Recent Advances in Drug Discovery of GPCR Allosteric Modulators

ADDEX Pharma S.A., Head of Core Chemistry Chemin Des Aulx 12, 1228 Plan-les-Ouates, Geneva, Switzerland Jean-Philippe Rocher, PhD

results in a number of differentiating factors. In fact, most Introduction allosteric modulators have little or no effect on receptor function until the active site is bound by an orthosteric The importance of the allosteric regulation of cellular ligand. Allosteric modulators therefore have multiple functions has been known for decades and even the word potential advantages compared to small molecule and “allosterome,” which describes the endogenous alloste- biologic orthosteric drugs. In particular, they offer new ric regulator molecules of a cell, has been proposed 1. chemistry possibilities allowing access to well known tar- Although best described as modulators of enzymes, gets that have been considered intractable to historical advances in molecular biology and robotic HTS technolo- small molecule approaches. For example, allosteric mod- gies recently allowed the discovery of small molecule allo- ulators may soon be developed for targets which hereto- steric modulators of various biological systems, including fore have been only successfully targeted with proteins GPCR and non-GPCR targets. Today, allosteric modula- and peptides. In other words, allosteric drugs with all the tors appear to be an emerging class of orally available advantages of small molecules － brain penetration, eas- therapeutic agents that can offer a competitive advantage ier manufacturing, distribution and oral administration over classical “orthosteric” drugs. This potential stems - may soon be viewed as the best life cycle management from their ability to offer greater selectivity and differ- strategy for protein therapeutics 2. Another potential ben- entiated control over disease mediating receptors. Most efit of allosteric drugs compared to orthosteric drugs is marketed drugs bind to receptors where the body’s own the greater ease of achieving higher selectivity for the tar- natural molecular activators (i.e. endogenous ligands) get. The allosteric sites, unlike the orthosteric sites, have bind specifically to a key part of each receptor’s anatomy been shown to display greater heterogeneity, in all like- called the “active site”. Orthosteric ligands are natural or liness because they have evolved with less evolutionary therapeutic molecules － including peptides and proteins pressure compared to the active sites, especially among － which bind to the active site of receptors. By contrast, closely related receptors that may share a common allosteric modulators bind to receptors at a different site endogenous ligand. In the case of metabotropic receptors and modify receptor function even if the endogenous for example, the glutamate binding orthosteric site is very ligand also binds to the active site at the same time. As a well conserved within 8 members of the family, rendering result, allosteric modulators are non-competitive, which the task of making subtype selective compounds highly

MEDCHEM NEWS No.3 AUGUST 2011 .7 challenging. Allosteric compounds discovered and devel- wide range of effects, starting with a change of the recep- oped in recent years have been shown to offer exquisite tor conformation from an inactive to an activated state. selectivity for one subtype vs other receptors of the same This leads to activation of the receptor associated with family (as well as other families). G-protein and initiates intracellular signaling cascades Allosteric drug discovery has been recently the topic of that mediate cellular responses. Allosteric modulators are increased interest, adressing various important target thought to stabilize or induce changes in the receptor classes besides G-protein coupled receptors (GPCRs): state causing a shift in their responsiveness to endoge- proteases 3, kinases 4, 5, phosphodiesterases 6 and ligand- nous ligands. gated ion channels 7.

Allosteric Modulators of GPCRs: Mechanism of Action and Detection

GPCRs are the largest family of integral membrane recep- tors; the ubiquitous distribution of GPCRs and their involvement in virtually all physiological processes make them extremely attractive targets for drug development. Allosteric modulation as a potential solution for the most challenging receptors in this class has been the topic of recent reviews 8, 9, 10. Allosteric modulators of GPCRs inter- act with binding sites that are topographically distinct

