Aberrant hepatic processing causes removal of activation peptide and primary polymerisation site from fibrinogen Canterbury (A alpha 20 Val > Asp)

Aberrant hepatic processing causes removal of

activation peptide and primary polymerisation

site from fibrinogen Canterbury (A alpha 20 Val

--> Asp).

S O Brennan, … , B Hammonds, P M George

J Clin Invest.

1995;

96(6)

:2854-2858.

https://doi.org/10.1172/JCI118356

.

A novel mechanism of molecular disease was uncovered in a patient with prolonged

thrombin time and a mild bleeding tendency. DNA sequencing of the fibrinogen A alpha

chain indicated heterozygosity for a mutation of 20 Val --> Asp. The molar ratio of

fibrinopeptide A to B released by thrombin was substantially reduced at 0.64 suggesting

either impaired cleavage or that the majority of the variant alpha chains lacked the A

peptide. The latter novel proposal arises from the observation that the mutation changes the

normal 16R G P R V20 sequence to R G P R D creating a potential furin cleavage site at

Arg 19. Synthetic peptides incorporating both sequences were tested as substrates for both

thrombin and furin. There was no substantial difference in the thrombin catalyzed cleavage.

However, the variant peptide, but not the normal, was rapidly cleaved at Arg 19 by furin.

Predictably intracellular cleavage of the Aalpha-chain at Arg 19 would remove

fibrinopeptide A together with the G P R polymerisation site. This was confirmed by

sequence analysis of fibrinogen Aalpha chains after isolation by SDS-PAGE. The expected

normal sequence was detected together with a new sequence (D V E R H Q S A-)

commencing at residue 20. Truncation was further verified by nonreducing SDS-PAGE of

the NH2-terminal disulfide knot which indicated the presence […]

Research Article

Find the latest version:

Aberrant Hepatic Processing Causes

Removal of Activation Peptide and

Primary Polymerisation Site from Fibrinogen Canterbury (Act 20 Val -+ Asp) Stephen 0. Brennan, Bruce Hammonds, and Peter M. George

Molecular Pathology Laboratory, Biochemistry Unit, Christchurch Hospital, Christchurch, New Zealand

Abstract

Anovel mechanism of molecular diseasewasuncovered in apatient with prolonged thrombin time andamild bleeding

tendency. DNA sequencing of the fibrinogen Aachain

indi-cated heterozygosity for a mutation of 20 Val -+ Asp. The

molarratio offibrinopeptide A toB released by thrombin

wassubstantially reducedat0.64suggesting either impaired cleavageorthat themajorityof the variant a-chainslacked

the A peptide. The latter novel proposal arises from the observation that the mutation changes the normal16RG P R V' _{sequence to R G P R D creating a potential furin}

cleavage site at Arg 19. Synthetic peptides incorporating bothsequencesweretestedassubstrates for both thrombin

andfurin.Therewas nosubstantialdifference inthe

throm-bin catalyzed cleavage. However, the variant peptide, but not the normal, was rapidly cleaved at Arg 19 by furin.

Predictably intracellular cleavage ofthe Aa-chain at Arg 19would remove fibrinopeptideAtogetherwith theG P R

polymerisationsite.Thiswasconfirmedbysequence

analy-sis of fibrinogen Aa chains after isolation by SDS-PAGE. The expected normal sequence wasdetected together with

a newsequence(DV E R HQSA-) commencingatresidue

20. Truncation was further verified by nonreducing SDS-PAGE of the NH2-terminal disulfide knot which indicated the presenceof aberrant homo- and heterodimers. (J. Clin Invest.1995.96:2854-2858.) Keywords:coagulation *furin *proprotein * protein synthesis dysfibrinogenemia

Introduction

Fibrinogenis a340-kD dimeric disulfide linked molecule with

eachhalfconsisting of threepolypeptide chains; Act,

_BP,

and

-y (1) with molecularweights of66,062, 54,358, and48,529,

respectively. Thefinal stepof the coagulationcascade and the first stepof fibrinpolymerisation involves the thrombin cata-lyzed cleavage of theArg-Glybond between residues 16 and 17of the ACt chain offibrinogen(2).The release of fibrinopep-tide AexposesanewGlyProArg NH2-terminalsequencewhich initiatespolymerisation by dockingtoapreexisting sitelocated

in theCOOH-terminal domain of thefibrinogen molecule, and

AddresscorrespondencetoStephen0.Brennan, Clinical Biochemistry

Unit, Canterbury Health Laboratories, Christchurch Hospital,

Christ-church, New Zealand. Phone: 64-3-364-0550; FAX: 64-3-364-0750; E-mail:[email protected]

