view - protein engineering design and selection

Protein Engineering vol.9 no.3 pp.283-29O, 1996

Locating the unpaired cysteine of tissue-type plasminogenactivator

Louis CSehl1, Hung V.Nguyen, Lea LBerleau,Patricia Arcila, William F.Bennett1 and Bruce A.Keyt2

Department of Cardiovascular Research, Genentech, Inc.,460 Point San Bruno Blvd, South San Francisco, CA 94080, USA

'Present address: COR Therapeutics, Inc., 256 East Grand Ave.,South San Francisco, CA 94080, USA

^ o whom correspondence should be addressed

Variants of tissue-type plasminogen activator (t-PA) wereconstructed with selected cysteines replaced by alanineto evaluate the role of an unpaired cysteine, which hasbeen presumed to be in the growth factor module. C75A,C83A, C84A and CC83-84AA variants of t-PA wereexpressed transiently in human embryonic kidney cells.The biochemical properties of these variants providedexperimental evidence to identify the unpaired cysteine int-PA. Assays of amidolytic activity, plasminogen activation(in the presence or absence of fibrinogen or fibrin), plasmaclot lysis, fibrin binding, clearance in mice, and interactionwith a panel of monoclonal antibodies were performed asthe basis for comparing these variants with wild-typet-PA. In all assays, C83A t-PA was biochemically equivalentto wild-type t-PA. C75A t-PA, C84A t-PA and CC83-84AAt-PA variants exhibited reduced activities in a variety offunctional assays. These variants displayed two- to threefoldlower activity in fibrinogen or fibrin stimulated plasmin-ogen activation, and fivefold reduced plasma clot lysisactivity compared with that of wild-type t-PA. The affinityof C75A t-PA and C84A t-PA for fibrin was decreasedmore than two orders of magnitude compared with C83At-PA or wild-type t-PA. Plasma clearance of C75A t-PAand C84A t-PA was reduced 2-fold in mice. The C75A,C84A and CC83-84AA variants displayed significantlydecreased reactivity with anti-tPA monoclonal antibodiesspecific for finger/growth factor domain epitopes. Thesedata are consistent with a disulfide linkage of Cys75 withCys84 and that Cys83 exists as an unpaired sulfhydryl. Thesignificance of the unpaired cysteine is as yet undeterminedsince C83A t-PA and wild-type t-PA are functionallyequivalent

Keywords: free sulfhydryl group/site-specific mutants/tissue-type plasminogen activator/unpaired cysteine

IntroductionTissue-type plasminogen activator (t-PA) converts plasminogento plasmin by a single proteolytic cleavage. t-PA is a multi-domain glycoprotein containing a region homologous to afibronectin type I module (F, finger), a growth factor-likedomain (GF), two kringle domains (Kl and K2) and a trypsin-like serine protease domain (P) (Banyai et al, 1983; Pennicaet al, 1983; Ny et al, 1984; Patthy, 1985). There are 35cysteine residues in t-PA, which form 17 disulfide bondswith one unpaired cysteine. The disulfide pairings, shown in

© Oxford University Press

Figure 1, were assigned based upon sequence homology withother proteins having known disulfide bonds (Banyai et al,1983; Pennica et al., 1983; Ny et al, 1984; Patthy, 1985).Human t-PA possesses adjacent cysteines at positions 83 and 84within the GF domain. Based upon homology with epidermalgrowth factor (EGF), Banyai et al (1983) assigned a disulfidebond linking Cys75 and Cys84, leaving cysteine at position83 unpaired. Interestingly, non-human forms of t-PA do nothave cysteine at position 83, as Figure 2 illustrates (Ny et al,1988; Rickles et al, 1988; Gardell et al, 1989; Feng et al,1990; Kratzschmar et al, 1991), which provides indirectsupport for a disulfide bond between cysteines at 75 and 84.However, no experimental evidence has been presented that(i) identifies the free sulfhydryl residue in human t-PA,(ii) assigns the disulfide linkages of t-PA or (iii) indicates thefunctional significance (if any) of the unpaired cysteine inhuman t-PA. Several researchers have used sulfhydryl-specificreagents to label t-PA, assuming the free sulfhydryl to be atposition 83 (Loscalzo, 1988; Nienaber et al, 1992; Stamleret al, 1992). However, classical chemical means have notsuccessfully determined the position of the unpaired cysteinein human t-PA.

Unpaired cysteines in extracellular proteins are relativelyunusual (Carter and Ho, 1994). In addition to the unpairedcysteine of t-PA, examples of free thiols in circulating plasmaproteins include Cysl25 of interleukin-2 (TL-2) (Robb et al,1984), Cysl7 of human granulocyte-colony stimulating factor(G-CSF) (Lu et al, 1989; Arakawa et al., 1993), Cys34 ofhuman serum albumin (HSA) (Peters, 1985; He and Carter,1992; Carter and Ho, 1994), Cys4057 in apolipoprotein(a)[apo(a)] and an unidentified cysteine in apolipoprotein B-100(apoB-100) (Koschinsky et al, 1993). The identities of theunpaired cysteine residues were established for these proteins,either by chemical modification with sulfhydryl-specificreagents or by site-directed mutagenesis. Cysl25 of TL-2 wasdemonstrated to be unpaired by chemical methods usingradiolabeled iodoacetic acid. Peptide mapping and sequencingindicated the location of the unpaired cysteine (Robb et al,1984). Mutagenesis studies confirmed the results of the chem-ical identification and examined the functional significance ofthe unpaired cysteine and the single disulfide bond in IL-2(Wang et al, 1984). For the much larger protein apo(a), thecDNA sequence indicated the presence of a free sulfhydryl atCys4057, which was believed to be unpaired on the basis ofsequence homology (McLean et al, 1987). The identity andfunctional role of an unpaired cysteine in apo(a) were evaluatedutilizing mutational analysis. The formation of a covalentcomplex with apoB-100 did not occur for the C4057S apo(a)mutant, confirming Cys4057 as the unpaired cysteine andthe site of intermolecular disulfide formation with apoB-100(Koschinsky et al, 1993).

