©Ferrata Storti Foundation

Home , Heparin cofactor II, Microglobulin, Neuroserpin, Transthyretin

Thrombosis • Molecular Basis of Disease Thrombosis as a conformational disease

Javier Corral Conformational diseases are a newly recognized group of heterogeneous disorders Vicente Vicente resulting from the conformational instability of individual proteins. Such instability Robin W. Carrell allows the formation of intermolecular linkages between β-sheets, to give protein aggregation and inclusion body formation. The serpin family of serine protease inhibitors provides the best-studied examples of the structural changes involved. Notably, mutations of α-1-antitrypsin result in its intracellular polymerization and accumulation in the liver leading eventually to cirrhosis. Here we consider how other conformational changes in another serpin, antithrombin, can cause its inactivation with consequent thrombosis. Thirteen different missense mutations in antithrombin are associated with either oligomer formation or with conversion of the active molecule into an inactive latent form. Each of these variant antithrombins is associated with an increased risk of thrombosis that typically occurs in an unexpectedly severe and sudden form. The trigger for this episodic thrombosis is believed to be the sudden conformational transition of the antithrombin with an accompanying loss of inhibitory activity. But what causes the transition? This is still unclear, though a likely contributor is the increased body temperature that occurs with infections hence the frequency of episodes associated with the urinary infections of pregnancy. The search for other causes is important, as the conformational perturbation of normal antithrombin is likely to be a contributory cause to the sporadic and apparently idiopathic occurrence of venous thrombosis. Key words: thrombosis, heterogeneous disorders, proteins. Haematologica 2005; 90:238-246 ©2005 Ferrata Storti Foundation

From the Universidad de he post-genomic era of the XXIst proteins. In eukaryotic cells, the variability Murcia/Centro Regional de Hemodonación, Spain (JC, VV); century has witnessed the develop- of protein products encoded by a single CIMR, University of Cambridge, UK Tment of technology that allows gene is due to both alternative translation (RWC). both the sequence and the expression of start sites and to alternative splicing. the whole genome to be analyzed. This Hence, the same nucleotide sequence is Correspondence: technology will be of fundamental rele- able to give a range of proteins with differ- Dr. Javier Corral, Centro Regional de Hemodonación, C/Ronda de Garay vance in understanding the role of genetics ent physiological functions. The final s/n, Murcia 30003, Spain. in human disease and already the results result of this strategy is a significant func- E-mail: [email protected] being obtained are changing classical con- tional diversity together with genetic cepts in science. The complexity of the economy. Thus proteomics is excelling human being was confidently expected to genomics. But, if the genome cannot be require a genetic potential much higher read in a single frame, it is questionable to than that carried by lower organisms. simplify the study of proteins based only However, the original estimate for the on their primary sequence. Furthermore, human of 100,000 genes coding for the functions arising from the same pri- 100,000 proteins has now been revised to mary sequence of amino acids may vary less than 35,000 genes; a number similar according to the folding of the expressed to that present in simpler species such as protein. Many proteins are inherently able the fruit fly and the mouse.1 This result is to change their conformation in a func- not only surprising, but also raises a new tionally relevant way in response to differ- question: how to explain the significant ent conditions or signals. Practically all discrepancy between the number of genes systems contain proteins that completely and proteins: 35,000 and 100,000, respec- change their functional state (active, inac- tively?2 An explanation for this discrepan- tive, or even to a new function) in cy is that single genes can encode multiple response to proteolysis, phosphorylation,

| 238 | haematologica/the hematology journal | 2005; 90(2) Thrombosis as a conformational disease

Figure 1: Schematic representation of aggregate formation of β-sheet-rich pro- Conformational change teins and its possible pathological consequences. Abnormal conformations allow β-sheet interactions between molecules, forming aggregates inside the Aggregates of cell (oval). These variants (especially conformational variants aggregates) are usually removed from Secretion New function the intracellular pathway followed by normal proteins, resulting in deficiency Disease 3 of the affected protein, which can be responsible for the development of one disease. In some cases, intracellular Biological function aggregates are very stable, being resistant to degradation. Their accumulation might be toxic and, after a long period of Aggregates are inefficiently secreted time, may result in cell death, causing a different late-onset disease. Finally, the Aggregates could be toxic abnormal conformation could lose the native function and acquire new functions that might be involved in a differ- Functional deficiency ent disease. The thick line or arrows rep- Cell death/Tissue damage resent different conformational states of Disease 1 a protein. Disease 2