Received for publication 20 April 1995 and accepted in revised

form24August1995.

centred close to y 363 Tyr (3). A similar thrombin catalysed reaction results in release of the B peptide from the

_BO3

chain. Thethree fibrinogen chains are synthesized in the liver and assembled into functional fibrinogen molecules before export from the hepatocyte (4). At least two intracellular endopro-teases are involved in the export and posttranslational modifica-tion of the molecule.Inthe endoplasmic reticulum signal pepti-dase removes the prepeptide from all three chains and a post-Golgi cleavage removes a'pro'peptide from the COOH-termi-nal of the Aa-chain (5). This fibrinogen convertase has not been identified but an hepatic convertase, first described in 1988, cleaves the propeptide from proalbumin at its Arg Arg (6) site and has been shownto haveapreference for cleavage atArgXYArg overX YArg Arg sequences (7). This endoge-nousCa2 -dependent microsomal serine protease has the same dibasicspecificity as recombinant furin (8),amicrosomal pro-teaseexpressedathighmRNAlevels in theliver. Indeed, furin and the endogenous protease also havethe samepH optimum and inhibitory spectrum. Thus the present evidence suggests thattheyare the same enzyme (9, 10). The Arg Pro Val Arg cleavage sequenceattheCOOH-terminal of the Aa-chain, con-formstothefurin processing motif (5, 9). After cleavage, the newly exposed P1 Arg would be expected to be removed by carboxypeptidaseH(11),or asimilarcirculatory carboxypepti-dase togive the maturechain with its COOH-terminal valine.

Similar dibasic sequences also mark the intracellular propro-tein processing sites in peptide hormones, growth factors, plasma proteins, receptors, bacterial toxins, viral envelope

gly-coproteins, and other liver-derivedcoagulation factors, includ-ingproprothrombin, profactors IX, X, and XII, and proproteins C, S, and Z (7, 9, 12). Human mutations that abolish these dibasic processing motifs result in diseases with pathologies dependentonthe function of thematureproduct.Forexample, mutations in the ArgProLys Argprocessing site of profactor

IX, at

-lArg

-+ _Ser (13) or

-4Arg

-- Gln (14), result in

hemophilia since the uncleaved precursors are unable to be activated

by

factorXIa. The mutation of -lArg -- _{Ser in the}

insulin proreceptor results in noninsulin dependent diabetes ( 15), while mutations of either the -1 or-2Arg ofproalbumin (10)result incirculatingproalbuminwithnoassociated

pathol-ogy. Most

importantly,

the mutation of+

lAsp

-- _{Val in}

proal-bumin Blenheimchangestheprocessing sequencefrom Val Phe Arg Arg Aspto Val PheArg Arg Val and this prevents

pro-cessingof theproalbuminprecursor(10, 16).

Herewereporta new disease mechanism in apatientwith aprolonged thrombintime. Themutation introducesan

inappro-priate proprotein cleavage sequence at residue 19 of the

ACt-chain of fibrinogen and is the first report ofa disease being

causedbythis mechanism. Here the mutation results in removal offibrinopeptideAand theprimaryGlyProArgpolymerisation

sequence before the fibrinogen moleculeentersthecirculation.

Methods

Materials. The proteaseinhibitors,E-64(thiol), o-phenanthroline (met-alo)andpepstatin (aspartyl)wereobtained fromBoehringerMannheim J. Clin.Invest.

©TheAmericanSocietyforClinicalInvestigation,Inc. 0021-9738/95/12/2854/05 $2.00

Biochemicals (Indianapolis, IN). Bothrops atrox venom (atroxin) was obtainedfrom Sigma Chemical Co. (St. Louis, MO) and stored at 5 mg/ml in water at-20'C.Humana-globinwasprepared by

chromatog-raphyof hemoglobin A globin on CM-cellulose in 6 M urea. Plasma analysis. Plasma was prepared fromtrisodiumcitrate antico-agulated whole blood andthrombin and reptilase clotting times were performed by standard protocols. Fibrinogen functional concentrations were determined using the Clauss method and molar concentrations were determined byquantification ofthe amount of fibrinopeptide B released from plasma on treatment with thrombin. Forfibrinopeptide

release assays 50bdof plasma and 50

pI

of 50 mM TrisHCl,50 mM NaCl, 10 mMCaCl2 were incubated with 5Fdof 10 mMphenanthroline

and 1 U of bovine thrombin (Parke Davis Thrombostat, Ann Arbor,

MI).After 5 min 100,IL of 49 mMphosphate buffer pH 2.9 wasadded,

the solutionboiled for 5min, microfuged,and 50 dlof the supernatant injected on to a Nova Pac C-18 column. The column was eluted with alineargradient from 25 to50% solvent B where the initial solvent,

A, was49 mM phosphate buffer pH 2.9 and solvent B was an equal mixture of A and acetonitrile(8).