In this paper, we describe a site-directed mutagenesisapproach to identify the unpaired cysteine of t-PA. C75A,C83A, C84A and CC83-84AA variants of t-PA were con-structed, expressed and evaluated for amidolytic activity and

283

Dow

nloaded from https://academ

ic.oup.com/peds/article-abstract/9/3/283/1502815 by guest on 10 April 2019

L.C-Sehl et aL

plasminogen activation. Based on the sequences of t-PA fromother species, C83R t-PA was also constructed. The cysteinesubstitution variants were characterized with respect to fibrinbinding, interaction with anti-tPA monoclonal antibodies andbinding of t-PA variants to hepatic receptors (as indicated byin vivo plasma clearance). Molecular modeling of the t-PAgrowth factor domain, derived from the murine epidermalgrowth factor structure (Montelione et aL, 1987), was usedto evaluate the feasibility of alternative disulfide linkages,such as Cys75-Cys83 versus Cys75-Cys84. The experimentalevidence presented in this paper indicates that Cys83 is theunpaired cysteine in human t-PA.

Materials and methodsMaterialsThe restriction endonucleases Kpnl and HindUI (and NEBufferfor Kpnl) were purchased from New England Biolabs(Beverly, MA). [a-^5S]d-ATP was from Amersham (ArlingtonHeights, IL). Human plasmin, //-D-isoleucyl-L-prolyl-L-arginine-p-nitroanilide dihydrochloride (S-2288) and H-D-valyl-L-leucyl-L-lysine-p-nitroanilide dihydrochloride (S-2251)were obtained from Kabi Pharmacia Hepar (Franklin, OH).//-D-tyrosyl-L-prolyl-L-arginyl chloromethyl ketone (YPRck)was purchased from Bachem (Philadelphia, PA). Humanfibrinogen and human thrombin were products of Calbiochem(La Jolla, CA). Cyanogen bromide-activated Sepharose 4Bresin and PD-10 columns were purchased from PharmaciaLKB (Piscataway, NT). Human plasma was obtained fromPeninsula Memorial Blood Bank (Burlingame, CA). Glu-plasminogen was purified from human plasma by a modifica-tion of the Deutsch and Mertz procedure (Deutsch and Mertz,1970; Bennett et aL, 1991).

Mutagenesis and dsDNA preparationOligonucleotide-directed mutants were constructed using themethod of Kunkel (Kunkel, 1985; Kunkel et aL, 1987) withthe Muta-Gene Ml3 in vitro mutagenesis kit from Bio-Rad(Richmond, CA). Mutagenic oligonucleotides were synthesizedby the Genentech DNA synthesis laboratory as describedpreviously (Bennett et aL, 1991). Double-strand DNA sampleswere prepared by Qiagen midi preparation procedures (Qiagen,Chatsworth, CA) from Escherichia coli tonAmm294 cells(Crowe, 1992). Successful mutagenesis was confirmed bydirectly sequencing double-strand DNA. Oligonucleotideprimers were synthesized as described previously (Bennettet aL, 1991). DNA strands were separated prior to sequencingwith sodium hydroxide according to a published procedure(Sambrook et aL, 1989). Sequencing was accomplished by thedideoxy method using United States Biochemical (Cleveland,OH) Sequenase reagents and protocols (Sanger et aL, 1977;Bennett et aL, 1991). DNA was labeled with [<x35S]d-ATP.

Transient expression of t-PAHuman embryonic kidney 293 cells were employed for transi-ent expression of the t-PA molecules as described previously(Gorman et aL, 1990; Bennett et aL, 1991). Expression levelswere typically in the range 1-5 ng/ml as determined using adual monoclonal enzyme-linked immunosorbent assay(ELISA). The monoclonal antibodies used in the ELISA, Mab672 and 621, were specific for antigenic determinants inkringle 2 and protease, respectively (Nguyen et al., 1993).

Enzymatic activity assaysConditioned media from transient expression of wild-typet-PA and t-PA variants were assayed for amidolytic activity284

(in both one- and two-chain form), plasminogen activatingactivity (unstimulated, fibrinogen and fibrin stimulated) andplasma clot lysis activity (one-chain form) as published previ-ously (Jones and Meunier, 1990; Bennett et aL, 1991). Thedata are reported as values normalized to activities obtainedwith wild-type t-PA.

Fibrin binding assaysYPRck was labeled with iodine-125 and covalently bound inthe active site of t-PA molecules as described previously (Keytet aL, 1992). Fibrin binding of [l25I]YPRck-labeled variantswas evaluated at various fibrinogen concentrations using previ-ously published methods (Bennett et aL, 1991).

Pharmacokinetic analysisClearance of [l25I]YPRck-labeled wild-type and variant formsof t-PA was determined in mice according to proceduresestablished for rats (Paoni et aL, 1993). Four-week-old ICRmice (Simonsen Labs, Gilroy, CA) weighing ~20 g each werechosen for the pharmacokinetic experiments. Wild-type t-PAor variants of t-PA (~80 ng of t-PA labeled with 0.1 ^Ci of[l25I]YPRck) were administered to groups of four mice byintravenous injection in the tail. Blood samples (70 \i\ each)were obtained at various times by orbital bleeds and precipitatedwith trichloroacetic acid (TCA) to a final concentration of10%. The TCA-precipitable radioactivity was quantitated usingan LKB gamma scintillation counter. Clearance from bloodwas calculated from the trapezoidal area under the curve fromzero to 40 min.

