A True Auto Activating Enzyme

10
A Tr ue Autoac ti vati ng Enzyme STRUCTURAL INSIGHT INTO MANNOSE- BINDIN G LECTI N-AS SOCIA TED SERINE PROTEASE-2 ACTIVATIONS * S Received for publication,June 2, 2005, and in revised form, July 13, 2005 Published, JBC Papers in Press, July 21, 2005, DOI 10.1074/jbc.M50 6051200 Pe ´ ter Ga ´ l ‡1 , Ver oni ka Harmat §1 , And rea Koc sis , Tu ¨ nde Bia ´ n , La ´ szl o ´ Bar na , Ge ´ za Ambrus , Barbara Ve ´ gh , Ju ´ lia Balczer , Robert B. Sim , Ga ´ bor Na ´ ray -Sz abo ´ § , and Pe ´ ter Za ´ vodszky ‡2 From the Ins tit ute of Enz ymo log y, Bio log ical Resear ch Cen ter , Hun gar ian Academy of Sciences, P.O. Box 7, Bud ape st H-1 518 , Hunga ry, the § Protein Modeling Group, Hungarian Academy of Sciences, Eo ¨ tvo ¨ s Lora ´ nd University, Pa ´zma ´ ny Pe ´ter st.1A, Budap est H-1117, Hunga ry, and Medi cal Resea rch Counci l Immun ochemistry Unit, Depar tment of Bioch emist ry, Universit y of Oxf ord, Oxf ord 0X1 3QU , Uni ted Kingdo m Few reports have described in detail a true autoactivation proc- ess, where no extrinsic cleavage factors are required to initiate the autoactivation of a zymogen. Herein, we provide structural and mecha nisti c insig ht into the autoac tivati on of a multi domai n seri ne protea se: manno se-bi ndin g lecti n-ass ociate d seri ne protea se-2 (MASP-2), the first enzymatic component in the lectin pathway of comple men t act iva tio n.Wecharacterize d the proe nzy me form of a MASP -2 cataly tic frag ment encompassing its C-ter mina l three domains andsolve d itscrystalstr uct ureat 2.4 Å res olu tio n. Sur pri s- ingly, zymogen MASP-2 is capable of cleaving its natural substrate C4, with an efficiency about 10% that of active MASP-2. Compari- son of the zymoge n andactiv e stru ctu res of MAS P-2 rev eal s tha t, in additi on to theactiv ati on domain , other loopsoftheserinepro tea se domain undergo significant conformational changes. This addi- tional flexibility could play a key role in the transition of zymogen MASP-2 into a proteolytically active form. Based on the three-di- mensional structures of proenzyme and active MASP-2 catalytic fragments, we present model for t he active zymogen MASP-2 com- plex and propose a mechanism for the autoactivation process. Extrinsic activating factor-initiated autoactivation of a zymogen is a classic textbook case. To date, however, few reports have described a true autoactivation process, where no extrinsic cleavage factors are required and the autoactivating capacity is an inherent property of the zymoge n. A phy siol ogic all y importantexample of true autoac tiv ationis the initiation of the complement cascade activation. The complement system is one of the proteolytic cascade systems found in the blood plasma of vertebrates. It provides the first line of immune defense against invading pathogens. The complement system is a sophisticated network of proteins (involving more than 30 compo- nents), which can be activated via three different routes: the classical, the lectin, and the alternative pathways (1). Activation of the comple- ment system culminates in the destruction and clearance of invading mic roo rga nis ms anddamage d or altered host cel ls. The cen tra l compo- nentsof thesyste m aremult idom ainserin e pro tea ses , whi ch areprese nt in zymogen forms and activate each other in a cascade-like manner (2). In the case of the classical and lectin pathways, a recognition mole- cule binds to a specific target, and this provides the activation signal that is transmitted to serine protease zymogens, which in turn initiate the cascade (3). Mannose-binding lect in (MBL) 3 is the recogniti on sub uni t of thelec- tin pathway (4). MBL binds to carbohydrate arrays (mainly to mannose and N -acetylglucosamine residues) on the surface of pathogens, which results in the autoa ctivation of MBL-a ssociat ed serin e protea se-2 (MASP-2) (5, 6). Activated MASP-2 then cleaves C4 and C2, the pre- cursors of the C3 convertase enzyme complex. MASP-2 is the only known MBL-associated protease that can directly initiate the comple- ment cascade, playing a key enzymatic role in the lectin pathway. The MBL-associated serine proteases together with C1r and C1s, the prot ease subc omp one ntsof thefirstcompon entof theclass ica l pat hwa y (C1), form a family of enzymes with identical domain organization (7). The C-terminal trypsin-like serine protease (SP) domain is p receded by five noncatalytic modules. At the N terminus, there is a C1r/C1s/sea urchin Uegf/bone morphogenic protein (CUB) domain followed by an epider mal gro wth factor(EG F)- like mod uleanda sec ond CUB domain. This N-terminal CUB1-EGF-CUB2 region is responsible for the inter- subunit inter actions (e.g. interaction between the prot eases and the rec - ognition subunits). The following two complement control protein modules (CCPs), which associate directly with the SP domain, stabilize the structure of the SP domain and are involved in the proteolytic process. In the case of C1s and MASP-2, which share almost the same sub- strate specificity, the CCPs were shown to provide accessory binding sites for the C4 substrate and thereby increase catalytic efficiency (8, 9). The CCPs, however, do not increase the efficiency of C2 cleavage, indi- cating that the two substrates bind to different regions of the enzymes. MASP-1, MASP-2, and C1r are capable of autoactivation, where the zymoge n protea ses become cleaved and activ ated withou t thecontribu- tion of any extrinsic cleavage factor. The autoactivation is an inherent property of the serine protease domains, the other modules are not involved in this process (9, 10). Our main priority was to characterize the structural background of the autoactivation process. Recently, the three-dimensional structure of the activated MASP-2 CCP2-SP fragment has been solved (11). The structure revealed the * This work was supported by Hungarian National Science Foundation (OTKA) Grants  T046444 and TS044730 , the Hungarian Ministry of Health (EU ¨ Tana ´c s 555/2003), and NatImmune A/SDenmar k.The costsof publ icat ionof thisarticle weredefr ayedin part by the payment of page charges. This article must therefore be hereby marked advertisement in acc ordanc e with 18U.S.C. Section 173 4 sol elyto indica te thi s fac t. Theatomiccoordin atesand struc turefactors(code1ZJK)have beendepositedin theProtein DataBank,ResearchCollabora toryfor Struc turalBioinfo rmati cs,RutgersUnive rsity , New Brunswick, NJ (http://www.rcsb.org/ ). S  The on-line version of this article (available at http://www.jbc.org) contains supple- mental Fig. 1 and TABLE ONE. 1  These two authors contributed equally to this work. 2  To whomcorresp onde nce shou ld be addr essed : Pe ´ter Za ´vodszkyInst. of Enzymolog y, Biological Research Center, Hungarian Academy of Sciences, P.O. Box 7, Budapest H-1518, Hungary. Tel.: 36-1-209 -3535; Fax: 36-1-466 -5465; E-mail: [email protected]. 3  The abbreviations used are: MBL, mannose-binding lectin; MASP-2, MBL-associated serine protease-2; SP, serine protease; CUB, C1r/C1s/sea urchin Uegf/bone morpho- genic protein domain; EGF, epidermal growth factor; CCP, complement control pro- tein module.  THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 280, NO. 39, pp. 33435–33444, September 30, 2005 © 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U .S.A. SEPTE MBER 30, 2005 VOLUME 280 • NUMB ER 39 JOURNAL OF BIOLOGICAL CHEMISTRY 33435   b  y  g  u  e  s  t   ,  o A  p r i  l  1  9  , 2  0 1 1 w w w .  j   b  c  o r  g D  o w n l   o  a  d  e  d f  r  o m  http://www.jbc.org/content/suppl/2005/08/02/M506051200.DC1.html Supplemental Material can be found at:

Transcript of A True Auto Activating Enzyme

Page 1: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 1/10

A True Autoactivating EnzymeSTRUCTURAL INSIGHT INTO MANNOSE-BINDING LECTIN-ASSOCIATED SERINE PROTEASE-2 ACTIVATIONS*□S

Received for publication,June 2, 2005, and in revised form, July 13, 2005 Published, JBC Papers in Press, July 21, 2005, DOI 10.1074/jbc.M506051200

