Subscriber access provided by NORTH CAROLINA A&T UNIV
Article
Receptor-based artificial metalloenzymes on living human cells Wadih Ghattas, Virginie Dubosclard, Arne Wick, Audrey Bendelac, Régis Guillot, Rémy Ricoux, and Jean-Pierre Mahy J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b04326 • Publication Date (Web): 18 Jun 2018 Downloaded from http://pubs.acs.org on June 18, 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 8 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
Journal of the American Chemical Society
Receptor-based artificial metalloenzymes on living human cells Wadih Ghattas,* Virginie Dubosclard, Arne Wick, Audrey Bendelac, Régis Guillot, Rémy Ricoux, Jean-Pierre Mahy* Laboratoire de Chimie Bioorganique et Bioinorganique (LCBB) Institut de Chimie Moléculaire et des Matériaux d’Orsay (ICMMO) UMR 8182 CNRS Univ Paris Sud, Université Paris-Saclay 91405 Orsay cedex, France ABSTRACT: Artificial metalloenzymes are known to be promising tools for biocatalysis, but their recent compartmentalization has led to compatibly with cell components thus shedding light on possible therapeutic applications. We prepared and characterized artificial metalloenzymes based on the A2A adenosine receptor embedded in the cytoplasmic membranes of living human cells. The wild type receptor was chemically engineered into metalloenzymes by its association with strong antagonists that were covalently bound to copper(II) catalysts. The resulting cells enantioselectively catalyzed the abiotic Diels-Alder cycloaddition reaction of cyclopentadiene and azachalcone. The prospects of this strategy lie in the organ confined in vivo preparation of receptor-based artificial metalloenzymes for the catalysis of reactions exogenous to the human metabolism. These could be used for the targeted synthesis of either drugs or deficient metabolites and for the activation of prodrugs, leading to therapeutic tools with unforeseen applications.
INTRODUCTION Artificial metalloenzymes are typically chemically prepared by incorporating synthetic transition metal catalysts into host proteins.1, 2 These hybrid biocatalysts are imparted with extraordinary catalytic properties provided by the non-native metal cofactor and with a protective and stereoselective environment provided by the hosting protein.3 The development of these state of the art biocatalysts is still in its infancy and despite several examples of efficient chemo-genetically optimized artificial metalloenzymes, to our knowledge these are not being used yet by the chemical industry.4 Their great potential is however driving their employment in looming therapeutic applications.5 Indeed, artificial metalloenzymes expand the repertoire of reactions catalyzed by natural enzymes and could be prepared in selected organs by using metal cofactors targeting specific host proteins. Subsequently, the formed artificial metalloenzymes could, for example, activate prodrugs by the catalysis of abiotic reactions. Ultimately, they could even be used in treatments to cure metabolic disorders by catalyzing the synthesis of deficient metabolites. For in vivo applications, metal cofactors ought to target their specific host protein in affected organs. Instead, most metal cofactors are damaged by cellular components,6 leading to their deactivation before formation of artificial metalloenzymes7. To overcome these limitations scientist have relied on metal cofactors that remain active in vivo8 or on compartmentalization in the periplasm in the case of Gram-negative Escherichia coli (E. coli) bacteria.9 The therapeutic potential of these approaches is however limited as human cells do not possess periplasm and as only a small number of metal cofactors resist to damage by cellular components. Herein, we report the preparation of an artificial metalloenzyme based on the wild-type human A2A Adenosine Receptor (AR) at the surface of living
Human Embryonic Kidney (HEK) cells, hence using human cell membranes as a compartment to shield metal cofactors from damage10 and provide artificial enzymes with promising therapeutic prospects. To our knowledge, membrane receptors were never used to form artificial enzymes thus far.11, 12 Our strategy targeted the wild type receptor for modifications by using one of its strong antagonists covalently bound to a copper catalyst. We evaluated the potential of the receptor based artificial metalloenzyme in the catalysis of a prototypereaction i.e. the Diels-Alder cycloaddition reaction, which is rarely catalyzed by natural enzymes13, 14 and not by any known human enzyme.15 The yield and stereoselectivity obtained for this carbon-carbon bond forming reaction were in agreement with the catalysis of the reaction by the artificial metalloenzyme assembled in the A2A AR embedded in the cytoplasmic membrane of the living cells. These conclusions were further corroborated by similar findings we obtained using the purified receptor-based artificial enzyme.