from binding sites of the endogeneous ligands (see Figure 1．Negative Allosteric Modulators diminish the signal of a Figure 1). Furthermore, positive allosteric modulators membrane-spanning GPCR. Positive Allosteric Modulators boost it. (PAM) generally do not activate receptors in the absence of an orthosteric ligand. In the presence of orthosteric Biophysical methods such as surface plasmon resonance ligand, PAM enhance the natural physiological activity of (SPR) and NMR that directly measure the GPCR-ligand the receptor imposing a “ceiling” at the magnitude of interaction have been investigated 11, 12, 13 in particular their allosteric effect; this property will limit the adverse for fragment based drug discovery (FBDD). These effects and also the desensitization that might be pro- approaches and the emerging structural biology technol- duced by an orthosteric agonist. Thus, by applying a non- ogies could be complementary biological methods, which competitive approach that is both more selective and, at may be useful for obtaining information on molecular the same time, able to preserve the physiological rhythms interaction. of endogenous ligand-GPCR signaling, it may be possible to show that therapeutic agents are safer than conven- The tools used to identify allosteric modulators have ben- tional competitive agonists or antagonists against the efited from modern molecular biology techniques such same targets. as complex cell engineering allowing functional expres- In addition, allosteric ligands have been identified that sion of targets of interest 14. Fluorescence-based assays bind to an allosteric site with high affinity without affect- measuring secondary messengers such as calcium ing the receptor function. These molecules are referred (Ca++) or cyclic adenosine monophosphate (cAMP) are to as neutral allosteric modulators or silent allosteric widely used in GPCR drug discovery. The miniaturization modulators (SAM). of the assay format from 96- to 384- and even 1536- to Binding of GPCRs by their endogenous ligand triggers a 3072-well plates and technological developments for assay

8. MEDCHEM NEWS No.3 AUGUST 2011 Recent Advances in Drug Discovery of GPCR Allosteric Modulators

readouts have made the screening of large corporate pharmacological profile of the allosteric molecule. chemical libraries in high throughput mode possible. Modulation of the functional activities of allosteric modu- However, conventional high throughput screening (HTS) lators by minor structural changes has been observed assays show a number of limitations. Firstly, most of these within several series, in particular with mGluR allosteric assays are endpoint assays, and completely ignore ligands 16. Figure 2 shows several examples of “PAM- dynamic changes linked to receptor activation. Secondly, NAM switches” in the mGluR5 ethynyl pyridine series. In most assays include a long incubation and/or revelation this MPEP series, partial and silent allosteric ligands of time, allowing potentially non-specific binding of candi- mGluR5 have been discovered. These molecules are dif- date compound to cellular components other than the ferentiated by the position of the methyl substituent on targeted receptor; and, finally, the signal measured is sev- the pyridine ring. 5-Methyl-6-(phenylethynyl)-pyridine eral steps removed in the signaling cascade, allowing (5MPEP) is a neutral allosteric ligand and it displaces potential up- or down-modulation of every step, leading to [3H]-3-Methoxy-PEPy from the MPEP binding site with a a loss of linearity between initial level of activation of Ki = 388 nM while exhibiting no functional response on receptor and the level of output measured. As a result of its own. 2-(2-(3-methoxyphenyl)ethynyl)-5-methylpyridine these factors, the potential for false positive and negative (M-5MPEP) is a partial antagonist and as such is partially in such assays is unacceptably high for successful alloste- inhibiting the functional response of the mGluR5 recep- ric drug discovery. tor to the glutamate. Such a profile, exhibiting activities At Addex we have developed proprietary novel whole-cell along the continuum between activation and antagonism assays that address the above issues. For example, the has only been described for allosteric modulators. ProxyLiteTM assay was designed to detect allosteric mod- Orthosteric antagonists are essentially binary in their ulatory activity with greater sensitivity than conventional functionalities and cannot function as partial antagonists. assays. It allows real-time dynamic measures of receptor A continuum of efficacy switching in a related series of activation by measuring signals that occur during the ethynyl pyridine carboxamides is also reported 17 (Figure receptor activation event 15. ProxyLite therefore bypasses 2). the downstream screening cascade and reduces potential of false positives and, perhaps more importantly, has revealed false negatives that occur during screening cam- paigns with less sensitive tools. At the same time, ProxyLite helps to generate robust activity data that facil- itates medicinal chemistry, and therefore drives greater efficiency during optimization. Figure 2．Functional switches in mGluR5 allosteric modulators series In addition to novel biological tools, Addex has assembled, a screening library which is biased towards the chemical Figure 3 depicts the [1,2,4]thiadiazole derivative SCH- features of allosteric ligands with the help of computa- 202676 which modulates the activity of many GPCRs tional chemistry. including dopaminergic, adrenergic, enkephalins, and muscarinic receptors 18. It appears as a promiscuous agent Allosteric modulators of GPCRs: and it supposes that certain allosteric sites are common a new challenge in medicinal chemistry across distinct receptors. As a result, the identification of such molecules during the early stages of the allosteric In the allosteric modulator field, the medicinal chemists drug discovery process, using functional selectivity have to optimize both potency and efficacy which are two screening, is crucial. independent important parameters contributing to the Allosteric modulation of adenosine receptors has been a