Peter Gal‡1, Veronika Harmat§1, Andrea Kocsis‡, Tunde Bian‡, Laszlo Barna‡, Geza Ambrus‡, Barbara Vegh‡,Julia Balczer‡, Robert B. Sim¶, Gabor Naray-Szabo§, and Peter Zavodszky‡2

From the‡Institute of Enzymology, Biological Research Center, Hungarian Academy of Sciences, P.O. Box 7, Budapest H-1518,Hungary, the §Protein Modeling Group, Hungarian Academy of Sciences, Eotvos Lorand University, Pazmany Peter st.1A,Budapest H-1117, Hungary, and ¶Medical Research Council Immunochemistry Unit, Department of Biochemistry,University of Oxford, Oxford 0X1 3QU, United Kingdom

Few reports have described in detail a true autoactivation proc-ess, where no extrinsic cleavage factors are required to initiate theautoactivation of a zymogen. Herein, we provide structural andmechanistic insight into the autoactivation of a multidomain serineprotease: mannose-binding lectin-associated serine protease-2(MASP-2), the first enzymatic component in the lectin pathway of 

complement activation. We characterized the proenzyme form of aMASP-2 catalytic fragment encompassing its C-terminal threedomains andsolved itscrystal structureat 2.4 Å resolution. Surpris-ingly, zymogen MASP-2 is capable of cleaving its natural substrateC4, with an efficiency about 10% that of active MASP-2. Compari-son of the zymogen andactive structures of MASP-2 reveals that, inaddition to theactivation domain, other loops of theserine proteasedomain undergo significant conformational changes. This addi-tional flexibility could play a key role in the transition of zymogenMASP-2 into a proteolytically active form. Based on the three-di-mensional structures of proenzyme and active MASP-2 catalyticfragments, we present model for the active zymogen MASP-2 com-plex and propose a mechanism for the autoactivation process.

Extrinsic activating factor-initiated autoactivation of a zymogen is aclassic textbook case. To date, however, few reports have described atrue autoactivation process, where no extrinsic cleavage factors arerequired and the autoactivating capacity is an inherent property of thezymogen. A physiologically important example of true autoactivation isthe initiation of the complement cascade activation.

The complement system is one of the proteolytic cascade systemsfound in the blood plasma of vertebrates. It provides the first line of immune defense against invading pathogens. The complement systemis a sophisticated network of proteins (involving more than 30 compo-nents), which can be activated via three different routes: the classical,

the lectin, and the alternative pathways (1). Activation of the comple-ment system culminates in the destruction and clearance of invading

microorganisms and damaged or altered host cells. The central compo-nents of thesystem aremultidomainserine proteases, which arepresentin zymogen forms and activate each other in a cascade-like manner (2).In the case of the classical and lectin pathways, a recognition mole-cule binds to a specific target, and this provides the activation signalthat is transmitted to serine protease zymogens, which in turn initiate

the cascade (3).Mannose-binding lectin (MBL)3 is the recognition subunit of the lec-

tin pathway (4). MBL binds to carbohydrate arrays (mainly to mannoseand N -acetylglucosamine residues) on the surface of pathogens, whichresults in the autoactivation of MBL-associated serine protease-2(MASP-2) (5, 6). Activated MASP-2 then cleaves C4 and C2, the pre-cursors of the C3 convertase enzyme complex. MASP-2 is the onlyknown MBL-associated protease that can directly initiate the comple-ment cascade, playing a key enzymatic role in the lectin pathway.

The MBL-associated serine proteases together with C1r and C1s, theprotease subcomponentsof thefirstcomponentof theclassical pathway(C1), form a family of enzymes with identical domain organization (7).The C-terminal trypsin-like serine protease (SP) domain is preceded by

five noncatalytic modules. At the N terminus, there is a C1r/C1s/seaurchin Uegf/bone morphogenic protein (CUB) domain followed by anepidermal growth factor (EGF)-like module and a second CUB domain.This N-terminal CUB1-EGF-CUB2 region is responsible for the inter-subunit interactions (e.g. interaction between the proteases and the rec-ognition subunits). The following two complement control proteinmodules (CCPs), which associate directly with the SP domain, stabilizethe structure of the SP domain and are involved in the proteolyticprocess.

In the case of C1s and MASP-2, which share almost the same sub-strate specificity, the CCPs were shown to provide accessory bindingsites for the C4 substrate and thereby increase catalytic efficiency (8, 9).The CCPs, however, do not increase the efficiency of C2 cleavage, indi-

cating that the two substrates bind to different regions of the enzymes.MASP-1, MASP-2, and C1r are capable of autoactivation, where thezymogen proteases become cleaved and activatedwithout the contribu-tion of any extrinsic cleavage factor. The autoactivation is an inherentproperty of the serine protease domains, the other modules are notinvolved in this process (9, 10). Our main priority was to characterizethe structural background of the autoactivation process.

Recently, the three-dimensional structure of the activated MASP-2CCP2-SP fragment has been solved (11). The structure revealed the

* This work was supported by Hungarian National Science Foundation (OTKA) Grants T046444 and TS044730, the Hungarian Ministry of Health (EU Tanacs 555/2003), andNatImmune A/SDenmark.The costsof publicationof thisarticle weredefrayedin partby the payment of page charges. This article must therefore be hereby marked“advertisement ” in accordance with 18U.S.C. Section 1734 solelyto indicate this fact.

Theatomiccoordinatesand structurefactors(code1ZJK)have beendepositedin theProteinDataBank,ResearchCollaboratoryfor StructuralBioinformatics,RutgersUniversity, New Brunswick, NJ (http://www.rcsb.org/).

□S  The on-line version of this article (available at http://www.jbc.org) contains supple-mental Fig. 1 and TABLE ONE.

1 These two authors contributed equally to this work.2 To whomcorrespondence should be addressed: Peter ZavodszkyInst. of Enzymology,

Biological Research Center, Hungarian Academy of Sciences, P.O. Box 7, BudapestH-1518, Hungary. Tel.: 36-1-209-3535; Fax: 36-1-466-5465; E-mail: [email protected].

3 The abbreviations used are: MBL, mannose-binding lectin; MASP-2, MBL-associatedserine protease-2; SP, serine protease; CUB, C1r/C1s/sea urchin Uegf/bone morpho-genic protein domain; EGF, epidermal growth factor; CCP, complement control pro-tein module.

 THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 280, NO. 39, pp. 33435–33444, September 30, 2005© 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

SEPTEMBER 30, 2005 • VOLUME 280 • NUMBER 39 JOURNAL OF BIOLOGICAL CHEMISTRY  33435

http://www.jbc.org/content/suppl/2005/08/02/M506051200.DC1.htmlSupplemental Material can be found at:

Page 2: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 2/10

background of some important physiological properties of the MASP-2enzyme. In this report, we characterize the zymogen catalytic fragment(CCP1-CCP2-SP) of MASP-2 and describe its x-ray structure. Basedupon the zymogen and active structures, we present models for theautoactivating complex and propose a mechanism for autoactivation.

MATERIALS AND METHODS

  Mutagenesis, Expression, and Purification of MASP-2 CCP1-

CCP2-SP R444Q Mutant —Mutagenesis was performed with theQuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA)according to the manufacturer’s instructions. Recombinant plasmid forexpression of wild type MASP-2 CCP1-CCP2-SP was used as template.Recombinant protein expression and renaturation were performed asdescribed earlier (9). The renatured protein solution was concentratedon ultrafiltration membrane (Millipore Corp., Bedford, MA), and it wascarried through its isoelectric point (pI 5.6) by dropping it into a 0.5 M

sodiumacetatebuffer(pH5.0).Thesolutionwasdialyzedagainst50mM

NaOAc,0.5mM EDTA(pH5.0)andfilteredona0.45-m nitrocellulosemembrane. The renaturated protein was purified on a Mono S HR 5/5column (Amersham Biosciences). It was eluted with a linear NaCl gra-dientfrom200to600mM. Thecollected fraction wasonce again carriedthroughtheisoelectricpointbydroppingitintoa1 M HEPES buffer (pH7.4), and it was dialyzed against 20 mM HEPES, 145 mM NaCl, 5 mM

EDTA (pH 7.4). The purification steps were monitored on SDS-PAGE.  Purification of C4—Human C4 was prepared from 20 ml of fresh

serum according to the methods of Dodds (12). The obtained proteinwas 70% pure, and it was dialyzed against 20 mM HEPES, 145 mM

NaCl, 0.5 mM EDTA (pH 7.4). Small C4 aliquots were frozen in liquidnitrogen and kept at 80 °C. They were thawed only once and used upwithin 3 days.