RESULTS AND DISCUSSION Strategy for targeting the A2A AR by metal cofactors. Extracellular adenosine activates four guanine nucleotide-binding protein-coupled receptor (GPCR) subtypes: A1, A2A, A2B and A316 and each of these four human adenosine receptors (ARs) plays an essential role in the response to adenosine.17 More specifically, the A2A AR is a guanine nucleotide-binding subunit (Gs) coupled receptor that, when activated, results in an increase of intracellular cyclic adenosine mono-phosphate (cAMP). This receptor subtype is a prime therapeutic target as it is implicated in immuno-oncological treatments as well as in Parkinson’s disease treatments among others.18 The seek for therapeutic applications yielded extensive libraries of reversible A2A ligands with a wide range of affinities. Both structure activity relationship (SAR) studies as well as the elucidation
ACS Paragon Plus Environment
Journal of the American Chemical Society 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
of over thirty-five 3-D crystal structures of free, ligand bound, and chimeric A2A AR have allowed to clarify its binding mode.19 Most A2A ligands consist of a core structure with variable substituents. On the one hand, the core gets bound inside the adenosine binding pocket and is responsible for the major contribution to the binding affinity. On the other hand, the variable substituents extend out of the binding pocket, pointing toward the extracellular medium, and can be used to tune the binding affinity and other physical properties relevant for drug development.20 Thus, to target the receptor a metal cofactor had to be attached to a substituent of a strong ligand of the A2A AR. Binding the metal complex to substituents of various lengths would allow to explore the effects that the depth of its insertion in the receptor could induce on the selectivity of the subsequent catalysis. Unlike agonists, antagonists do not induce a cellular response i.e. increase of intracellular cAMP and the resulting acceleration of the internalization of the receptor.21 Consequently, receptors containing antagonist bound metal complexes would remain for a relatively longer amount of time on the cell surface than their analogues containing agonist bound metal complexes. Preladenant (1) and SCH 412348 (2) are strong antagonists of the A2A AR (Ki = 1.1 nM and 0.6 nM, respectively) that were potential drugs developed for the treatment for Parkinson's disease.22 They possess related chemical structures as they share the same 2furanyl-7H-pyrazolo[4,3-e][1,2,4]triazolo[1,5-c]pyrimidin-5amine adenosine-like core but differ by the terminal N-aryl substituent on the piperazine ring (Figure 1). It is noticeable that this substituent weakly affects the binding affinity since both antagonists exhibit comparable inhibition constants in the nanomolar range. This was confirmed by massive SAR studies, which proved that this core structure was suitable for developing A2A antagonists upon substitution with different residues leading to derivatives such as SCH 58261 (3) and SCH 4442416 (4) with Ki = 1.9 nM and 11.1 nM, respectively (Figure 1).23 Furthermore, the core of the antagonist was substituted by bulky groups such as an alexa-fluorophore and resulted in derivative (5) that still showed a high affinity for the receptor (Figure 1).24 NH 2 O
N N
NH 2 N N
O
N
O
N
N N 1, Ki = 1.1 nM
N
N N
N
N
N
N N
O 3, Ki = 1.9 nM NH 2
NH 2 O
frequently used as the active site of artificial metalloenzymes by us and others,25, 26, 27, 28, 29, 30, 31 since it was found to be able to catalyze cycloadditions efficiently under mild conditions i.e. in buffered water pH∼7 at 4 ˚C. Hallmark substrates cyclopentadiene (6) and 2-aza-chalcones (7a-c) reacted at 2.5 mol% loading of catalyst and afforded up to 50 % yield of isomer products (8a-c) in 72 h. Under the same conditions, mechanistic probes chalcone (7d) and 4-aza-chalcone (7e) that cannot coordinate copper(II) in a bidentate mode did not react demonstrating that a reactive electrophilic diene was not obtained unless both the pyridine and the carbonyl group of the substrate were bound to Cu(II) (Scheme 1A). The devised molecules (12) and (13) were prepared by the reaction of tosylate ester (6) with two N-monosubstituted piperazine derivatives (10) and (11) that also contained a 1,10phenanthroline moiety (Scheme 1B).
F
N
O
N N
F
N
N N 2, Ki = 0.6 nM
N
N
N
N
N N 4, Ki = 11.1 nM O
Scheme 1. A) Diels-Alder cycloaddition leading to four isomer products. B) Synthesis of prospected 12 and 13 possessing a a residual 1,10-phenanthroline group. Copper(II)phenanthroline complex or artificial metalloenzyme.
NH 2 NH 2 O
N N N
N
O
O
O HN
SO 3
O NH
NH
O N N 5, Ki = 111 nM
O SO 3
2
NH 2
Figure 1. Chemical structures and Ki of various antagonists sharing the 2-furanyl-7H-pyrazolo[4,3-e][1,2,4]triazolo[1,5c]pyrimidin-5-amine core structure.