Fibrinogen analysis. Fibrinogen was purified by precipitation with 20% saturatedammoniumsulphate (20°C),after standing (4°C) for 30 minthe precipitate was collectedby centrifugation and washed twice with 20% saturated ammoniumsulphate before being dissolved in water at370C.

DNAamplification. Genomic DNA was isolatedfromwhite blood cells andoligonucleotide primersweresynthesizedon anDNA

synthe-sizer(390;Applied Biosystems,Inc., FosterCity, CA).The Aaprimer pair targeted nucleotides 1111-1130 (5'ATT GCT GTT GCT CTC TTTTG3')and 1290-1309(5'AATCTCCTG CTT CCC CCG CT

3'), and the

BP3

pair targeted 3187-3206 (5'GGG TGT TGG AAT AGT TAC AT 3') and 3495-3515 (5'CTG CCA TGA CTA CAG

GCTTT3')(17). Theseprimersflank the secondexonof each chain. After a hot start(98°C, 10 min) Taqpolymerase(Boehringer Mannheim

Biochemicals)wasaddedtotheamplificationmixture which contained 100 ng of DNA, 50pmol primers,and 200

1LM

dNTPs. 30amplification cycleswereperformed of denaturation(940C,30 s),annealing (60°C,

30 s),and extension(72°C,30s),withafinal extension of 7 min. The PCRproduct wasgel purified andsequenced directly bythedideoxy

chain termination method using [33P] ATP and Taq DNA

polymer-ase(18).

Proteolytic cleavage of syntheticfibrinopeptides. Synthetic peptides

of sequenceDF LAEG G G V R G P R V V E R HWandD FL

AEGG G V R G P R D V E R HW, corresponding tofibrinogen

Aa-residues7-25, were obtained from Chiron Mimotopes (Vic, Austra-lia) and contained blocked NH2 (acetyl) and COOH (amide) terminals. Trp rather than the normalGlnwasincorporated at residue 25 (number-ing based on the NH2-terminal of the Aa-chain rather than that of the

peptide) tofacilitate the rapididentification of the NH2- and COOH-terminal cleavage fragments. Thepurity and sequence fidelity of the

peptidesweredetermined/verified bythemanufacturerusingion spray

triple quadrupole mass spectrometry. Their measuredmasses (MH+) were2178.4 and 2194.3D,respectively.

Forthrombindigests, 1 [lIbovine thrombin(0.02 U)wasaddedto

0.7 nmol ofpeptidein 5plof 50mMTris/HClpH 7.4, 50 mM NaCl.

Separate tubes wereincubatedfor2.5, 5, 10, 15, 20,and 30 min and dilutedwith 50 tlIof 49mMphosphatebuffer pH 2.9 beforeanalysis by reverse phaseHPLC (8). The amountofpeptidecleaved ateach timepointwasdeterminedas apercentage basedonthe appearance of the COOH-terminal product and disappearance of the parent peptide. Forfurindigests, the purifiedA704mutant soluble form of the mouse enzyme wasused (10). The furin, 0.3

_tLI

(4 mU), was added to 0.7 nmolof peptide in 2.6

fII

50 mMMES, 1 mMCaCl2,pH 5.5 containing 50MM,phenanthroline, 20fsMpepstatin, 0.1 mg/ml E-64, and 1 mg/ ml a-globin. Individual tubes were incubated at 30°C for up to 6 h before being diluted with 50

pl

of 49 mM phosphate buffer, pH 2.9 and analyzed byreversephase HPLC (8).Asfor thrombin digests the

effluentwasmonitoredat254aswellas215 nm.

Protein sequenceanalysis. Fibrinogen (20

Mg)

wasseparatedinto

its three constituent chains by SDS-PAGE in a 7.5% gel under reducing conditions and electroblotted onto a polyvinylidene difluoride mem-brane. Directsequencing was carried out on anApplied Biosystems471 instrument using a Problot cartridge.