Immunochemical reactivityMonoclonal antibodies, raised against recombinant human t-PA(Activase), were used to evaluate the immunochemical reac-tivity of cysteine substitution variants of t-PA. Antigenic speci-ficity of the monoclonal antibodies was initially characterizedby the interaction of the antibodies with domain deletionvariants of t-PA (Sinicropi et al., 1989). A higher resolutionepitope map of t-PA was obtained by comparing the immuno-chemical reactivity of 35 monoclonal antibodies with a seriesof alanine scan variants of t-PA (Nguyen et aL, 1993). Themethods utilized in these reports were modified as follows forthe evaluation of cysteine to alanine t-PA variants. Microtiterplates (96-well, polycarbonate, from Dynatech) were coatedwith 100 (il per well of affinity-purified goat polyclonal anti-tPA at 0.625 (ig/ml in 0.05 M sodium carbonate (pH 9.6) for16 h at 4°C. The plates were washed with PBS containing0.01% Tween 20. The plates were incubated with bovineserum albumin at 5 mg/ml in PBS for 1 h at 25°C, thenemptied and washed with PBS containing Tween 20. Aliquotsof samples containing t-PA or t-PA variants, in conditionedmedium (100 (xl) from transiently transfected 293 cells, werediluted to 10 ng/ml (by the ELISA described above), addedto the wells of the microtiter plate and incubated for 2 h at25CC with gentle agitation. The microtiter plates were emptiedand washed with PBS containing Tween 20. Various anti-tPAmonoclonal antibodies (100 nl) were added to microtiter platewells at specified concentrations for each monoclonal antibody.Anti-tPA monoclonal antibodies were titrated, in an ELISAformat using purified t-PA at 10 ng/ml, to determine theconcentration of each antibody that yielded absorbance inthe range 1.0-1.2 at 492 nm. The absorbance values for eacht-PA variant were normalized to the values obtained with wild-type t-PA for each antibody to give the relative bindingdata reported.

Dow



Tissue-type plasminogen activator

Fig. 1. The primary structure of human t-PA. The growth factor domain,described as exon V in the t-PA gene (Ny et aL, 1984), includes the regionfrom SerSO to Asp87. Proposed disulfides are represented as black linesbetween cysteines. Cysteines at 75, 83 and 84 are identified by bold-facednumerals. N-linkcd glycosylation sites at Asnll7, -184 and -448 areindicated by Y-shaped black bars. Fucosylated threonine is represented by adiamond at position 61. The conversion of one-chain t-PA to two chain t-PAby plasmin occurs with cleavage of the Arg275-Ile276 peptide bond(arrow). Active site residues His322, Asp371 and Ser478 are indicated byasterisks.

Human tPA

HaratwtPA

UtxaetPA

RUtPABUtPA

M O U M E Q F

S C S E PS C J E P

S C S E PS C S E PIC I

t lc IPS S Y DC

R C F « f 6 T C 0 0 A L Y F S D F V C o ( t ^ J I J H JI C F * t 6 0 C O O A L Y F S J F Y C O e P D S F Y S t l C B II C F I l t l C l O U Y F i D> C F D C ( T

i c F « s 6 T cf i lo A I T J I F I D F V C o c f [[H» t t [ * l q n H I E J LD J Y J

F f C U r l i F J U I C B

C O I l A L Y F S D F V C O C P D I F V C ( > C O I

Fig. 2. Alignment of t-PA sequence with the growth factor domain. Thesequence corresponding to amino acids 50-90 of human t-PA is alignedwith the growth factor sequences from hamster, mouse, rat and vampire batforms of t-PA. Cysteines at positions 75, 83 and 84 (indicated by anasterisk) are shaded. Murine epidermal growth factor sequence, residues 5 -48, is also aligned with the t-PA growth factors. Homologous, sharedsequences are enclosed by boxes.

Molecular modelingModels of the GF domain of t-PA were generated with theInsight n, and Discover software from Biosym Technologies(San Diego, CA). The models of the wild-type domain andthe C84A variant of the GF domain were based on the NMRstructure of murine EGF solved by Montelione et al. (1987).The sequences of murine EGF and human t-PA were alignedto determine which residues in EGF were to be replaced ordeleted. The side chains of 35 amino acids were substitutedwith those residues found in t-PA GF and the carboxy-terminalfour residues of the murine EGF structure were deleted. Toaccommodate the t-PA GF sequence, three residues weredeleted from the loop containing Cys6 through to Cysl4 inEGF (Figure 2). The loop conformation was chosen via theloop search protocol of insight II and energy minimized usingthe Discover program. During the minimization, only the

loop residues (residues Cys51 through to Cys56 using t-PAnumbering) were flexible while the remainder of the moleculewas constrained. In a final series of energy minimizations, theside-chain atoms of the entire GF domain and all atoms in theloop were allowed to adopt new conformations with respectto the original EGF structure. Other than the loop from Cys51to Cys56, the backbone conformations of EGF and the GFmodel are identical. The C84A model of t-PA growth factordomain was made by first substituting alanine for Cys84 andsubsequently forming a new disulfide bond between Cys75and Cys83. The new disulfide was optimized by energyminimization allowing only the Cys75 and Cys83 side chainatoms to move.

ResultsWild-type and variant forms of t-PA were expressed bytransient transfection of human fibroblasts and the conditionedmedia was quantified by a dual monoclonal antibody-basedELJSA. The monoclonal antibodies used in the ELISA, Mab672 and Mab 621, were specific for antigenic determinants inthe kringle 2 and protease domains, respectively, as previouslydetermined with domain deletion variants of t-PA (Sinicropiet al., 1989). As such, the concentration of variant t-PA proteinin transfected 293 cell conditioned media was accuratelydetermined by this ELISA, and varied from 1 to 5 |ig/ml.

Enzymatic activities of the t-PA mutantsAmidolytic activity, plasminogen activation (in the presenceor absence of fibrin) and plasma clot lysis activity data for theC75A, C83A, C83R, C84A and CC83-84AA mutants arepresented in Table I. These data indicate that C83A, C83Rand wild-type tPA exhibit similar properties in all assays. Thedata also show that all of the mutants are similar to wild-typetPA with respect to one- and two-chain amidolytic activities.Non-stimulated plasminogen activation for C83A, C83R andC84A mutants is normal, whereas the activities of C75A andCC83-84AA mutants are moderately to severely decreased(30-50% loss of wild-type activity, respectively). In theplasminogen activation assays involving a stimulator, C75AC84A and CC83-84AA are two- to threefold less active thanwild-type. The same mutants are more severely impaired(decreased 5-fold) in the plasma clot lysis system. As a whole,the data indicate that each of the mutants is a fully functionalserine protease and that C83A and C83R are functionallysimilar to wild-type t-PA. However, the C75A, C84A andCC83-84AA variants are impaired with respect to their abilityto be stimulated by fibrin, particularly in a plasma clotlysis system.