 Purification of Wild Type Zymogen MASP-2 CCP1-CCP2-SP —Theexpression, solubilization, renaturation, and dialysis were performedaccording to Ref. 9. The protein waspurified at thesame conditions as itwas in the case of the R444Q mutant except that the entire procedurewas carried out at 4 °C. Zymogen aliquots (1.93 M) were frozen inliquid nitrogen and kept at 20 °C. Right before usage, zymogen ali-quots, thawed and kept on ice and were consumed within 24 h.

  Autoactivation of Wild Type Zymogen MASP-2 CCP1-CCP2-SP —Autoactivation experiments were carried out under physiological conditions.Theconcentrationof zymogenMASP-2CCP1-CCP2-SPwas1.93M. 12–14samples weretaken atvarious time points within 60min from thebeginning of incubation.Anestimatedhalf-lifewasgiven forthezymogen bymeasuringthediminutionoftheCCP1-CCP2-SPchainandtheappearanceoftheSPdomainonreducing SDS-PAGE.Thequantification of thesedatawas madebyusingaGELDOC 1000 instrumentand MolecularAnalystsoftwarefordensitometriccalculations (Bio-Rad). 2–4 parallel experiments were analyzed to determinethe half-lifeof wild type zymogen.

 Activation of MASP-2 CCP1-CCP2-SP R444Q Mutant —The R444Q

(2.95 M) mutant was activated by thermolysin as described in Ref. 13.C4 Cleavage—To measure the kinetic parameters of the C4 cleavage

by MASP-2 CCP1-CCP2-SP R444Q mutant and by the thermolysin-activated mutant, they were incubated with C4 at 37 °C. Serial dilutionswere made from mutant and substrate to find the optimal, well charac-terizableconditions. The concentration was 2.85 108

M,6.11 109

M, and 6.08 107M for the uncleaved R444Q mutant, the thermoly-

sin-activated R444Q mutant, and the C4 substrate, respectively. Typi-cally,11–13samplesweretakenwithin55minfromthebeginningofthereaction at various times. Data from 2–4 independent measurementswere used for the calculations.The kinetic parameters were determinedby visualizing and measuring the diminution of the chain of C4 onCoomassie-stained SDS-PAGE using a GEL DOC 1000 instrument and

Molecular analyst software (Bio-Rad) for densitometric calculations.The reactions were assumed to be of the Michaelis-Menten type. Thekinetic constants k cat, K 

m,and k cat/ K 

mwere estimatedby unbiased,non-

linear regression methods, regressing the data on the following equa-tion: t  ([S

0] [S] K 

m ln([S

0]/[S]))/(k cat [E

0]). In the presence of 

C1 inhibitor, no cleavage could be detected, so this reaction was used asa negative control.

The Effect of R444Q Mutant on the Autoactivation of Zymogen

  MASP-2 CCP1-CCP2-SP —Wild type zymogen MASP-2 CCP1-

CCP2-SP (0.064 M) was incubated with a 10-fold excess of R444Q(0.64 M) mutant. Serial dilutions of the wild type zymogen were madeto find a concentration at which the rate of autoactivation was slowenough. The incubations were carried out in 20 mM HEPES, 145 mM

NaCl, 0.5 mM EDTA (pH 7.4) buffer at 37 °C for 3 h. All of the sampleswere visualized on SDS-PAGE and were analyzed by densitometry asdescribed above. The appearance of the SP domain was followed.

Cleavage of Synthetic Substrate—The cleavage rates of MASP-2CCP1-CCP2-SP fragments on the synthetic substrate benzyloxycar-bonyl-Gly-Arg-S-benzyl (MP Biomedicals Inc., Aurora, OH) wereobtained as described in Ref. 11.

  Differential Scanning Calorimetry—Calorimetric measurementswereperformed on a VP-DSC (MicroCal)differential scanning calorim-

eter. Denaturation curves were recorded between 20 and 80 °C at apressure of 2,5 atm, using a scanning rate of 1 °C/min. The proteinconcentration was set to 0.2 mg/ml. Samples were dialyzed against 20mM Hepes (pH 7.4), 145 mM NaCl, and the dialysis buffer was used as areference. Heat capacities were calculated as outlined by Privalov (14).

Crystallographic studies on zymogen MASP-2 R444Q Mutant —Crys-tals of the zymogen MASP-2 CCP1-CCP2-SP R444Q mutant fragmentwere grown using the hanging drop method at 20 °C. Crystals wereobtained by mixing 2 l of reservoir solution and 2 l of protein solu-tion. The reservoir solutioncontained 20%polyethyleneglycol6000,0.2M NaCl, 10% glycerol, and 0.1 M Tris-HCl, pH 8.0. The protein solutioncontained1mg/mlMASP-2fragmentinabufferof20m M Tris/HCl,pH7.4, and 0.03% NaN

3.

Data were collected using the ID14 EH4 beam line at the EuropeanSynchrotron Radiation Facility at cryogenic temperatures. Data wereprocessed with the XDS program package; they were scaled, merged,and reduced with XSCALE (15).

The structure was solved by molecular replacement using the pro-gram MOLREP (16) of Collaborative Computing Project 4 (17). The SPdomain of the MASP-2 activated structure (11) (Protein Data Bankaccession code 1Q3X) was used as a search model. Refinement wascarried out with the REFMAC5 program (18), using restrained maxi-mum likelihood refinement and TLS refinement (19). ARP (20) wasused for automatic solvent building. Model building was carried outusing the O program (21). The final model contains protein residues296–686 with the exception of residue 661. The stereochemistry of the

structure was assessed with PROCHECK (22). Data collection andrefinement statistics are shown in TABLE ONE.

 Molecular Modeling of the Enzyme-Substrate Complex of MASP-2—Wedocked the P4-P3 (Thr441–Gly447) segment of zymogen MASP-2 to thesub-stratebindingsiteoftheactivestructure.TheAutoDock3.05(23)programhasbeen used for docking the zymogen MASP-2 heptapeptide fragment to theactive form. The charges were assigned to the structures by using SYBYL 6.5(Tripos Associates Inc., St. Louis, MO). The initial conformation of the frag-ment was built and optimized by SYBYL 6.5. AutoDockTools was used todefine therotablebondsof thefragment. Weused a Lamarckian geneticalgo-rithm for the docking with local search, parameterized according to previoussystematic optimization studies (24) (ga_run 200; generation 27.000;ga_num_evals 25.000.000; ga_pop_size 100, number of active torsion

 Autoactivation of MASP-2

33436 JOURNAL OF BIOLOGICAL CHEMISTRY  VOLUME 280• NUMBER 39• SEPTEMBER 30, 2005

Page 3: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 3/10

18). Inorderto fitthe zymogenMASP-2 to the dockedfragment,wemutatedbackthestructure in silico accordingto the wild type (Q444R).To have a real-istic model, we built a conformational library of the flexible loop of zymogenMASP-2 that contains our heptapeptide fragment (Arg439–Gly448) by SwissPDBviewer(25).A selection of energetically favorable loopconformationshasbeen root mean square deviation-fitted to the docked heptapeptide. TheGly442–Tyr446peptidefragmentofzymogenMASP-2wasreplacedbythecor-respondingdockedfragment.The obtainedcomplexstructureofzymogenandactivated MASP-2 was energy-minimized, relaxed at 310 K by molecular

dynamics simulation in water using GROMACS (26). Coordinates of theMASP-2-MASP-2 complexmodel areavailableuponrequest.

RESULTS AND DISCUSSION

Characterization of the Recombinant Zymogen CCP1-CCP2-SP 

Fragment of MASP-2

 Design of Stable Zymogen MASP-2 CCP1-CCP2-SP —The purpose of the present study was to investigate the autoactivation mechanism of the MASP-2 zymogen at various structural levels. MASP-2 possessesinherent autoactivating capacity and requires no extrinsic enzymatic

factors for the autoactivation to occur. Wild type MASP-2, similarly toother trypsin-like proteases, can be activated through the cleavage of anArg-Ile bond at the N-terminal region of the catalytic SP domain. Wedesigned, constructed, and expressed an R444Q mutant of the catalyticfragment of MASP-2, thereby ensuring that the protein does notundergo autoactivation during the enzymatic characterization, crystal-lization, and structure determination processes. Gln was chosen on thebasis of isomorphic replacement (27) andwas predictedto have theleastprobability to interfere with the protein fold.