CNBrdigestion. 1 mg of fibrinogen and 2 mg of CNBr were

incu-batedovernight in 75fr of 70% formic acid. Afterdrying(overNaOH)

peptideswereredisolved in100ydof 1%SDS,6 Mureaandanalyzed

directly bynonreducingSDS-PAGE. ForcleavagewithBothropsatrox

venom(atroxin) 20Mg ofCNBrdigestwasdiluted into 50Ml 25 mM TrisHCOpH7.4, 25 mM NaCl, and 25Mgof venom was added.Samples wereincubated at20'C for 20 min and again analyzed bynonreducing

SDS-PAGE.

Results

Casehistory. The propositus, a 45-y male vegan with Type-III hyperlipidemia, was initially seen at a specialized lipid clinic. Fibrinogen was measured as part of a routine panel of cardiovas-cular risk factors. On specific questioning, he admitted a ten-dency to prolonged bleeding, lastingup to 15 min, from even minorcuts. Healso noted that thesecutswouldreadilyresume bleedingafter minorbumps. He hadnohistoryofmajorbleeds despite several episodes of surgery includinganamputationof one armaftertraumasustainedin a motor vehicle accident. He denied any easy bruising although soft tissue injuries such as a sprained ankle often resulted in large hematomas. He hastwo siblings andtwochildren,who were notavailable fortesting, but was not awareof any familialbleedingorthrombotictendency. Fibrinogen levels measured functionally by thrombin clot-ting gave values of 0.8 and 0.7 mg/mlon twooccasions while quantitation based on the amount of fibrinopeptide B release was 1.3 mg/ml for both these samples. The prothrombin time (INR) andAPPtime were normal at 1.2 and 32 s, respectively (normal ranges 0.8-1.2 and 26-37 s). The prolonged thrombin andreptilase time of 32s(normal 18-22 s) and 29 s (normal 18-22 s) however, suggested an impairment of fibrinopeptide releaseand/or fibrinpolymerisation.

Fibrinopeptide release. Reverse phase thrombin catalysed

fibrinopeptiderelease assays(Fig. 1)wereperformedonwhole plasma and incorporated 10 mM phenanthroline to inhibit the carboxypeptidase B mediated conversion ofpeptide B to its des-Arg form. Analysis of 10 different normal plasmas indicated that the average fibrinopeptide A to B ratio, based on peak areas, was0.92. Analysis of three separate cases (Brennan,S. O., B. Hammonds and P. M. George unpublished data) with the

fibrinogen

Christchurch mutation(Bf3 14Arg--

Cys)

_gave

averageA to Bratiosof 2.1 indicating, asexpected, that theB peptidewas notbeing removed from - 50% ofthe

BP

chains.

In contrast, repeated analysis of plasma from the present case (Fig. 1 b) showed adecrease in fibrinopeptideA release with

anA:Bratio of0.58. This resultexplainstheobservedprolonged

thrombinandreptilasetimes and suggested that either the pro-positus is heterozygous foramutation that preventsthe throm-bin catalysed release of the A peptide, orthat the A peptide was missing from approximately half of the Aa chains in the circulating fibrinogen.

A

B

C

b

a Thrombin

R C

N

A

Figure 1. Reverse phase HPLC analysis of fibri-nopeptide release. (a) normal control; (b)

pro-positus with fibrinogen Canterbury; and (c)

con-trolpatient with

fibrino-genChristchurchB,614

I

Arg- _Cys.50

MI

of

plasmawasincubated with thrombin and the

re-leasedpeptides (A and

_+_

" | " | B) separated byreverse phase HPLC. Gradient 6

1

2*

25-50% B over14

min.

indicates that the propositusisheterozygous foramutation of

Aa 20 Val -- _{Asp (GTT-+GAT).} This result was confirmed

by sequencingthe complementary strand andrepresents a new

mutation; fibrinogen Canterbury.

Cleavage ofsyntheticfibrinopeptides. Theprolonged throm-bin time and decreased yield of fibrinopeptide A could be readily explained ifthis mutation inhibited thrombin cleavage

as is reported to occur in other fibrinogens with mutations at

Arg

Gly

Pro

CG

G T G

C

cc

A

Arg G

G G

AspVal

ANT

T

Va

'GT

G

A

T

C

G

A

T

C

G

Figure2. DNAsequencingof thefibrinogenAagene, exon2.Normal control (right)anda_heterozygoteforfibrinogen Canterbury (left), showingthat the affected individual isheterozygousfor A and Tatthe second base of the codon for amino acid 20. The abnormalsequence encodesamutationof Aa 20 Val toAsp.