de-phosphorylation, or binding to another protein. Such secondary modifications provide a successful Conformational diseases way to regulate the function of a single protein, as they allow the activity of a protein to be controlled The conformation of proteins is maintained by the by switching it on or off, both in time and also in tight packing of their amino acid side-chains. The localization in cells or tissues. The main advantage of replacement of a single amino acid, subtle modifica- this strategy is its rapidity, since changes in the fold- tions of pH or temperature, or any other conditions ing of proteins do not require synthesis of new pro- able to modify the hydrogen-bond network of the teins, and can be achieved instantly. Thus the com- protein, can be sufficient to cause a conformational plexity of a system does not necessarily require a change, especially in proteins with flexible structures. large number of genes and proteins, since the tissue- Recent results demonstrate that inappropriate specific requirements of higher organisms can be met changes in the conformation of an underlying protein by localized conformational changes. often result in intermolecular linkage of the unstable This recent development in our understanding of protein. In such conformational diseases,4 the molec- the genome has also resulted in significant changes in ular instability results in aberrant intermolecular link- our understanding of the molecular basis of diseases. ages, in which single peptide strands become aligned In past decades, the recognition of the association of to form highly stable β-sheets. The resulting β-linked a gene with a disease was dependent on the finding structures have at least two different pathological of significant mutations causing either the deficiency consequences. The first is that because these aberrant or the inactivation of the encoded protein. As expect- β-linked proteins are unable to follow the usual intra- ed, mutations resulting in premature stop codons cellular pathways, their cellular localization and pro- (nonsense mutations and frameshift mutations), and cessing is disrupted. Thus, conformational transfor- those affecting relevant functional domains readily mation of the protein leading to β-linked aggregation explained a number of genetic disorders. However, it causes a loss-of-function of the normal protein was difficult in many cases to explain the severe (Figure 1). A second consequence is that, once they effect of single missense mutations that primarily have formed, these β-linked aggregates are quite sta- affect neither the expression nor the function of pro- ble, and hence are resistant to intracellular degrada- teins. Recently, however, an increasing number of tion, with a consequent accumulation that leads to papers provide evidence that such apparently minor formation of inclusion bodies. This accumulation of mutations can modify the folding of the proteins to aggregated protein adversely affects the cell and can cause a decreased conformational stability that con- lead to its premature death, to give the long-term sequently results in disease.3 damage that characterizes the conformational dis-

haematologica/the hematology journal | 2005; 90(2) | 239 | J. Corral et al.

Table 1. Main human diseases caused by aberrant linkage Reactive center loop between -sheets of conformationally unstable proteins. P1-P1’ β (protease binding site) Protein Disease C sheets Hemoglobin Sickle cell anemia Hinge Unstable hemoglobin inclusion-body hemolysis Drug-induced inclusion body hemolysis B sheets Shutter Prion protein Creutzfeld-Jakob disease Bovine spongiform encephalopathy Gerstmann-Straussler-Scheinker disease Fatal familial insomnia Kuru A sheets Glutamine repetitions Huntington’s disease Spinocerebelar ataxin Dentato-rubro-pallido-Luysian atrophy Machado-Joseph-atrophy Heparin binding domain β-amyloid peptide Alzheimer’s disease Down’s syndrome Familial Alzheimer’s--amyloid precursor Figure 2. Schematic crystal structure of native antithrombin. α Presenilins 1&2 helix (ribbons) and β-sheets (arrows) are shown. The partially inserted reactive loop is also shown, pointing out the amino acids cut by the protease (P1-P1’). The mobility of this loop, which is α-synuclein Familial Parkinson’s disease able to insert inside the central β-sheet A forming a new β-sheet Tau protein Frontotemporal dementia (Pick disease) (s4A), is crucial for the efficient inhibitory anticoagulant function of antithrombin. Three regions have special relevance in the struc- Immunoglobulin light chain Amyloidosis tural flexibility and function of antithrombin: A. The shutter region Systemic or nodular AL amyloidosis is centrally located and lies at the intersection of the major secondary structural elements of the β-sheet A. Along with the adja- Serum amyloid A protein Reactive systemic amyloidosis cent breach region, the shutter facilitates sheet opening and Chronic inflammatory disease accepts the conserved hinge of the reactive center loop. B. The hinges of the reactive center. The structural relevance of shutter Transthyretin Senile systemic amyloidosis and hinges is emphasized by the high homology displayed in Familial amyloid neuropathy these regions by the family of serpins. C. The heparin-binding Familial cardiac amyloid domain, specific to antithrombin and heparin co-factor II.

β2-microglobulin Hemodialysis amyloidosis Prostatic amyloid Apolipoprotein AI Familial amyloid polyneuropathy form encephalopathies. However, the grouping is Familial visceral amyloid much more extensive than this and includes the amy- Cistatin C Hereditary (Icelandic) visceral angiopathy loidoses, as well as familial emphysema and thrombosis, amongst many other disorders (Table 1). These Lysozyme Familial visceral amyloidosis diseases, although heterogeneous in origin, share Serpins Serpinopathies common features in their pathological mechanisms 1-antitrypsin Emphysema/cirrhosis with, in each case, the conformational change in an C-1 inhibitor Angioedema Neuroserpin Familial dementia underlying protein leading to its intermolecular link- Antichymotrypsin Chronic obstructive bronchitis age and aggregation. As a consequence, lessons Antithrombin Thrombosis learned from the study of each individual disease have a relevance to the group as a whole and, in particular, to the development of the design of therapies to prevent the protein aggregation and accumulation. The best studied examples of conformational diseases (Figure 1). The occurrence of cell death is eases, in terms of the structural changes involved, are cumulative and time-dependent, explaining why provided by the serpin family of SERine Protease many conformational diseases have a slow onset and Inhibitors.5 More than 250 different serpins have are developed at older ages. Finally, in some cases, now been identified in diverse organisms including the abnormal form may not only lose its native func- viruses, mammals, plants, insects and most recently tion but may achieve a new modified function that in bacteria.6 All the members of the family share a can be reflected in the development of atypically tightly conserved structure with more than 30% severe disease. sequence identity and a common framework tertiary The term conformational diseases covers a group structure.7 Crystallographic structures of many ser- of heterogeneous disorders, which include the social- pins have now been solved and all are formed of ly and medically relevant dementias of Alzheimer’s some 400 residues folded into three β-sheets (A-C), disease, Parkinson’s disease and the prion spongi- and nine β-helices (A-I). A feature of each structure,