MEDCHEM NEWS No.3 AUGUST 2011 .9 pockets 13. These results may boost structure-based drug discovery approaches of classical GPCR ligands; how- ever, it should be noted that allosteric modulators have added complexity since a molecule could show affinity for its target protein with an efficacy ranging from null (SAM) or partial to full. The rhodopsin-like class A includes Figure 3．Diff erential activity of allosteric modulators fairly large receptors families like the adenosine receptor and the metabotropic cholinergic receptor families. Many topic of intense efforts towards the identification of new structurally distinct adenosine receptor PAM have been chemotypes 19. The 2-amino-3-benzoyl thiophene series reported, however, there are still efforts to be made to has been a prolific lead series of adenosine A1 PAM. The identify more potent and subtype-selective molecules 26. modes of action of these particular classes of compounds Recent progress in the discovery of selective allosteric have been deeply investigated and research indicates that modulators of muscarinic receptors M1 and M4 suggest they have a dual mode of action. For example, VCP520 that allosteric activation of these receptors has procogni- behaves as a PAM and VCP 333 shows inhibition of the tive and antipsychotic potential 27, 28. CYM2503, a selective GPCR pathway 20. These examples illustrate some of the Galanin R2 receptor PAM having anticonvulsant effect in complexities medicinal chemists faced during allosteric animal models has been disclosed 28. This illustrates the lead optimization and highlight the need for tools that can potential of PAM as a non-peptidergic approach to address measure functional activities. medicinal needs.

Recent developments The secretin-like receptor family or class B GPCR family is a family of peptide-binding receptors including Numerous examples of allosteric modulators have been corticotropin releasing factor (CRF), glucagon, glucagon- reported in the cited reviews; in this article, we would like like peptide-1 (GLP-1), and calcitonin receptors, all of to stress some recent discoveries and new avenues in the which are involved in major physiological functions. field 21. Besides CRF1 non-competitive antagonists, few small We have focused this article on small molecules; how- molecule modulators of this receptor class have been dis- ever, lipidated fragments of intracellular GPCR loops are covered 29. A sulfonyl quinoxaline derivative allosteric developed as a novel pharmacological approach of pep- agonist of GLP-1 receptor has been reported by Novo tide based therapeutics with long half life. These lipopep- Nordisk. This molecule which modulate GLP-1R response tides so called pepducins 22 modulate GPCR activity using in a peptide-agonist dependent manner has shown in an allosteric mechanism; a recent pepducin agonist of the vitro glucose dependent potentiation of insulin secretion chemokine receptor CXCR4 demonstrates in vivo activity in pancreatic islet cells 30. Such studies highlight the in chemotaxis model 23. potential of small-molecules modulators of this receptor class and the complexity of the pharmacology. The GPCR family A has been an area of intense activity with regard to discovery and characterization of Family C GPCRs have been proven to be the most novel allosteric modulators; the recent success of X-ray amenable to allosteric modulation, in particular PAM. β 24 crystallography of active-state GPCR structure with 2 Glutamate is the major excitatory amino acid neurome- and adenosine A2a 25 receptors has allowed descriptions diator in the brain; allosteric modulators of mGluR have a of family A GPCR binding sites in the antagonist and ago- very attractive therapeutic potential; many small drug- nist conformational states, including putative allosteric like molecules have been developed against mGluR1 31,

10. MEDCHEM NEWS No.3 AUGUST 2011 Recent Advances in Drug Discovery of GPCR Allosteric Modulators

mGluR2 32, mGluR4 33 and mGluR5 34, 35. Table 1 illustrates candidates of these two drugs is very competitive. Besides the recent progress leading to clinical candidates in these drugs, several allosteric modulators have reached mGluR5 NAM and mGluR2 PAM series. clinical proof of concept. NBI30775, a CRF1 NAM discov- ered at Janssen Pharmaceuticals and Neurocrine The Most Advanced GPCR Biosciences showed efficacy in a major-depression clini- Allosteric Modulators cal trial 36. The mGluR5 NAM ADX10059 was the first compound of this class reported to improve the clinical Table 1 also shows two allosteric modulators of GPCR, symptoms in GERD patients 37. This molecule has also which are currently on the market, cinacalcet and maravi- shown a significant benefit for the acute treatment of roc, and several clinical leads. The search for follow-up migraine 38. AFQ056 has shown a positive outcome in