I I -\

6 8 10 12 14 16

I I

6 8 10 12 14

Figure3. Reverse phase separation of thrombin(a)and furin (b)digests

of synthetic peptides incorporating the normal, A F L A E G G G V

R16G P R V 20 VERH Wsequence (peptide A) and the variant

peptide with the 20 Val- _Aspsubstitution(peptideR).The thrombin digests shownarefora5-min incubation and the gradient is from 14

to55% B. The furindigestsshownwerefor 2 hand gradientwasfrom 14to62% B. N and C peptidesarederived from cleavageatArg 16 andN' andC' from cleavageatArg 19, respectively.

or nearthecleavagesite(19, 20);e.g. fibrinogens Ledyard 16

AcArg-+_CysandAarhus 19Aa

Arg

--Gly(21, 22).Totest

this possibility, synthetic peptides spanning residues 7-25 of the normal andfibrinogen Canterburysequence wereincubated

withthrombin and the reactionmonitoredovertime byreverse

phase HPLC. The normal (A) peptide D F L A E G G G V R'6 G P R V V E R H Wwas cleaved exclusively atArg 16 to give two more polar products (Fig. 3 a). Sequence and

composition analysis establishedthat theseproductswereD F

L A EG G GV Rand G P R V V E R H W(peptidesNand C, respectively). Analysisof thrombindigestsof thefibrinogen Canterbury (R) peptideD FL A E G G G V R G P R D V E R H Wgave asimilar result withcomposition analysis confirm-ing that thepeak at 8 min corresponded to the new

COOH-terminalpeptide G P R D V E R H W and that the 14.5-min peak correspondedto theNH2-terminal fragment, D F L A E G G G V R.However,while bothpeptideswerecleavedatthe expectedsite,therewas asmalldecrease in therateofcleavage of the abnormalpeptide.The relativerateofcleavage (abnormal to normal) was 0.56 with a standard deviation of 0.08 in 8

separate analyses. This small change appeared insufficient to explainthe decrease infibrinopeptide Arelease since thiswas

measuredby anendpointassay.

Fibrinogen is also subjected to cleavage by intracellular endoproteases. Signal peptidase catalyses cotranslational

re-moval of thepresequenceand thehepaticconvertase, whichis

thought to be the protease furin is active in the Golgi. This

enzymehasaspecificityforX Y R R ZorR X Y R Zsequences

withaspecialcondition that Z, the

PI'

residue,cannotbe valine

(16). We therefore postulated thatthe mutation in fibrinogen Canterbury, which converts thenormal R G P R V sequence

to R G P RD, introducesafurincleavage site. Thiswas

con-firmed when thepeptide containingthisnewsequencewas

incu-bated witharecombinant soluble formof furin and theproducts separated by reverse phase HPLC. Two fragments were

ob-b Furin

R

served (Fig. 3 b) and sequence/composition analysis estab-lished that N' was D F L A E G G GVRG P R and thatC' wasD V E R HWconfirming that cleavagewas occurring at Arg19. The rate of thisreaction, 0.6 nmol/min perunit offurin, wascomparableto that observed forthecleavage ofapeptide RGVFRRDAHKSEVAW (0.5 nmol/min per unit) which incor-porates theprocessing site of proalbumin (8). There was, how-ever, no detectablecleavage of the normal fibrinogen peptide evenafter prolonged incubations of 6 h.

Analysis of

fibrinogen.

Fibrinogenwas purified by ammo-niumsulphate precipitation and analyzed by agarose gel electro-phoresis, nonreducing SDS-PAGE in 4% gels and reducing 7.5% SDS-PAGE gels. Normal patterns were observed in all cases, specifically the expected 340- and 305-kD forms were observed innonreducing SDS gels (23), andonreducing gels

theindividual Aa,

BP,

and y chainscomigrated with the corre-sponding chains from normal individuals. However,protein se-quence analysis of the individual chains, after transfer to a polyvinylidene difluoride membrane, showed two major se-quences in theAaband of thepropositus. Onewastheexpected NH2-terminal sequence of Ala Asp SerGly Glu Gly Asp Phe Leu-and the secondwasAsp Val GluArg His Gln(Ser)

Ala-.Two PTH amino acids wereclearly detectable ateach cycle

except for residueseven whereonly Asp wasdetected and the otherresidue waspresumed tobe Ser. The aberrant sequence commencesatthe Aa20 Val--

Asp

mutation siteand

proceeds

with anormal primary structure beyond residue 21. The two products werepresent in the ratio 60:40, normal to truncated variant.