Fibrin bindingThe importance of the amino-terminal domains in t-PA forfibrin binding has been established (Banyai et al., 1983; vanZonneveld et al, 1986a,b; Verheijen et al, 1986; Bennettet al, 1991). Structural alterations in these regions of themolecule are likely to result in reduced fibrin affinity. Asdisplayed by the binding curves in Figure 3, the C83A variantis similiar to wild-type t-PA in fibrin affinity, while the affinitiesof C75A and C84A are reduced by approximately two ordersof magnitude. Significantly decreased fibrin affinity observedfor both C75A and C84A mutants is consistent with thedata for plasminogen activation in the presence of fibrin orfibrinogen (especially plasma clot lysis). It is also consistentwith data for t-PA variants in which presumed disulfidepairs were eliminated from the amino-terminal region of the

285

Dow



LX-Sehl et al

Table I.

Assay*

Enzymatic activities of t-PA variants

Activity of variant

C83A

normalized to

C84A

that of wild-type t-PA

C75A CC83-84AA C83R

1-ch 22882-ch 22881-ch non-stim 22512-ch non-stim 22511-ch fg stim 22512-ch fgstim 22511-ch fn stim 22512-ch fn stim 2251Plasma clot lysis

0.941.081.211.091.021.030.951.151.34

1.141.160.991.060.620-510.640490.17

1.220.840.640.840.460.340.310.320.18

1.281.000.550.550.370.400.460.620.17

1.080.950.930.770.750.980.811.00n.d.

"Assays: 1-ch and 2-ch refer to results obtained with t-PA or t-PA variants assayed in the predominantly single-chain or in the two-chain form, respectively.2288 and 2251 identify the chromogenic substrates for t-PA and plasmin, respectively. The co-factor status in the plasminogen activation assay is either 'non-stim' (absence of fibrin co-factor), 'fg stim' (presence of fibnnogen) or 'fn stim' (presence of fibrin).

100

10"' 10° 10' 10!

Fibrinogen (ng/ml)

10*

Fig. 3. Fibrin binding of radiolabeled t-PA and t-PA variants. Fibnn clotswere formed with fibrinogen at concentrations which varied from 0.5 Jig/mlto 10 mg/ml. The binding curves for radiolabeled t-PA and t-PA variants (at1 ng/ml) are represented by the percentage bound as shown for wild-typet-PA ( • ) , C75A t-PA ( • ) , C83A t-PA (O) and C84A t-PA (A).

molecule. The putative disulfide bonds were individuallyeliminated by substitution of disulfide-linked cysteine pairs toalanines. The t-PA variants C6A, C36A; C34A, C43A; C51A,C62A; C56A, C73A and C75A, C84A were expressed transi-ently in human kidney 293 cells. Assay of their fibrin-bindingproperties indicated that each of these mutants was severelycompromised with respect to fibrin binding (data not shown).Mutations of the GF domain cysteines (51, 62; 56, 73 and 75,84), particularly the C75A, C84A mutation, were associatedwith the greatest effect on fibrin binding. Variants with disulfidedeletions in the growth factor domain exhibited decreasedfibrin affinity by two to three orders of magnitude. Similiareffects of the single mutations, C75A and C84A, comparedwith the putative disulfide deletions, suggest that the singlemutations disrupt the 75-84 disulfide bond, altering the struc-ture of t-PA in the F/GF region of the molecule. The lack ofan effect for the C83A mutation is consistent with Cys83being unpaired and not required for the structural integrity ofthe amino-terminal region of the protein.

Pharmacokinetic analysisMutations of t-PA with decreased fibrin binding have beenassociated, in some cases, with reduced plasma clearance

286

Iw2a.

10 20 30

Time (minutes)

Fig. 4. Clearance of radiolabeled t-PA and t-PA variants in mice. Datashown were determined with sets of four mice for each sample of t-PAtested, as indicated for wild-type t-PA (• ) , C75A t-PA (• ) , C83A t-PA (O)and C84A t-PA (A).

740 742 363 371 369 668 387 672 383 362 728

Finger/Growth Factor Kringle 1 Kringle 2 Protease

Specific Anti-tPA Monoclonal Antibodies

Fig. 5. Relative reactivity of t-PA and t-PA variants with selected anti-tPAmouse monoclonal antibodies. Binding of the t-PA variants to four groupsof monoclonal antibodies which are specific to finger/growth factor (mAbs740, 742, 363, 371, 369), kringle I (mAbs 668 and 387), kringle 2 (mAbs672 and 383) or protease (mAbs 362 and 728) domains. The data presentedas bar graphs are normalized to the antigenic activity observed with wild-type t-PA.

Dow



Tissue-type plasminogen activator

Fig. 6. Computer-generated stereoscopic diagram of t-PA growth factor based on the NMR-derived structure of murine epidermal growth factor (Montelioneet al., 1987). Oval ribbon diagrams of t-PA growth factor models are displayed with the expected disulfide bonds indicated as yellow 'ball-and-stick'connections between the appropriate carbon and sulfur atoms. The disulfides Cys51-Cys62, Cys56-Cys73 and Cys75-Cys84 appear at the top, center andbottom of the GF model, respectively. In the model of wild-type growth factor structure, the unpaired sulfhydryl is indicated as a short yellow 'ball-and-stick'at position 83. With respect to the variant C84A t-PA, a hypothetical disulfide bond connecting Cys74 and Cys83 is shown in red, near the yellow disulfideconnecting Cys75 and Cys84. The amino and carboxy termini of the growth factor domain are indicated with N and C.

(Ahern et al, 1990; Yahara et al, 1992). Cysteine-to-alanine mutants in this study with decreased fibrin binding,C75A t-PA and C84A t-PA, also exhibit decreased plasmaclearance (Figure 4). C83A t-PA, which exhibited normal fibrinbinding, demonstrates the same rate of plasma clearance in miceas wild-type t-PA (28.7 ± 0.9 and 28.1 ± 5.6 ml/min.kg,respectively), while C75A t-PA and C84A t-PA are clearedapproximately twofold more slowly (14.2 ± 0.7 and 13.4 ±1.5 ml/min.kg, respectively). These data indicate that thehepatic receptors that mediate the rapid plasma clearance oft-PA have reduced in vivo binding of the C75A and C84Amutants of t-PA.