Stability of Zymogen MASP-2 CCP1-CCP2-SP R444Q Mutant —Weexpressed the R444Q mutant of the catalytic CCP1-CCP2-SP fragmentof MASP-2 in E. coli cells. The wild type activated form of this fragmentof MASP-2 has already been successfully expressed in the same expres-sionsystem, and its biochemical properties have been characterized (9).After renaturation, the R444Q mutant was purified to homogeneity byionexchangechromatography. Thepurified proteinmigratedas a singleband of 44 kDa on the reducing SDS-PAGE (Fig. 1), indicating that nocleavage occurred within the polypeptide chain during expression andpurification. In contrast to the R444Q mutant, the wild type fragmentwasfullyactivatedafterthesametreatment.Toassessthestabilityofthepurified R444Q mutant, it was labeled with 125I and was incubated witheither buffer or human plasma at 37 °C for 24 h. The samples were runon SDS-PAGE,and theproteinswere visualizedby autoradiography.Nocleavageproductcouldbeobserved(datanotshown)demonstratingthestability of the R444Q zymogen even upon prolonged incubation atphysiological conditions.

  Folding of Zymogen MASP-2 CCP1-CCP2-SP R444Q Mutant —Tofurther characterize the folding and stability of our stable zymogenR444Q mutant, we measured the melting profile of the active wild typeand the zymogen R444Q mutant by differential scanning calorimetry.As the melting curves demonstrate (Fig. 2), both fragments show asharp, cooperative melting transition, indicating a compact, foldedstructure. The melting point of the active fragment (50.8 °C) is 2.6 °Chigher than that of the zymogen form (48.2 °C), and the calorimetricenthalpy change is also larger in the case of the active species. Thisdifference can be explained by the stabilization effect of the activationprocess. During activation, the loosely bound, flexible loops of the acti-

 vation domain become part of the more compact activated structure.The zymogen mutantMASP-2 fragment showed no detectable activ-

ity on synthetic substrate (benzyloxycarbonyl-Gly-Arg-S-benzyl) evenat very high levels of enzyme concentration (200 M), indicating thatthe catalytic machinery is disrupted in the zymogen and there is noactive trypsin-like serine protease contamination in the purifiedmaterial.

To confirm that the zymogen mutant MASP-2 CCP1-CCP2-SP frag-ment is correctly folded and can be converted into an active enzyme, itwas treated with thermolysin (a non-trypsin-like) protease to specifi-cally cleave the Gln444-Ile445 bond (13). The R444Q mutant was acti-

 vated using limited proteolysis by thermolysin (Fig. 1), giving rise to anactive MASP-2 species with activity on a synthetic and a protein sub-

TABLE ONE

Data collection and refinement statistics of the zymogen R444Qmutant form of MASP-2 CCP1-CCP2-SP fragment

Crystal parameters

Space group P212121Cell constants a 47.665 Å, b 72.689 Å,

c 110.989 Å

Data quality 

Resolution range (last resolution shell) 28.868–2.18Å (2.25–2.18 Å)

 Rmeas

a 0.099 (0.575)

Completeness 88.1% (43.1%)

No. of observed/unique 151,887/18,330 (1508/800)

 I / ( I ) 15.13 (1.90)

Refinement residuals

 R 0.207

 Rfree

b 0.253Model quality 

Root mean square bond lengths (Å) 0.005

Root mean square bond angles(degrees)

0.875

Root mean square general planes (Å) 0.002

Ramachandran plot: residues in core/allowed/disallowed regions

272/51/0

Model contents

Protein residues 390

Protein atoms/water molecules 2910/84

Residues in dual conformations 1

Residues with disordered side chains 23

Disordered residues 1a Rmeas (h(n/(n 1))0.5 j Ih I hj )/(hj  I hj ) with  I h ( j  I hj )/n j .b 5.1% of the reflections in a test set for monitoring the refinement process.

FIGURE 1. Coomassie-stainedSDS-PAGE of wildtypeandR444QmutantMASP-2CCP1-CCP2-SPfragment. Lane 1, molecular massmarkers; lane 2,wildtype activatedMASP-2CCP1-CCP2-SP; lane 3,zymogen MASP-2 CCP1-CCP2-SP R444Q mutant;lane 4, thermolysin-activated MASP-2 CCP1-CCP2-SP R444Q mutant.

 Autoactivation of MASP-2

SEPTEMBER 30, 2005 • VOLUME 280 • NUMBER 39 JOURNAL OF BIOLOGICAL CHEMISTRY  33437

Page 4: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 4/10

stratecomparable with that of the wild type MASP-2 fragment (TABLETWO). The recovery of enzymaticactivityof theR444Qmutant follow-ing thermolysin cleavage underlined that it is suitable for studying thestructural and functional properties of zymogen MASP-2.

The Cleavage of C4 by Zymogen MASP-2 CCP1-CCP2-SP R444Q

 Mutant —Previous studies demonstrated that a zymogen MASP-2(S633A) mutant was able to form a complex with C4, a natural proteinsubstrate of wild-type MASP-2 (29). This is most probably due to theaccessory C4 binding sites on the CCP2 module of MASP-2 (9). Weincubatedour stable zymogenR444Q MASP-2with human C4 at 37 °C.To our surprise, zymogen MASP-2 was able to cleave C4 with a highefficiency (Fig. 3, TABLE THREE). This activity was completely abol-ished in the presence of C1 inhibitor, indicating that it was mediated byzymogen MASP-2 and not by other (potentially contaminating) prote-ase. The fact that the K 

m values for the zymogen and the active enzyme

are in the same range indicates that the accessory C4 binding site ispresent on both forms of the enzyme, and it is not affected by theconformation change of MASP-2 activation. Since zymogen MASP-2showed no activity on synthetic substrate but was shown to cleave C4,we argue that the one-chain zymogen form of MASP-2 can adopt an

active-like conformation, and this conformational change may beinduced by the large protein substrate C4. Previously, it was shown thattrypsinogen can be converted into an active state upon strong ligandbinding (e.g. pancreatic trypsin inhibitor and Ile-Val dipeptide) withoutproteolytic cleavage (30). A physiologically important example of a pro-teolytically active serine protease zymogen is tissue-type plasminogenactivator, which has significant (10–20%) activity relative to the two-chain form (31). The proteolytic activity of zymogen MASP-2 could beresponsible for the first step of the autoactivation process, where azymogen MASP-2 molecule cleaves and activates another zymogenMASP-2 molecule.

 Autoactivation of MASP-2—The observed rate of MASP-2 autoacti- vation is concentration-dependent. At low concentrations (0.1 M)during renaturation, the catalytic fragment of wild type MASP-2remains zymogen for several days. However, rapid autoactivationoccurs during the subsequent concentration and purification steps (9).To prepare wild type zymogen MASP-2, we adjusted the pH to 5.0immediately after renaturation and performed the subsequent chro-matographic steps at 4 °C. At pH 5.0, the histidine residue in the cata-lytic triad becomes protonated, and the rate of theproteolysis decreases

TABLE TWO

Cleavage rates of MASP-2 CCP1-CCP2-SP and its R444Q mutant on benzyloxycarbonyl-Gly-Arg-S-benzyl(Z-Gly-Arg-S-Bzl) synthetic substrateandon C4 protein substrate

Enzyme Z-Gly-Arg-S-Bzl k cat

/ K m

C4 k cat

/ K m

 M 1 s1

MASP-2 CCP1-CCP2-SP 9.40 105 6.2 104 5.50 105 5104 a

MASP-2 CCP1-CCP2-SP R444Q activated by thermolysin 1.30 105 1.7 103 8.50 105 2.5105

MASP-2 CCP1-CCP2-SP R444Q —b 7.36 104 1.0104

a Data from Ambrus et al. (9).b Below the detection limit of the assay.

FIGURE 2. DSCmelting profiles of wild type and

R444Q mutant MASP-2 CCP1-CCP2-SP frag-ment. Shown is excess transition heat capacity of the wild type (solid line) and R444Q mutant(dashed line) fragments. Melting temperaturesareindicated.