The six NH2 termini of the pairs of Aa,

BP

and y chains that make up the central domain of thefibrinogen molecule,are heldrigidly together by aseries ofdisulfide bonds. This 'NH2-terminal disulfide knot' can be liberated as a unit by CNBr cleavage. Itconsists of residues (Aa 1-51,

_BPf3

1-118, and y 1-78) x 2 and, containing494 amino acids,it is the largest

ex-pectedCNBrfragment (2) apart fromhigherMrforms which result fromincomplete cleavage,possibly resulting from partial demethylation of the methionines. Fig. 4 A shows the SDS-PAGE pattern of CNBrdigests of nonreduced fibrinogen. The propositus (lanes I and 4) showed the expected normal 494 residuefragment(a) together withtwolowermolecularweight forms (b and c) which were notpresent in controls (lanes 2, 3, and5). As expected the (a) band runsabove the albumin marker(lane6)which, because of its intense disulfidebonding,

has an apparent Mr of 50 kD in nonreducing gels (24, 25). Incubation with Bothrops atrox venomconfirmed that band a wasthe494 residueNH2-terminaldisulfide knot sinceremoval of the two Apeptides resulted in adecrease in Mr (Fig. 4B, lanes I and2). Similar incubation oftheCNBrdigest from the propositus resulted in normalisation of the band pattern (lanes 4 and 3). The atroxin cleavage

products

all migrated below band b and in the position of the minor c band. These data confirm theamino acid sequence data which indicated trunca-tion ofthe Aa chain and further show that the circulating fi-brinogencontainsbothhomo- andheterodimers oftheabnormal Aa chains. Similarly homo- and heterodimers have been ob-served withfibrinogen Birmingham (26).

Discussion

DNA sequence analysis readily established that the propositus was

heterozygous

foramutation of Aa 20 Val --

_Asp.

_Since

B

Figure4.(A)SDS-PAGE of CNBrdigests ofpurified fibrinogen;lanes 1 and 4, propositus; lanes 2, 3, and 5 normal control individuals. Lane 6 shows markers of serum albumin,pepsin, trypsinogen,andlysozyme.

Nonreducing 10% gel. (B) CNBr digestsoffibrinogenbefore andafter

incubation with Bothrops atrox venom; normalfibrinogen,lanes I

(be-fore)and 2(after);andfibrinogen Canterbury,lanes 4(before) and3 (after),respectively.Nonreducing7.5%gel.

this substitution is close to thethrombin cleavage site at Arg 16, it initially seemed plausible that the prolonged thrombin time might be explained by defective thrombin binding and cleavage. This has been reported for other variants (19-22) and isconsistent with the observation that fibrinopeptide A was notreleased from some 40% of theAa chains after treatment with excess thrombin.

Failure of thrombin cleavage was, however, not verified in experiments with synthetic peptides which showed that both the variantand normalpeptidewerecleaved with similar efficienc-ies (0.5 and 0.9 nmol/min per unit, respectively). The predicted

change in amino acid sequence from Arg GlyPro Arg Val to ArgGly Pro ArgAsp suggestedanalternativeexplanation for the prolonged thrombin time; that the variant Aa chain was being cleaved, within the intracellular export pathway, by the

hepatic proproteinconvertase (7). This novel suggestion was substantiated in three ways: (a) by sequence analysis of the Aa chains fromcirculating fibrinogen which showed that 40% of the molecules commenced atthe new aspartic acid at residue 20; (b) by the demonstration of truncation in the NH2-terminal disulfide knot after CNBr cleavage; and (c) by the demonstra-tionthatpeptides incorporating the new sequence were rapidly cleaved bypurifiedrecombinant furin the proposed in situ he-paticconvertase.

The sequencerequirements for cleavage by the endogenous KEX2-like hepatic convertase (7) and furin (8, 10) are identi-cal, with absolute requirements for paired basic residues in ei-theran XYArg Arg orpreferably anArgX Y Arg sequence. Forprocessing to occur howeverthe

PI'

residue immediately after the cleavage sitemust notbe valine. Theclearest demon-stration of this precise requirementcomes from the observed failureofproalbuminprocessinginindividualswiththe proal-bumin Blenheim(+1

Asp

-- _{Val) mutation. Here the}

_existing

Val Phe Arg Arg Asp site is replaced by an ineffective Val PheArg ArgValsequence(16). Infibrinogen Canterbury the conversemutationoccursand thus the new Aasequence ofArg

GlyProArgAsp wouldbeexpectedtopermitthe intracellular removal of the Apeptide together with theGly Pro Arg

the surface of the molecule and thus accessible to proteases. However, this appears to be the case since in prolonged diges-tions with thrombin normal fibrinogen can be cleaved at this site.