Table 1．GPCR Allosteric Modulators Pipeline

Target Drug Structure Stage and indications

Cinacalcet Launched; CaSR PAM （Amgen） hyperparathyroidism

Maraviroc Launched; CCR5 NAM （Pfi zer） AIDS

Reparixin PhaseII/III; CXCR1/CXCR2 NAM （Dompe） Reperfusion injury in lung/kidney transplantation

AFQ056 Phase IIb completed; （Novartis） PD-LID, Fragile X

Dipraglurant Phase II; （Addex） PD-LID, Dystonia mGluR5 NAM STX107 Phase II & III for Fragile X; Undisclosed （Seaside Therapeutics） Phase II in autism RO4917523 Phase II ; Undisclosed （Hoff mann-La Roche） Depression, Fragile X

Fenobam Phase II; （Neuropharm） Fragile X

ADX71149 mGluR2 PAM Undisclosed Phase IIa in Schizophrenia （Addex & Ortho McNeil-Janssen） AZD8529 Phase IIa; mGluR2/3 PAM Undisclosed Astra Zeneca Schizophrenia

MEDCHEM NEWS No.3 AUGUST 2011 .11 patients with Parkinson’s disease levodopa-induced dys- 6. A.B. Burgin, O.T. Magnusson, J. Singh, P. Witte, B.L. kinesia (PD-LID) in two phase IIa clinical trial and one Staker, J.M. Bjornsson, M. Thorsteinsdottir, S. Hrafnsdottir, T. Hagen, A. Kiselyov, L.J. Stewart and M.E. Gurney Nat. Phase IIb trial when given in combination with levodopa 39; Biotechno. 2010, 28, 63-72. In addition to AFQ056, the Michael J. Fox Foundation has 7. R.C Hogg, B. Buisson and D. Bertrand Biochem Pharm recognized ADX48621 specifically and mGluR5 NAM 2005, 70, 1267-1276 generally as a promising approach for the treatment of 8. M. De Amici, C. Dallanoce, U. Holzgrabe, C. Tränkle and K. Mohr Med Res Rev 2010, 30, 463-549. PD-LID. Addex initiated Phase IIa testing of ADX48621 in 9. P.J. Conn, A, Christopoulos and C.W. Lindsley Nat Rev PD-LID patients earlier this year. mGluR5 NAM appears Drug Disc 2009, 8, 41-54. also as an interesting therapeutic approach towards 10. K.J. Gregory, C. Valant, J. Simms, P.M. Sexton and A. Fragile X syndrome which is an inherited cause of mental Christopoulos in GPCR Molecular Pharmacology and Drug Targeting A Gilchrist J Wiley Sons 2010 276 retardation and autism; a Phase II clinical study with feno- . , . & , , - 299. bam suggests beneficial effect 40. Addex also announced 11. I. Navratilova, J. Besnard and A.L. Hopkins ACS Med Chem earlier this year the entry into a Phase IIa clinical schizo- Let May 16 2011, DOI: 10.1021/ml2000017, http://pubs. phrenia trial for ADX71149, which is a mGluR2 PAM, acs.org. developed in collaboration with Ortho-McNeil-Janssen 12. F.M. Assadi-Porter, M. Tonelli, E. Maillet, K. Hallenga, O. Benard, M. Max and J.L. Markley J. Am. Chem. Soc. 2008, Pharmaceuticals, which discovered and developed ris- 130, 23, 7212-7213. peridone (Risperdal), a leading anti-psychotic. 