Immunochemical reactivity of cysteine substitution variantsTo detect structural disruptions due to misfolding of themutant t-PA molecules, cysteine substitution variants werescreened for differential immunoreactivity (with respect towild-type t-PA) using a series of 35 monoclonal antibodies(Mabs) raised against t-PA. The antigenic specificities ofthese antibodies were initially determined using domain dele-tion variants of t-PA (Sinicropi et al, 1989). The reactivitiesof C75A t-PA, C83A t-PA, C84A t-PA and CC83-84AA t-PAwere evaluated with the panel of monoclonal antibodies.These results indicate that C83A t-PA was only minimallydisrupted with respect to the reactivity of the entire panel ofmonoclonal antibodies. A selected set of results, shown inFigure 5, were obtained with five Mabs having anti-finger and/or growth factor specificity (anti-F/GF) and two representativeMabs directed against each of the other domains (anti-Kl,anti-K2 or anti-P). Three antibodies (363, 740 and 742)with anti-F/GF specificity display a set of shared antigenic

determinants as indicated by similar reactivity with charged-to-alanine mutants (Nguyen et al, 1993). These Mabs demon-strated similar reactivity with the t-PA cysteine-to-alaninemutants of t-PA (especially Mabs 740 and 742), although Mab363 displayed greater reactivity for all cysteine mutants (Figure5). Antibodies 740 and 742 exhibited a complete loss ofreactivity with cysteine variants C75A, C84A and CC83-84AA. The binding of C83A t-PA was decreased approximatelytwofold with Mabs 740 and 742. In contrast, the C83A mutantappeared equivalent to wild-type t-PA with respect to Mab363 binding, while C75A, C84A and CC83-84AA mutantswere 7 to 10-fold decreased relative to wild-type t-PA. Anadditional pair of F/GF specific antibodies (Mabs 369 and371) displayed decreased reactivity with all the cysteinemutants of t-PA; especially CC83-84AA t-PA. C75A, C83Aand C84A mutants exhibited reduced binding to Mabs 369and 371.

Anti-Kl, anti-K2 and anti-protease specific t-PA antibodieswere also used to evaluate the overall structural integrity ofcysteine mutants of t-PA. Included in Figure 5 are the bindingdata for representative monoclonal antibodies specific to Kl(Mabs 668 and 387), K2 (Mabs 672 and 383) and the proteasedomain (Mabs 362 and 728) of t-PA (data not shown for 24additional Mabs against non-F/GF epitopes). The results ofthese studies indicate that antibodies to non-F/GF regionsof t-PA do not distinguish any significant immunochemicaldifference between wild-type t-PA and the C75A, C83A, C84Aand CC83-84AA mutants of t-PA. The structural disruptionsdue to cysteine-to-alanine mutations appear to be restrictedto the F/GF region of the molecule. These data suggest thatthe antigenic structure of C83A t-PA is similar to wild-type

287

Dow



L.CSeW et aL

t-PA as detected by antibodies specific to the F/GF region.In addition, C75A t-PA and C84A t-PA are recognized assignificantly different from wild-type t-PA by three F/GFspecific antibodies. Furthermore, CC83-84AA t-PA is immuno-chemically distinct from wild-type t-PA, as indicated by lowreactivity with all five F/GF specific monoclonal antibodies.

Molecular modeling of the GF domain of t-PAThe GF domain of t-PA was modeled using the structure ofmurine EGF determined by Montelione et al. (1987). Figure6 shows two models of the GF domain that are superimposedfor comparison. One model is the wild-type GF whichpossesses three disulfides including Cys75 and Cys84. Toexamine the possibility of alternate disulfide pairing in the GFdomain, another model was constructed of a variant containingthe C84A mutation and a disulfide bond between Cys75—Cys83. Other than the alternative disulfide linkages and theC84A mutation, the models are identical. These models illus-trate that a 75-83 disulfide bond is not likely to form withoutsignificant perturbation of the GF structure. The modeled 75-83 disulfide was considerably different from the presumed'normal' 75-84 disulfide bond. The 75-84 disulfide possessesa bond length of 1.99 A and a x3 side chain dihedral angle of90.8°, which are consistent with expected disulfide bond values(Richardson, 1981; Thornton, 1981; Katz and Kossiakoff,1986). The values for the length and x3 angle of the 75-83disulfide bond (2.16 A and -106.1°, respectively) are notwithin the expected ranges. The %3 angle for this bond has anopposite twist compared with the 75-84 bond. Additionally,the x l ' and *$.' side chain dihedral angles for the 75-83disulfide indicate a strained system which was created in anattempt to accommodate the bond length and x3 requirementsfor disulfide formation. The strained nature of the 75-83disulfide as compared to the normal disulfide structure repres-ented by the 75-84 bond is in agreement with the experimentaldata. The modeled GF domain provides a. reasonable explana-tion for the observed differences among the variants of t-PAand further supports the conclusion that Cys83 is unpaired int-PA. These data suggest that t-PA GF domain contains aCys75-Cys84 bond which is necessary for t-PA structureand function.

Discussion

In addition to its substrate, plasminogen, t-PA interacts with avariety of other molecules. Fibrin(ogen) stimulates plasmin-ogen activation by t-PA (Hoylaerts et al, 1982; Bennett et al,1991). t-PA also binds to fibrin (Banyai et al, 1983; vanZonneveld et al, 1986a,b; Bennett et al, 1991). Plasminogenactivator inhibitor-1 (PAI-1) rapidly inhibits circulating t-PA(Wiman et al, 1984; Vaughn et al, 1989). t-PA is clearedfrom circulation via one or more hepatic receptors (Nilssonet al, 1985; Otter et al, 1991; Bu et al, 1994). Lysine andlysine analogs also interact with t-PA via a site on the K2domain (Rijken et al, 1982; Qeary et al, 1989; Bennettet al, 1991). The numerous intermolecular interactions of t-PAprovide the basis for a variety of functional assays which canbe used to evaluate the effects of discrete mutations. Forexample, residues in the 296-299 region of the molecule affectboth fibrin specificity of t-PA and the rate of inhibition byPAI-1 (Madison et al., 1989, 1990; Bennett et al, 1991;Eastman et al, 1992; Paoni et al., 1992, 1993). Residues 67-69 in the growth factor domain have been implicated inclearance of t-PA (Anderson et al, 1988; Browne et al, 1990;Bassel-Duby et al, 1992). Certain residues in the finger domain