It also appears that fibrinogen is exposed to furin in the hepatic export pathway since the cDNA for the Aa chain en-codes anadditional 15 residue propeptide not found in the circu-lating molecule (5). In this instance, cleavage occurs at an Arg Pro Val Arg Gly-sequence and is followed by carboxypeptidase H exoproteolytic removal of the exposed Arg to generate the observed mature COOH-terminal sequence of Arg Pro Val. When fibrinogen is expressed in kidney cells (5) which, like thehepatocyte, contain high levels of furin (27), the secreted protein contains two species of Act chain: one with the extension and one with the mature circulating sequence. As would be expectedforfurin mediated cleavage, mutation of the

P1

Arg-+

Gly totallyprevented fibrinogen cleavage in the baby hamster kidney expression system. Further investigation showed that thisextended molecule underwent normal assembly and secre-tion, and that therewas noalteration in itsabilitytoclot in the presence of thrombin or crosslink in the presence of factor

XIIIa.Similarly intracleavage of thenew'propeptide' sequence atthe NH2-terminal offibrinogen Canterburydoesnotappearto have amajoreffect on assemblyorsecretion since the truncated moleculeconstitutes 40%of thecirculatingachains and homo-dimers

(a*B,8yy)2

aswell asheterodimers weredetected after CNBrcleavage. However,removal of theNH2-terminaltwenty residues has apredictably major impacton function since the intracellularcleavageremovestheGly Pro Arg sequence which is theprimary initiator ofpolymerisation.

This mutation reveals a new disease mechanism but also implies a general constraint onthe sequence of secreted pro-teins. Specifically valine residues that follow exposed dibasic sequencesmight beimportantinpreventingunwantedcleavage

of any protein that transits the constitutive export pathway. Similarlyotherlarge

alkyl

PI'

residues,noticeablyleucine and isoleucine, also inhibit proalbumin cleavage(28) andhistidine is notallowed at the P2position (8). Thus, in the Bl chain,

histidine 19, which occursinasequence ofArg Gly His Arg, might have the specific role ofpreventing similar intracellular cleavage, and mutationsatthis residuemightallow similar intra-cellular cleavage of theB/3chain.

Acknowledgments

We thank Dr. KazuhisuNakayama forproviding theA704furin and

ChristineHicktonfor hematological results.

This workwassupportedbytheCanterburyMedical Research Foun-dation andthe Health Research Council of New Zealand.

References

1.Doolittle,R. F. 1994. The molecularbiologyof fibrin. In The Molecular Basis of Blood Disease. G.Stamatoyannopoulos,A. W.Nienhuis,P. W.Majerus,

and H. Varmus, editors. W. B. SaundersCo.,Philadelphia. 701-723. 2.Henschen, A.,andJ.McDonagh. 1986.Fibrinogen, fibrin and factor XIII, InBloodCoagulation,F.F. A. Zwaal and H. C.Hemker,editors. Elsevier Scienve

Publishers,Amsterdam. 171-241.

3. Yamazumi, K., and R. F. Doolittle. 1992. Photoaffinity labeling of the

primary fibrin polymerization site: localization of the lable to y-chain Tyr 363. Proc.Nadl. Acad. Sci. USA. 89:2893-2896.

4.Hartwig, R.,andK.Danishefsky.1991. Studies on the assembly and secre-tion offibrinogen.J.Biol. Chem. 266:6578-6585.

5.Farrell,D., S. Huang, and E. W. Davies. 1993. Processing of the carboxyl 15-amino acid extension in the chain offibrinogen. J.Biol.Chem. 268:10351-10355.

6. Brennan, S. O., and R. J. Peach. 1988. KEX2-like protease in heptic secre-tory vesicles converts proalbumin. FEBS Letn. 229:167-170.

7.Brennan,S.O.,and R. J. Peach. 1991. Theprocessingof humanproinsulin

and chickenproalbumin byrathepaticvesicles suggests a convertase specific for X Y Arg Arg or Arg X Y Arg sequences. J. Biol. Chem. 266:21504-21508.

8.Brennan,S.O.,and K.Nakayama.1994.Cleavageof proalbuminpeptides byfurin revealsunexpectedrestrictions at theP2andPI'_{sites. FEBS Letu.}

347:80-84.

9. Chiron,M.F., C. M. Fryling, and D. J. FitzGerald. 1994.Cleavageof pseudomonas exotoxin and diphtheria toxin by a furin-like enzyme from beef liver. J. Biol. Chem.269:18167-18176.