| 240 | haematologica/the hematology journal | 2005; 90(2) Thrombosis as a conformational disease

12 3 4 5

AT D D D ++ + E + E ++ E Cleared HBD + F H + F F + + + G ++ G + G H H + H

TAT

Figure 3. Conformational changes displayed by antithrombin (AT) during the suicide mechanism of inhibition of proteases. (i) The circulating antithrombin has the reactive loop partially inserted inside the molecule. This conformation has low inhibitory capacity. (ii) The binding of the essential pentasaccharide present in heparin and other glycosaminogycans (H) to the heparin-binding domain (HBD) results in a structural change in antithrombin that allows the release of the reactive loop. This process increases the inhibitory capacity of antithrombin up to 300-fold. (iii) Antithrombin binds thrombin (T) or another target protease by the reactive loop, which cuts it between P1-P1’. This proteolysis causes a significant conformational change in the antithrombin molecule. The cut reactive loop (P1- P15) is inserted between β-sheet sA3 and sA5, forming a new β-sheet (sA4). The consequences of these movements are crucial for protease and serpin. (iv) First, the insertion of the reactive loop also translocates the protease to the other pole of the serpin. The protease displays conformational changes that affect the reactive domain, losing proteolytic capacity. (v) Moreover, new interactions can be formed between protease and serpin allowing the formation of an almost irreversible complex (thrombin-antithrombin complex -TAT-) that is rapidly cleared from circulation. Moreover, the formation of the s4A also affects the heparin binding domain, releasing the heparin. as depicted in Figure 2, is the presence at the top of that directly generates the fibrin clot and activates the molecule of an exposed and mobile peptide loop platelets, in two parallel actions leading to the forma- containing the reactive center of the inhibitor. It is tion of the clot that prevents blood loss. In order to this loop that determines the specificity of inhibition provide the rapid response needed to prevent hemor- and consequently the sequence of the loop varies rhage, the hemostatic mechanism involves a cascade from member to member of the family.8 The flexible of proteolytic reactions leading to activation of pro- structure of this loop is essential for the inhibitory teases that rapidly amplify the response and generate action, which leads to an irreversible destruction of ample amounts of thrombin. This rapid onset of the target protease. This occurs through a suicidal coagulation is balanced by equivalent anticoagulation change in the serpin’s conformation which disrupts and fibrinolytic mechanisms that have to be equally and destroys the protease (Figure 3). However, this rapid and efficient in their response. The key to main- structural flexibility also makes serpins especially taining optimal hemostasis is therefore provided by vulnerable to conformational changes. Therefore, the range of serpins that control each of these series even minor genetic modifications or subtle environ- of proteolytic cascades, including heparin cofactor II, mental changes can severely modify the flexibility of the plasminogen activator inhibitors (PAI-1 and 2), the molecule with consequent disadvantageous con- α2-antiplasmin and, especially, antithrombin. formational change. Even small conformational Antithrombin is the most important endogenous changes of this type can have disastrous affects, due anticoagulant, as demonstrated by the thrombotic to the key role of serpins in regulating critical physi- risk associated with patients with heterozygous defi- ological processes, including coagulation, fibrinoly- ciency and the fact that homozygous deficiency is sis, complement activation, angiogenesis, apoptosis, fatal.10,11 Antithrombin has the typical three-dimen- inflammation, as well as neoplasia and viral diseases sional structure of all the serpins, but displays two (Table 1).9 special features. The first is that the reactive center loop of circulating native antithrombin is partially inserted into the body of the molecule, at the same Thrombosis as a conformational disease time obscuring the arginine (denoted P1) that is the key amino acid determining the specificity of inhibi- An efficient hemostatic system requires a complex tion to thrombin and factor Xa. The second special mechanism able to respond immediately to vascular feature of antithrombin is a site that binds a specific damage with the quick release of thrombin. pentasaccharide present in heparin and other vascu- Thrombin is the final protease of the clotting cascade lar glycosaminoglycans (Figure 2). The inhibitory

haematologica/the hematology journal | 2005; 90(2) | 241 | J. Corral et al.