13. M. Congreve, C.J. Langmead, J.S. Mason and F.H. Marshall J. Med. Chem. 2011, 54, 13, 4283-4311. In conclusion, allosteric modulation of GPCRs is benefit- 14. N.T. Burford, J. Watson, R. Bertekap and A. Alt Biochem Pharmacol 2011, 81, 6, 691-702. ing from major advances in the understanding of the func- 15. R. Lütjens 9th Annual Congress: GPCRs in Drug Discovery tioning, detection and optimization of allosteric modula- 22-23 March 2011, Berlin. tors. The first marketed GPCR allosteric modulators and 16. M.R. Wood, C.R. Hopkins, J.T. Brogan, P.J. Conn and C.W. the growing number of clinical proof of concept studies Lindsley Biochemistry 2011, 50, 2403-2410. using allosteric molecules suggests that PAM and NAM 17. A. Graven Sams, G. Kobberoe Mikkelsen, R.M. Brodbeck, X. Pu and A. Ritzen Bioorg. Med. Chem. Lett. 2011, 21, 11, are perceived as differentiated and that they are likely to 3407-3410. become more common therapeutic agents. 18. T. Kenakin Molecular Interventions 2004, 4, 4, 222-229. 19. A. Göblyös and A.P. IJzerman Biochim. Biophys. Acta, Aknowledgement 2011, 1808, 5, 1309-1318. 20. C. Valant, L. Aurelio, V.B. Urmaliya, P. White, P.J. Scammells, P.M. Sexton and A. Christopoulos, Mol. Pharmacol. 2010, I thank Dr Robert Lütjens (Head of Core Biology) and Mr 78, 3, 444-455. Chris Maggos for assistance and helpful discussions dur- 21. M. Rocheville and S.L. Garland Drug Discov Today Tech 7 ing the preparation of this manuscript. 2010, , 1, 87-94. 22. L. Covic, A.L. Gresser, J. Talavera, S. Swift, A. Kuliopulos Proc Natl Acad Sci USA 2002, 99, 643-648. Reference list 23. B. Tchernychev, Y. Ren, P. Sachdev, J.M. Janz, L. Haggis, 1 JE Lindsley and J Rutter Proc Natl Acad Sci USA 2006 . . . . , A. O’Shea, E. McBride, R. Looby, Q. Deng, T. McMurry, 103 28 10533 10535 , , - . M.A. Kazmi, T.P. Sakmar, S. Hunt and K.E. Carlson Proc 2 V Mutel and B Bettler Current Neuropharmacology 2007 . . . , Natl Acad Sci 2010, 107, 51, 22255-22259. 5 3 148 , , . 24. S.G.F. Rasmussen, H-J. Choi, J.J. Fung, E. Pardon, P. 3 A Shen Mol. Biosyst 2010 6 1431 1443 . . . , , - . Casarosa, P.S. Chae, B.T. DeVree, D.M. Rosenbaum, F.S. 4 L Garuti M Roberti and G Bottegoni Curr. Med. Chem. . . , . . Thian, T.S. Kobilka, A. Schnapp, I. Konetzki, R. K. Sunahara, 2010 17 25 2804 2821 , , , - . S. H. Gellman, A. Pautsch, J. Steyaert, W.I. Weis and B.K. 5 JA Lewis EP Lebois and C W Lindsley Curr. Opin. . . . , . . . . Kobilka Nature 2011, 469, 175－180. Chem. Biol 2008 12 269 280 , , , - . 25. G. Lebon, T. Warne, P.C. Edwards, K. Bennett, C.J. Langmead,