FIGURE 3. Cleavage of C4 by MASP-2 CCP1-CCP2-SP R444Q mutant. Incubation times areindicatedin minutes.The and chainsofC4 andthe digestion product, , are indicated by thearrows.

 Autoactivation of MASP-2

33438 JOURNAL OF BIOLOGICAL CHEMISTRY  VOLUME 280• NUMBER 39• SEPTEMBER 30, 2005

Page 5: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 5/10

dramatically. Using this purification strategy, we managed to preparewild type zymogen MASP-2, which remains relatively stable at pH 5.0even at a relatively high concentration (3 M). At 4 °C, it autoactivatesslowly(its half-life is approximately 2 weeks), butat 37 °C andpH 7.5, itshalf-life is only 26 min. The activation curve of the wild type zymogenMASP-2 shows the typical features of a true autoactivation process; ithas a sigmoid shape with a lag phase at the beginning (Fig. 4). In the lagphase of the autoactivation process the first reaction step dominateswhere zymogen molecules cleave zymogen molecules. It is a slow andconcentration-dependent process. In the exponential phase of the acti-

 vation curve, the second reaction step dominates where active MASP-2

moleculescleavezymogenMASP-2molecules.Thesecondstepismuchmore efficient than the first one. We used the unactivated wild typeenzymeand the R444Q mutant to prove unambiguously that in the firststep of autoactivation a zymogen molecule cleaves another zymogenmolecule. We used the wild type zymogen MASP-2 at a low concentra-tion, where the rate of autoactivation is slow, and added a 10-fold molarexcess of stable zymogen R444Q mutant (Fig. 5). At low concentration(0.07 M), the autoactivation of wild type MASP-2 is very slow; after a3-h incubation at 37 °C, only 11% of the protease is activated. The pres-ence of a 10-fold excess of the R444Q mutant, however, multiplies therate of activation. After 3 h of incubation, 30% of the wild type MASP-2

TABLETHREE

Kinetic parameters of MASP-2 CCP1-CCP2-SPand itsR444Q mutant on C4 substrate

Enzyme k cat

K m

k cat

/ K m

 s1  M M 1 s1

MASP-2 CCP1-CCP2-SP 0.90 0.4 1.60 10 6 5 106 5.50 105 5 104 a

MASP-2 CCP1-CCP2-SP R444Q 0.026 0.006 3.77 10 7 1.5 107 7.36 104 1.0 104

a Data from Ambrus et al . (9).

FIGURE 4. Time courseof autoactivation of wildtype zymogen MASP-2 CCP1-CCP2-SP frag-ment. Incubation was carriedout at 37 °C.f, con-centration of the zymogen starting from 1.84 M;Œ, concentration of the SP domain from activeMASP-2 CCP1-CCP2-SP.

FIGURE 5. Autoactivation of wild type zymogenMASP-2 CCP1-CCP2-SP in the presence of itsR444Q mutant. Concentration of the SP domainof activated wild type zymogen MASP-2 CCP1-CCP2-SP was determined after a 3-h incubation at37 °C.

 Autoactivation of MASP-2

SEPTEMBER 30, 2005 • VOLUME 280 • NUMBER 39 JOURNAL OF BIOLOGICAL CHEMISTRY  33439

Page 6: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 6/10

is activated. Thisclearly shows that zymogenMASP-2 molecules exces-sively contribute to the enzymatic steps of the autoactivation process.Furthermore, since no extrinsic cleavage factors were detected in ourpreparations, it also indicates that a zymogen MASP-2 cleavage byanother zymogen is the first step in the autoactivation of MASP-2.

The facts that theautoactivationof thewild type MASP-2fragmentisstrongly concentration-dependent and that a large excess of R444Qmutant is needed to detect the proteolytic activity of the zymogenMASP-2 against zymogen MASP-2 substrate suggest that, unlike C4,

there is no accessory binding site on the zymogen CCP1-CCP2-SPfrag-ment for the zymogen MASP-2 substrate. It should be noted, however,that the full-length MASP-2 molecules most probably exhibit higherrates of autoactivation as they dimerize through their CUB-EGFregions, and in the MBL-MASP-2 complex their catalytic domains areplaced in a proper orientation to facilitate the autoactivation process(29, 32).

Structural Studies

Crystal Structure of the CCP1-CCP2-SP Fragment of Zymogen R444Q

 Mutant MASP-2—The structure of the zymogen mutant form of MASP-2 CCP1-CCP2-SP fragment was solved by molecular replace-ment and refined to 2.4 Å resolution (for electron density, see supple-mental Fig. 1) (28). All of the amino acid residues could be built in theelectron density maps except for two regions. One of themis the first 12residues of the sequence, including the Ala-Ser-Met extra tripeptide onthe N terminus and residues Thr287–Ser295. The second one is residue661 (c221) of the activation domain. (Throughout this work, MASP-2numbering is used in comparisons with homologous proteins, togetherwith chymotrypsin numbering (marked with “c”) for the SP domain.)

Some amino acid side chains are missing from the final model due tolack of electron density. These indicate flexible or destabilizedregions inthe molecule. Seven of them are grouped at the free end of the CCP1module, probably due to lack of intermodule contacts with the CUB2module.Theremaining14missingsidechainsarelocatedonthesurfaceof the SP domain, most of them in the loops of the activation domain.

The overall structure is mace-like, showing the two CCP domainsconcatenated in a rodlike structure attached to the globular SP domain(Fig. 6 A).

Structure of the Complement Control Protein Modules of Zymogen

  R444Q Mutant MASP-2—The fold of the CCP1 module with six-strands is highly similar to that of C1r (33), except for two segments(root mean square distance for the backbone atoms: 0.92 Å). These areloop 350–354 and region 306–309. A striking feature of CCP1 moduleof MASP-2 is a large hydrophobic patch on the molecular surfaceformed by residues 299–307 and 357–363. On the contrary, the spatialdistribution of the polar and charged groups is more uniform to thecorresponding domain of C1r. The possible role of this hydrophobicregion in the function of MASP-2 is unknown.

The conformation of the CCP2 module is similar to that found in thewild type active MASP-2 structure (11) (Protein Data Bank accessioncode 1Q3X). The root mean square distances of backbone atoms are asfollows: 0.47 and 0.77 Å for molecule A and B. This underlines the roleof the CCP2 module in C4 binding. Our solution experiments revealedthat zymogen andactivated MASP-2bind C4 with similar affinity.Sincethe structure of the SP domain undergoes major conformationalchanges during autoactivation, including the structure of the classicalsubstrate binding region, it is very likely that the C4 binding site on theunchanged CCP2 module ensures the similar C4 binding affinity of thezymogen and the activated forms of MASP-2.

The region 360–366 containing the linker between the two CCPsforms a -strand. It forms backbone hydrogen bonds and further con-

tacts with both CCPs, conserved across C1r and MASP-2. This is thereasonwhytherelativeorientationsofthetwoCCPsaresosimilarintheMASP-2 and C1r structures. Although the amino acid sequence of thelinker region shows only 50% similarity in MASP-2, C1r, and C1s, thesegments contacting this linker (residues 330–335, 414– 416, and 387–388 in MASP-2) show 67% similarity in the three enzymes, which ishigher than thesimilarity of thewhole CCP1-CCP2region(57%). Theseresults support a model of the CCP1-CCP2-SP with a quasirigid CCP1-CCP2 junction rather than one with a flexible CCP1-CCP2 junction.Therefore, we suggest that the CCP1-CCP2 junction has only a minorrole in orienting the SP domains of these proteins in their functionalcomplexes during the process of autoactivation and/or cleaving theirsubstrates.

The relative orientation of the CCP2 and SP domains in the zymogenMASP-2 structure is somewhat different from those observed in theMASP-2 active structures. The topology of hydrogen bonds and othercontacts of the CCP2 and SP domains is similar to that of the moleculeB in the structure of the MASP-2 active form (11). The CCP2-SP hingeregion of MASP-2, C1r, and C1s shows higher variability than theCCP1-CCP2 junction (34), suggesting that its role is to orient the SPdomain prior to its enzymatic action.