Activated Figure 4. Conformational changes of native antithrombin observed spontaneously or in response to different condi- n i r tions or mutations. The circulating native a TAT p e antithrombin has the reactive loop par- h

o t tially inserted and five -sheets A (five- g β n i stranded conformation). Binding of d n bin i rom heparin allows complete release of the B to t ing Bind reactive loop, resulting in a five-stranded conformation with increased inhibitory Polymer function. Proteolytic rupture of the reac- Cleaved tive loop by target or non-target proteas- Proteolysis es, results in the insertion of the reactive Den loop, forming a cleaved six-stranded atu rati structure (alone or complexed with the ng a gen protease). The insertion of an intact reac- ts/m uta tion tive loop inside the molecule also gener- Native s ates a six-stranded structure that is called latent antithrombin. This structure forms dimers by a C-sheet linkage with a native molecule. Finally, the insertion of the intact reactive loop in the central β- Other ? sheet A of other molecules, allows forma- Latent Latent-native dimer tion of polymers.

activity of antithrombin requires a flexible structure. the molecule sensitive to mutations or other change The binding of this DEFGH pentasaccharide to resulting in its dysfunction (Figure 4). The integration antithrombin triggers a minor, but functionally rele- of the reactive loop inside the molecule can also vant change in conformation of the protein that result from its proteolytic cleavage by non-target pro- allows the reactive center loop to be completely teases such as elastase. This change in the cleaved released (Figure 3). The new conformation has a 300- antithrombin, from a five-stranded stressed state to a fold higher affinity for the coagulation protease, fac- six-stranded relaxed-state, is accompanied by an tor Xa.12 The binding of the protease to the reactive extraordinary increase in thermal stability and a loss loop results in its cleavage between the P1 arginine in heparin affinity. and the adjacent P1’ serine. In doing so, the protease It has been shown that a similar A-sheet transition is linked by a covalent bond to the cleaved loop from a five-stranded to a hyperstable six-stranded which moves from its external exposed position to form can occur in serpins even if the reactive loop is form a new central β-sheet s4A in the main β-sheet, intact. Thus, the uncleaved reactive loop of transforming the five-stranded molecule into a six- antithrombin can be inserted into its own β-sheet A, stranded structure. The protease, tethered by its to form what is called the latent structure. bonding to the loop, is in this way translocated to the Alternatively, the opening of β-sheet A of one mol- other pole of the serpin. These movements have ecule can allow the entry of the uncleaved reactive direct structural and functional consequences that loop of another molecule to form a dimer, which affect both the protease and the serpin. then extends into long chains of polymers. As with The protease suffers a gross loss of ordered struc- the latent form, polymeric antithrombins also have a ture, with total disruption of its active site, to give six-stranded structure with an accompanying low irreversible inhibition of the protease. The heparin- affinity for heparin. binding site of the antithrombin is modified, return- Transformation of native antithrombin to the ing to a low heparin affinity conformation that monomeric latent structure takes place spontaneous- allows the release of the heparin. ly under physiological conditions, with some 3% of The movement of the reactive loop of the serpin the total circulating antithrombin changing to the and the translocation of the protease allow new latent form each day.13,14 Polymers, however, do not interactions between the protease and antithrombin spontaneously form in the plasma but only appear leading to the formation of a stable thrombin- under severe denaturing conditions such as high tem- antithrombin complex (TAT). The anticoagulant peratures (60ºC) or low pH (less than 6.5).15 As will be function of antithrombin is completely fulfilled as the shown, the transformations to the latent and poly- TAT complexes are quickly removed from circulation meric forms are facilitated by missense mutations (Figure 3). However, the movement of the reactive that affect relevant mobile regions of the antithrom- 13 center loop and the central opening of β-sheet A, bin molecule. Such transformations of native although essential for inhibitory activity, also renders antithrombin to a cleaved, latent or polymeric con-

| 242 | haematologica/the hematology journal | 2005; 90(2) Thrombosis as a conformational disease

E428K: Polymer P443L: ??? hinges of the reactive center loop, or in the region involved in the shutter-like opening of the main β- sheet of the molecule required for insertion of the F229L: polymer reactive loop into the sheet. The modification of just a single amino acid in these sensitive regions can dis- A382TL: trimer G424R: Dimer S-S turb the network of interactions in the whole mole-