Inspired by this family of strong antagonists we devised molecules that comprised their core structure and residual substituents containing a 1,10-phenanthroline moiety that is known to be able to bind copper(II), leading to catalysts for Diels-Alder cycloadditions based on the Lewis acidity of the metal cation. The Cu(II)-1,10-phenanthroline complex was
Binding of the cofactor into the A2A AR on membrane debris and on cells. Commercial membrane preparations of A2A were used to study the binding of the [3H]CGS21680 radioligand (Kd = 27 nM) in competition with increasing concentrations of either 12 or 13. The results indicated that both molecules maintained a high affinity for the receptor as extracted inhibition constants of 3.8 ± 0.5 nM and 5.2 ± 0.9 nM were found for 12 and 13 respectively, using the Cheng–Prusoff equation (Figure 2A).32 Functional assays that measure intracellular cAMP concentration were also used to probe the effect of 12 and of 13 on HEK-293 cells that expressed A2A (HEK-A2A). Incubation of HEK-A2A cells with increasing concentrations of either 12 or 13 had no effect on the concentration of intracellular cAMP, whereas, in contrast, a concen-
ACS Paragon Plus Environment
Page 2 of 8
Page 3 of 8 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
Journal of the American Chemical Society tration dependent production of cAMP was induced by incubation with increasing concentrations of agonist 5′-(Nethylcarboxamido)adenosine (NECA). Furthermore, the production of cAMP decreased when HEK-A2A cells were incubated with a mixture of 1.25 µM NECA i.e. the concentration that increases intracellular cAMP to 80 %, and increasing concentrations of either 12 or 13 (Figure 2B). It is noteworthy that NECA, 12 and 13 had no effect on intracellular cAMP production by HEK-293 cells that did not express A2A. These results indicated that 12 and 13 were antagonists with a strong affinity for A2A. To form artificial metalloenzymes in the cytoplasmic membrane at the surface of living cells, the antagonists 12 and 13 had to be combined with copper(II) and then incubated with HEK-A2A cells. Therefore, copper(II) complexes were prepared under the conditions that were thereafter used for catalysis, i.e. by incubation of 40 nM 12 or 13 with 1 equiv. copper(II) nitrate in buffer A (MOPS 75 mM, pH 7.4, glucose 17 mM, NaCl 50 mM), and were characterized in solution by high resolution electrospray ionization mass spectrometry (HR ESI-MS) before incubation with HEK-A2A cells. The obtained molecular masses and isotopic patterns (Figure 2 C and D) corresponded to those calculated for [12+Cu2++Cl-]+ and [13+Cu2++Cl-]+ ions respectively, which demonstrated the formation of 12Cu and 13Cu complexes containing a copper(II) ion coordinated by 1,10phenanthroline,33 and a chlorine anion from the buffer solution. Complexes 12Cu and 13Cu appeared to be as good antagonists of the A2A AR on cells as 12 and 13, leading to identical affinity constants, respectively within the experimental error (Figure 2A and B). These results indicated that copper(II) was coordinated in the antagonists away from their core structure, most probably in the 1,10-phenanthroline group as expected and in accordance with the HR ESI-MS results.
Figure 2. A) inhibition of the binding of radio ligand [3H]CGS21680 by A2A in the presence of increasing concentrations of 12 ( ), 13 ( ), 12Cu ( ) and 13Cu ( ). B) Quenching of TR-FRET signal of an antibody of cAMP by increasing concentrations of NECA ( ), and increase of the TRFRET signal with addition of increasing concentrations of 12 ( ), 13 ( ), 12Cu ( ) and 13Cu ( ) in the presence of 1.2 µM NECA. HR ESI-MS spectra of C) 12Cu, and of D) 13Cu.
Furthermore, single crystals of complex 12Cu could be obtained by slow diffusion of ether in a solution of 12Cu in methanol and were suitable for X-ray diffraction analysis. The solved crystal structure proved that indeed only the nitrogen atoms of the 1,10-phenanthroline residue were bound to copper(II), while the remainder of the antagonist was available for
receptor binding (Figure 3). The copper(II) cation was also found to be coordinated by two chlorine anions and one methanol molecule in a typical Jahn-Teller distorted square pyramidal geometry. The 1,10-phenanthroline and chlorine ligands formed the distorted coordination square plane while the methanol ligand was found in the elongated axial position. The (1,10-phenanthroline)N-Cu(II) (2.041 ± 0.004 Å and 2.033 ± 0.004 Å), and Cu(II)-Cl (2.2445 ± 0.014 Å and 2.2583 ± 15 Å) bond lengths are similar to those observed for the antagonist free copper(II)(1,10-phenanthroline)dichloride, typically employed in preparing artificial Diels-Alderases.34 In the proposed mechanism for the catalysis of the Diels-Alder cycloaddition reaction, the 1,10-phenanthroline ligand remains metal bound while the replacement of the chloride ligands by the bidentate coordination of substrates 7 on the copper(II) cation leads to reactive electrophilic dienes.35
Figure 3. Graphical representation of the 3-D crystal structure of complex 12Cu. Two co-crystallized methanol molecules were removed for the sake of more clarity.