Structure of the Zymogen R444Q Mutant SP Domain—The SPdomain shows some typical features of zymogen structures of the chy-motrypsin family (35). Compared with the active SP domain of 

FIGURE 6. Crystal structure of the zymogen R444Q mutant form of MASP-2 CCP1-CCP2-SP fragment. A, overall structure represented as a ribbon diagram. The zymogenstructure(lightblue, withN andC terminilabeled)is superimposedoverthe SPdomainof theMASP-2activeform(magenta, ProteinData Bankcode1Q3X, moleculeA). Regionsof theSP domainunchangeduponactivationare shown in gray . Theactivesite residuesareshown as sticks and labeled a.s. B, stereo view of the SP domain with yellow arrowsshowingthe direction ofloop movements uponactivation. (Loops ofthe substrate bind-ing region labeled as in Ref. 37). C , stereo view of the activation domain. The catalytictriad and further residues in key positions are shown as sticks with chymotrypsin num-

bering.Oxygen, nitrogen, andsulfuratoms arecoloredred , blue,and green, respectively. The figure was created using MOLSCRIPT (40).

 Autoactivation of MASP-2

33440 JOURNAL OF BIOLOGICAL CHEMISTRY  VOLUME 280• NUMBER 39• SEPTEMBER 30, 2005

Page 7: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 7/10

MASP-2, it shows more residues in thegenerously allowed region of theRamachandran map, whereas the number of residues in the mostallowed region is smaller. The activation domain involving the activa-tionpeptide and loops1, 2,andD asdefined byHuber and Bode (36)(seeTABLE FOUR; labels as in Ref. 37) shows a high degree of flexibilityrelative to the other parts of the structure, indicated by weaker electrondensity and higher B-factors, especially in the case of the first threeloops.

Further comparison of the structure of the zymogen SP domain withthat of the active form of MASP-2 reveals that the core structure of thechymotrypsin fold with the two six-stranded-barrel domains is virtu-ally equivalent in the two forms (Fig. 6 B). The catalytic triad possessestheir activeconformation, butthe oxyanion hole andthe substratespec-ificity pocket are missing (Fig. 6C ). Major structural differences areobserved in the loop conformations of the activation domain usualamong proteins of the chymotrypsin family. We detected further signif-icant conformational changes in the neighboring loop 3 as well as loopslining the substrate binding groove from the other side (loops A, B, andC) (Fig. 6 B). These loops include regions loosely associated with theremaining part of the enzyme. These observations indicate that the SPdomainof zymogen MASP-2has a very flexible structure. Loops A, B, C,and 3 are usually unchanged upon activation of the chymotrypsin-likeenzymes, but they show major conformation changes or they are par-tially disordered in another autoactivating enzyme, C1r (34). This extraflexibility must be a general phenomenon in the case of the zymogenserine proteases with autoactivation ability, where the one-chain formof the enzyme must transiently convert from an inactive to an activeconformation. We propose that the loops of the activation domaintogether with loops A, B, C, and3 form an “autoactivation domain.” Thehigh flexibility of this region and the coordinated conformationalchanges of the loops make possible the formation of the proteolyticallyactive zymogen. TABLE FOUR shows the backbone segments with dif-ferentconformations in thezymogen andactive forms of theSP domainMASP-2. These differences reveal that upon activation, loops 3, A, B,and C move toward the activation domain, thereby making the struc-ture more compact. In the activation domain, loop D moves towardloops 1 and 2, opening up the substrate binding groove on the leavinggroup side (Fig. 6 B). These large scale movements result in a structurestabilized by more interatomic contacts.

Detailed structural analysis shed light on some remarkable features of the zymogen structure (Fig. 6C ), which, we believe, are related to thecharacteristic enzymatic activity of the zymogen enzyme.

The activation peptide is detached from the enzyme surface, havingno specificinteractions. Thisfeature wasreported for other members of the chymotrypsin family. This segment is also one of the regions withelevated B-factors. The mutated side chain of Gln444 (c15) is packedabove a hydrophobic cluster of side chains (Ile567 c135, Pro591 c161, andLeu621 c185). In the native enzyme, Arg444 could not form favorable

interactions with those, and it is probably surrounded by the solvent.Thus, the activation peptide contains an approximately 8-residue-longflexible segment ready to clamp to the substrate binding subsites of theactive enzyme.

Loop 1, which precedes the active site serine, possesses a uniqueconformation. All but one residue of the 622–632 (c185a-c194) seg-ment undergo major backbone conformation changes upon activation.The zymogen enzyme lacks the oxyanion hole and the S1 pocket as

detected in several zymogenstructures.The sidechain of Asp632

(c194),which forms a salt bridge with the amino terminus of the cleaved acti- vation peptide in the active enzyme, is hydrogen-bonded to the amideNH of residues 573 (c141) and 574 (c142) of loop D. The side chain of the specificity-determining Asp627 (c189) turns out in the solvent simi-larly to that of zymogen C1r. It is disordered, since it has few contactswiththerestoftheresidues.Surprisingly,theS1pocketisblockedbythebasic side chain of Arg630 (c192). A basic residue in this position in anenzyme with trypsin-like substrate specificity is unusual in the chymo-trypsinfamily. Arg630 (c192)in thezymogen structureinhibits thebind-ing or adequate positioning of the P1 basic residue of the substrate. Thisrepulsive force could be compensated for if the substrate has anextended binding surface on the enzyme. This can be the reason of thedetected difference in the activities observed on large substrates(extended binding surface) and a small one (limited binding surface).

Loop 2 linked by a disulfide bridge to loop 1 shows only minor con-formational changes. Itsends seem to be more stabilized in thezymogenstructure than in the active structure via elongation of the two adjacent-strands. The middle part of loop 2 of the zymogen enzyme lacks all of its hydrogen bond interactions with the rest of the protein due to slightchanges in the main chain conformation of the ends and the great dif-ference in the structure of loop 1. Wesuggest that loop 2 is floppy in thezymogen enzyme, since it can move without great conformational rear-rangements facilitating the conformational transition of the zymogenenzyme to the catalytically effective form.

The fourth part of the activation domain, loop D, is stabilized byseveral contacts in a highly different conformation in the active andzymogen structureof MASP-2. Itsconformation is somewhat similar tothat of zymogen proteinase C (38). The N-terminal part of it stabilizesthe side chain of Asp632 (c194)by hydrogen bonds formed with residues573 (c141) and 574 (c142). A few other enzymes (C1r, chymotrypsin,zymogen E (39), and proteinase C) also establish at least one of thesehydrogen bonds in their zymogen forms. Loop D of MASP-2, however,also shields Ser633 (c195) NH by a hydrogen bond with Thr576 (c144)OH. The conformation of the 573–576 (c141–c144) segment stericallyprevents the flipping of residue Asp632, and consequently it inhibits theformation of the oxyanion hole.

The Model of the Enzyme-Substrate Complex of Activated and Zymo-

 gen MASP-2—In order to study the structural details of the mechanismof MASP-2autoactivation, we dockedthe zymogenMASP-2 SP domain

TABLEFOUR

Surface segments of MASP-2SP domainwith differentbackbone conformationin the zymogen mutant andactive forms (D> 1.5 Å)

Different Cposition

Overlaps with loopin active form

Part of activation domainin zymogen form Characteristic to the conformation change

440–447 Activation peptide (436–441) Yes Major conformation change

465–467 A (463–469) Bend of loop

489–494 B (485–496) Major conformation change

526 C (524–528) Minor tilt of loop

575–581 D (575–582) Yes Major conformation change

604–610 3 (594–611) Bend of loop621–631 1 (621–625) Yes Major conformation change

658–663 2 (657–665) Yes Major bend and minor conformation change

 Autoactivation of MASP-2

SEPTEMBER 30, 2005 • VOLUME 280 • NUMBER 39 JOURNAL OF BIOLOGICAL CHEMISTRY  33441

Page 8: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 8/10

to the substrate binding region of the activated SP domain and relaxedthe structure by short molecular dynamics simulation. This model (Fig.7, A–C ) demonstrates the second step of the autoactivation process,when an activated MASP-2 cleaves a zymogen MASP-2 molecule. Theresulting complex structure possesses an extended binding surfacebetween active and zymogen MASP-2 burying 1517 and 1665 Å2 of thesurface areas of the zymogen and active structures, respectively. Theintermolecular interactions include 19 hydrogen bonds and four salt

bridges.Theinterfaceisbuiltupbyresiduesofloops1andtheactivationloop as well as regions 564–567 (c132–c135), 587–591 (c157–c161),and 640–645 (c202–c205) of the substrate (zymogen) structure as wellas loops 1, 2, 3, A, B, C, and D of the active enzyme (supplementalTABLE ONE). The central S1-P1 salt bridge is surrounded by regions of the two molecules that form mostly hydrophobic interactions. On theedge of the binding region of both molecules, four isolated clusters of charged and polar residues form predominantly electrostatic interac-tions (labeled in Fig. 7C ).