P80S/T: Dimer S-S cule, resulting in overall conformational changes that C95R: Dimer S-S can affect all the functions of the molecule, including T85M/K:Lat/Pol N187D: Lat/Pol its inhibitory activity and its heparin binding affinity. But most importantly, such conformational changes can result in a loss of stability and facilitate the formation of intermolecular linkages. The significance Figure 5. Conformational mutations and the structural conse- of this group of mutations is emphasized by their quence identified in antithrombin (white) and heparin cofactor II repeated identification in different members of the (gray). The P443L affecting the heparin cofactor II should have a strong conformational effect as the variant is retained inside the serpin family, being associated in each case with the cell associated with the GRP78/BiP chaperone, but no oligomers late-onset of pathology that characterizes conforma- have been identified. S-S: disulfide linked; Lat: latent; Pol: polymer. tional diseases. For example, mutations in the hinge or shutter region of both α1-antitrypsin and neuroserpin are associated with the formation of polymers. The polymerization of each of these serpins formation have direct functional consequences, as apparently occurs by the same mechanism, with the each of these forms is inactive. They are not able to opening of the β-sheet of one molecule allowing the bind proteases, and hence there is a loss of the anti- insertion of the reactive loop of the next to give loop- coagulant capacity of the native molecule. Therefore, sheet linkages. These polymers are retained inside mutations or other conditions that favor spontaneous the cells, causing plasma deficiency and also cell transformation of antithrombin to its latent or poly- damage due to the toxicity of aggregates, with result- meric forms will predictably result in an increased ant liver cirrhosis and dementia, respectively.9 risk of thrombosis. Although such loop-sheet polymers are the best char- acterized consequence of conformational instability within the serpins, the range of β-interlinkages is Conformational mutations in antithrombin and more extensive than this one form.16 In particular, we other hemostatic serpins showed that P80S and G424R mutations affecting the shutter region of antithrombin, identified in two The world-wide genetic analysis of patients with Spanish families with severe deficiency of this anti- congenital deficiency of antithrombin has identified coagulant, resulted in the formation of dimers with some 130 different mutations affecting the gene both non-covalent and disulfide linkages of two encoding this protein.11 Many of these mutations intact molecules of the mutant antithrombin.17 In have an obvious effect that explains the associated contrast to the polymers formed with homologous phenotype. For example, non-sense mutations and mutations identified in α1-antrypsin or neuroserpin, insertions or deletions causing a frameshift explain the mutated closed-dimer identified with these vari- the associated type I deficiency. Moreover, missense ant antithrombins is not only hyperstable and has no mutations affecting the heparin binding site or the inhibitory activity, but also, unlike standard loop- reactive center disturb these functional domains, sheet dimers, does not undergo transition to poly- explaining the associated type II deficiency. mers. The results with these unusual mutants indi- However, recent attention has focused on a group of cate that the variant accumulates intracellularly in the mutations that are characteristically associated with hepatocytes, as recently supported by studies with particularly severe and episodic thrombotic disease. antithrombin Morioka. In this variant, the mutation Some 40 different missense mutations have now (C95R) causes the loss of one of the three disulfide been found to result in a severe type I deficiency bonds of antithrombin and results in a dimeric form often associated with pleiotropic consequences. The with an intermolecular disulfide bond.18 However, study of these variants raised the likelihood that the even if these mutant dimeric antithrombins accumu- unusually severe disease resulted from an aberrant late within the hepatocyte, the much slower overall conformational change rather than just the loss of rate of synthesis of antithrombin as compared to that activity (Figure 5). These deductions were strength- of α1-antitrypsin or neuroserpin, explains the ened by the finding that the mutations occurred in absence of cellular damage due to the retention of the mobile regions of antithrombin, mainly in the antithrombin dimers. Thus, the prime pathological