were demonstrated to be important for fibrin binding; othersfor clearance (Ahem et al, 1990). In the current study, cysteineresidues in the GF domain of t-PA have been mutated toalanine. The properties of these variants were used to identifythe unpaired cysteine in t-PA and to determine what effect, ifany, its replacement would have on the properties of themolecule. The results of this study demonstrate that removalof the Cys83 sulfhydryl has no effect on the enzymatic activityof t-PA. C83A t-PA also binds fibrin like wild-type t-PA andis cleared in mice at the same rate as wild-type. Mutation ofthe adjacent cysteine, Cys84, to alanine yielded a t-PA variantwith reduced clearance and greatly decreased fibrin binding,fibrin stimulation and plasma clot lysis. These results areconsistent with Cys83 being unpaired and Cys84 beinginvolved in a disulfide bond with Cys75.

Cysteine mutations were chosen on the basis of disulfidepair assignments made by sequence homologies between t-PAand other proteins with known disulfide bond patterns (Banyaiet al, 1983; Pennica et al, 1983; Ny et al, 1984; Patthy,1985). In view of the sequence homology, it has been suggestedthat Cys83 is the unpaired cysteine of the 35 such residues int-PA (Banyai et al, 1983). Experimental evidence providedby a number of researchers has indicated that a single titratablecysteine is present in t-PA (Loscalzo, 1988; Nienaber et al,1992; Stamler et al, 1992). These researchers have reported0.8-1.0 mol of reactive cysteine per mole of t-PA. However,no experimental evidence has as yet identified the site of labelincorporation. Typically, identification of an unpaired cysteinewould involve reaction of the protein with a sulfhydryl-specificlabel and subsequent isolation of the labeled residues. Byvarious chemical approaches, identifying the unpaired cysteineof t-PA has proven fruitless. The difficulties include enzymaticdigestion of native t-PA and the presence of consecutivecysteines at positions 83 and 84. t-PA is relatively resistant toprotease digestion unless denatured, reduced and carboxyme-thylated (Ling et al, 1991), hence the assignment of disulfidebonds presents an extremely challenging problem. The currentinvestigation utilized the approach of replacing cysteines withalanines and studying the properties of the variant t-PAmolecules produced by those mutations.

As described above, the C83A mutation had little or noeffect on the properties of t-PA and the data are consistentwith Cys83 being unpaired with an unknown function. Theeffects of the C75A C84A and CC83-84AA mutations on theproperties of t-PA are consistent with a 75-84 disulfide bondin the wild-type molecule. Such mutations would eliminatethe 75—84 disulfide bond, and the decreased activity of thecorresponding t-PA variants supports that hypothesis. Themonoclonal antibody binding data for those mutants areindicative of structural defects; likewise, their decreasedfibrin affinity is consistent with data obtained for doublecysteine mutations which delete individual disulfide bonds inthe finger and GF domains. C6A, C36A; C34A, C43A; C51A,C62A; C56A, C73A and C75A, C84A variants of t-PA wereexpressed and evaluated with respect to fibrin binding. All ofthe disulfide deletion mutants exhibited significantly decreasedfibrin affinity. These observations of the properties of C75A,C83A, C84A and CC83-84AA t-PA variants, and also thesequence homology information, are indicative of the existenceof a Cys75-Cys84 bond.

C84S, another t-PA mutant which disrupts the 75-84 disul-fide bond, was reported by Adachi et al. (1992). The C84Smolecule, designated E6010, possesses significantly less clot

288

Dow



Tissue-type pJasmlnogen activator

lysis activity in vitro and is cleared more slowly in vivo thanwild-type t-PA. The data for C84S and C84A mutants of t-PAare consistent with molecules that are cleared more slowlyowing to the disruption of a key disulfide bond and partialunfolding of the GF domain. According to Adachi et al.(1992), the molecule shows promise as a thrombolytic agentdespite a reduction in clot lysis activity in vitro. Ahern et al.(1990) suggested that a t-PA variant with reduced clearancecould be an effective thrombolytic drug, although in vitroactivity was lower than that of the wild-type enzyme.

The properties of C83A t-PA indicate that the unpairedcysteine of t-PA may not contribute to t-PA function in spiteof its apparent solvent accessibility to sulfhydryl-specificreagents. For some plasma proteins the functions of free thiolsare known, whereas for other proteins, including t-PA, the freethiol groups have no known function. An example of afunctionally characterized unpaired thiol is Cys34 in HSA,which is essential for complex formation with cysteine orglutathione, ligation of metal ions and metal-containing com-plexes and dimerization of the molecule as it ages (Carter andHo, 1994). Human plasma fibronectin contains at least oneunpaired cysteine which is fully exposed by partial denaturation(Wagner and Hynes, 1979). The free sulfhydryl group hasbeen implicated in the formation of covalent multimers offibronectin (Mosher and Johnson, 1983). Cys4057 of apo(a)is necessary for the formation of an intermolecular disulfidebond with apoB-100, a covalent interaction in the formationof the lipoprotein(a) particle (Koschinsky et al., 1993). Incontrast, the function of Cys17 of G-CSF is not known, althoughit is associated with decreased protein stability (Arakawa et al.,1993). Interestingly, IL-12 has an unpaired cysteine (Cys252)that has been demonstrated to be complexed with cysteine orthioglycolic acid (Tangarone et al, 1995).

With respect to the current study, Cys 125 of IL-2 is aninteresting and pertinent example of an unpaired cysteine inan extracellular protein. This free sulfhydryl group was identi-fied by using chemical modification (Robb et al., 1984) andmutagenesis approaches (Wang et al., 1984), has no known role,and participates in alternative disulfide pairs when denatured(Browning et al, 1986) or in certain IL-2 mutants (Rong et al.,1992). C125S IL-2 possesses full biological activity (Wanget al, 1984; Liang et al, 1986; Rong et al, 1992), much likethe C83A variant of t-PA. Native IL-2 contains a disulfidebond between Cys58 and CyslO5 (Robb et al, 1984). In theC58S variant of IL-2, a 105-125 disulfide bond is formed.However, the biological activity of this form of IL-2 is greatlyreduced (Wang et al, 1984; Liang et al, 1986; Rong et al,1992). Investigation of disulfide pairing for IL-2 under denatur-ing conditions also demonstrated the formation of a 105-125disulfide bond and that the conformation of the IL-2 in thepresence of such a disulfide bond is significantly different thanthat of native IL-2 (Browning et al, 1986).