Accommodation of the P1 arginine (Arg444, c15) of zymogenMASP-2in theS1 pocket of theactive serineproteaseis characteristic tothe trypsin-like serine proteases. The extended P5-P2 segment of theactivation loop of the zymogen molecule forms several contacts of thesubstrate binding groove of the active enzyme (Fig. 7 A). In addition tothe canonical hydrogen bonds of the P1 and P3 residues (oxyanion holeand antiparallel -sheet formation with the enzyme), the backbone of theP5P1 andP2 residues is also hydrogen-bonded to theenzyme. TheP2–P5 residues with their small side chains are bound to a flat hydro-phobic regioncontacted predominantly by loopC (S5 andS4 sites),loop3(S4andS3sites)andloop2(S3andS2sites).Attheleavinggroupside,Leu445 (c16, P1) and Tyr446 (c17, P2) form several hydrophobic con-tacts with His483 (c57) and residues of loops 1 and D (supplemental

TABLE ONE). The extension of the S5-S2 binding region containsmostly nonpolar side chains. Outside these hydrophobic regions, therearefour major hydrophilic regions with intermolecular hydrogen bondsand intermolecular salt bridges and further complementary electro-static interactions (Fig. 7C ).Inpolarregion1,Arg630 (c192) of theactiveenzyme establishes a salt bridge with Asp589 (c159) of the substrate, butits side chain carbon atoms form contacts with the P2-P1 region of thesubstrate. Polar region 2 of the complex includes favorable interactionsbetween Asp564, Asp565 (c123–133, substrate), and Arg578 (c148,enzyme). The central interaction of polar region 3 is the salt bridgeformed by Arg439 (c10, P6 residue of the substrate) and Asp526 (c96a,enzyme). Polarregion4 includes weak electrostatic interactionbetweenGlu473 (c204a, substrate) and Lys604 (c173a, enzyme).

 Proposed Autoactivation Mechanism—The zymogen MASP-2 mole-cule, exhibiting enzymatic activity in solution, should possess differentconformation from that detected in the crystal structure. The zymogenstructure must undergo conformational changes to be able to bind theactivation peptide of the other molecule, especially the P1 arginine, andto form the oxyanion hole.

Our model of the activated and zymogen MASP-2 complex can alsoprovide information on this initial step of autoactivation. Merely over-layingthe activatedmoleculeof themodel with thezymogen SP domainreveals that the favorable intermolecular contacts of the active-zymo-gen complex would be only partially destroyed in a zymogen-zymogencomplex with similar orientationof the two molecules;large continuouscontact regions are practically unchanged, including the binding region

of the N-terminal end of the activation loop, most of the largest hydro-philic binding patch and the fourth hydrophilic region (Fig. 7 D). Theseregions possibly play an importantrole in preorienting thetwo SP headsand forming an initial complex between them.

The remaining parts of the binding region undergo conformationalchanges during activation or during a conformational transition fromthe zymogen to the active-like state. Some of those make steric clashesand unfavorable contacts for the overlaid zymogen structure (Fig. 7 D).Theseregionsofthebindingsurfacemighthavearoleinthezymogentoactive-like conformational transition of MASP-2 in two possible ways;theconformational changemay be driven by another MASP-2moleculein a substrate position through an induced fit mechanism and/or bystabilizing the active-like conformation. One of the key points of the

FIGURE 7. Molecular modeling of the enzyme-substrate complex of MASP-2. A, ste-reoviewof thebindingof theP5-P2 segment (Thr440–Tyr446) ofthe zymogen molecule(light blue carbon atoms) by the activated enzyme (magenta carbon atoms). The figurewas prepared using a representative structure of the molecular dynamics simulation of thecomplex ofSP domains.Oxygenandnitrogenatomsare coloredred and blue, respec-tively, and hydrogen bonds are represented as shaded green lines. Chymotrypsin num-bering was applied for labeling residues. B, the enzyme-substrate complex of activatedMASP-2 (molecular surface colored magenta) and zymogen MASP-2 (molecular surfacecolored light blue) CCP1-CCP2-SP fragments. C , molecular surface of the enzyme-sub-strate complex of activated(left ) andzymogenMASP-2(right ; shouldbe rotatedby 180°around the vertical axis to form the complex). The surface is colored based on electro-static potential (blue, positive; red , negative). Four regions at the edge of the bindingregion have polar character, and these are labeled. D, contact surfaces of the zymogen-zymogen MASP-2 complex. The protein molecules in the role of the enzyme ( left ) andsubstrate(right ) areoriented similarlyto those in C . Thelowerpartof thecontact surface(green) forms favorable interactions for both zymogene SP domains. The upper part of the contact surface (orange) with several unfavorable contacts (some shown as red 

arrows) is formed by loops of the activation domain of the SP domain on the left . It islikely, therefore, that a main driving force in the inactive-active conformational changeof zymogenMASP-2is to avoid theseunfavorable interactions( left ). A was createdusingPymol (41), B was created using VMD (42), and C  and D were created using Insight IIversion 2000L (AccelrysSoftware Inc.). Electrostatic potentialwas calculated by the Del-phi module of Insight II.

 Autoactivation of MASP-2

33442 JOURNAL OF BIOLOGICAL CHEMISTRY  VOLUME 280• NUMBER 39• SEPTEMBER 30, 2005

Page 9: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 9/10

zymogen to active-like transition is the S1/P1 interaction. In the zymo-gen-like conformation, the interaction between Arg630 (c192) blockingthe S1 pocket and the approaching P1 arginine of thesubstrateis clearlyunfavorable. This electrostatic repulsion, however, may initiate a con-formation change of loop 1 and the formation of the oxyanion hole andS1 pocket. There are two other regions of the zymogen enzyme thatwould make bad contacts with thesubstrateprior to theconformationaltransition: loop D (the side chains Arg578, Phe580, and Thr576) and loop3 (Pro605 and Pro606). Although there are some other parts of theMASP-2 surface changing, these three regions should undergo mostdrastic conformational shift to form favorable interactions withthe sub-strate.

To sum up the above observations, we propose the following frame-work for theautoactivation process when a zymogen MASP-2moleculecleaves and activates another zymogen MASP-2 molecule. The firstevent is recognition andpreorientation of a zymogen MASP-2moleculeby another MASP-2 molecule (Fig. 8 A). During this process, extendedcontact regions are formed between the two SP domains that remain

 virtually identical during the autoactivation event. Subsequently, thezymogen to active-like conformational transition of the first moleculeoccurs, driven by forming more favorable interactions with the sub-strate. This transition is likely to be initiated in its activation domain,since this region develops more favorable contacts with the substrate in

the active-like conformation. Thezymogen cleaves thesubstrate, andinthe last step it releases the activated substrate molecule.

We propose that thesecondstep of theautoactivationprocess, wherethe activated MASP-2 molecule cleaves the zymogen one resembles thecanonical way of substrate cleaving by serine proteases (Fig. 8 B). Com-parison of activated and zymogen MASP-2 structures and our model of the enzyme-substrate complex formed by them revealed that both theenzymeand thesubstratestructures arepracticallypreformedfor form-ing the canonical Michaelis complex.

Conclusion

The availability of the crystal structure of both the zymogen andactivatedform of thecatalytic region of MASP-2enabled us to study thestructural details of the autoactivation mechanism of this serine prote-ase. Although the crystal structure of the zymogen shows an inactivespecies with disrupted oxyanion hole and substrate binding pocket, oursolution studies provided clear evidence that the proenzyme form of MASP-2 can exhibitproteolytic activity on protein substrates. Onepre-requisite of the zymogen to active transition is the flexibility of thezymogen SP domain structure. This notion was supported by the sig-nificant conformational change of several loops (the activation domainandloopsA,B,C,and3)andwasreinforcedbyDSCmeasurements.The

structure of the CCP2 module, however, which associates directly withthe SP domain and most probably contains accessory binding site(s) forthe large C4 substrate, does not change during activation. This canexplain the relatively high efficiency of zymogen MASP-2 cleaving C4.By themolecular modeling of theactive MASP-2and zymogen MASP-2enzyme-substrate complex, we provided a detailed analysis of the inter-actions between the two molecules at the atomic level during the auto-activation process. The model describes a rather extended contact sur-face between the two molecules involving both hydrophobic andcharged regions. The large contact surface could explain why zymogenMASP-2 can cleave protein substrates, whereas it is inactive on a smallsynthetic one. The zymogen MASP-2 protein substrate interaction caninduce and/or stabilize the transiently active conformation of the one-

chain form.