haematologica/the hematology journal | 2005; 90(2) | 243 | J. Corral et al. consequence is an effective lack of secretion, which observed to be associated with thrombosis in patients results in an apparent type I deficiency, with a severe with some of these antithrombin variants.17,20-22 The loss of circulating anticoagulant activity and associat- hemostatic consequences of such conformational ed increased thrombotic risk. mutations are not just restricted to antithrombin. Two There is also a second, rather perplexing, group of interesting examples have been recently described in mutations which have a milder conformational effect heparin co-factor II, another anticoagulant serpin that but are associated with even more severe disease con- also inhibits thrombin. In one variant, a missense sequences. Characteristically these mutations result in mutation (P443L) in the reactive loop of heparin co-fac- a change in stability that does not greatly effect their tor II results in a heterozygous deficiency.27 This expression such that the variants are classified as being appears to have key structural consequences as the of type II. Also, because they are often secreted in a mutant variant is retained inside the cell associated normally folded and active form, there are not neces- with the GRP78/BiP chaperone, one element of the sarily any gross changes in functional activity. It is quality control of the cell.27 Another conformational only subsequent to the secretion of these variants in mutation of heparin co-factor II has been identified in the plasma that major conformational changes occur a family with homozygous deficiency.28 This missense with consequent pleiotropic changes and the onset of change (E428K) affects a conserved glutamate at the what is often severe episodic thrombosis. Examples critical hinge of a mobile peptide loop, 17 residues dis- are the A382T mutation allowing formation of circu- tant (P17) from the reactive center at P1.29 Interestingly, lating trimers19 and other mutations that facilitate the the same P17 mutation of a glutamate to a lysine is also formation of polymers20-22 or transition to the latent responsible for the archetypal example of a conforma- 21,22 form. All of these aberrational forms of antithrom- tional disease, the deficiency of the plasma serpin α1- bin (dimers, trimers, polymers, as well as the latent antitrypsin (commonly present in people of European form) have reactive loops that are buried by insertion descent). The consequence of this mutation in α1- into their own or another molecule. Hence, all are inac- antitrypsin is an instability of folding such that the tive with loss of the anticoagulant properties of the reactive loop of one molecule can insert into a β-sheet native conformation, and with a resultant increased of another to give sequentially the formation of long thrombotic risk. Thrombosis in these patients often bead-like polymers of the abnormal 1-antitrypsin. The occurs at a relatively early age as life-threatening or combination of the misfolding and the intracellular recurrently severe episodes. Thus, the severity of these accumulation of polymers results in grossly reduced phenotypes is greater than would be expected for a secretion of the protein and hence in the plasma defi- mere loss of function or inefficiency of secretion. It ciency.30 Recently, it was demonstrated that a disease seems then, that conformational mutations of in the fruit fly Drosophila was due to the same P17 Lys antithrombin may have an additional pathological mutation in a serpin that controls the fly’s immune effect that increases the severity and risk of venous response.31 Evidence that the same mechanism of poly- thrombosis. In the case of mutations that facilitate the merization occurs in all these variants is provided by latent transition there is an additional loss of anticoag- engineered expression in E. coli of mutant heparin co- ulant activity, as the abnormal latent molecule then factor II, with either a P17 Lys or Ala substitution forms a dimer with a normal native molecule23 causing (Corral et al., unpublished results). The absence, with the an additional loss of activity (Figure 4). This propagat- P17 Lys mutant of heparin co-factor II, of any apparent ed loss of activity is compounded by the preference of effects on liver function is not surprising. Heparin the latent molecule to dimerize with the most active cofactor II, like antithrombin, is synthesized in the anticoagulant form of antithrombin, the α-antithrom- liver to concentrations reaching only a fraction of those bin glycoform. of α1-antitrypsin. So it is not surprising that the mis- There is also increasing evidence that the inactive folding and accumulation of heparin co-factor II in conformational forms of the plasma serpins in general, hepatocytes has no apparent effect on liver function. and of antithrombin in particular, have potent messen- The conformational consequences of this mutation in ger functions. Recently, it has been demonstrated that heparin co-factor II seem to be restricted to the gross- the latent conformation, as well as causing a loss of ly reduced secretion of the protein and hence its defi- anticoagulant activity, also gains a function, as a signif- ciency in plasma. The surprising finding, compatible icant anti-angiogenic agent.24 Similarly, circulating with that of others, is the absence of any evident sig- 25 polymerized α1-antitrypsin induces chemotaxis, and nificant pathological consequences of the plasma defi- proteolytically cleaved variants of antitrypsin display ciency of heparin co-factor II, an observation that is pro-inflammatory properties.26 Again in a way that is supported by a recent mice knockout model.32 not understood, the putative gain-of-function of these However, heparin co-factor II deficiency might conformational variants may be exacerbated by specif- increase thrombotic risk when combined with other ic situations such as pregnancy or infection, frequently thrombophilic defects.29,33 The conclusion of all this is

| 244 | haematologica/the hematology journal | 2005; 90(2) Thrombosis as a conformational disease

Table 2. Conformational mutations identified in antithrombin (AT) and heparin co-factor II (HCII): phenotype and clinical consequences. Mutation Deficiency Carriers Thrombosis Age Recurrence Infection Pregnancy Reference of onset AT P80S Type I 15 53% 18 Yes Yes Yes 17 G424R Type I 21 62% 30 Yes Yes Yes 17, 35 P80T Type I 2 100% NR NR NR NR 36 C95R Type I 8 12.5% <42 Yes NR No 18, 37 A382T Type II 6 50% 18 Yes NR Yes 19 F229L Type II 5 (1 +/+) 40% 19 No Yes No 20 N187D Type II 2 100% 24 Yes Yes Yes 21 T85M Type II 4 75% 42 Yes NR NR 22 T85K Type II 2 50% 10 No Yes NR 22 HC II P443L Type I 7 2 arterial thrombosis 68 No No No 27 E428K Type I 15 (2 +/+) 1 venous thrombosis* 22 Yes No No 29