Binding stability on cells over time. The binding stability of the antagonists in the receptor binding pocket was investigated for 72 h at 4 ˚C using a commercial BOron-DIPYrromethene (BODIPY) fluorophore covalently bound to the AR antagonist Xanthine Amine Conger (BODIPY-XAC ; Kd = 31 nM). It was previously shown that cytoplasmic membranes were stained when a similar BODIPY-XAC conjugate was incubated at 22 ˚C with Chinese Hamster Ovary (CHO) cells that express the A1 adenosine receptor subtype.36 However, after 90 min, the fluorophore was detected inside the cells, which reflected its internalization. We thus decided to incubate either HEK-293 or HEK-A2A cells at 4 ˚C with 5 nM BODIPYXAC. After washing with buffer A, microscopic imaging revealed that a 2 min incubation was sufficient to stain the cytoplasmic membranes of HEK-A2A cells (Figure 4A), while HEK-293 cells remained unstained. More importantly, the stained HEK-A2A cells were maintained in buffer A for 72 h at 4 ˚C and microscopic imaging revealed that more than 90 % of BODIPY-XAC remained at the surface of the HEK-A2A cells (Figure 4B) while in contrast, at 22 ˚C internalization was observed after only about 10 min (Figure 4C). These results demonstrated that when HEK-A2A cells were kept at 4 ˚C, antagonists could be maintained bound in the A2A AR for at least 72 h. Finally, the incubation at 4˚C of HEK-A2A cells with a mixture of 5 nM BODIPY-XAC and increasing concentrations of 12Cu gradually prevented the staining of cytoplasmic membranes indicating that 12Cu was binding the A2A AR instead of BODIPY-XAC (Figures 4 D1 to D3).
ACS Paragon Plus Environment
Journal of the American Chemical Society 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
Figure 4. Fluorescent Spinning disk confocal laser microscopic (FSDCLM) imaging of, in blue 5 µg/mL Hoechst 33342 and in red 5 nM BODIPY-XAC stained living HEK-A2A cells A) after incubation for 2 min at 4 ˚C and washing in buffer A, B) after further incubation for 72 h at 4 ˚C, C) after further incubation for 10 min at 22 ˚C. FSDCLM imaging of, in blue 5 µg/mL Hoechst 33342 and in red 5 nM BODIPY-XAC stained living HEK-A2A cells in the presence of D1) 25 nM, D2) 2.5 µM, and D3) 25 µM 12Cu and incubation for 2 min at 4 ˚C followed by washing in buffer A.
Catalysis with the artificial enzymes at the surface of cells. The artificial enzymes resulting from combining either 12Cu or 13Cu with A2A, at the surface of living HEK-A2A cells suspended in buffer A, were used at 2.5 mol% to study the catalysis of the Diels-Alder cycloaddition of 6 and 7a-e at 4˚C. In addition to the catalysis with the enzymes, positive and negative controls were performed i.e. the catalysis using either 12Cu or 13Cu in absence of cells, and the catalysis with HEK-293 and HEK-A2A cells in the absence of the metal complexes, respectively. Similar experiments were also performed using either 12Cu or 13Cu combined with HEK-293 that do not express A2A. Reactions were stopped after 12, 24, 48 and 71 h by extraction with ethyl acetate and samples were analyzed by chiral-HPLC with MS/MS detection (Figure S2, Table 1). After 72 h, positive controls showed that both 12Cu and 13Cu catalyzed the reaction with similar yields. Starting from 7a, 8a was obtained with about 50 % yield (20 TON), 84/16 endo/exo ratio and no enantiomeric excess (ee). Compounds 7b and 7c were less reactive and yielded 40 % of 8b, 83/17 endo/exo ratio and 33 % of 8c with 65/35 endo/exo ratio, respectively and both without ee. Negative controls yielded 9 ± 2 % (3.6 TON) of 8a, 4 % of 8b and < 3 % of 8c. This background reaction could be assigned to the catalysis of the reaction by template effects that could occur in the different hydrophobic cavities at the surface of cells including those in membrane proteins.37 It could also have resulted from the remainders of dead cells including leaked transition metals possessing Lewis acid properties. Catalysis with either 12Cu or 13Cu combined with HEK-293 cells provided similar low yields with a maximum of 16 % (6.4 TON) of 8a, suggesting that the catalysis with the metal complexes was mostly inhibited by the cells via non-specific binding. The catalysis with 12Cu-based artificial metalloenzymes at the surface of living HEK-A2A cells yielded up to 42 ± 4 % (17 TON) in the case of 8a and an ee of up to 28 ± 0.5 % in the case of 8c. The endo/exo ratio of products 8a-c also was affected by the selectivity of the artificial enzyme as the endo isomer was further favored. This was remarkable with 8c where endo/exo ratio reached 80/20. As the complex remained active, it demonstrated the protective effect of the embedment of the catalyst inside the A2A AR. A decline in yields vs. positive controls was de-