12. MEDCHEM NEWS No.3 AUGUST 2011 Recent Advances in Drug Discovery of GPCR Allosteric Modulators

A.G. W. Leslie and C.G. Tate, Nature, 474, 521-525. 20, 3, 441-445. 26. C. La Motta, S. Sartini, M. Morelli, S. Talliani and F. Da 35. J.-P. Rocher, B. Bonnet, C. Boléa, R. Lütjens, E. Le Poul, S. Settimo Curr Top Med Chem 2010, 10, 10, 976-992. Poli, M. Epping-Jordan, A.-S. Bessis, B. Lüdwig, V. Mutel 27. T.M. Bridges, E.P. LeBois, C.R. Hopkins, M.R. Wood, C.K. Curr Top Med Chem 2011, 11, 680-695. Jones, P.J. Conn and C.W. Lindsley Drug News Perpectives 36. K.A. Emmitte ACS Chem Neurosci April 15 2011, DOI: 10. 2010, 23, 4, 229-240. 1021/cn2000266, http://pubs.acs.org. 28. S.D. Kuduk, R.K. Chang, C.N. Di Marco, D.R. Pitts, 37. C. Chen and D.E. Grigoriadis Drug Develop Res 2005, 65, T.J. Greshock, L. Ma, M. Wittmann, M.A. Seager, K.A. 216-226. Koeplinger, C.D. Thompson, G.D. Hartman, M.T. Bilodeau 38. F. Zerbib, S. Bruley des Varannes, S. Roman, R. Tutuian, and W.J. Ray J Med Chem 2011, 54, 13, 4773-4780. J-P. Galmiche, F. Mion, J. Tack, P. Malfertheiner and C. 29. X. Lu, E. Roberts, F. Xia, M. Sanchez-Alavez, T. Liu, R. Keywood Aliment Pharmacol Ther 2011, 33, 8, 911-921. Baldwin, S. Wu, J. Chang, C.G. Wasterlain and T. Bartfai 39. P. J. Goadsby and C.G. Keywood 61st meeting of the Proc Natl Acad Sci 2010, 107, 34, 15229-15234. American Academy of Neurology, 2009, Seattle, USA. 30. S.R.J. Hoare Curr Neuropharm 2007, 5, 3, 168-179. 40. F. Stocchi, A. Destee, N. Hattori, R.A. Hauser, A.E. Lang, 31. C. Koole, D. Wootten, J. Simms, C. Valant, R. Sridhar, O.L. W. Poewe, O. Rascol, M. Stacy, E. Tolosa, C. Trenkwalder, Woodman, L.J. Miller, R.J. Summers, A. Christopoulos H. Gao, J. Nagel, B. Gomez-Mancilla, M. Merschhemke, and P.M. Sexton Mol Pharmacol 2010, 78, 3, 456-465. S. Tekin and W. Abi-Saab 15th International Congress of 32. D.R. Owen ACS Chem Neurosci March 10 2011, DOI: Parkinson’s Disease and Movement Disorders, 2011, 10.1021/cn2000124, http://pubs.acs.org. Toronto, Canada. 33. A.A. Trabanco, J.M. Cid, H. Lavreysen, G.J. Macdonald 41. E. Berry-Kravis, D. Hessl, S. Coffey, C. Hervey, A. Schneider, and G. Tresadern Curr Top Med Chem 2011, 18, 47-68. J. Yuhas, J. Hutchison, M. Snape, M. Tranfaglia, D.V. Nguyen 34. S.P. East and K. Gerlach Expert Opin Ther Patents 2010, and R.A. Hagerman J Med Genet 2009, 46, 266-271.

NEXT ISSUE 次号予告〔Vol.21 No.4 2011 年 11 月 1 日発行〕

◆巻頭言 21 世紀の「知」（Intelligence）を如何に読み解くか 兼松 顕（九州大学高等研究院・名城大学） ◆創薬最前線 三和化学の創薬研究 城森 孝仁（三和化学研究所三重研究パーク） ◆WINDOW インドにおける新薬開発の現状と今後の展望 蔵方 慎一（第一三共株式会社研究開発本部） 学術振興会の留学生としての留学体験記 佐々木 道子（広島大学大学院医歯薬学総合研究科） ◆ESSAY 医薬基盤研究所の目指すところ 松田 岳彦（医薬品基盤研究所戦略企画部） 特集：エピジェネティクスと医薬品の標的分子 ・ エピジェネティクスと癌 岩間 厚志（千葉大学大学院医学研究院細胞分子医学分野） ・エピジェネティクスと免疫疾患 生田 宏一（京都大学ウイルス研究所生体応答学研究部門） ・ エピジェネティクスと中枢神経系疾患 久保田 健夫（山梨大学大学院医学工学総合研究部） ・ エピジェネティクスと代謝性疾患 亀井 康富・小川 佳宏（東京医科歯科大学難治疾患研究所分子代謝医学分野） ◆DISCOVERY Discovery of alogliptin, a novel DPP-4 inhibitor Dr. Stephen Gwaltney（Takeda Sun Diego） ◆REPORT FMC2011 参加報告 河南 三郎（田辺三菱製薬株式会社創薬第二研究所） ◆Coff ee Break 芦澤 一英（エーザイフード・ケミカル株式会社） ◆記事・他 ※編集上の都合により、タイトル、掲載等の変更があることがあります

MEDCHEM NEWS No.3 AUGUST 2011 .13