Age of onset: mean age of the first thrombotic episode in all carriers. Infection or pregnancy associated with the onset. NR: not reported. +/+ Homozygous subjects. *This patient also carried a stop mutation affecting the antithrombin gene. that it is likely that the secretion of conformationally Conclusions unstable serpins into the plasma has pathological consequences due to mechanisms that are as yet ill under- A new pathological mechanism resulting from the stood.34 Table 2 summarizes the phenotypic and clini- conformational instability of hemostatic serpins has cal consequences of all conformational variants identi- been discovered to be involved in thrombosis. We fied in antithrombin and heparin co-factor II.10,17-22,27,29,35-37 reviewed several cases of unexpectedly severe and sudden thrombosis associated with variant antithrombins whose defects in function relate to the instability of the Conformational changes caused by other protein and the tendency to aggregate, especially at the factors elevated body temperature accompanying infection. Unfortunately, these variants are difficult to detect in The perplexing feature of the second group of unsta- plasma by simple laboratory tests. However, the mis- ble antithrombin variants is their often sudden transi- sense mutations responsible for these variants affect tion to a conformationally inactive form with the onset key structural regions of the serpin. Two practical con- of severe thrombosis. The external factors that trigger clusions may be drawn from this scenario. First, the such inactivating transitions are not well understood. simple phenotypic measurement of antithrombin activ- The transition could reflect an interaction with other ity performed in regular clinical laboratories should be prothombotic factors such as factor V Leiden, or pro- complemented by a genetic analysis, which may help thrombin 20210A. One factor that is significant is body to identify conformational variants with a strong temperature, because even the slight increases from 37 thrombotic risk. Second, infection might increase the to 40°C that occurs in patients with fever are sufficient risk of thrombosis, not only in patients with these to trigger the conformational transition in vitro and pre- unstable variants, but probably in all patients, especial- sumably in vivo. Interestingly, there are several cases of ly with a deficiency of antithrombin. Finally, it is the onset of severe vena caval thrombosis in males, important to search for other factors able to cause con- often as young adults, in conjunction with infectious formational changes of normal antithrombin in order to pneumonia and fever. The onset of thrombosis in identify new risk factors for thrombosis. females, however, seems predominantly to occur during pregnancy and in association with urinary infec- We appreciate the secretarial assistance of Anita Rutherford. Javier Corral is Ramón y Cajal Scientist of the University of tions. In both these examples in males and females, the Murcia. This study was supported by FIS 01/1249 and increase in body temperature will undoubtedly exacer- SAF2003-00840 (MCYT & FEDER). bate pathological thrombosis. However, there are Manuscript received October 26, 2004. Accepted December 29, 2004. clearly other factors involved as well. For example, it is not only pregnancy and infection that predispose to thrombosis in females, but also the raised estrogen lev- els that result from standard contraceptive pill usage. Why and how this happens is still unknown.

haematologica/the hematology journal | 2005; 90(2) | 245 | J. Corral et al.