tected with the three substrates and could be caused by the hindering of the access of the substrate to the copper center by A2A protein residues. More importantly, because A2A protein residues orchestrated the positioning of the substrate around the copper center, they should also be responsible for the obtained endo/exo ratios and ees. Similarly, 13Cu-based metalloenzymes on cells were active and provided yields and selectivities similar to those obtained in the positive control experiments. Bearing in mind that 13Cu has a comparable affinity for the receptor as 12Cu, and that in 13Cu the 1,10phenanthroline residue is covalently bound further away from the core of the antagonist than it is in 12Cu, it can be assumed that 13Cu was also bound into the receptor at the surface of cells but its catalytic center was predominantly exposed to the extracellular buffer medium. Therefore, 13Cu remained active when combined with HEK-A2A cells but did not induce selectivity for any of the four products. Mechanistic probes 7d-e did not react under all assayed conditions and only 7a-c that could bind Cu(II) in a bidentate mode, were reactive substrates. This implied that the reaction was indeed taking place only on metal centers. On the contrary, if any template effect had catalyzed the reaction in the case of 7a-c it should have also catalyzed the cycloaddition with 7d-e. In fact, results showed no background catalysis with 7de in the presence of cells. This indicates that the main background catalysis observed in the case of 7a-c was most probably due to transition metals leaked by dead cells. Finally, 1000 equiv. of XAC (Kd = 2.75 nM)38 were added to 12Cu and then the mixture was incubated with HEK-A2A cells and used for catalysis. After 72h, yields of 12 % ± 2 of 8a, 9 % ± 2 of 8b and 5 % ± 2 of 8c without ee, were obtained, hence similar to the results of the catalysis with metal complexes in the presence of HEK-293 cells. The reason for this is most likely the fact that XAC occupied the receptor instead of 12Cu, thus preventing the assembling of the artificial metalloenzyme and leaving the complex interacting non-specifically with cell components, which lead to the inhibition of the catalysis. The kinetics of the reaction was followed in the presence of 12Cu as catalyst combined with either HEK-293 or HEK-A2A cells. Using 12Cu combined with HEK-293 cells, the reaction quickly slowed down and the yield reached 14 % after 24 h and only 16 % after 72 h. Additionally, there was no significant ee in favor of any of products 8a. These results indicate that the catalysis was inhibited by HEK-293 cells, most probably via non-specific binding of the complex. In the presence of 12Cu inserted into the receptor at the surface of HEK-A2A cells, the reaction was faster, yielding 27 % after 24 h and 42 % after 72 h, which demonstrates the protective effect of the protein on the catalyst. An ee was observed from the beginning of the reaction (8 % ee after 12 h) indicating that the catalysis was taking place within the artificial enzyme and was not only due to background catalysis. The ee increased with time to reach 14 % after 72 h, suggesting that the catalysis by the artificial enzyme was dominant. Furthermore, the initial rate of the reaction catalysed by the artificial enzyme was similar to that of the reaction catalysed by the Cu(II) catalyst in solution in absence of cells. (Figure 5). These results further confirm the binding assays, the functional assays and the imaging data to prove that the artificial enzyme was assembled at the surface of living cells expressing the A2A receptor, which simultaneously protected the metal complex and induced stereoselectivity in the catalysis.
ACS Paragon Plus Environment
Page 4 of 8
Page 5 of 8 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
Journal of the American Chemical Society Table 1. Results of the catalysis of the Diels-Alder cycloaddition reaction by different catalysts. Catalyst
Yield (%) 8a
8b
8c
8a
12