10.Brennan,S.O.,and K.Nakayama. 1994.Furin has the proalbumin sub-stratespecificity and serpin inhibitory properties of an in situ hepatic convertase. FEBS Lett. 338:147-151.

11.Davison, H.W.,and J. C. Hutton. 1987. The insulin secretorygranule

carboxypeptidase H: purification and demonstration of involvement in proinsulin processing. Biochem. J. 245:572-582.

12. Halbon, P. A., and J. -C. Irminger. 1994. Sorting and processing of secretoryproteins.Biochem. J. 299:1-18.

13. Diuguid, D. L., M. J. Rabiet, B. C. Furie, H. A. Liebman, and B. Furie. 1968. Molecular basis of hemophilia B: defective enzyme due to a mutation in thefactor IX precursor. Proc.Nati.Acad. Sci. USA. 83:5803-5807.

14. Bentley, A. K., D. J. C. Rees, C. Rizza, and G. G. Brownlee. 1986. Defectivepropeptide processing of blood clotting factor IX caused by mutation of Arg- Gln at position 4. Cell. 45:343-348.

15.Yoshimasa, Y.,S. Seino, J. Whittaker, T. Kakehi, A. Kosaki, H.Kuzuya,

H.Imura, G. I. Bell, and D. F. Steiner. 1988. Insulin resitant diabetes due to a

pointmutation that prevents insulin proreceptorprocessing.Science(Wash. DC).

240:783-787.

16. Brennan, S. O., R. J. Peach, and D. R. Boswell. 1989. Novel human proalbumin variant with intact dibasic sequence facilitates identification of its converting enzyme. Biochim.Biophys.Acta. 993:48-50.

17.Chung, D. W., J. E.Harris,and E. W. Davie. 1990. Nucleotide sequence ofthe three genescodingfor humanfibrinogen.Adv. Exp. Med. Biol.

281:39-48.

18.Murray,V. 1989. Improved double stranded DNA sequencingusingthe linearpolymerasechainreaction. Nucleic Acids Res. 17:8889.

19. Soria, J.,C. Soria,M. Samama, A.Henschen, and C. Southan. 1982. Detection of abnormalities indysfibrinogenaemia: specialreport on fibrinogen

Metzcharacterisedbyaminoacidsubstitution locatedatthepeptidebond cleaved by thrombin. In Fibrinogen: Recent Biochemical and Medical Aspects. A.

Henschen,H.Graeff,and F.Lottspeich,editors.Walter deGruyten,Berlin. 129-143.

20.Higgins,D.L.,and J. A. Shafer. 1981.Fibrinogen Petoskey,a

dysfibrino-genaemiacharacterisedby replacementofArgAa 16 byahistidylresidue. J. Biol. Chem.256:12013-12017.

21.Lee,M.H.,E.Kaczmarek,T. D.Chin,A.Oda,S.McIntosh,K. A. Bauer, L.P. Clyne,and J.McDonagh. 1991.Fibrinogen Ledyard(Aa 16Arg- _Cys).

Biochemical andphysiolgoicalcharacterization. Blood. 78:1744-1752. 22.Hessel, B.,S.Stenberg,J.Dyr,B.Kudryk,L.Therkildsen,and B. Blom-back, 1986.FibrinogenAarhus:a newcaseofdysfibrinogenaemia. Thromb. Res. 42:21-37.

23.Holm,B.,D. W. T.Nilsen,P.Kierulf,and H. C. Godal. 1985. Purification andcharacterisation of 3 differentfibrinogenswith different molecularweights

obtained from normal humanplasma. Thromb.Res. 37:165-176.

24.Brennan,S.O.,and A. P. Fellowes. 1993. Albumin HawkesBay:alow level variant causedby lossofasulphydrylgroupatposition177.FEBSLett.

1182:46-50.

25.Andre,C.M.,A. R.McGiven,andS.McQuilken.1980.Immunological investigationof renaldisease. In Practical Methods in ClinicalImmunology.A.R.

McGiven,editor. ChurchillLivingstone,New York.76-91.

26.Siebenlist, K. R.,J. T.Prchal,and M. W. Mosesson. 1988.Fibrinogen Birmingham:aheterozygousdysfibrinogenaemia (Aal6 ArgtoHis) containing

heterodimeric molecules. Blood.71:613-618.

27.Seidah,N.G.,M.Chretien,and R.Day. 1994. Thefamilyof subtilisin/ kexin likeproproteinandprohormoneconvertases:divergentand shared functions. Biochimie(Paris).76:197-209.

28.Oda, K.,Y.Misumi,M.Sohda,N.Takami,Y.Sakaki,and Y.Ikehara,