) MultiBac: Expanding the research toolbox for multiprotein complexes. Trends Biochem. Sci. 37, 49–57. (34) Adams, P. D., Afonine, P. V., Bunkóczi, G., Chen, V. B., Davis, I. W., Echols, N., Headd, J. J., Hung, L. W., Kapral, G. J., Grosse-Kunstleve, R. W., McCoy, A. J., Moriarty, N. W., Oeffner, R., Read, R. J., Richardson, D. C., Richardson, J. S., Terwilliger, T. C., and Zwart, P. H. (2010) PHENIX: A comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. Sect. D Biol. Crystallogr. 66, 213–221. (35) Chock, S. P., and Huang, C. Y. (1984) An optimized continuous assay for cAMP phosphodiesterase and calmodulin. Anal. Biochem. 138, 34–43. (36) Biasini, M., Bienert, S., Waterhouse, A., Arnold, K., Studer, G., Schmidt, T., Kiefer, F., Cassarino, T. G., ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 34

Bertoni, M., Bordoli, L., and Schwede, T. (2014) SWISS-MODEL: Modelling protein tertiary and quaternary structure using evolutionary information. Nucleic Acids Res. 42, 252–258. (37) Tsujishitaj, H., Moriguchi, I., and Hirono, S. (1993) Potential-Scaled Molecular Dynamics and Potential Annealing: Effective Conformational Search Techniques for Biomolecules. J. Phys. Chem. 97, 4416–4420. (38) Sinko, W., Miao, Y., De Oliveira, C. A. F., and McCammon, J. A. (2013) Population based reweighting of scaled molecular dynamics. J. Phys. Chem. B 117, 12759–12768. (39) Mollica, L., Decherchi, S., Zia, S. R., Gaspari, R., Cavalli, A., and Rocchia, W. (2015) Kinetics of proteinligand unbinding via smoothed potential molecular dynamics simulations. Sci. Rep. 5, 11539. (40) Pronk, S., Páll, S., Schulz, R., Larsson, P., Bjelkmar, P., Apostolov, R., Shirts, M. R., Smith, J. C., Kasson, P. M., Van Der Spoel, D., Hess, B., and Lindahl, E. (2013) GROMACS 4.5: A high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics 29, 845–854. (41) Frisch, M.J., et al. (2009) Gaussian 09. (42) Bayly, C. I., Cieplak, P., Cornell, W., and Kollman, P. A. (1993) A well-behaved electrostatic potential based method using charge restraints for deriving atomic charges: the RESP model. J. Phys. Chem. 97, 10269– 10280. (43) Decherchi, S., Bottegoni, G., Spitaleri, A., Rocchia, W., and Cavalli, A. (2018) BiKi Life Sciences: a New Suite for Molecular Dynamics and Related Methods in Drug Discovery. J.Chem.Inf.Model. 58, 219–224. (44) Piana, S., Lindorff-Larsen, K., and Shaw, D. E. (2011) How robust are protein folding simulations with respect to force field parameterization? Biophys. J. 100, L47–L49. (45) Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., and Klein, M. L. (1983) Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926–935. (46) Darden, T., Perera, L., Li, L., and Lee, P. (1999) New tricks for modelers from the crystallography toolkit: The particle mesh Ewald algorithm and its use in nucleic acid simulations. Structure 7, 55–60.

ACS Paragon Plus Environment

Page 33 of 34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

(47) Bussi, G., Donadio, D., and Parrinello, M. (2007) Canonical sampling through velocity rescaling. J. Chem. Phys. 126. (48) Card, G. L., England, B. P., Suzuki, Y., Fong, D., Powell, B., Lee, B., Luu, C., Tabrizizad, M., Gillette, S., Ibrahim, P. N., Artis, D. R., Bollag, G., Milburn, M. V., Kim, S. H., Schlessinger, J., and Zhang, K. Y. J. (2004) Structural basis for the activity of drugs that inhibit phosphodiesterases. Structure 12, 2233–2247. (49) Carzaniga, L., Amari, G., Rizzi, A., Capaldi, C., De Fanti, R., Ghidini, E., Villetti, G., Carnini, C., Moretto, N., Facchinetti, F., Caruso, P., Marchini, G., Battipaglia, L., Patacchini, R., Cenacchi, V., Volta, R., Amadei, F., Pappani, A., Capacchi, S., Bagnacani, V., Delcanale, M., Puccini, P., Catinella, S., Civelli, M., and Armani, E. (2017) Discovery and Optimization of Thiazolidinyl and Pyrrolidinyl Derivatives as Inhaled PDE4 Inhibitors for Respiratory Diseases. J. Med. Chem. 60, 10026–10046. (50) Brullo, C., Ricciarelli, R., Prickaerts, J., Arancio, O., Massa, M., Rotolo, C., Romussi, A., Rebosio, C., Marengo, B., Adelaide, M., Hagen, B. T. J. Van, Goethem, N. P. Van, Ursi, P. D., Orro, A., Milanesi, L., Guariento, S., Cichero, E., Fossa, P., Fedele, E., and Bruno, O. (2016) New insights into selective PDE4D inhibitors : 3- ( Cyclopentyloxy ) -4- oxime ( GEBR-7b ) structural development and promising activities to restore memory impairment. Eur. J. Med. Chem. 124, 82–102. (51) Müller, K., Faeh, C., and Diederich, F. (2007) Fluorine in pharmaceuticals: looking beyond intuition. Science 317, 1881–6. (52) Huang, A., Wo, K., Lee, S. Y. C., Kneitschel, N., Chang, J., Zhu, K., Mello, T., Bancroft, L., Norman, N. J., and Zheng, S. L. (2017) Regioselective Synthesis, NMR, and Crystallographic Analysis of N1-Substituted Pyrazoles. J. Org. Chem. 82, 8864–8872. (53) Schrödinger. The PyMOL Molecular Graphics System.

ACS Paragon Plus Environment

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

TOC

ACS Paragon Plus Environment

Page 34 of 34