In the light of the report of alternate disulfide pairing inmutant IL-2 and other molecules, the possibility for t-PA, thateither Cys83 or Cys84 could form a disulfide bond with Cys75was considered and addressed by molecular modeling. Thisanalysis indicated that a 75—83 disulfide bond would be verydifferent from a 75-84 disulfide bond if each formed in thecontext of the GF domain structure (Figure 6). A 75-83 bondwould be strained unless the GF structure was altered toaccommodate a more favorable disulfide. According to thework of Katz and Kossiakoff (1986), a strained system doesnot preclude disulfide bond formation. However, in the case

of t-PA GF domain, the experimental data indicate that (i) a75-83 bond does not form, or (ii) it forms in an altered GFdomain structure and the functional integrity of the domain iscompromised in the presence of a C84A mutation. Combinedwith the experimental data, differences between the predicted75-84 and the 75-83 disulfide bonds provide a strong indicationthat a 75-83 bond does not occur, or that the structure of theGF domain is significantly altered if such a bond does occur.Although the structure of a recombinant finger/growth factordomain pair has been reported (Smith et al, 1994), it doesnot address this issue since the cysteine at position 83 wasmutated to serine to simplify the expression of the polypeptide.

C83A t-PA is essentially identical with wild-type t-PA inall of the data described above, suggesting that 83 is theunpaired cysteine of t-PA. These data also suggest that Cys83is not important for t-PA function. Stamler et al. (1992) haveproposed that the unpaired sulfhydryl group of t-PA may benitrosylated. Since the homologous plasminogen activators inmouse, rat, hamster and bat do not contain cysteines atanalogous positions in their sequences (Ny et al, 1988; Ricklesetal, 1988; Gardell etal, 1989; Feng et al, 1990; Kratzschmaret al, 1991), nitrosylation and its possible function would beunique to human t-PA. Referring to Figure 2, the bat moleculehas a glutamine at position 83 (with respect to the humansequence) whereas the other species of t-PA have an arginineat that position. t-PA molecules from other species do notcontain an unpaired cysteine and a function of the enzymethat is specific to humans is not currently known. However,this does not eliminate the possibility that Cys83 of t-PA maybe involved in an as yet unknown covalent intermolecularinteraction. With regard to the current data, the conclusionmust be that Cys83 is the unpaired cysteine in t-PA and thatit has no currently defined structural or functional role.

AcknowledgmentsThe authors thank the DNA synthesis group for oligonucleotides and theAssay Services group for quantitation of t-PA by ELJSA. Special thanks aredue to Carlos Duarte and Kerne Andow for advice on computer graphicsdisplay and to Mary Sliwkowski for helpful discussions.

ReferencesAdachi.H., Nagaoka,N., Nomoto.K., Yuzuriha,T. and Shoji.T. (1992) Jpn. J.

Pharmacol., 58,309-319.Ahem.TJ. et al (1990) J. Biol. Chem., 265, 5540-5545.Anderson.S and Keyt,B.A. (1989) US Pat., 5,270,198.ArakawaX, Prestrelski.SJ., Narhi.L.O., Boone.T.C. and Kenney.W.C. (1993)

J. Protein Chem., 12, 525-531.Banyai.L., Varadi,A. and Patthy,L. (1983) FEBS Lett., 163, 37^tl.Bassel-Duby.R. et al. (1992) J. Biol. Chem., 267, 9668-9677.Bennett,W.F., Paoni.N.F, Keyt,B.A., Botstein.D., Jones.AJ.S., Presta,L.,

Wurm.F.M. and Zoller.MJ., (1991) /. Biol. Chem., 266, 5191-5201.BrowncMJ., Chapman.C.G., Dodd,I., Esmail.A.F. and RobinsonJ.H. (1990)

Thromb. Res., 59, 687-692.BrowningJ.L., Mattaliano.RJ., Chow.E.P, Liang,S.-M., Allet,B., RosaJ. and

SmarU-E. (1986) Anal. Biochenu, 155, 123-128.Bu,G., Warshawsky.I. and Schwartz,A.L. (1994) Blood, 83, 3427-3436.Carter.D.C. and HoJ.X. (1994) Adv. Protein Chem., 45, 153-203.Cleary.S., MulkerrinJvl.G. and Kelley.R.F. (1989) Biochemistry, 28,

1884-1891.Crowe J.(ed.) (1992) Qiagen Plasmid Handbook. Qiagen Technical Publishing,

Chatsworth, CA, pp. 9-15.Deutsch£>.G. and Mertz,E.T. (1970) Science, 170, 1095-1096.Eastman.D., Wurm,F.M., van Reis.R. and Higgins.D.L. (1992) Biochemistry,

31, 419-422.Feng.P, Ohlsson.M. and Ny,T., (1990) J. Biol Chem., 265, 2022-2027.Gardell.SJ. et al (1989) J. Biol. Chem., 264, 17947-17952.

289

Dow



L.CSehl et al

Gorman,C.M., Gies.D.R. and McCray,G. (1990) DNA Protein Engng Tech.,2, 3-10.

He.X.M. and Carter.D.C. (1992) Nature, 358, 209-215.Hoylaerts,M., Rijken.D.C, Lijnen.H.R. and Collen.D. (1982) J. Bwl. Chem.,

257,2912-2919.Jones.AJ.S. and Meunier,A.M. (1990) Thwmb. Haemostas., 64, 455-463.Katz.B.A. and KossiakoffA (1986) J. Biol. Chem., 261, 15480-15485.Keyt,B.A., Berleau.L.T., Nguyen.H.V. and Bennett,W.F., (1992) Anal.

Biochem., 206, 73-83.Koschinsky,M.L., Cote\G.R, Gabel.B. and van der Hoek.Y.Y. (1993) /. Biol.