 Acknowledgments—We greatly acknowledge the European Synchrotron Radi-

ation Facility for provision of synchrotron radiation facilities, and we thank 

 Edward Mitchell for assistance in using beamline ID 14 4.

REFERENCES

1. Morley,B.J.,andWalport,M.J.(eds)(1999) Complement Factsbook , Academic Press,London

2. Sim, R. B., and Tsiftsoglou, S. A. (2004) Biochem. Soc. Trans. 32, 21–273. Gal, P., and Ambrus, G. (2001) Curr. Protein Pept. Sci. 2, 43–594. Holmskov, U., Thiel, S., and Jensenius, J. C. (2003) Annu. Rev. Immunol. 21, 547–5785. Thiel, S., Vorup-Jensen, T., Stover, C. M., Schwaeble, W., Laursen, S.B., Poulsen, K.,

Willis,A.C.,Eggleton,P.,Hansen,S.,Holmskov,U.,Reid,K.B.M.,andJensenius,J.C.(1997) Nature 386, 506–510

6. Schwaeble, W., Dahl, M. R., Thiel, S., Stover, C., and Jensenius, J. C. (2002) Immuno-

biology 205, 455–4667. Volanakis, J.E., andArlaud, G.J. (1998) in The Human Complement System in Health

and Disease (Frank, M., and Volanakis, J. E., eds) pp. 49 –81, Marcel-Dekker, NewYork

8. Rossi, V., Bally, I., Thielens, N. M., Esser, A. F., and Arlaud, G. J. (1998) J. Biol. Chem.

273, 1232–12399. Ambrus, G., Gal, P., Kojima, M., Sz ilagyi, K., Balczer, J., Antal, J., Graf, L., Laich, A.,

Moffat, B. E., Schwaeble, W., Sim, R . B., and Zavodszky, P. (2003) J. Immunol. 170,1374–1382

10. Kardos, J., Gal, P., Szi lagyi, L., Thielens, N. M., Szilagyi, K., Lorincz, Z., Kulcsar, P.,Graf, L ., Arlaud, G. J., and Zavodszky, P. (2001) J. Immunol. 167, 5202–5208

11. Harmat, V., Gal, P., Kardos, J., Szilagyi, K., Ambrus, G., Vegh, B., Naray-Szabo, G., andZavodszky, P. (2004) J. Mol. Biol. 342, 1533–1546

12. Dodds, A. W. (1993) Methods Enzymol. 223, 46–6113. Lacroix, M., Ebel, C., Kardos, J., Dobo, J., Gal, P., Zavodszky, P., Arlaud, G. J ., and

FIGURE 8. Schematic diagram of the proposed autoactivation mechanism of MASP-2. TheCCP1-CCP2 moiety is shown as dotted ellipses, andtheSP domain isshownas a gray blob with the activation peptide represented as a black loop. The flexibleCCP2/SP junction helps in orienting the SP domains ( gray arrows) in the physiologicalMBLMASP-2 complex. A, during “step 1” the zymogen molecule (left ) cleaves anotherzymogen molecule (right ). In their initial complex, in addition to favorable contacts(whitearrows), unfavorableinteractions(blackarrows) arealso present,and theinactive/active-like conformational transition of the enzyme SP domain ( from light gray to gray striped ) ispromoted.Aftertheenzyme reaction,theproduct(dark gray SP ) is released. B,

during “step 2,” the activated (left , dark gray SP ) and zymogen (right , light gray SP )MASP-2 play the roles of the enzyme and the substrate, respectively. In contrast to step1, here the binding surface of the enzyme and the substrate is preformed to make thecanonical Michaelis complex prior to the enzyme reaction.

 Autoactivation of MASP-2

SEPTEMBER 30, 2005 • VOLUME 280 • NUMBER 39 JOURNAL OF BIOLOGICAL CHEMISTRY  33443

Page 10: A True Auto Activating Enzyme

8/7/2019 A True Auto Activating Enzyme

http://slidepdf.com/reader/full/a-true-auto-activating-enzyme 10/10

Thielens, N. M. (2001) J. Biol. Chem. 276, 36233–3624014. Privalov, P. L. (1979) Adv. Protein Chem. 33, 167–24115. Kabsch, W. (1993) J. Appl. Crystallogr. 26, 795–80016. Vagin, A. A., and Teplyakov, A. (1997) J. Appl. Crystallogr. 30, 1022–102517. Collaborative Computational Project 4 (1994) Acta Crystallogr. Sect. D 50, 760–76318. Murshudov, G. N., Vagin, A. A., and Dodson, E. J. (1997) ActaCrystallogr. Sect.D 53,

240–25519. Winn, M., Isupov, M., and Murshudov, G. N. (2001) Acta Crystallogr. Sect. D 57,

122–13320. Lamzin, V. S., and Wilson, K. S. (1997) Methods Enzymol. 277, 269–30521. Jones,T. A.,Zou,J.-Y.,Cowan, S.W., andKjeldgaard, M.(1991) ActaCrystallogr.Sect.

 A 47, 110–11922. Laskowski, R. A., MacArthur, M. W., Moss, D. S., and Thornton, J. M. (1993) J. Appl.

Crystallogr. 26, 283–29123. Morris, G. M., Goodsell, D. S., Halliday, R. S., Huey, R., Hart, W. E., Belew, R. K., and

Olson, A. J. (1998) J. Comput. Chem. 19, 1639–166224. Thomsen, R. (2003) BioSystems 72, 57–7325. Guex, N., and Peitsch, M. C. (1997) Electrophoresis 18, 2714–272326. Lindahl, E., Hess, B., and van der Spoel, D. (2001) J. Mol. Mod. 7, 306–31727. Tudos, E., Cserzo, M., and Simon, I. (1990) Int. J. Pept. Protein Res. 36, 236–23928. Esnouf, R. M. (1999) Acta Crystallogr. Sect. D 55, 938–94029. Chen, C.-B., and Wallis, R. (2004) J. Biol. Chem. 279, 26058–26065

30. Bode, W., Schwanger, P., and Huber, R. (1978) J. Mol. Biol. 118, 99–11231. Renatus, M., Engh, R. A., Stubbs, M. T., Huber, R., Fischer, S., Kohnert, U., and Bode,

W. (1997) EMBO J. 16, 4797–480532. Thielens, M. N., Cseh, S., Thiel, S., Vorup-Jensen, T., Rossi, V., Jensenius, J. C., and

Arlaud, G. J. (2001) J. Immunol. 166, 5068–507733. Budayova-Spano, M., Lacroix, M., Thielens, N. M., Arlaud, G. J., Fontecilla-Camps,

J. C., and Gaboriaud C. (2002) EMBO J. 21, 231–23934. Budayova-Spano,M., Grabarse,W., Thielens,N. M.,Hillen, H.,Lacroix,M., Schmidt,

M., Fontecilla-Camps, J. C., Arlaud, G. J., and Gaboriaud, C. (2002) Structure 10,1509–1519

35. Bode, W., and Huber, R. (1978) FEBS Lett. 90, 265–26936. Huber, R., and Bode, W. (1978) Acc. Chem. Res. 11, 114–12237. Perona, J. J., and Craik, C. S. (1997) J. Biol. Chem. 272, 29987–2999038. Gomis-Ruth, F. X., Gomez-Ortiz, M., Vendrell, J., Ventura, S., Bode, W., Huber, R.,

and Aviles, F. X. (1995) EMBO J. 14, 4387–439439. Pignol,D., Gaboriaud, C., Michon, T.,Kerfelec,B., Chapus, C., and Fontecilla-Camps,

J. C. (1994) EMBO J. 13, 1763–177140. Kraulis, P. (1991) J. Appl. Crystallogr. 24, 946–95041. DeLano, W. L. (2002) The PyMOL Molecular Graphics System, DeLano Scientific,

San Carlos, CA42. Humphrey, W., Dalke, A., and Schulten, K. (1996) J. Mol. Graphics 14, 33–38

 Autoactivation of MASP-2

33444 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 280• NUMBER 39• SEPTEMBER 30, 2005