2003;278:15116-22. during E. coli sepsis and fails to pre- References 16. Carrell RW, Gooptu B. Conformational vent the associated coagulopathy in a changes and disease-serpins, prions and primate model. Thromb Haemost Alzheimer’s. Curr Opin Struct Biol 1998;80:816-21. 1. Venter JC, Adams MD, Myers EW, Li 1998;8:799-809. 27. Kanagawa Y, Shigekiyo T, Aihara K, PW, Mural RJ, Sutton GG, et al. The 17. Corral J, Huntington JA, González- Akaike M, Azuma H, Matsumoto T, et sequence of the human genome. Conejero R, Mushunje A, Navarro M, al. Molecular mechanism of type I Science 2001;291:1304-51. Marco P, et al. Mutations in the shutter congenital heparin cofactor (HC) II 2. Tyers M, Mann M. From genomics to region of antithrombin result in forma- deficiency caused by a missense muta- proteomics. Nature 2003;422:193-7. tion of disulfide-linked dimers and tion at reactive P2 site: HC II Toku- 3. Kopito RR, Ron D. Conformational severe venous thrombosis. J Thromb shima. Thromb Haemost 2001;85: diseases. Nature Cell Biol 2000;2: Haemost 2004;2:931-9. 101-7. E207-9. 18. Tanaka Y, Ueda K, Ozawa T, Sakura- 28. Villa P, Aznar J, Vayá A, Espana F, 4. Carrell RW, Lomas DA. Conforma- gawa N, Yokota S, Sato R, et al. Ferrando F, Mira Y, et al. Hereditary tional disease. Lancet 1997; 350:134-8. Intracellular accumulation of anti- homozygous heparin cofactor II defi- 5. Ye S, Goldsmith EJ. Serpins and other thrombin Morioka (C95R), a novel ciency and the risk of developing covalent protease inhibitors. Curr mutation causing type I antithrombin venous thrombosis. Thromb Haemost Opin Struct Biol 2001;11:740-5. deficiency. J Biol Chem 2002;277: 1999;82:1011-4. 6. Irving JA, Steenbakkers PJ, Lesk AM, 51058-67. 29. Corral J, Aznar J, Gonzalez-Conejero Op den Camp HJ, Pike RN, Whisstock 19. Lindo VS, Kakkar VV, Learmonth M, R, Villa P, Minano A, Vaya A, et al. JC. Serpins in prokaryotes. Mol Biol Melissari E, Zappacosta F, Panico M, et Homozygous (Glu428Lyus) type I Evol 2002;19:1881-90. al. Antithrombin-TRI (Ala382 to Thr) deficiency of heparin cofactor II. 7. Whisstock J, Skinner R, Lesk AM. An causing severe thromboembolic ten- Relevance of P17 residue in serpins atlas of serpin conformations. Trends dency undergoes the S-to-R transition and relationship with conformational Biochem Sci 1998;23:63-7. and is associated with a plasma-inac- diseases. Circulation 2004;110:1303-7. 8. Stein PE, Carrell RW. What do dys- tive high-molecular-weight complex 30. Carrell RW, Lomas DA. α1-antitrypsin functional serpins tell us about molec- of aggregated antithrombin. Br J Hae- deficiency. A model for conformation- ular mobility and disease? Nat Struct matol 1995;89:589-601. al diseases. N Engl J Med 2002;346:45- Biol 1995;2:96-113. 20. Picard V, Dautzenberg MD, Villoutreix 53. 9. Lomas DA, Carrell RW. Serpinopathies BO, Orliaguet G, Alhenc-Gelas M, 31. Green C, Brown G, Dafforn TR, and the conformational dementias. Nat Aiach M. Antithrombin Phe229Leu: a Reichhart JM, Morley T, Lomas DA, et Rev Genet 2002;3:759-68. new homozygous variant leading to al, Drosophila necrotic mutations mir- 10. Bayton T, Lane D. Antithrombin spontaneous antithrombin polymer- ror disease-associated variants of mutation database. Department of ization in vivo associated with severe human serpins. Development 2003; Haematology. Imperial College childhood thrombosis. Blood 2003; 130:1473-8. Faculty of Medicine. Charing Cross 102:919-25. 32. He L, Vicente CP, Westrick RJ, Eitzman Hospital Campus. Hammersmith. 21. Bruce D, Perry DJ, Borg JY, Carrell RW, DT, Tollefsen DM. Heparin cofactor II London. UK. Wardell MR. Thromboembolic dis- inhibits arterial thrombosis after endo- 11. Ishiguro K, Kojima T, Kadomatsu K, eases due to conformational changes thelial injury. J Clin Invest 2002;109: Nakayama Y, Takagi A, Suzuki M, et of antithrombin Rouen VI (187 Asn 213-9. al. Complete antithrombin deficiency Asp). J Clin Invest 1994;94:2265-74. 33. Bernardi F, Legnani C, Lunghi B, in mice results in embryonic lethality. J 22. Beauchamp NJ, Pike RN, Daly M, Lunghi B, Ferraresi P, Palareti G, et al. Clin Invest 2000;106:873-8. Butler L, Makris M, Dafforn TR, et al. A heparin cofactor II mutation (HCII 12. Whisstock JC, Pike RN, Jin L, Skinner Antithrombins Wibble and Wobble Rimini) combined with factor V R, Pei XY, Carrell RW, et al. Conform- (T85M/K): archetypal conformational Leiden or type I protein C deficiency ational changes in serpins: II. The diseases with in vivo latent-transition, in two unrelated thrombophilic sub- mechanism of activation of antithrom- thrombosis, and heparin activation. jects. Thromb Haemost 1996;76:505-9. bin by heparindagger. J Mol Biol 2000; Blood 1998;92:2696-706. 34. Carrell RW, Corral J. What can Droso- 301:1287-305. 23. Zhou A, Huntington JA, Carrell RW. phila tell us about serpins, thrombosis 13. Carrell RW, Huntington JA, Mushunje Formation of the antithrombin het- and dementia? Bioessays 2004;26:1-5. A, Zhou A. The conformational basis erodimer in vivo and the onset of 35. Jochmans K, Lissens W, Vervoort R, of thrombosis. Thromb Haemost thrombosis. Blood 1999;94:3388-96. Peeters S, De Waele M, Liebaers I. 2001; 86:14-22. 24. O’Reilly MS, Pirie-Shepherd S, Lane Antithrombin-Gly 424 Arg: a novel 14. Corral J, Rivera J, Martínez C, WS, Folkman J. Antiangiogenic activi- point mutation responsible for type 1 González-Conejero R, Miñano A, ty of the cleaved conformation of the antithrombin deficiency and neonatal Vicente V. Detection of conformation- serpin antithrombin. Science 1999; thrombosis. Blood 1994;83:146-51. al transformation of antithrombin in 285:1926-8. 36. Millar DS, Wacey AI, Ribando J, blood by crossed immunoelectro- 25. Parmar JS, Mahadeva R, Reed BJ, Melissari E, Laursen B, Woods P, et al. phoresis: new application for a classi- Farahi N, Cadwallader KA, Keogan Three novel missense mutations in the cal method. Effect of temperature in MT, et al. Polymers of α(1)-antitrypsin antithrombin III (AT3) gene causing plasmatic antithrombin; role in senes- are chemotactic for human neu- recurrent venous thrombosis. Hum cence of antithrombin and thrombo- trophils: a new paradigm for the Genet 1994;94:509-12. sis. J Lab Clin Med 2003;142:298-315. pathogenesis of emphysema. Am J 37. Ozawa T, Takikawa Y, Niiya K, 15. Zhou A, Stein PE, Huntington JA, Respir Cell Mol Biol 2002;26:723-30. Fujiwara T, Suzuki K, Sato S, et al. Carrell RW. Serpin polymerization is 26. Harper PL, Taylor FB, DeLa Cadena Antithrombin Morioka (Cys 95-Arg): a prevented by a hydrogen bond net- RA, Courtney M, Colman RW, Carrell novel missense mutation causing type work that is centered on his-334 and RW. Recombinant antitrypsin Pitts- I antithrombin deficiency. Thromb stabilized by glycerol. J Biol Chem burgh undergoes proteolytic cleavage Haemost 1997;77:403.

| 246 | haematologica/the hematology journal | 2005; 90(2)