Chem., 268, 19819-19825.KratzschmarJ., Haendler.B., Langer,G., Boidol.W., Bringmann.P., Alagon.A.,

Donner,P. and Schleuning,W.-D. (1991) Gene, 105, 229-237.Kunkel.T.A. (1985) Proc. Natl Acad. Sci. USA, 82, 488-492Kunkel.T.A., RobertsJ.D. and Zabour,R.A. (1987) Methods Enzymoi, 154,

367-382.Liang,S.-M., Thatcher.D.R., Liang,C.-M. and Allet.B. (1986) J. Biol. Chem.,

261, 334-337.Ling.V., GuzzenxA.W., Canova-Davis.E., StultsJ.T., Hancock.W.S.,

Covey.T.R. and Shushan,B.I. (1991) Anal. Chem., 63, 2909-2915.LoscalzoJ. (1988)/ Clin. Invest., 82, 1391-1397.Lu,H.-S., Boone.T.C, Souza.L.M. and Lai,P.-H. (1989) Arch. Biochem.

Biophys., 268, 81-92.Madison,E.L., Goldsmith.E.J., GeranLR.D., Gething,M.-J. and SambrookJ.F.

(1989) Nature, 339, 721-724.Madison.E.L., Goldsmith.EJ., Gerard.R.D., Gething,M.-J., SambrookJ.F. and

Bassel-Duby,R.S. (1990) Proc. Natl Acad. Sci. USA, 87, 3530-3533.McLeanJ.W., TomlinsonJ.E., Kuang,W.-J., Eaton,D.L., Chen.E.Y.,

Fless.G.M., Scanu.A.M. and Lawn.R.M. (1987) Nature, 330, 132-137.Montelione.G.T., Wuthrich.K., Nice.E.C, Burgess.A.W. and Scheraga,H.A.

(1987) Proc. Natl Acad. Sci. USA, 84, 5226-5230.Mosher.D.F. and Johnson.R.B. (1983) /. Biol. Chem., 258, 6595-6601.Nguyen.H., ArcilaJ1., Clark,K., Fendly.B., Bennett,W. and KeyuB. (1993)

Thwmb. Haemostas., 69, 826.Nienaber,V.L., Young.S.L., BirktoftJJ., Higgins.D.L. and Berhner.L. J. (1992)

Biochemistry, 31, 3852-3861.Nilsson.S., Einarsson,M., Ekvahm.S., Haggroth,L. and Mattsson.C. (1985)

T^romb. Res., 39, 511-521.Ny.T., r'vhf. and Lund.B. (1984) Proc. Natl Acad. Sci. USA, 81, 5355-5359.Ny.T., Leo.iardsson.G. and Hsueh,AJ.-W. (1988) DNA, 7, 671-677.Otter,M., Barrett-Bergshoeff.M.M. and Rijken.D.C. (1991) /. Biol. Chem.,

266, 13931-13935.Paoni,N.F. et aL (1992) Protein Engng, 5, 259-266.Paoni.N.F. et aL (1993) Thwmb. Haemostas., 70, 307-312.Patthy.L. (1985) Cell, 41, 657-663.Pennica, D. et al. (1983) Nature, 301, 214-221.Peters.T.Jr (1985) Adv. Pwtein Chem. 37, 161-245RichardsonJ. (1981) Adv. Pwtein Chem., 34, 167-339.Rickles.RJ., DarrowAL. and Strickland.S. (1988) / Biol. Chem., 263,

1563-1569.Rijken.D.C, Hoylaerts>l. and CollenJ3. (1982) J. Biol. Chem., 257,

2920-2925.Robb,RJ., Kutny.R.M., Panico.M., Morris.H.R. and Chowdry.V. (1984) Proc.

Natl Acad. Sci. USA, 81, 6486-6490.Rong.Y., Lee,N. and Liang,S.-M. (1992) Biochem. Biophys. Res. Commun.,

188, 949-955.SambrookJ., Fritsch,E.M. and Maniatis.T. (1989) Molecular Cloning: A

Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory Press, ColdSpring Harbor, NY, pp. 1370-1372.

Sanger.F., NikJen.S. and CoulsonAR- (1977) Proc. Natl Acad. Sci. USA, 74,5463-5467.

Sinicropi, D. et al. (1989) Thwmb. Haemostas., 62, 395.Smith.B.O., Downing,A.K., Dudgeon.TJ., Cunningham.M., Driscoll,P.C. and

Campbell,I.D. (1994) Biochemistry, 33, 2422-2429.StamlerJ.S., Simon.D.I., Jaraki.O., OsbomeJ.A., Francis.S., Mullins.M.,

Singel.D. and LoscalzoJ. (1992) Pwc. Natl Acad. Sci. USA, 89, 8087-8091.Tangarone.B., VathJ., Nickbarg.E., Yu,W., HarrisA and Scoble.H. (1995)

Pwtein Sci., 4, Suppl. 2, 150.ThomtonJ.M. (1981) / MoL Biol., 151, 261-287.van ZonneveldA-J-, Veerman.H. and Pannekoek,H. (1986a) Pwc. Natl Acad.

Sci. USA, 83, 4670-4674.van Zonneveld,A.-J., Veerman,H. and Pannekoek.H. (1986b) J. Biol. Chem.,

30, 14214-14218.Vaughan,D.E., DeClerck,PJ., De Mol.M. and Collen.D. (1989)/ Clin. Invest.,

84, 586-591.VerheijenJ., Caspers.M.P.M., Chang.G.T.G., de Munk,G.A.W., Pouwels.P.H.

and Enger-Valk.B. E (1986) EMBO J., 5, 3525-3530.

Wagner.D.D. and Hynes.R.O. (1979) J. Biol. Chem., 254, 6746-6754.WangA, Lu,S.-D. and Mark.D.F. (1984) Science, 224, 1431-1433.Wiman,B., ChmielewskaJ. and R4nby,M. (1984) J. Biol. Chem., 259,

3644-3647.Yahara,H. et al. (1992) Thwmb. Haemostas., 68, 672-677.

Received August 9, 1995; revised November 7, 1995; accepted November13, 1995

290

Dow



view - protein engineering design and selection

Documents